---
type: index
title: "Manufacturing Processes"
categories: 4
processes: 118
url: https://manufacturingprocesses.org/processes
---

# Manufacturing Processes

Every process on manufacturingprocesses.org, organized into four families: forming, cutting, joining, finishing.

## Forming (60)

Hub: https://manufacturingprocesses.org/processes/forming.md

- [3D Thermal Laminating](https://manufacturingprocesses.org/processes/forming/3d-thermal-laminating.md): 3D thermal laminating bonds a decorative foil or veneer to a contoured substrate with heat and vacuum so it wraps around the profiled edges.
- [Binder Jetting](https://manufacturingprocesses.org/processes/forming/binder-jetting.md): Binder jetting prints liquid binder into a powder bed to hold each layer, producing green parts that are later sintered or infiltrated.
- [Blow Molding](https://manufacturingprocesses.org/processes/forming/blow-molding.md): Blow molding inflates a heated plastic tube or preform against the walls of a mold with compressed air, producing hollow one-piece containers and tanks.
- [CNC Wire Bending](https://manufacturingprocesses.org/processes/forming/cnc-wire-bending.md): CNC wire bending feeds round or flat wire through a programmable head that bends it in sequence into a finished 2D or 3D form.
- [Centrifugal Casting](https://manufacturingprocesses.org/processes/forming/centrifugal-casting.md): Centrifugal casting pours molten metal into a spinning mold so centrifugal force packs it against the wall and drives inclusions toward the bore.
- [Ceramic Injection Molding (CIM)](https://manufacturingprocesses.org/processes/forming/ceramic-injection-molding-cim.md): Ceramic injection molding molds a ceramic powder and binder feedstock, then debinds and sinters it into a dense technical ceramic part.
- [Ceramic Slip Casting](https://manufacturingprocesses.org/processes/forming/ceramic-slip-casting.md): Ceramic slip casting pours liquid clay slip into a porous plaster mold that draws off water, leaving a solid layer that becomes the part.
- [Clay Throwing](https://manufacturingprocesses.org/processes/forming/clay-throwing.md): Clay throwing forms a rotationally symmetric vessel by hand from a spinning ball of plastic clay on a potter's wheel.
- [Composite Laminating](https://manufacturingprocesses.org/processes/forming/composite-laminating.md): Composite laminating lays reinforcing fiber into a mold and consolidates it with resin, curing into a stiff, light monocoque part.
- [Compression Molding](https://manufacturingprocesses.org/processes/forming/compression-molding.md): Compression molding presses a measured charge of rubber or thermoset into an open heated mold that closes to cure it into its final shape.
- [DMC and SMC Molding](https://manufacturingprocesses.org/processes/forming/dmc-and-smc-molding.md): DMC and SMC molding compression-molds a pre-mixed dough or sheet of chopped fiber and thermoset resin into a stiff, dimensionally stable part.
- [Deep Drawing](https://manufacturingprocesses.org/processes/forming/deep-drawing.md): Deep drawing pushes a sheet metal blank through a die with a punch so the metal flows into a seamless cup or box deeper than its own diameter.
- [Die Casting](https://manufacturingprocesses.org/processes/forming/die-casting.md): Die casting injects molten aluminum, zinc or magnesium into a hardened steel die under high pressure, producing thin-walled net-shape metal parts.
- [Dip Molding](https://manufacturingprocesses.org/processes/forming/dip-molding.md): Dip molding withdraws a heated former from liquid plastisol or latex, leaving a coating that cures into a flexible open-ended part such as a grip or cap.
- [Direct Metal Laser Sintering (DMLS)](https://manufacturingprocesses.org/processes/forming/direct-metal-laser-sintering-dmls.md): Direct metal laser sintering fuses metal powder layer by layer with a laser, building dense metal parts anchored to a supported build plate.
- [Directed Energy Deposition (DED)](https://manufacturingprocesses.org/processes/forming/directed-energy-deposition-ded.md): Directed energy deposition feeds metal wire or powder into a moving melt pool, building up large features or repairing existing parts.
- [Electroforming](https://manufacturingprocesses.org/processes/forming/electroforming.md): Electroforming grows a metal shell by electrodeposition onto a mandrel that is afterwards removed, producing thin parts with sub-micron detail.
- [Electron Beam Melting (EBM)](https://manufacturingprocesses.org/processes/forming/electron-beam-melting-ebm.md): Electron beam melting fuses metal powder with an electron beam in vacuum at high preheat temperature, which cuts residual stress in titanium parts.
- [Filament Winding](https://manufacturingprocesses.org/processes/forming/filament-winding.md): Filament winding wraps resin-impregnated continuous fiber onto a rotating mandrel in a controlled pattern, producing tubes and pressure vessels.
- [Forging](https://manufacturingprocesses.org/processes/forming/forging.md): Forging deforms solid metal under compressive load in a hammer, press or dies, so the grain flow follows the part contour and fatigue strength rises.
- [Fused Deposition Modeling (FDM)](https://manufacturingprocesses.org/processes/forming/fused-deposition-modeling-fdm.md): Fused deposition modeling extrudes molten thermoplastic filament along programmed paths so each layer welds to the one beneath it.
- [Glassblowing](https://manufacturingprocesses.org/processes/forming/glassblowing.md): Glassblowing gathers molten glass on a blowpipe and inflates it, by hand or by machine, into hollow ware such as bottles, jars and vessels.
- [Heat Treatment](https://manufacturingprocesses.org/processes/forming/heat-treatment.md): Heat treatment cycles metal through controlled heating and cooling to change its hardness, strength, ductility or internal stress state.
- [Hydroforming](https://manufacturingprocesses.org/processes/forming/hydroforming.md): Hydroforming presses sheet or tube into a single-sided die with high-pressure fluid, producing smooth seamless shapes that would need several stampings.
- [Injection Molding](https://manufacturingprocesses.org/processes/forming/injection-molding.md): Injection molding forces molten thermoplastic into a steel mold under high pressure, producing high-volume identical plastic parts.
- [Insert Molding](https://manufacturingprocesses.org/processes/forming/insert-molding.md): Insert molding loads metal inserts into the cavity and molds polymer around them, capturing threads, terminals or stampings in a single shot.
- [Investment Casting](https://manufacturingprocesses.org/processes/forming/investment-casting.md): Investment casting builds a ceramic shell around a wax pattern, melts the wax out and pours metal into the cavity, giving fine detail in almost any alloy.
- [Lampworking](https://manufacturingprocesses.org/processes/forming/lampworking.md): Lampworking softens glass rod and tube in a bench torch and forms it directly, producing scientific glassware, neon tube and small artwork.
- [Liquid Silicone Rubber (LSR) Molding](https://manufacturingprocesses.org/processes/forming/liquid-silicone-rubber-lsr-molding.md): LSR molding injects two-part liquid silicone into a heated mold where it cures, producing flexible parts with a very wide service temperature range.
- [Masked Stereolithography (MSLA / DLP)](https://manufacturingprocesses.org/processes/forming/masked-stereolithography-msla-dlp.md): Masked stereolithography cures a whole resin layer at once through an LCD or DLP image, so build time depends on height rather than part count.
- [Material Jetting (PolyJet)](https://manufacturingprocesses.org/processes/forming/material-jetting-polyjet.md): Material jetting sprays droplets of photopolymer and cures them layer by layer, allowing several materials and colors in one build.
- [Metal Extrusion](https://manufacturingprocesses.org/processes/forming/metal-extrusion.md): Metal extrusion pushes heated billet through a shaped die, producing continuous aluminum or copper profile of constant cross-section.
- [Metal Injection Molding](https://manufacturingprocesses.org/processes/forming/metal-injection-molding.md): Metal injection molding molds a powder-and-binder feedstock exactly like a plastic, then debinds and sinters it into a dense small metal part.
- [Metal Spinning](https://manufacturingprocesses.org/processes/forming/metal-spinning.md): Metal spinning presses a rotating metal disc against a mandrel with a roller or hand tool, forming it into an axially symmetric hollow shape.
- [Metal Stamping](https://manufacturingprocesses.org/processes/forming/metal-stamping.md): Metal stamping shears and forms sheet metal between matched dies in a press, producing flat and shallow-formed parts at very high rates.
- [Multi Jet Fusion (MJF)](https://manufacturingprocesses.org/processes/forming/multi-jet-fusion-mjf.md): Multi Jet Fusion prints a fusing agent onto nylon powder and drives it with infrared energy, fusing whole layers at once for high throughput.
- [Overmolding](https://manufacturingprocesses.org/processes/forming/overmolding.md): Overmolding molds a second material, usually a soft elastomer, directly onto an already-molded rigid substrate so the two bond permanently.
- [Panel Beating](https://manufacturingprocesses.org/processes/forming/panel-beating.md): Panel beating shapes sheet metal by hand with hammers, dollies and an English wheel, producing one-off compound-curved panels without any dies.
- [Paper Pulp Molding](https://manufacturingprocesses.org/processes/forming/paper-pulp-molding.md): Paper pulp molding draws a slurry of cellulose fiber onto a perforated screen tool by vacuum, then dries the fiber mat into a rigid formed part.
- [Plastic Extrusion](https://manufacturingprocesses.org/processes/forming/plastic-extrusion.md): Plastic extrusion pushes molten polymer through a shaped die and cools it, producing continuous profile, sheet, pipe or film of constant cross-section.
- [Powder Metallurgy (Press and Sinter)](https://manufacturingprocesses.org/processes/forming/powder-metallurgy-press-and-sinter.md): Press-and-sinter powder metallurgy compacts metal powder in a rigid die and sinters it below its melting point into a finished net-shape part.
- [Press Braking](https://manufacturingprocesses.org/processes/forming/press-braking.md): Press braking bends sheet metal by driving a punch into a V-die, forming one straight-line bend at a time.
- [Press Molding Ceramics](https://manufacturingprocesses.org/processes/forming/press-molding-ceramics.md): Press molding ceramics forms clay between a die and a rotating profile tool or matched mold, producing flatware and repeat shapes quickly.
- [Pultrusion](https://manufacturingprocesses.org/processes/forming/pultrusion.md): Pultrusion pulls continuous fiber through a resin bath and a heated die, curing it into a constant-section composite profile of unlimited length.
- [Rapid Prototyping](https://manufacturingprocesses.org/processes/forming/rapid-prototyping.md): Rapid prototyping builds a part layer by layer directly from a CAD model, with no tooling and no minimum order quantity.
- [Reaction Injection Molding](https://manufacturingprocesses.org/processes/forming/reaction-injection-molding.md): Reaction injection molding mixes two liquid reactants that polymerize inside a low-pressure mold, producing large tough polyurethane parts on light tooling.
- [Roll Forming](https://manufacturingprocesses.org/processes/forming/roll-forming.md): Roll forming passes coil stock through a sequence of contoured roller stands that bend it progressively into a constant cross-section profile.
- [Rotation Molding](https://manufacturingprocesses.org/processes/forming/rotation-molding.md): Rotational molding tumbles powdered polymer inside a heated mold on two axes so it fuses into a seamless, stress-free hollow part with a uniform wall.
- [Sand Casting](https://manufacturingprocesses.org/processes/forming/sand-casting.md): Sand casting pours molten metal into a cavity formed in bonded sand around a pattern, then breaks the mold away to release the part.
- [Selective Laser Melting (SLM)](https://manufacturingprocesses.org/processes/forming/selective-laser-melting-slm.md): Selective laser melting fully melts metal powder with a laser under inert gas, producing parts at close to wrought density.
- [Selective Laser Sintering (SLS)](https://manufacturingprocesses.org/processes/forming/selective-laser-sintering-sls.md): Selective laser sintering fuses nylon powder layer by layer with a laser, so the surrounding loose powder supports the part and no supports are needed.
- [Sheet Metal Fabrication](https://manufacturingprocesses.org/processes/forming/sheet-metal-fabrication.md): Sheet metal fabrication cuts, bends and joins flat stock into a finished assembly, typically laser cutting followed by press braking, hardware and welding.
- [Steam Bending](https://manufacturingprocesses.org/processes/forming/steam-bending.md): Steam bending plasticises solid timber with saturated steam so it can be bent around a former and held until it dries into shape.
- [Stereolithography (SLA)](https://manufacturingprocesses.org/processes/forming/stereolithography-sla.md): Stereolithography cures liquid photopolymer layer by layer with a scanning UV laser, giving the smoothest surfaces in additive manufacturing.
- [Superforming](https://manufacturingprocesses.org/processes/forming/superforming.md): Superforming heats a superplastic aluminum alloy sheet and forms it with gas pressure over a single-sided tool, producing deep, complex panels.
- [Swaging](https://manufacturingprocesses.org/processes/forming/swaging.md): Swaging hammers or presses the end of a tube or bar inward against a die to reduce or taper its diameter without removing any material.
- [Thermoforming](https://manufacturingprocesses.org/processes/forming/thermoforming.md): Thermoforming heats a plastic sheet until pliable and pulls or presses it over a single-sided mold, producing large thin-walled parts with low tooling cost.
- [Tube and Section Bending](https://manufacturingprocesses.org/processes/forming/tube-and-section-bending.md): Tube and section bending draws or presses metal tube and profile around a former, with an internal mandrel wherever the wall would otherwise collapse.
- [Vacuum Casting (Urethane Casting)](https://manufacturingprocesses.org/processes/forming/vacuum-casting.md): Vacuum casting, known in the US as urethane casting, pours polyurethane resin into a silicone mold under vacuum to copy a master pattern for short runs.
- [Wood Laminating](https://manufacturingprocesses.org/processes/forming/wood-laminating.md): Wood laminating glues thin veneers or kerfed strips over a form so the assembly holds a curve permanently once the adhesive has cured.

## Cutting (15)

Hub: https://manufacturingprocesses.org/processes/cutting.md

- [5-Axis CNC Machining](https://manufacturingprocesses.org/processes/cutting/5-axis-cnc-machining.md): 5-axis CNC machining tilts and rotates the part or the head so the tool reaches every face and stays short and rigid on complex surfaces.
- [CNC Machining](https://manufacturingprocesses.org/processes/cutting/cnc-machining.md): CNC machining removes material from solid stock with a programmed cutting tool, producing accurate parts directly from a CAD model with no tooling.
- [CNC Milling](https://manufacturingprocesses.org/processes/cutting/cnc-milling.md): CNC milling removes material with a rotating multi-flute cutter moved along programmed paths, producing prismatic parts from solid stock.
- [CNC Turning](https://manufacturingprocesses.org/processes/cutting/cnc-turning.md): CNC turning rotates the workpiece against a stationary tool, producing round parts with excellent concentricity and short cycle times.
- [Die Cutting](https://manufacturingprocesses.org/processes/cutting/die-cutting.md): Die cutting presses a shaped steel rule or rotary die through sheet stock, cutting flat outlines in paper, board, foam, film and textile.
- [Electrical Discharge Machining](https://manufacturingprocesses.org/processes/cutting/electrical-discharge-machining.md): EDM erodes conductive metal with controlled electrical sparks across a dielectric gap, cutting hardened material and sharp internal corners.
- [Glass Scoring](https://manufacturingprocesses.org/processes/cutting/glass-scoring.md): Glass scoring runs a hardened wheel across the surface to create a controlled fracture line, then breaks the sheet cleanly along it.
- [Laser Cutting](https://manufacturingprocesses.org/processes/cutting/laser-cutting.md): Laser cutting melts, burns or vaporises a narrow kerf through flat stock with a focused beam, producing profiles from a 2D file with no tooling.
- [Photochemical Machining](https://manufacturingprocesses.org/processes/cutting/photochemical-machining.md): Photochemical machining masks sheet metal photographically and etches away the unmasked areas, cutting burr-free flat parts with no hard tooling.
- [Plasma Cutting](https://manufacturingprocesses.org/processes/cutting/plasma-cutting.md): Plasma cutting blows an ionised gas jet through conductive metal, melting and ejecting a kerf far faster than a laser on thick plate.
- [Punching and Blanking](https://manufacturingprocesses.org/processes/cutting/punching-and-blanking.md): Punching and blanking shear sheet metal between a punch and a die, producing holes and flat outlines at very high speed.
- [Sawing](https://manufacturingprocesses.org/processes/cutting/sawing.md): Sawing separates stock with a toothed blade — band, circular or cold saw — and is the default first operation on bar, tube and plate.
- [Swiss Screw Machining](https://manufacturingprocesses.org/processes/cutting/swiss-screw-machining.md): Swiss screw machining feeds bar stock through a guide bushing so cutting happens right next to the support, holding tight tolerances on slender parts.
- [Tube and Profile Laser Cutting](https://manufacturingprocesses.org/processes/cutting/tube-and-profile-laser-cutting.md): Tube laser cutting rotates and feeds tube or profile through a laser head, cutting holes, slots and joint geometry along its length.
- [Water Jet Cutting](https://manufacturingprocesses.org/processes/cutting/water-jet-cutting.md): Water jet cutting drives a high-pressure jet of water and abrasive through the material, cutting almost anything with no heat-affected zone.

## Joining (18)

Hub: https://manufacturingprocesses.org/processes/joining.md

- [Adhesive Bonding](https://manufacturingprocesses.org/processes/joining/adhesive-bonding.md): Adhesive bonding joins parts with a cured polymer layer that spreads load over the whole bond area instead of concentrating it at fastener points.
- [Arc Welding](https://manufacturingprocesses.org/processes/joining/arc-welding.md): Arc welding melts and fuses metal with an electric arc between an electrode and the work, shielded from the atmosphere by gas or flux.
- [Friction Welding](https://manufacturingprocesses.org/processes/joining/friction-welding.md): Friction welding rubs two parts together under load until the interface plasticises and forges together, with no melting and no filler metal.
- [Hot Plate Welding](https://manufacturingprocesses.org/processes/joining/hot-plate-welding.md): Hot plate welding presses two thermoplastic parts against a heated platen to melt the joint faces, then clamps them together to fuse.
- [Joinery](https://manufacturingprocesses.org/processes/joining/joinery.md): Joinery cuts interlocking geometry into timber so the parts locate and hold each other, with or without adhesive or fasteners.
- [Laser Plastic Welding](https://manufacturingprocesses.org/processes/joining/laser-plastic-welding.md): Laser plastic welding passes a beam through a transmissive part into an absorbing one, melting only the interface between them.
- [Mechanical Fastening](https://manufacturingprocesses.org/processes/joining/mechanical-fastening.md): Mechanical fastening joins parts with screws, inserts, clinch hardware or snap fits so the joint can be assembled and taken apart again.
- [PCB Assembly (SMT Reflow)](https://manufacturingprocesses.org/processes/joining/pcb-assembly-smt-reflow.md): PCB assembly prints solder paste, places components and reflows the board through a heated oven so every joint forms in one pass.
- [Power Beam Welding](https://manufacturingprocesses.org/processes/joining/power-beam-welding.md): Power beam welding fuses metal with a laser or electron beam focused to a very small spot, producing deep, narrow welds with little distortion.
- [Resistance Welding](https://manufacturingprocesses.org/processes/joining/resistance-welding.md): Resistance welding passes a heavy current through the joint so its own electrical resistance melts a nugget between the sheets.
- [Riveting](https://manufacturingprocesses.org/processes/joining/riveting.md): Riveting sets a rivet through aligned holes and upsets its tail so the joint clamps permanently, with no heat and no access to both sides for blind types.
- [Soldering and Brazing](https://manufacturingprocesses.org/processes/joining/soldering-and-brazing.md): Soldering and brazing join metals with a molten filler that wets the joint by capillary action while the parent metal stays solid.
- [Staking](https://manufacturingprocesses.org/processes/joining/staking.md): Heat staking softens a molded plastic boss with hot air or ultrasonic energy and reforms it into a head that traps the mating part.
- [Timber Frame Structures](https://manufacturingprocesses.org/processes/joining/timber-frame-structures.md): Timber frame structures assemble large solid or engineered timber members into a load-bearing frame using cut joints and steel connectors.
- [Ultrasonic Welding](https://manufacturingprocesses.org/processes/joining/ultrasonic-welding.md): Ultrasonic welding applies high-frequency vibration through a horn to melt a thermoplastic joint locally in well under a second.
- [Upholstery](https://manufacturingprocesses.org/processes/joining/upholstery.md): Upholstery builds a padded surface from suspension, foam, wadding and textile over a frame, fixed with staples, ties and stitching.
- [Vibration Welding](https://manufacturingprocesses.org/processes/joining/vibration-welding.md): Vibration welding rubs two thermoplastic parts together in-plane at low frequency under pressure until the interface melts and fuses.
- [Weaving](https://manufacturingprocesses.org/processes/joining/weaving.md): Weaving interlaces cane, rattan or synthetic fiber over a frame to build a structural surface out of flexible material.

## Finishing (25)

Hub: https://manufacturingprocesses.org/processes/finishing.md

- [Abrasive Blasting](https://manufacturingprocesses.org/processes/finishing/abrasive-blasting.md): Abrasive blasting propels media at a surface with compressed air to clean it, strip scale, or produce a uniform matte texture.
- [Anodizing](https://manufacturingprocesses.org/processes/finishing/anodizing.md): Anodizing grows a hard porous oxide layer on aluminum electrochemically, which can then be dyed and sealed for wear and corrosion resistance.
- [Bead Blasting](https://manufacturingprocesses.org/processes/finishing/bead-blasting.md): Bead blasting propels glass or ceramic beads at a surface to produce an even satin matte texture without cutting into it aggressively.
- [Black Oxide](https://manufacturingprocesses.org/processes/finishing/black-oxide.md): Black oxide converts the steel surface to magnetite in a hot alkaline bath, giving a black finish that adds essentially no thickness.
- [CNC Engraving](https://manufacturingprocesses.org/processes/finishing/cnc-engraving.md): CNC engraving cuts text and artwork into a surface with a small pointed or ball-nose tool following a programmed path.
- [Cerakote Ceramic Coating](https://manufacturingprocesses.org/processes/finishing/cerakote-ceramic-coating.md): Cerakote sprays a thin ceramic-filled polymer coating onto a prepared part and cures it, giving wear and corrosion resistance in a very thin film.
- [Chemical Conversion Coating](https://manufacturingprocesses.org/processes/finishing/chemical-conversion-coating.md): Chemical conversion coating reacts the metal surface with a chemical bath to form a thin protective, conductive or paint-adhering film.
- [E-Coating (Electrophoretic Deposition)](https://manufacturingprocesses.org/processes/finishing/e-coating-electrophoretic-deposition.md): E-coating deposits paint electrically from a bath so it plates evenly into every recess, then cures to a uniform film.
- [Electroplating](https://manufacturingprocesses.org/processes/finishing/electroplating.md): Electroplating deposits a metal coating from solution onto a conductive part by passing current through an electrolyte bath.
- [Electropolishing](https://manufacturingprocesses.org/processes/finishing/electropolishing.md): Electropolishing reverses plating: current dissolves the surface fastest at the peaks, leaving a bright, deburred, passive finish.
- [Foil Blocking and Embossing](https://manufacturingprocesses.org/processes/finishing/foil-blocking-and-embossing.md): Foil blocking and embossing press a heated die into the surface to transfer metallic foil, raise a relief, or do both at once.
- [Galvanizing](https://manufacturingprocesses.org/processes/finishing/galvanizing.md): Galvanizing coats steel with zinc, usually by hot dipping, so the zinc corrodes sacrificially and protects the steel beneath it.
- [Grinding, Sanding and Polishing](https://manufacturingprocesses.org/processes/finishing/grinding-sanding-and-polishing.md): Grinding, sanding and polishing remove a thin layer of material with bonded or coated abrasive to correct form or improve surface finish.
- [Hydro Transfer Printing](https://manufacturingprocesses.org/processes/finishing/hydro-transfer-printing.md): Hydro transfer printing floats a printed film on water, activates it, and wraps it around a submerged part to cover it in a pattern.
- [Laser Marking and Engraving](https://manufacturingprocesses.org/processes/finishing/laser-marking-and-engraving.md): Laser marking alters the surface with a focused beam — annealing, etching or engraving it — to make permanent, contact-free marks.
- [PVD Coating](https://manufacturingprocesses.org/processes/finishing/pvd-coating.md): PVD coating vaporises a solid target inside a vacuum chamber so it condenses on the part as a very thin, extremely hard film.
- [Pad Printing](https://manufacturingprocesses.org/processes/finishing/pad-printing.md): Pad printing lifts ink out of an etched plate with a silicone pad and transfers it onto a curved or recessed surface.
- [Passivation](https://manufacturingprocesses.org/processes/finishing/passivation.md): Passivation removes free iron from a stainless steel surface with an acid bath so the chromium oxide layer can reform unbroken.
- [Photo Etching](https://manufacturingprocesses.org/processes/finishing/photo-etching.md): Photo etching masks a surface photographically and etches shallow detail into it for decoration, marking or texture.
- [Powder Coating](https://manufacturingprocesses.org/processes/finishing/powder-coating.md): Powder coating sprays electrostatically charged dry polymer powder onto an earthed part, then bakes it into a tough continuous film.
- [Screen Printing](https://manufacturingprocesses.org/processes/finishing/screen-printing.md): Screen printing pushes ink through a stencilled mesh onto the surface below, laying down one opaque color per pass.
- [Spray Painting](https://manufacturingprocesses.org/processes/finishing/spray-painting.md): Spray painting atomises liquid coating onto a prepared surface, building a film that cures into a decorative and protective layer.
- [UV Inkjet Printing](https://manufacturingprocesses.org/processes/finishing/uv-inkjet-printing.md): UV inkjet printing jets UV-curable ink directly onto a flat substrate and cures it instantly under UV light.
- [Vacuum Metalizing](https://manufacturingprocesses.org/processes/finishing/vacuum-metalizing.md): Vacuum metalizing evaporates aluminum inside a vacuum chamber so it condenses as a mirror-bright film on the part surface.
- [Vibratory Tumbling and Mass Finishing](https://manufacturingprocesses.org/processes/finishing/vibratory-tumbling-and-mass-finishing.md): Vibratory tumbling agitates parts with abrasive media in a bowl so edges are broken and surfaces smoothed in bulk, without hand work.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes)*


<!-- ---------------- -->

---
type: category
name: "Forming"
processes: 60
url: https://manufacturingprocesses.org/processes/forming
---

# Forming

60 manufacturing processes in the forming family.

Forming processes shape material without removing it. Molten, softened or
solid stock is pushed into the geometry of a mold, a die or a tool, and
the part emerges at or near net shape.

This is where tooling cost and part cost diverge most sharply: a forming
process usually demands an expensive tool up front and then produces parts
for pennies, which is why volume is the first question to ask of any
forming route.

## Processes

| Process | Tolerances | Typical volumes | Lead time |
| --- | --- | --- | --- |
| [3D Thermal Laminating](https://manufacturingprocesses.org/processes/forming/3d-thermal-laminating.md) | Set by the machined substrate; the laminate itself adds roughly 0.008–0.024 in (0.2–0.6 mm) per laminated face depending on foil or veneer thickness. | 50–500,000 panels; economical at any quantity because no mold is required | Days rather than weeks — there is no tooling to build, only a router program. Press cycles run roughly 40–90 seconds per load; rotary profile wrapping runs continuously at 30–200 ft/min (10–60 m/min). |
| [Binder Jetting](https://manufacturingprocesses.org/processes/forming/binder-jetting.md) | Roughly ±0.5% of dimension on sintered metal (about ±0.008 in on a 1.5 in feature), with ±0.3% achievable on well-characterized geometry. Binder-jetted foundry sand molds hold approximately ±0.012 in (±0.3 mm). | 50–50,000 parts for sintered metal; 1 and upward for foundry sand molds and cores | 7–15 business days for sintered metal, most of it furnace scheduling and post-machining. Binder-jetted sand molds are frequently produced in 2–5 business days. |
| [Blow Molding](https://manufacturingprocesses.org/processes/forming/blow-molding.md) | ±0.020 in (±0.5 mm), roughly ±1%, on blown body dimensions; injection-molded neck finishes hold about ±0.005 in (±0.13 mm). Wall thickness is not directly controllable and commonly varies 30% or more across a part. | 1,000 to many millions; rotary lines run tens of thousands of containers per hour | 3–8 weeks for an aluminum EBM tool; 6–12 weeks for an ISBM preform tool plus blow mold; hours per production run thereafter |
| [CNC Wire Bending](https://manufacturingprocesses.org/processes/forming/cnc-wire-bending.md) | Roughly ±0.020 in (±0.5 mm) on leg lengths and ±1–2° on bend angles, tightening to about ±0.010 in and ±1° with first-article correction; tolerances accumulate along multi-bend parts | 100–1,000,000+ parts; a few hundred is enough to absorb the setup | 1–3 weeks including programming and first articles; days for repeat orders. No tooling lead time |
| [Centrifugal Casting](https://manufacturingprocesses.org/processes/forming/centrifugal-casting.md) | Outside diameter, cast against a metal die, holds substantially tighter than the bore, which is uncontrolled and always machined; wall thickness is set by pour volume and carries the loosest tolerance of the three | 1–10,000 parts; a single large ring is economic on a sand-lined mold | 2–6 weeks for a run against existing mold sizes; longer if a new permanent mold is required |
| [Ceramic Injection Molding (CIM)](https://manufacturingprocesses.org/processes/forming/ceramic-injection-molding-cim.md) | ±0.3–0.5% of dimension as fired (about ±0.006–0.010 in on a 2 in feature); diamond-ground features hold ±0.0002 in (±0.005 mm) or better. | 10,000–1,000,000+ parts per year | 12–20 weeks for tooling and shrink-factor development, then 2–4 weeks per production batch, most of it furnace scheduling. |
| [Ceramic Slip Casting](https://manufacturingprocesses.org/processes/forming/ceramic-slip-casting.md) | Roughly ±1–2% of dimension on the fired part, dominated by variation in drying and firing shrinkage rather than by the mold. | 100–50,000 pieces per year with plaster molds; higher with pressure casting | 3–8 weeks for model and mold making. In production, plaster molds typically yield one or two castings per mold per day, so output is set by mold count. |
| [Clay Throwing](https://manufacturingprocesses.org/processes/forming/clay-throwing.md) | — | 1–500 pieces | Days to weeks per batch. Throwing is minutes per piece, but drying takes days and pieces are held until a full kiln load is ready for each of two firings. |
| [Composite Laminating](https://manufacturingprocesses.org/processes/forming/composite-laminating.md) | ±0.010 in (±0.25 mm) on tool-side surfaces. Laminate thickness varies ±10–20% on open (bag-side) surfaces; matched tooling or RTM is required to control both faces. | 1–5,000 parts per year, depending on route: wet layup at the low end, RTM at the high end | 2–6 weeks including tooling for prototype work; 1–5 days per part once tooling exists, dominated by layup labor and cure cycle. |
| [Compression Molding](https://manufacturingprocesses.org/processes/forming/compression-molding.md) | About ±0.005 in (±0.13 mm) on features contained within one mold half; ±0.010 in (±0.25 mm) or more on dimensions spanning the parting line, since flash thickness varies with charge weight. Rubber tolerance classes follow ISO 3302-1. | 100–50,000 parts per year; above that, injection or transfer molding usually takes over | 3–8 weeks for tooling; cycles of 1–10+ minutes depending on section thickness, plus deflash and any post-cure |
| [DMC and SMC Molding](https://manufacturingprocesses.org/processes/forming/dmc-and-smc-molding.md) | About ±0.1–0.3% of dimension, roughly ±0.005 in/in, thanks to the near-zero mold shrinkage of low-profile compounds. Tighter than most thermoplastic molding, and dimensionally stable with temperature. | 1,000–100,000 parts per year | 10–20 weeks for steel tooling; cycle times of 60–180 seconds per part in production. |
| [Deep Drawing](https://manufacturingprocesses.org/processes/forming/deep-drawing.md) | Roughly ±0.010 in (±0.25 mm) on drawn diameters and depth; ironed walls hold tighter, near ±0.002 in (±0.05 mm). Wall thickness varies 10–25% from the blank gauge by design | 25,000–10,000,000+ parts; below about 5,000 a single-sided process is usually cheaper | 8–16 weeks for a multi-station draw die; 1–3 weeks per production run thereafter |
| [Die Casting](https://manufacturingprocesses.org/processes/forming/die-casting.md) | ±0.010 in (±0.25 mm) on the first inch, then ±0.002 in per additional inch (NADCA standard linear tolerance, aluminum); dimensions crossing the parting line need added allowance | 10,000–1,000,000+ parts per die; below about 5,000/year the tooling rarely pays back | Roughly 10–16 weeks for die build and sampling; 2–4 weeks per production release thereafter |
| [Dip Molding](https://manufacturingprocesses.org/processes/forming/dip-molding.md) | Inside diameter follows the former closely, about ±0.005 in (±0.13 mm); wall thickness is much looser at roughly ±0.010–0.030 in (±0.25–0.75 mm) and varies along the dip axis. | 100 to several million parts; economical at both ends because tooling is so inexpensive | 1–3 weeks for formers; hours per rack thereafter, with cycle time driven by fusion oven residence |
| [Direct Metal Laser Sintering (DMLS)](https://manufacturingprocesses.org/processes/forming/direct-metal-laser-sintering-dmls.md) | About ±0.004 in (±0.1 mm) on features up to 1 in (25 mm), then roughly ±0.2% of nominal. Suppliers commonly quote ±0.003 in (±0.076 mm) plus ±0.001 in/in for well-controlled geometry; machined features are far tighter. | 1–500 parts; economical wherever the geometry cannot be machined or cast | 5–15 business days. The print itself is typically 1–4 days, with stress relief, plate removal, support removal, and any machining or HIP adding most of the balance. |
| [Directed Energy Deposition (DED)](https://manufacturingprocesses.org/processes/forming/directed-energy-deposition-ded.md) | As-deposited surfaces are held to roughly ±0.02–0.04 in (±0.5–1.0 mm) at best, and large WAAM structures to considerably less. Final tolerances come from the machining operation that follows, not from the deposition. | 1–100 parts; repairs and one-off large structures are the core of the market | 1–4 weeks depending on size, heat treatment, and the machining that follows. Repairs of existing components are often turned around in days. |
| [Electroforming](https://manufacturingprocesses.org/processes/forming/electroforming.md) | The mandrel-contacting surface replicates the mandrel closely, so its accuracy is the mandrel's accuracy; wall thickness typically varies ±10–25% across a part unless shields and current thieves are used | 1–10,000 parts; single pieces are entirely practical, and permanent mandrels amortize over hundreds | 4–10 weeks including mandrel manufacture and process development; plating alone runs days per part for thick walls |
| [Electron Beam Melting (EBM)](https://manufacturingprocesses.org/processes/forming/electron-beam-melting-ebm.md) | About ±0.012–0.016 in (±0.3–0.4 mm), looser than laser powder bed fusion because of the thicker layers and coarser powder. Machined interfaces hold normal machining tolerances. | 1–10,000 parts per year; unusually well suited to serial production of a repeating titanium part | 7–20 business days. Build time is competitive, but the controlled cooldown from process temperature adds many hours to every cycle. |
| [Filament Winding](https://manufacturingprocesses.org/processes/forming/filament-winding.md) | Inside diameter ±0.005–0.010 in (±0.13–0.25 mm), set by the mandrel. Wall thickness and outside diameter carry roughly ±0.5–1% because thickness builds in whole layers. | 100–500,000 parts per year | 6–16 weeks for mandrel and pattern development on a new part; hours per part in production once the winding program exists. |
| [Forging](https://manufacturingprocesses.org/processes/forming/forging.md) | Commercial impression-die forgings hold roughly ±0.030 in (±0.8 mm) on small parts, widening with size, plus separate allowances for die wear, mismatch, and die closure; machine anything needing better than about ±0.010 in | 1–100 pieces open die; 1,000–500,000+ per year impression die | 8–16 weeks to sink an impression die and produce first articles; 3–8 weeks per production run. Open-die work in 2–6 weeks with no tooling |
| [Fused Deposition Modeling (FDM)](https://manufacturingprocesses.org/processes/forming/fused-deposition-modeling-fdm.md) | ±0.020 in (±0.5 mm) or ±0.5% of nominal, whichever is greater, on desktop-class machines. Industrial systems hold ±0.008 in (±0.2 mm) or ±0.002 in/in. Drilled and reamed features are far tighter. | 1–500 parts | 1–5 business days; small parts print in 1–4 hours and a full plate in 10–30 hours. No tooling, so the first part ships as fast as the last. |
| [Glassblowing](https://manufacturingprocesses.org/processes/forming/glassblowing.md) | — | 1–1,000 pieces hand blown; hundreds of thousands to hundreds of millions machine formed | Hours to days per piece in a studio, plus annealing time that scales with section thickness. Container mold sets take weeks to months; machine production is continuous. |
| [Heat Treatment](https://manufacturingprocesses.org/processes/forming/heat-treatment.md) | — | 1 piece to millions; furnace loads are priced by weight, so batching drives the unit cost | 3–10 business days for standard atmosphere hardening and tempering; longer for carburizing, and 1–3 weeks for long nitriding cycles |
| [Hydroforming](https://manufacturingprocesses.org/processes/forming/hydroforming.md) | Roughly ±0.010–0.030 in (±0.25–0.75 mm) on formed features; sections calibrated hard against the die hold better than regions that are only partly expanded | 1,000–250,000 parts per year; above roughly 500,000, stamping's cycle time wins | 8–16 weeks for die design, build, and process development; 2–6 weeks per production run thereafter |
| [Injection Molding](https://manufacturingprocesses.org/processes/forming/injection-molding.md) | ±0.005 in (±0.127 mm) commercial on features under 1 in; ±0.002 in (±0.05 mm) precision; ±0.1–0.2% of dimension on larger parts. Add tolerance for anything spanning the parting line or a side action. | 1,000–1,000,000+ parts; economics are strongest above 10,000 | 2–4 weeks for an aluminum bridge tool, 6–12 weeks for a hardened steel production tool; parts in hours to days once the tool is sampled |
| [Insert Molding](https://manufacturingprocesses.org/processes/forming/insert-molding.md) | Molded features hold normal injection molding tolerances, about ±0.005 in (±0.127 mm). Insert position relative to molded features is looser and depends on how positively the insert is located — budget the insert's own tolerance plus the locating clearance. | 1,000–1,000,000+ parts; below 1,000, post-mold insert installation is usually cheaper | 6–12 weeks for tooling including insert locating features and presence sensing; cycles run several seconds longer than the equivalent plain molding |
| [Investment Casting](https://manufacturingprocesses.org/processes/forming/investment-casting.md) | About ±0.005 in (±0.13 mm) on the first inch plus roughly ±0.003 in per additional inch; flatness and straightness are usually the limiting characteristics on long or thin parts | 100–50,000 parts per year; single prototypes possible with printed wax patterns | 6–12 weeks for wax tooling and first article, then 4–8 weeks per production run; printed-wax prototypes in 2–4 weeks with no tool |
| [Lampworking](https://manufacturingprocesses.org/processes/forming/lampworking.md) | — | 1–1,000 pieces; unit cost is flat with quantity because there is no tooling | Hours to days per piece, plus a kiln annealing cycle sized to the thickest section. Complex laboratory assemblies are quoted in bench hours rather than lead weeks. |
| [Liquid Silicone Rubber (LSR) Molding](https://manufacturingprocesses.org/processes/forming/liquid-silicone-rubber-lsr-molding.md) | About ±0.002–0.004 in (±0.05–0.10 mm) on small precision features; ISO 3302-1 class M1 is achievable on tightly controlled dimensions. Larger dimensions scale with the 2–3.5% shrinkage allowance. | 1,000–1,000,000+ parts; below about 500, compression molding avoids the tooling cost | 6–12 weeks for a production LSR tool with cold runner and vacuum sealing; cycles of 20–60 seconds thereafter, plus any post-cure |
| [Masked Stereolithography (MSLA / DLP)](https://manufacturingprocesses.org/processes/forming/masked-stereolithography-msla-dlp.md) | About ±0.004–0.008 in (±0.1–0.2 mm) on general geometry; ±0.002 in (±0.05 mm) is achievable on small parts with fine-pitch panels and a compensated first article. | 1–5,000 small parts; cost per part falls in proportion to how densely the plate is filled | Same day to 3 business days. A 4 in (100 mm) tall build runs roughly 3–8 hours plus wash and post-cure, regardless of how many parts are on the plate. |
| [Material Jetting (PolyJet)](https://manufacturingprocesses.org/processes/forming/material-jetting-polyjet.md) | About ±0.004 in (±0.1 mm) on parts up to 4 in (100 mm), and roughly ±0.008 in (±0.2 mm) or ±0.1% of nominal on larger geometry. | 1–50 parts | 1–3 business days. Printing is typically 4–12 hours, with support removal adding anywhere from minutes to several hours depending on internal geometry. |
| [Metal Extrusion](https://manufacturingprocesses.org/processes/forming/metal-extrusion.md) | Standard mill tolerances per published aluminum extrusion practice — roughly ±0.010 in (±0.25 mm) on small cross-section dimensions, wall thickness held as a percentage, plus separate straightness, twist, and flatness allowances | Several hundred pounds minimum per die run, up to millions of pounds; unit cost is nearly flat across that range | 2–4 weeks for a new die and first run; 2–6 weeks per production order thereafter |
| [Metal Injection Molding](https://manufacturingprocesses.org/processes/forming/metal-injection-molding.md) | About ±0.3% of the dimension as standard (roughly ±0.003 in on a 1 in feature), with ±0.1–0.2% achievable on controlled features or after sizing; machine anything tighter | 20,000–1,000,000+ parts per year; below about 5,000 machining is usually cheaper | 8–14 weeks for tooling and first articles; 4–8 weeks per production run thereafter, set largely by furnace scheduling |
| [Metal Spinning](https://manufacturingprocesses.org/processes/forming/metal-spinning.md) | Roughly ±0.010–0.030 in (±0.25–0.75 mm) on diameters, tighter on CNC machines with hard mandrels; wall thickness varies through the part in conventional spinning and follows the sine law in shear spinning | 1–5,000 parts; above about 5,000 deep drawing usually wins on cycle time | 1–3 weeks including mandrel; days for repeat work on an existing mandrel |
| [Metal Stamping](https://manufacturingprocesses.org/processes/forming/metal-stamping.md) | ±0.005 in (±0.13 mm) typical on blanked and pierced features; ±0.001–0.002 in (±0.025–0.05 mm) achievable in precision dies. Formed dimensions spanning several stations hold looser than single-station features | 25,000–10,000,000+ parts per year; below about 10,000 laser cutting and press braking usually win | 10–20 weeks for progressive die design, build, and tryout; 1–3 weeks per production release thereafter |
| [Multi Jet Fusion (MJF)](https://manufacturingprocesses.org/processes/forming/multi-jet-fusion-mjf.md) | ±0.008 in (±0.2 mm) or ±0.3% of nominal, whichever is greater. Tighter features are machined or reamed after the build. | 10–10,000 parts per year; unit cost is close to flat across that range when builds are nested well | 3–7 business days. A full build runs roughly 10–15 hours of printing plus a comparable controlled cooling cycle before depowdering. |
| [Overmolding](https://manufacturingprocesses.org/processes/forming/overmolding.md) | Substrate features hold normal injection molding tolerances, about ±0.005 in (±0.127 mm). Dimensions taken over the elastomer are looser and compressible; in insert-transfer overmolding, add the substrate loading clearance to the position tolerance of the second shot. | 5,000–1,000,000+ parts; two-shot tooling pays back above roughly 100,000 | 8–14 weeks for a two-shot rotating tool; 6–10 weeks for a pair of insert-transfer tools; cycles are one conventional molding cycle plus the elastomer shot |
| [Panel Beating](https://manufacturingprocesses.org/processes/forming/panel-beating.md) | — | 1–50 pieces; a form block for jig chasing becomes worthwhile above a handful of repeats | Days to weeks per panel, driven entirely by shape difficulty and finish requirements |
| [Paper Pulp Molding](https://manufacturingprocesses.org/processes/forming/paper-pulp-molding.md) | ±0.020–0.060 in (±0.5–1.5 mm) for thick-wall and transfer molded parts; ±0.010–0.020 in (±0.25–0.5 mm) for thermoformed precision pulp. | 10,000–10,000,000+ parts per year | 4–10 weeks for tooling, substantially faster than plastic molding tools. Production cycles run from seconds to a few minutes per part depending on grade and thickness. |
| [Plastic Extrusion](https://manufacturingprocesses.org/processes/forming/plastic-extrusion.md) | ±0.005 in (±0.13 mm) on small dimensions held against a calibrator; ±0.020–0.060 in (±0.5–1.5 mm) on free-extruded surfaces of large profiles; angles ±1–2°. Bow and twist are specified separately per unit length. | Quoted by weight — minimum runs of several hundred to a few thousand pounds; annual volumes reach millions of feet | 3–6 weeks for a custom profile die plus calibration tooling, including one or more die correction iterations; days per production run thereafter |
| [Powder Metallurgy (Press and Sinter)](https://manufacturingprocesses.org/processes/forming/powder-metallurgy-press-and-sinter.md) | Roughly ±0.001–0.002 in per inch on dimensions perpendicular to the pressing direction, controlled by die and core rod steel; dimensions along the pressing axis hold considerably looser unless the part is sized after sintering | 25,000–5,000,000+ parts per year; below about 5,000–10,000 machining is usually cheaper | 12–20 weeks for die set design, build, and qualification; 3–6 weeks per production run thereafter |
| [Press Braking](https://manufacturingprocesses.org/processes/forming/press-braking.md) | ±0.010 in (±0.25 mm) on a single bend dimension, accumulating roughly ±0.010 in per additional bend; bend angle ±1°, or ±0.5° with in-process angle measurement | 1–25,000 parts; above about 25,000 identical parts, hard tooling usually wins | 1–10 business days in a job shop; same-day for simple parts with material in stock. No tooling lead time |
| [Press Molding Ceramics](https://manufacturingprocesses.org/processes/forming/press-molding-ceramics.md) | Roughly ±0.5–1% of dimension for dry-pressed parts and ±1–2% for plastic-formed jiggered and ram-pressed ware. Ceramic tile dimensional classes are defined in ISO 13006 and ANSI A137.1. | 1,000–10,000,000+ pieces per year depending on variant | 6–16 weeks for steel dry-pressing tooling; 2–6 weeks for ram press dies or jigger molds. Production cycles run about one second (dry pressing) to a minute (ram pressing). |
| [Pultrusion](https://manufacturingprocesses.org/processes/forming/pultrusion.md) | Governed by ASTM D3917 for glass-reinforced pultruded shapes (EN 13706 in Europe). Wall thickness is held closest; overall width, straightness, and twist carry progressively looser bands. Do not assume machining-class tolerances on any pultruded dimension. | Thousands to millions of linear feet; a custom die requires a long run to amortize | 10–20 weeks for a new die; standard catalog profiles ship from stock. Production runs are measured in feet per minute rather than parts per hour. |
| [Rapid Prototyping](https://manufacturingprocesses.org/processes/forming/rapid-prototyping.md) | Process-dependent: ±0.004 in (±0.1 mm) for SLA, MSLA, PolyJet, and metal powder bed fusion; ±0.008–0.012 in (±0.2–0.3 mm) for MJF and SLS; ±0.020 in (±0.5 mm) for desktop FDM. Machined features are tighter than any of these. | 1–1,000 parts; unit cost is essentially flat with quantity | 1–5 business days for plastic parts; 5–15 business days for metal, where heat treatment and machining dominate. No tooling lead time in any case. |
| [Reaction Injection Molding](https://manufacturingprocesses.org/processes/forming/reaction-injection-molding.md) | About ±0.020 in (±0.5 mm) on features under 12 in; on large panels expect roughly ±0.1–0.2% of the dimension. Thermal expansion of polyurethane is high, so specify the inspection temperature. | 250–10,000 parts per year; below that use urethane casting, above roughly 25,000 consider injection molding | 4–10 weeks for cast aluminum or nickel shell tooling; 1–5 minute cycles thereafter, plus trim and paint |
| [Roll Forming](https://manufacturingprocesses.org/processes/forming/roll-forming.md) | ±0.010–0.030 in (±0.25–0.75 mm) on cross-section dimensions and about ±1° on formed angles; cut-length tolerance depends on whether cutoff is flying or stationary | 20,000+ linear feet per profile; below that, press braking is usually cheaper | 12–20 weeks to design, cut, and debug a roll set; days to weeks per production run thereafter |
| [Rotation Molding](https://manufacturingprocesses.org/processes/forming/rotation-molding.md) | About ±1–2% of the dimension on mold-formed features; wall thickness commonly varies ±10–20% around nominal. Polyethylene shrinkage of 3.0–3.5% must be built into the tool. | 10–10,000 parts per year; the sweet spot is large parts at low to medium volume | 4–10 weeks for tooling depending on size and whether it is fabricated steel or cast aluminum; a 10–60 minute cycle per part thereafter |
| [Sand Casting](https://manufacturingprocesses.org/processes/forming/sand-casting.md) | About ±0.030 in (±0.8 mm) on the first inch plus roughly ±0.003 in per additional inch in green sand; no-bake and shell molds hold tighter. Allow up to 0.030 in (0.8 mm) mismatch across the parting line | 1–100,000 parts per year; economical from a single piece | 1–4 weeks for a pattern, then 1–3 weeks per casting run; prototype castings from printed patterns in 1–2 weeks |
| [Selective Laser Melting (SLM)](https://manufacturingprocesses.org/processes/forming/selective-laser-melting-slm.md) | About ±0.004 in (±0.1 mm) up to 1 in (25 mm), then roughly ±0.2% of nominal. Machined features hold normal machining tolerances. | 1–500 parts | 5–15 business days including stress relief, cut-off, support removal, and any machining or hot isostatic pressing. |
| [Selective Laser Sintering (SLS)](https://manufacturingprocesses.org/processes/forming/selective-laser-sintering-sls.md) | ±0.012 in (±0.3 mm) up to about 4 in (100 mm), then ±0.3% of nominal. Critical bores are normally reamed after the build. | 1–5,000 parts; cost per part is nearly flat with quantity | 3–7 business days. A full-height build is 20–40 hours of print time plus 8–24 hours of controlled cooling before parts can be removed. |
| [Sheet Metal Fabrication](https://manufacturingprocesses.org/processes/forming/sheet-metal-fabrication.md) | ±0.005 in (±0.13 mm) on laser-cut features; about ±0.010 in (±0.25 mm) per bend, accumulating across bends; welded assemblies are substantially looser unless fixtured or machined after welding | 1–10,000 parts per year; above about 10,000 stamping or roll forming usually wins | 3–10 business days typical, 1–3 days for quick-turn simple parts; no tooling lead time |
| [Steam Bending](https://manufacturingprocesses.org/processes/forming/steam-bending.md) | — | 1–5,000 pieces per year | Days per batch: roughly one hour of steaming per inch (25 mm) of thickness, a minute to bend, then 1–7 days held on the former while the piece dries and sets. |
| [Stereolithography (SLA)](https://manufacturingprocesses.org/processes/forming/stereolithography-sla.md) | ±0.006 in (±0.15 mm) over the first inch, then about ±0.001 in/in (±0.1% of nominal); ±0.004 in (±0.1 mm) on small, well-supported features. Vendors quote between ±0.1% and ±0.2% of dimension. | 1–100 parts; a few hundred when the SLA part serves as a master for urethane casting | 1–3 business days for typical prototypes, same day on desktop machines. A 6 in (150 mm) tall build runs 8–20 hours plus 1–2 hours of washing and post-cure. |
| [Superforming](https://manufacturingprocesses.org/processes/forming/superforming.md) | — | 10–10,000 parts per year; above roughly 10,000 conventional stamping wins on cycle time | 8–16 weeks for tooling and process development; 4–10 weeks per production run, constrained by the long cycle time |
| [Swaging](https://manufacturingprocesses.org/processes/forming/swaging.md) | Roughly ±0.002–0.005 in (±0.05–0.13 mm) on swaged outside diameter; inside diameter holds a comparable tolerance only when a mandrel is used | 100–1,000,000+ parts; practical in dozens on a manual machine | 2–6 weeks including a new die set; days for repeat work on existing tooling |
| [Thermoforming](https://manufacturingprocesses.org/processes/forming/thermoforming.md) | ±0.030 in (±0.75 mm) on vacuum-formed tool-side features; ±0.015 in (±0.4 mm) pressure formed and on CNC-trimmed edges. Non-tool-side dimensions carry all sheet thickness variation. | 50–50,000 parts per year for heavy gauge; thin-gauge roll-fed packaging runs into the millions | 1–2 weeks for prototype wood or epoxy tooling; 2–4 weeks for a production aluminum tool; days per run thereafter |
| [Tube and Section Bending](https://manufacturingprocesses.org/processes/forming/tube-and-section-bending.md) | Roughly ±1° on bend angle and plane of bend, and about ±0.030 in (±0.75 mm) on leg lengths; tolerances accumulate along a multi-bend part, so dimension from a single datum end | 1–500,000 parts; economical from a single piece if tooling already exists | Days to 2 weeks against existing tooling; 4–8 weeks if a new radius tool set is required |
| [Vacuum Casting (Urethane Casting)](https://manufacturingprocesses.org/processes/forming/vacuum-casting.md) | ±0.3% of the nominal dimension, with a minimum of about ±0.010 in (±0.25 mm). Accuracy depends on the master pattern and drifts slightly as the silicone mold ages. | 10–100 parts; 15–25 parts per silicone mold before it must be replaced | 5–15 working days from master pattern to first parts, including silicone cure; a few parts per mold per day thereafter |
| [Wood Laminating](https://manufacturingprocesses.org/processes/forming/wood-laminating.md) | Springback is typically 1–3% of the bend curvature with a rigid adhesive, against 5–10% for a steamed bend. Forms are made slightly overbent and calibrated on the first assembly. | 1–5,000 pieces per year | 1–3 weeks including form making. In production, cure time per glue-up is several hours at room temperature or minutes with hot pressing or radio-frequency curing. |

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/forming)*


<!-- ---------------- -->

---
type: process
name: "3D Thermal Laminating"
category: "Forming"
subcategory: "Composites"
materials: ["Composite", "Wood"]
tolerances: "Set by the machined substrate; the laminate itself adds roughly 0.008–0.024 in (0.2–0.6 mm) per laminated face depending on foil or veneer thickness."
volumes: "50–500,000 panels; economical at any quantity because no mold is required"
lead_time: "Days rather than weeks — there is no tooling to build, only a router program. Press cycles run roughly 40–90 seconds per load; rotary profile wrapping runs continuously at 30–200 ft/min (10–60 m/min)."
url: https://manufacturingprocesses.org/processes/forming/3d-thermal-laminating
---

# 3D Thermal Laminating

3D thermal laminating bonds a decorative foil or veneer to a contoured substrate with heat and vacuum so it wraps around the profiled edges.

- **Category**: [Forming](https://manufacturingprocesses.org/processes/forming.md)
- **Family**: Composites
- **Materials**: Composite, Wood
- **Typical tolerances**: Set by the machined substrate; the laminate itself adds roughly 0.008–0.024 in (0.2–0.6 mm) per laminated face depending on foil or veneer thickness.
- **Typical volumes**: 50–500,000 panels; economical at any quantity because no mold is required
- **Lead time**: Days rather than weeks — there is no tooling to build, only a router program. Press cycles run roughly 40–90 seconds per load; rotary profile wrapping runs continuously at 30–200 ft/min (10–60 m/min).

## Overview

3D thermal laminating bonds a thermoplastic foil or a thin veneer to a profiled panel using heat and vacuum, so the surface material wraps continuously over the face and around every routed edge in one operation. The result is a component with no visible edge band and no seam at the profile — the reason it dominates painted-look kitchen and bathroom cabinet doors.

Two variants cover the field. 3D laminating (3DL), also called membrane pressing or thermofoiling, uses a heated silicone membrane in a vacuum press to draw foil down over a routed MDF panel in roughly a minute per cycle. 3D rotary laminating (3Dr), or profile wrapping, feeds a continuous linear profile through a train of rollers that fold the foil around it at line speeds of roughly 30–200 ft/min (10–60 m/min), which is how skirting, door casing, and moldings are surfaced.

The economic argument is tooling: neither variant needs a mold. The only "tool" is the router program that shapes the substrate, so a new door profile costs a CNC program rather than a die.

## How it works

1. **Machine the substrate.** MDF is the standard substrate because it routes to a dense, uniform, sealed profile edge that particleboard cannot match. The panel is profiled on a CNC router and sanded — any tearout or fuzz telegraphs straight through the foil.
2. **Apply adhesive.** A reactive polyurethane (PUR) or EVA hot-melt is roller-coated or sprayed onto the face and profile and allowed to flash off. The choice matters in service: EVA softens at a much lower temperature than PUR, which is why foil doors adjacent to ovens and dishwashers are specified with PUR.
3. **Load the press (3DL).** Panels are laid on the press table with spacing between them so the membrane can reach down the sides of each. The foil sheet is laid over the whole table.
4. **Heat and draw down.** The silicone membrane descends, the foil is heated to its forming range — for typical PVC and PET foils this is in the region of 250–290 °F (120–145 °C) — and vacuum is drawn beneath while pressure is applied above. The membrane forces the softened foil into every profile detail and around the panel edges. Cycle time is roughly 40–90 seconds.
5. **Cool and separate.** Panels are removed and the excess foil between them is trimmed, usually with a router or knife following the panel edge on the underside.
6. **Profile wrapping (3Dr).** For linear profiles the sequence is continuous instead: the profile is fed through a glue applicator and then through a series of shaped pressure rollers that progressively fold and press the foil around the section, with a trimming station at the end.
7. **Back-face and finish.** Membrane-pressed panels are laminated on the reverse with a backing foil or balancing laminate to control moisture movement, and the assembled component is drilled and hardware-fitted.

## Design guidelines

### Radii, not arrises
Foil will not conform to a sharp corner: it thins, bridges, or splits. Keep external radii at 0.08–0.12 in (2–3 mm) minimum and internal radii at 0.16 in (4 mm) or more. Deep, narrow profile detail is the most common cause of foil bridging.

### Profile depth and undercut
The membrane presses from above, so any part of the profile it cannot reach will not be wrapped. Avoid undercuts entirely, and keep grooves wide enough relative to their depth that the membrane can enter — a narrow deep groove leaves an unbonded bridge over it.

### Substrate quality is the finish
Foil is thin and takes the shape of what it covers. A machining mark, a fuzzy routed edge, or a swollen MDF region shows through as a defect. Sand the profile properly and dust-extract before glue.

### Panel spacing in the press
Panels must be spaced far enough apart on the table for the membrane to reach the full depth of each edge. Crowding the table is what produces poorly wrapped edges and is a production-planning constraint as much as a design one.

### Specify the adhesive for the service temperature
PUR reactive hot-melt gives substantially better heat and moisture resistance than EVA. Any door or panel near an oven, hob, dishwasher, or steam source should be specified PUR, with a heat deflector where the appliance manufacturer requires one.

### Balance the panel
Laminate the back face as well as the front. An MDF panel surfaced on one side only takes up moisture asymmetrically and cups.

### Edge and end-grain sealing
The routed profile edge of MDF is more absorbent than the face. This is why the wrap must be continuous over it — an unwrapped or poorly bonded edge is where moisture ingress and delamination begin.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| External radius | 0.12 in (3 mm) | 0.08 in (2 mm) | Foil thins and splits on sharp corners |
| Internal radius | 0.20 in (5 mm) | 0.16 in (4 mm) | Membrane must reach the corner |
| Undercuts | None | — | Membrane presses from one direction |
| Groove width:depth | 2:1 or wider | 1:1 | Membrane bridges narrow grooves |
| Substrate | MDF | Particleboard edges too coarse | Profile edge quality is the finish |
| Adhesive | PUR near heat sources | EVA general purpose | EVA softens at low temperature |
| Back face | Balancing laminate | — | One-sided panels cup |

## Variants

- 3D Laminating (3DL)
- 3D Rotary Laminating (3Dr)

## Cost drivers

The defining cost feature is the absence of tooling. A new profile requires a router program and a cutter, not a mold or a die, so design changes are inexpensive and short runs are viable. That is why the process took over cabinet doors from painted and veneered construction across most of the market.

Running cost is press cycle time and foil. A membrane press cycles in roughly 40–90 seconds and can hold multiple panels per cycle, so throughput depends on how efficiently the table is packed — which is in tension with the spacing needed for good edge wrap. Foil is priced per square meter, and the trimmed material between panels is waste.

Labor sits in the machining, sanding, glue application, and trimming steps rather than in the pressing itself. Rework is expensive because a defect usually means stripping and re-machining rather than patching.

Volume breakpoints: 3DL is economical from tens of panels upward and scales to hundreds of thousands. There is no volume at which it needs a tooling investment, which is its structural advantage over molded or [compression molded](/processes/forming/compression-molding) alternatives.

1. Design profiles with generous radii — foil conformability, not the router, sets the limit.
2. Standardize on a small number of profiles so the press table can be packed with mixed work.
3. Specify PUR adhesive wherever heat or moisture is credible; the failure mode is delamination in service.
4. Always balance the back face.
5. Use rotary wrapping (3Dr) for anything with a constant linear section — it is far faster than pressing.

## FAQ

### What is the minimum radius for a foil-wrapped profile?

Roughly 0.08–0.12 in (2–3 mm) on external corners and 0.16 in (4 mm) or more on internal ones. Foil is drawn over the profile by a membrane and thins as it stretches, so a sharp arris either splits the foil or leaves it unbonded and prone to lifting.

### Why is MDF used as the substrate?

MDF routes to a dense, uniform, closed profile edge that takes adhesive evenly. Particleboard edges are coarse and porous, so the foil telegraphs the surface texture and bonds unevenly. Since the foil is thin and takes the shape of whatever it covers, substrate quality is the finish quality.

### What is the difference between 3DL and 3Dr?

3D laminating (membrane pressing) presses foil over a shaped panel in a vacuum press, wrapping the face and all four profiled edges in one 40–90 second cycle. 3D rotary laminating, or profile wrapping, runs a constant linear section continuously through shaped rollers at roughly 30–200 ft/min (10–60 m/min) and is used for moldings, skirting, and casing.

### Why do thermofoil cabinet doors delaminate near ovens?

Almost always because an EVA hot-melt adhesive was used and its softening temperature was exceeded. Reactive polyurethane (PUR) adhesive has substantially higher heat and moisture resistance and should be specified for any panel adjacent to an oven, hob, dishwasher, or steam source.

### Does a foil-laminated panel need finishing on the back?

Yes. MDF surfaced on one face only absorbs and releases moisture asymmetrically and cups. A backing foil or balancing laminate on the reverse is standard practice, not an optional cosmetic step.

### What tooling does 3D thermal laminating require?

None beyond a router cutter and CNC program for the substrate profile, plus the press itself. That is the process's central economic advantage: a new door profile costs a program rather than a mold, so short runs and frequent design changes are inexpensive.

## Alternative processes

- [Spray Painting](https://manufacturingprocesses.org/processes/finishing/spray-painting.md): Spray painting atomises liquid coating onto a prepared surface, building a film that cures into a decorative and protective layer.
- [Powder Coating](https://manufacturingprocesses.org/processes/finishing/powder-coating.md): Powder coating sprays electrostatically charged dry polymer powder onto an earthed part, then bakes it into a tough continuous film.
- [Thermoforming](https://manufacturingprocesses.org/processes/forming/thermoforming.md): Thermoforming heats a plastic sheet until pliable and pulls or presses it over a single-sided mold, producing large thin-walled parts with low tooling cost.
- [Wood Laminating](https://manufacturingprocesses.org/processes/forming/wood-laminating.md): Wood laminating glues thin veneers or kerfed strips over a form so the assembly holds a curve permanently once the adhesive has cured.
- [Compression Molding](https://manufacturingprocesses.org/processes/forming/compression-molding.md): Compression molding presses a measured charge of rubber or thermoset into an open heated mold that closes to cure it into its final shape.

## Related processes

- [Wood Laminating](https://manufacturingprocesses.org/processes/forming/wood-laminating.md): Wood laminating glues thin veneers or kerfed strips over a form so the assembly holds a curve permanently once the adhesive has cured.
- [Thermoforming](https://manufacturingprocesses.org/processes/forming/thermoforming.md): Thermoforming heats a plastic sheet until pliable and pulls or presses it over a single-sided mold, producing large thin-walled parts with low tooling cost.
- [CNC Machining](https://manufacturingprocesses.org/processes/cutting/cnc-machining.md): CNC machining removes material from solid stock with a programmed cutting tool, producing accurate parts directly from a CAD model with no tooling.
- [Adhesive Bonding](https://manufacturingprocesses.org/processes/joining/adhesive-bonding.md): Adhesive bonding joins parts with a cured polymer layer that spreads load over the whole bond area instead of concentrating it at fastener points.
- [Compression Molding](https://manufacturingprocesses.org/processes/forming/compression-molding.md): Compression molding presses a measured charge of rubber or thermoset into an open heated mold that closes to cure it into its final shape.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/forming/3d-thermal-laminating)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Binder Jetting"
category: "Forming"
subcategory: "Additive Manufacturing"
materials: ["Metal", "Ceramic", "Composite"]
tolerances: "Roughly ±0.5% of dimension on sintered metal (about ±0.008 in on a 1.5 in feature), with ±0.3% achievable on well-characterized geometry. Binder-jetted foundry sand molds hold approximately ±0.012 in (±0.3 mm)."
volumes: "50–50,000 parts for sintered metal; 1 and upward for foundry sand molds and cores"
lead_time: "7–15 business days for sintered metal, most of it furnace scheduling and post-machining. Binder-jetted sand molds are frequently produced in 2–5 business days."
url: https://manufacturingprocesses.org/processes/forming/binder-jetting
---

# Binder Jetting

Binder jetting prints liquid binder into a powder bed to hold each layer, producing green parts that are later sintered or infiltrated.

- **Category**: [Forming](https://manufacturingprocesses.org/processes/forming.md)
- **Family**: Additive Manufacturing
- **Materials**: Metal, Ceramic, Composite
- **Typical tolerances**: Roughly ±0.5% of dimension on sintered metal (about ±0.008 in on a 1.5 in feature), with ±0.3% achievable on well-characterized geometry. Binder-jetted foundry sand molds hold approximately ±0.012 in (±0.3 mm).
- **Surface finish**: Ra 200–500 µin (5–13 µm) on sintered metal as built, dependent on powder size and orientation; machining or tumbling improves it. Sand molds are far rougher and impart their own texture to the casting.
- **Typical volumes**: 50–50,000 parts for sintered metal; 1 and upward for foundry sand molds and cores
- **Lead time**: 7–15 business days for sintered metal, most of it furnace scheduling and post-machining. Binder-jetted sand molds are frequently produced in 2–5 business days.

## Overview

Binder jetting prints a liquid binder from inkjet heads into a powder bed, gluing each layer into a fragile "green" part that is later cured, depowdered, and then sintered or infiltrated to full strength. Nothing is melted during printing, which is the key to the process: there is no melt pool, no residual stress, no supports, and no requirement that the material be weldable.

Because whole layers are printed rather than scanned, binder jetting is among the fastest additive processes, with layers of 0.002–0.004 in (50–100 µm) for metals and 0.008–0.016 in (200–400 µm) for foundry sand. Metal parts shrink 15–20% linearly during sintering, and managing that shrinkage — rather than printing — is where the engineering effort goes.

The three commercial branches are metal (316L, 17-4 PH, low-alloy steels, copper, tungsten carbide), foundry sand molds and cores for [sand casting](/processes/forming/sand-casting), and full-color gypsum models. Volumes range from one sand mold to tens of thousands of small sintered metal parts.

## How it works

1. **Recoat.** A roller or blade spreads a layer of powder — 0.002–0.004 in (50–100 µm) for metal, 0.008–0.016 in (200–400 µm) for silica sand.
2. **Print binder.** A piezo inkjet carriage traverses the bed, jetting binder only where the cross-section is solid. Print speed depends on carriage passes, not on cross-sectional complexity, so a bed full of parts prints little slower than one part.
3. **Cure in place.** For metal systems the bed is warmed (commonly around 180–200 °C / 355–390 °F) to polymerize the binder and give the green part enough strength to be handled. Sand systems cure at or near room temperature with a furan or phenolic binder and an activator.
4. **Depowder.** The green part is dug out and cleaned with brushes and low-pressure air. Green strength is comparable to chalk; this is the most fragile point in the whole workflow and drives minimum wall thickness.
5. **Debind and sinter (metal only).** The part is set on ceramic setters and run through a furnace cycle that first burns out the binder and then sinters the metal — around 1,350–1,400 °C (2,460–2,550 °F) for 316L. The part shrinks 15–20% in every direction and typically reaches 96–99.5% density.
6. **Compensate and verify.** The printed geometry is scaled up by the expected shrinkage before printing, and thin or overhanging sections need setters or sacrificial supports to stop them sagging under gravity at sintering temperature. First-article iteration on the scaling factors is normal.
7. **Finish.** Sintered parts can be machined, plated, or polished conventionally. Sand molds go straight to the foundry; color models are infiltrated with cyanoacrylate or resin for strength.

## Design guidelines

### Design for sintering, not for printing
The printer will make almost anything. The furnace will not. Long unsupported spans, heavy sections cantilevered off thin ones, and large flat plates all sag or distort at sintering temperature. Keep sections uniform, support long horizontal features on setters, and expect the first article to reveal distortion the CAD model did not.

### Uniform wall thickness
Non-uniform sections shrink at different rates and pull the part out of shape. Aim for constant wall thickness the way you would for [metal injection molding](/processes/forming/metal-injection-molding) — the sintering physics is essentially the same.

### Wall thickness
Green strength sets the floor: 0.040 in (1 mm) minimum for metal, and 0.080 in (2 mm) if the part is large or will be handled much before sintering. Sand molds work at far heavier sections.

### Shrinkage and tolerance
Plan for 15–20% linear shrinkage in metal, applied as a scaling factor. Achievable tolerance is proportional — roughly ±0.5% of dimension after sintering, sometimes ±0.3% on well-characterized geometry. Any feature needing better than that must be machined afterward, so add stock.

### Holes and threads
Sintering closes small features. Keep holes at 0.060 in (1.5 mm) or larger, and machine or tap all threads after sintering — see the [tap drill chart](/charts/tap-drill-chart) for pilot sizes, remembering the pilot must be scaled up for shrinkage before printing.

### Powder removal
Enclosed cavities hold un-bonded powder that must be removed while the part is still green and fragile. Provide generous access ports of 0.20 in (5 mm) or more.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Wall thickness (metal) | 0.080 in (2 mm) | 0.040 in (1 mm) | Green strength during depowdering |
| Hole diameter | 0.12 in (3 mm) | 0.060 in (1.5 mm) | Sintering closes small features |
| Section uniformity | Within 2:1 | 3:1 | Uneven shrinkage distorts the part |
| Unsupported span | 0.8 in (20 mm) | Support on setters | Gravity sag at sintering temperature |
| Powder escape port | 0.30 in (8 mm) | 0.20 in (5 mm) | Green part cannot be shaken hard |
| Machining stock | 0.020 in (0.5 mm) | — | Sintered tolerance is proportional |

## Cost drivers

Binder jetting has the best throughput economics of any metal additive process at quantity, because printing is fast, nothing is scanned point-by-point, and the bed can be packed in all three dimensions with no supports. Per-part cost falls steeply with nesting density and then flattens.

Against that, the furnace is a fixed cost and a fixed risk. Sintering runs take many hours, setters and sacrificial supports are consumable, and yield loss from distortion is a real line item on new geometry until the shrinkage compensation is dialed in. Development cost for a first article is therefore higher than for laser powder bed fusion, and unit cost at volume is lower.

For foundry sand, the economics are different again: binder jetting replaces pattern and core box tooling entirely, so a one-off casting that would have needed a wooden pattern becomes economic, and complex cores that no pattern shop could draft become routine.

Volume breakpoints: metal binder jetting is generally uncompetitive below about 50 parts (development and furnace overhead) and increasingly attractive from a few hundred to tens of thousands, where it competes directly with [metal injection molding](/processes/forming/metal-injection-molding) — without MIM's tooling cost.

1. Design uniform sections; distortion, not printing, drives scrap.
2. Nest densely — you pay for bed volume and furnace space.
3. Batch parts of the same alloy so they share a sintering run.
4. Accept ±0.5% tolerance everywhere you can, and machine only the features that need better.
5. For casting work, use binder-jetted sand to delete pattern tooling entirely rather than to copy an existing pattern.

## FAQ

### How much do binder jetted metal parts shrink?

15–20% linearly during sintering, in all three axes. The printed geometry is scaled up by that factor beforehand, but shrinkage is only uniform if wall sections are uniform — which is why design for sintering matters more than design for printing in this process.

### What tolerance can metal binder jetting hold?

About ±0.5% of the dimension after sintering, and ±0.3% on geometry whose shrinkage behavior has been characterized through first articles. That is proportional rather than absolute, so large parts have large tolerance bands. Critical features are machined afterward.

### Does binder jetting need support structures?

Not during printing — the surrounding loose powder holds everything, and there is no melt pool or residual stress. Sintering is a different matter: long unsupported spans sag under their own weight at furnace temperature and need ceramic setters or sacrificial supports.

### How dense are binder jetted metal parts?

Typically 96–99.5% after sintering, depending on alloy and cycle. That is below laser powder bed fusion, so fatigue properties are lower, but it is comparable to metal injection molding and adequate for a wide range of structural and consumer applications.

### Why is binder jetting used for foundry sand?

It prints molds and cores directly from CAD with no pattern or core box, which removes pattern tooling from the casting cost entirely and allows core geometry that could not be drawn from a box. Layers of 0.008–0.016 in (200–400 µm) make it fast enough for full-size molds.

### How does binder jetting compare with metal injection molding?

The debinding and sintering steps are nearly identical, and so are the design rules and achievable densities. The difference is up front: MIM needs an injection tool and pays back over tens of thousands of parts, while binder jetting needs none and is competitive from a few hundred parts upward.

### Why are binder jetted green parts so fragile?

Only polymer binder holds the powder together before sintering — green strength is on the order of chalk. That is why minimum walls are 0.040–0.080 in (1–2 mm), why depowdering is done gently by hand, and why the cure step before depowdering is not optional.

## Alternative processes

- [Metal Injection Molding](https://manufacturingprocesses.org/processes/forming/metal-injection-molding.md): Metal injection molding molds a powder-and-binder feedstock exactly like a plastic, then debinds and sinters it into a dense small metal part.
- [Direct Metal Laser Sintering (DMLS)](https://manufacturingprocesses.org/processes/forming/direct-metal-laser-sintering-dmls.md): Direct metal laser sintering fuses metal powder layer by layer with a laser, building dense metal parts anchored to a supported build plate.
- [Investment Casting](https://manufacturingprocesses.org/processes/forming/investment-casting.md): Investment casting builds a ceramic shell around a wax pattern, melts the wax out and pours metal into the cavity, giving fine detail in almost any alloy.
- [Powder Metallurgy (Press and Sinter)](https://manufacturingprocesses.org/processes/forming/powder-metallurgy-press-and-sinter.md): Press-and-sinter powder metallurgy compacts metal powder in a rigid die and sinters it below its melting point into a finished net-shape part.
- [Sand Casting](https://manufacturingprocesses.org/processes/forming/sand-casting.md): Sand casting pours molten metal into a cavity formed in bonded sand around a pattern, then breaks the mold away to release the part.

## Related processes

- [Metal Injection Molding](https://manufacturingprocesses.org/processes/forming/metal-injection-molding.md): Metal injection molding molds a powder-and-binder feedstock exactly like a plastic, then debinds and sinters it into a dense small metal part.
- [Sand Casting](https://manufacturingprocesses.org/processes/forming/sand-casting.md): Sand casting pours molten metal into a cavity formed in bonded sand around a pattern, then breaks the mold away to release the part.
- [Direct Metal Laser Sintering (DMLS)](https://manufacturingprocesses.org/processes/forming/direct-metal-laser-sintering-dmls.md): Direct metal laser sintering fuses metal powder layer by layer with a laser, building dense metal parts anchored to a supported build plate.
- [Powder Metallurgy (Press and Sinter)](https://manufacturingprocesses.org/processes/forming/powder-metallurgy-press-and-sinter.md): Press-and-sinter powder metallurgy compacts metal powder in a rigid die and sinters it below its melting point into a finished net-shape part.
- [Multi Jet Fusion (MJF)](https://manufacturingprocesses.org/processes/forming/multi-jet-fusion-mjf.md): Multi Jet Fusion prints a fusing agent onto nylon powder and drives it with infrared energy, fusing whole layers at once for high throughput.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/forming/binder-jetting)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Blow Molding"
category: "Forming"
subcategory: "Plastics and Rubber"
materials: ["Plastic"]
tolerances: "±0.020 in (±0.5 mm), roughly ±1%, on blown body dimensions; injection-molded neck finishes hold about ±0.005 in (±0.13 mm). Wall thickness is not directly controllable and commonly varies 30% or more across a part."
volumes: "1,000 to many millions; rotary lines run tens of thousands of containers per hour"
lead_time: "3–8 weeks for an aluminum EBM tool; 6–12 weeks for an ISBM preform tool plus blow mold; hours per production run thereafter"
url: https://manufacturingprocesses.org/processes/forming/blow-molding
---

# Blow Molding

Blow molding inflates a heated plastic tube or preform against the walls of a mold with compressed air, producing hollow one-piece containers and tanks.

- **Category**: [Forming](https://manufacturingprocesses.org/processes/forming.md)
- **Family**: Plastics and Rubber
- **Materials**: Plastic
- **Typical tolerances**: ±0.020 in (±0.5 mm), roughly ±1%, on blown body dimensions; injection-molded neck finishes hold about ±0.005 in (±0.13 mm). Wall thickness is not directly controllable and commonly varies 30% or more across a part.
- **Surface finish**: Smooth gloss to matte or grained, set by the mold surface; the pinch-off seam is always visible on extrusion blow molded parts
- **Typical volumes**: 1,000 to many millions; rotary lines run tens of thousands of containers per hour
- **Lead time**: 3–8 weeks for an aluminum EBM tool; 6–12 weeks for an ISBM preform tool plus blow mold; hours per production run thereafter

## Overview

Blow molding inflates a hot thermoplastic tube or a reheated preform against the walls of a two-part mold with compressed air, producing hollow, seamless containers in one piece. It makes essentially every plastic bottle, jerry can, fuel tank, duct, and air intake in production today, overwhelmingly in HDPE, PP, PET, and PVC.

Three families cover almost all of it. Extrusion blow molding (EBM) clamps a continuously extruded parison and inflates it, and is the route for handleware, industrial containers, and automotive ducting. Injection blow molding (IBM) molds a preform onto a core pin and blows it at a second station, giving precise neck finishes on small pharmaceutical bottles. Injection stretch blow molding (ISBM) reheats an injection-molded preform and stretches it axially while blowing, biaxially orienting PET into clear, pressure-capable bottles.

Inflation pressure runs roughly 25–150 psi (1.7–10 bar) for extrusion blow molding and 500–600 psi (35–40 bar) for PET stretch blow molding.

## How it works

**Extrusion blow molding**

1. **Extrude the parison.** A single-screw extruder pushes melt through an annular die to form a hanging tube. HDPE melt runs about 350–420 °F (175–215 °C). Parison wall is programmed point by point along its length so the regions that will draw thinnest start out thicker.

2. **Clamp.** The two mold halves close on the parison. Pinch-off edges at top and bottom weld the tube shut and shear away the tail and the moil.

3. **Blow.** A blow pin or needle admits air at roughly 25–150 psi (1.7–10 bar). The parison expands to the cavity in well under a second. Blow-up ratio — cavity diameter divided by parison diameter — is normally kept at or below 3:1.

4. **Cool.** The part solidifies against the mold wall. Cooling is the cycle bottleneck exactly as in injection molding; a 1 L HDPE bottle typically cycles in 10–20 seconds, a large tank in minutes.

5. **Deflash and trim.** Pinch-off flash at top and bottom is removed, in line or in a separate operation.

**Injection stretch blow molding**

1. **Injection mold the preform.** PET is dried hard — typically 4–6 hours at 300–350 °F (150–175 °C) — and injection molded into a test-tube-shaped preform with the finished neck thread already on it.

2. **Condition.** The preform is reheated through infrared ovens into PET's orientation window, roughly 210–250 °F (100–120 °C): above the glass transition but well below the melting point.

3. **Stretch and blow.** A stretch rod drives down the axis while pre-blow air, then final blow air at 500–600 psi (35–40 bar), expands the preform against the cavity. Typical ratios are 2–3× axial and 3.5–4.5× hoop, for a planar orientation around 10–16×.

4. **Set and eject.** Biaxial orientation is what gives PET bottles their clarity, stiffness, and gas barrier. An unoriented PET bottle of the same weight is hazy and creeps under carbonation pressure.

### Which variant fits which part?

EBM handles handles, integral spouts, and thick industrial walls but leaves a pinch seam and holds looser tolerances. IBM produces no pinch scrap and a precise neck, but is limited to small parts. ISBM is the only route to biaxially oriented PET and dominates beverage packaging.

## Design guidelines

### Wall thickness is an output, not an input

You do not specify blow-molded wall thickness directly. You specify a shape and a shot weight, and the material distributes itself according to how far each region has to stretch. Corners and the deepest draw thin the most. Container walls typically land at 0.012–0.040 in (0.3–1.0 mm); industrial tanks and ducts at 0.080–0.250 in (2–6 mm). Design so the thinnest predicted section still meets the requirement, and let parison programming even out the rest.

### Blow-up ratio

Keep the ratio of maximum cavity diameter to parison diameter at or below 3:1, and aim for 2:1–2.5:1. Above 3:1 the material thins unpredictably at the extremes and corners go translucent and brittle. If one region demands a much larger blow-up than the rest, split the part or reprogram the parison for that zone.

### Radii

Generous radii everywhere. An internal radius of at least one wall thickness is the floor; 2–3× wall is far safer. Sharp corners are where material is stretched hardest and therefore ends up thinnest — the opposite of the injection-molded case, where sharp corners run thick and hot.

### Draft

0.5–1° per side works on smooth surfaces because the part shrinks away from the cavity as it cools — less than injection molding needs. Textured surfaces want 1–1.5°. The draft-versus-texture-depth relationship tabulated on [/charts/injection-molding-design-guidelines](/charts/injection-molding-design-guidelines) applies to blow molds as well, and that chart's wall thickness table governs the injection-molded preform in any IBM or ISBM job.

### Ribs and features

Blow molding cannot fill a deep, narrow rib — available pressure is one to two orders of magnitude below injection pressure. Stiffen with broad, shallow features instead: swage rings, domed panels, and recessed label areas. Depth beyond about 1× the local wall rarely forms cleanly.

### Parting line and pinch-off

Put the parting line where flash removal is easy and cosmetically acceptable. The pinch-off weld at the base of an EBM part is the weakest section in the container — never route a load path or place a mounting feature across it.

### Neck finishes

In IBM and ISBM the neck is injection molded rather than blown, so it holds injection-molding tolerances of about ±0.005 in (±0.13 mm) while the blown body does not. Put every closure, seal, and thread requirement in the neck and keep tight dimensions off the body. Standard finishes (the SP-400 and SP-410 series) let you buy stock closures.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Blow-up ratio | 2:1 to 2.5:1 | 3:1 | Beyond this, corners thin out and go brittle |
| Internal radius | 2–3 × wall | 1 × wall | Corners stretch most and end up thinnest |
| Draft, smooth | 0.5–1° per side | 0.25° | Part shrinks away from the cavity as it cools |
| Draft, textured | 1–1.5° per side | — | Texture depth adds required draft |
| Rib or emboss depth | ≤ 0.5 × wall | about 1 × wall | Low blow pressure cannot force material into deep detail |
| Container body wall | 0.012–0.040 in (0.3–1.0 mm) | — | Typical bottle range |
| Industrial EBM wall | 0.080–0.250 in (2–6 mm) | — | Tanks, ducts, cases |
| Tight-tolerance features | Neck finish only | — | Necks are injection molded; bodies are blown |

## Variants

- Extrusion Blow Molding (EBM)
- Injection Blow Molding (IBM)
- Injection Stretch Blow Molding (ISBM)

## Cost drivers

Tooling is cheap relative to injection molding, because a blow mold is a cooled shell that sees clamp force but never 20,000 psi of melt pressure — aluminum is the normal mold material for EBM. The expensive items in a blow molding program are machine time and, in ISBM, the preform tool, which is a full injection mold with all the cost that implies.

Volume breakpoints:

- Under about 1,000 parts: single-cavity EBM in a machined aluminum tool is possible, but setup and purge scrap dominate the bill.
- 1,000–100,000: single or twin cavity EBM, or two-stage ISBM running purchased stock preforms.
- Over 1,000,000: multi-cavity rotary EBM wheels or rotary stretch-blow lines, which run at tens of thousands of containers per hour.

Cost reduction, in order of leverage:

1. **Lightweight the part.** Resin is the dominant per-part cost in high-volume containers. One gram off a bottle at high annual volume is the entire conversation, and parison programming is how you take it off without thinning the corners past their limit.
2. **Use a stock neck finish.** Standard finishes let you buy off-the-shelf closures and, in ISBM, off-the-shelf preforms instead of paying for a preform tool.
3. **Design for a single parting line.** Every moving core or insert for a handle, spout, or undercut adds cycle time and maintenance.
4. **Reduce the blow-up ratio.** A squatter part distributes material more evenly, which lets you take shot weight out without dropping below minimum wall.
5. **Keep tight tolerances in the neck.** Body dimensions are cheap to loosen and expensive to tighten.

## FAQ

### What is the difference between extrusion blow molding and stretch blow molding?

Extrusion blow molding clamps a hot extruded tube (a parison) and inflates it immediately, which suits HDPE and PP handleware, tanks, and ducts. Injection stretch blow molding reheats a previously injection-molded PET preform and stretches it axially while blowing, biaxially orienting the polymer. That orientation is what gives PET bottles their clarity, stiffness, and carbonation resistance.

### What blow-up ratio can blow molding handle?

Keep the ratio of maximum cavity diameter to parison diameter at or below 3:1, and design toward 2:1–2.5:1 where you can. Higher ratios stretch the material unevenly and leave corners thin, translucent, and prone to cracking.

### What wall thickness does blow molding produce?

Consumer container walls typically land at 0.012–0.040 in (0.3–1.0 mm) and industrial tanks and ducts at 0.080–0.250 in (2–6 mm). Wall thickness is an outcome of shot weight and local stretch rather than something you specify directly, so design to the thinnest predicted section.

### Can blow molded parts have ribs?

Only shallow ones. Blow pressure is roughly 25–150 psi in extrusion blow molding, one to two orders of magnitude below injection pressure, so deep narrow ribs will not fill. Stiffen instead with broad swage rings, domed panels, and recessed label areas no deeper than about half the wall thickness.

### What tolerance can blow molding hold?

Expect about ±0.020 in (±0.5 mm), or roughly ±1% of the dimension, on blown body features. Necks made by injection blow or injection stretch blow molding are the exception — they are injection molded and hold about ±0.005 in (±0.13 mm), so put sealing and thread requirements there.

### Why is the pinch-off seam the weak point on a blow molded part?

At the pinch-off, the mold shears the parison closed and the two inside surfaces weld together while already partly cooled. That weld never reaches the strength of the surrounding wall, so mounting features, load paths, and drop-impact zones should be kept away from it.

### Which blow molding variant should I use for a small pharmaceutical bottle?

Injection blow molding. It molds the preform onto a core pin and blows it at a second station, so there is no pinch-off scrap and the neck finish holds injection-molding tolerances of about ±0.005 in (±0.13 mm). The trade is that it is limited to small parts and cannot produce integral handles.

## Alternative processes

- [Rotation Molding](https://manufacturingprocesses.org/processes/forming/rotation-molding.md): Rotational molding tumbles powdered polymer inside a heated mold on two axes so it fuses into a seamless, stress-free hollow part with a uniform wall.
- [Thermoforming](https://manufacturingprocesses.org/processes/forming/thermoforming.md): Thermoforming heats a plastic sheet until pliable and pulls or presses it over a single-sided mold, producing large thin-walled parts with low tooling cost.
- [Injection Molding](https://manufacturingprocesses.org/processes/forming/injection-molding.md): Injection molding forces molten thermoplastic into a steel mold under high pressure, producing high-volume identical plastic parts.
- [Reaction Injection Molding](https://manufacturingprocesses.org/processes/forming/reaction-injection-molding.md): Reaction injection molding mixes two liquid reactants that polymerize inside a low-pressure mold, producing large tough polyurethane parts on light tooling.

## Related processes

- [Plastic Extrusion](https://manufacturingprocesses.org/processes/forming/plastic-extrusion.md): Plastic extrusion pushes molten polymer through a shaped die and cools it, producing continuous profile, sheet, pipe or film of constant cross-section.
- [Injection Molding](https://manufacturingprocesses.org/processes/forming/injection-molding.md): Injection molding forces molten thermoplastic into a steel mold under high pressure, producing high-volume identical plastic parts.
- [Rotation Molding](https://manufacturingprocesses.org/processes/forming/rotation-molding.md): Rotational molding tumbles powdered polymer inside a heated mold on two axes so it fuses into a seamless, stress-free hollow part with a uniform wall.
- [Thermoforming](https://manufacturingprocesses.org/processes/forming/thermoforming.md): Thermoforming heats a plastic sheet until pliable and pulls or presses it over a single-sided mold, producing large thin-walled parts with low tooling cost.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/forming/blow-molding)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "CNC Wire Bending"
category: "Forming"
subcategory: "Metal"
materials: ["Metal"]
tolerances: "Roughly ±0.020 in (±0.5 mm) on leg lengths and ±1–2° on bend angles, tightening to about ±0.010 in and ±1° with first-article correction; tolerances accumulate along multi-bend parts"
volumes: "100–1,000,000+ parts; a few hundred is enough to absorb the setup"
lead_time: "1–3 weeks including programming and first articles; days for repeat orders. No tooling lead time"
url: https://manufacturingprocesses.org/processes/forming/cnc-wire-bending
---

# CNC Wire Bending

CNC wire bending feeds round or flat wire through a programmable head that bends it in sequence into a finished 2D or 3D form.

- **Category**: [Forming](https://manufacturingprocesses.org/processes/forming.md)
- **Family**: Metal
- **Materials**: Metal
- **Typical tolerances**: Roughly ±0.020 in (±0.5 mm) on leg lengths and ±1–2° on bend angles, tightening to about ±0.010 in and ±1° with first-article correction; tolerances accumulate along multi-bend parts
- **Surface finish**: Retains the incoming wire finish; cut ends carry a small shear burr unless chamfered or ground
- **Typical volumes**: 100–1,000,000+ parts; a few hundred is enough to absorb the setup
- **Lead time**: 1–3 weeks including programming and first articles; days for repeat orders. No tooling lead time

## Overview

CNC wire bending feeds round, square, or flat wire from a coil through a programmable bending head that forms it into a finished 2D or 3D shape, then cuts it off. There is no part-specific tooling: the geometry lives entirely in the program, so a new part number is a setup rather than a tool build.

Wire diameters typically run 0.020–0.500 in (0.5–12 mm), with heavier machines going further, in low-carbon steel wire, music and spring wire, 302/304 stainless, aluminum, copper, and brass. Simple parts come off at cycle times measured in seconds.

It is the process behind display and retail racks, appliance and dishwasher baskets, seat and furniture frames, clips, hooks, handles, and formed springs. Because tooling is programmable, it covers a volume range no die-based process can — economic in hundreds and still competitive in the millions — and the whole part is made in one operation, with no weld or joint in the formed length.

## How it works

1. **Payoff and straightening.** Wire is drawn from a coil through a set of straightening rollers in two perpendicular planes. Getting this right is fundamental: residual coil curvature that survives the straightener shows up as bowed legs on every part.
2. **Feed.** Servo-driven rollers or a gripper feed the wire a programmed length. Feed accuracy sets the leg length tolerance directly.
3. **Bend.** A bending head forms the wire against a pin or die. On a 2D machine the head bends in a single plane; on a 3D machine the head rotates around the wire axis, or the wire itself is rotated between bends, so bends can be placed in any plane.
4. **Springback compensation.** The machine overbends by a programmed amount. Springback is a function of the wire's yield strength and diameter, and it varies between material lots — which is why a first article is checked and the program trimmed for each new coil.
5. **Secondary in-line operations.** Many machines integrate end forming, coiling, thread rolling, chamfering, punching, or flattening within the cycle.
6. **Cut off.** The finished part is sheared free. The cut end carries a small burr and is often chamfered or ground in a secondary operation if it will be handled.
7. **Post-processing.** Resistance or projection welding for assemblies, then zinc plating, powder coating, e-coating, or plastic dip.

Because there is no die, the process's accuracy is a function of feed servo precision, wire straightness, and material consistency rather than of tooling. That makes wire diameter and temper consistency an important purchasing specification, not an afterthought — a coil with tensile strength varying along its length will produce parts with varying angles no matter how good the machine is.

## Design guidelines

### Bend radius

Inside bend radius should be at least 1–2× wire diameter for annealed low-carbon steel and 2–3× for spring tempers and stainless. Tighter radii crack the outer fiber, and in spring wire the failure is sudden. The radius is set by the bend pin, so a design specifying several different radii needs several pins and, potentially, a tool change mid-cycle.

### Distance between bends

Leave at least 2–3× wire diameter of straight length between bends. The bending head needs physical clearance around the wire, and adjacent bends interfere with tooling before they interfere with each other.

### Tolerance stacking

Every bend adds angular error, and every leg adds feed error. Dimension from one end rather than chaining dimensions along the part, and put the tight tolerance on the feature that matters — usually a hook opening or a mounting hole spacing — rather than on the overall envelope.

### 2D before 3D

A part that can be formed in a single plane runs faster and holds tighter than one requiring rotation between bends. Where a design can be flattened into one plane without losing function, it should be.

### Material and temper

Specify the alloy and temper, not just "steel wire." Annealed low-carbon wire forms to tight radii and stays where it is put; spring tempers hold load but need larger radii and show much more springback variation. See the [material properties chart](/charts/material-properties) for a comparison of the base materials.

### Closed forms and overlaps

A wire form that closes on itself — a ring, a closed rectangle — cannot be bent shut without the free end colliding with the machine head. Closed shapes are formed open and then welded, which is a separate operation with its own tolerance.

### Design for the weld, if there is one

Resistance and projection welding of wire assemblies needs contact geometry: crossed wires weld well, parallel wires touching along a length do not. Where a wire form is welded into an assembly, make sure the joints are point contacts.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Inside bend radius | 2× wire diameter | 1× diameter, annealed steel | Outer fiber cracks at tighter radii |
| Bend radius, spring temper | 3× wire diameter | 2× diameter | Hard tempers have little outer-fiber ductility |
| Straight between bends | 3× wire diameter | 2× diameter | Bending head needs physical clearance |
| Number of distinct radii | One | Each radius needs its own bend pin | Tool changes interrupt the cycle |
| Geometry | Single plane where possible | 3D needs rotation between bends | Planar parts run faster and hold tighter |
| Closed loops | Form open, weld closed | Cannot bend a loop shut in the head | The free end collides with the tooling |
| Leg length tolerance | ±0.020 in (±0.5 mm) | ±0.010 in (±0.25 mm) | Set by feed servo accuracy and wire straightness |
| Bend angle tolerance | ±1–2° | ±1° with first-article correction | Springback varies between material lots |

## Cost drivers

There is no tooling to amortize, so cost is setup plus cycle time plus wire. Setup is the dominant term at low volume: threading and straightening the coil, loading the program, and correcting springback on first articles takes a few hours regardless of quantity. Cycle time is seconds per part, so at volume the wire itself and the secondary operations — welding, plating, coating — usually cost more than the bending.

Volume breakpoints: a few hundred parts is enough to absorb a setup. From 5,000 to 500,000, wire bending is at its most competitive, beating stamping outright because there is no die to build and no skeleton scrap. Very high volumes of a stable design may still move to a dedicated four-slide or multi-slide machine, which is faster but requires hard tooling.

1. **Reduce the bend count.** Each bend is cycle time on every part.
2. **Keep the part planar.** 3D bends require rotation, which slows the cycle and widens the tolerance.
3. **Use one wire diameter across an assembly.** Changing diameters means changing straightener and feed setup.
4. **Avoid welded closures.** A form that needs a closing weld doubles the operation count; redesigning it as an open form with a hooked overlap can remove the weld entirely.
5. **Choose a standard coating.** Zinc plating and powder coating on wire are commodity operations; specialty finishes cost disproportionately on a low-mass part.

## FAQ

### What is the minimum bend radius for CNC wire bending?

About 1–2× wire diameter for annealed low-carbon steel and 2–3× for spring tempers and stainless. Below that the outer fiber cracks, and in spring wire that failure is sudden rather than progressive. The radius comes from the bend pin, so using one radius throughout avoids tool changes.

### How close together can two bends be?

At least 2–3× wire diameter of straight length between them. The limit is physical clearance for the bending head, not the metal — adjacent bends run into the tooling before they run into each other. Very close bends may require a different machine or a secondary forming operation.

### What tolerance can CNC wire bending hold?

Roughly ±0.020 in (±0.5 mm) on leg lengths and ±1–2° on bend angles, improving to about ±0.010 in and ±1° once first articles have been used to correct the program. Because tolerances accumulate along the part, dimension from one end rather than chaining dimensions bend to bend.

### Does CNC wire bending require tooling?

No part-specific tooling. The geometry lives in the program, and the only hard tooling is standard bend pins and feed rollers. That is why the process is economic in the hundreds — there is no die to amortize — and why a design revision costs a reprogram rather than a tool build.

### Can a closed wire loop be bent in one operation?

No. The free end would collide with the bending head before the loop closed. Closed shapes such as rings and closed rectangles are formed open and then joined by resistance or projection welding, which is a separate operation carrying its own tolerance.

### Why do parts from a new coil come out at different angles?

Springback is a function of the wire's yield strength, which varies between coils and even along a coil. The machine compensates by overbending a programmed amount, so a first article is checked and the program trimmed each time material changes. Specifying alloy and temper tightly reduces the variation.

## Alternative processes

- [Metal Stamping](https://manufacturingprocesses.org/processes/forming/metal-stamping.md): Metal stamping shears and forms sheet metal between matched dies in a press, producing flat and shallow-formed parts at very high rates.
- [Tube and Section Bending](https://manufacturingprocesses.org/processes/forming/tube-and-section-bending.md): Tube and section bending draws or presses metal tube and profile around a former, with an internal mandrel wherever the wall would otherwise collapse.
- [CNC Machining](https://manufacturingprocesses.org/processes/cutting/cnc-machining.md): CNC machining removes material from solid stock with a programmed cutting tool, producing accurate parts directly from a CAD model with no tooling.
- [Laser Cutting](https://manufacturingprocesses.org/processes/cutting/laser-cutting.md): Laser cutting melts, burns or vaporises a narrow kerf through flat stock with a focused beam, producing profiles from a 2D file with no tooling.

## Related processes

- [Press Braking](https://manufacturingprocesses.org/processes/forming/press-braking.md): Press braking bends sheet metal by driving a punch into a V-die, forming one straight-line bend at a time.
- [Swaging](https://manufacturingprocesses.org/processes/forming/swaging.md): Swaging hammers or presses the end of a tube or bar inward against a die to reduce or taper its diameter without removing any material.
- [Tube and Section Bending](https://manufacturingprocesses.org/processes/forming/tube-and-section-bending.md): Tube and section bending draws or presses metal tube and profile around a former, with an internal mandrel wherever the wall would otherwise collapse.
- [Roll Forming](https://manufacturingprocesses.org/processes/forming/roll-forming.md): Roll forming passes coil stock through a sequence of contoured roller stands that bend it progressively into a constant cross-section profile.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/forming/cnc-wire-bending)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Centrifugal Casting"
category: "Forming"
subcategory: "Metal"
materials: ["Metal"]
tolerances: "Outside diameter, cast against a metal die, holds substantially tighter than the bore, which is uncontrolled and always machined; wall thickness is set by pour volume and carries the loosest tolerance of the three"
volumes: "1–10,000 parts; a single large ring is economic on a sand-lined mold"
lead_time: "2–6 weeks for a run against existing mold sizes; longer if a new permanent mold is required"
url: https://manufacturingprocesses.org/processes/forming/centrifugal-casting
---

# Centrifugal Casting

Centrifugal casting pours molten metal into a spinning mold so centrifugal force packs it against the wall and drives inclusions toward the bore.

- **Category**: [Forming](https://manufacturingprocesses.org/processes/forming.md)
- **Family**: Metal
- **Materials**: Metal
- **Typical tolerances**: Outside diameter, cast against a metal die, holds substantially tighter than the bore, which is uncontrolled and always machined; wall thickness is set by pour volume and carries the loosest tolerance of the three
- **Typical volumes**: 1–10,000 parts; a single large ring is economic on a sand-lined mold
- **Lead time**: 2–6 weeks for a run against existing mold sizes; longer if a new permanent mold is required

## Overview

Centrifugal casting pours molten metal into a mold spinning about its own axis, so centrifugal force throws the metal outward against the mold wall and holds it there while it freezes. Density does the quality control: dense metal is driven to the outside diameter while slag, oxide, and gas — all lighter than the melt — collect at the bore, where they are machined away.

The result is a hollow cylindrical part with exceptionally sound outer-wall metal, no cores, and no risers. It is the standard process for cast iron pipe, bronze bushings and bearing sleeves, cylinder liners, mill rolls, and large rings in stainless and nickel alloys, from a few inches to several feet in diameter.

Geometry is the constraint: the process only makes bodies of revolution, and the bore is set by how much metal is poured rather than by tooling, so it is always machined. Metal yield is high because there is no gating system to cut off and remelt.

## How it works

Two arrangements cover most work.

**True centrifugal casting** spins a cylindrical mold — often a water-cooled steel die, sometimes sand-lined — about a horizontal or vertical axis and pours metal into it with no central core. The bore forms itself as a free liquid surface. A horizontal axis is used for long parts such as pipe and liners; a vertical axis suits short rings, where the bore takes a slight parabolic taper because gravity competes with rotation.

**Centrifuge (semi-centrifugal) casting** arranges several conventional mold cavities around a central sprue on a spinning table, using rotation to pressurize the feed into shapes that are not themselves bodies of revolution.

The cycle:

1. **Mold prep.** The permanent mold is coated with a refractory wash or a sand lining, then preheated. Coating thickness controls how fast the shell chills, and therefore the outer-wall microstructure.
2. **Spin up.** The mold reaches speed before pouring. Speed is set by the G-force required at the mold wall rather than by rpm directly — G-factors quoted in foundry practice for horizontal work generally fall in the range of 60–100 — so a small-diameter mold spins far faster than a large one.
3. **Pour.** Metal is delivered along the length of the mold from a launder or trough. Pour rate and traverse speed set wall uniformity.
4. **Solidification under force.** Freezing runs from the outer wall inward. Because the pressure gradient continuously feeds the freezing front, no risers are needed and shrinkage porosity is pushed toward the bore rather than into the wall.
5. **Segregation.** Lower-density inclusions, oxide film, and dross migrate to the free inner surface. This is the process's defining metallurgical advantage.
6. **Extraction and machining.** The casting is stripped once solid. The bore is bored out — commonly 0.125–0.25 in (3–6 mm) of radius or more — to clear the segregated layer, and the OD is turned to size.

Bi-metal parts are made by pouring a second alloy after the first has partly frozen, which is how bronze-lined steel bushings and hard-faced rolls are produced.

## Design guidelines

### Bodies of revolution only

The economic geometry is a cylinder, sleeve, ring, or pipe. Flanges and steps on the outside diameter are possible in a stepped die; features on the bore are not, because the bore is a free liquid surface. Anything non-round is either a centrifuge casting or a different process entirely.

### Machine the bore, always

Plan on removing 0.125–0.25 in (3–6 mm) of radius from the bore, and more on large castings or less clean melts. That layer holds the inclusions and dross the process deliberately pushed there. A design that specifies an as-cast bore discards the main advantage of the process.

### Wall thickness

Keep the wall above roughly 0.25 in (6 mm) for iron and steel, with 0.375 in (10 mm) a more comfortable minimum on long castings. Uniform wall along the length is easiest; a tapered wall requires a controlled pour traverse and adds cost.

### Wall tolerance is the loosest dimension

Wall thickness is set by the volume of metal poured, not by tooling, so it carries a looser tolerance than the die-controlled outside diameter. Tolerance the OD, machine the bore concentric to it, and let the wall fall where it falls.

### Length-to-diameter ratio

Long horizontal castings need a traversing pour to keep the wall uniform, and wall control gets progressively harder as L/D rises. Short rings are easier and are usually cast on a vertical axis.

### Alloy selection

Bronze bearing alloys, gray and ductile iron, stainless, and nickel alloys all cast well centrifugally. Alloys with wide freezing ranges and strong density differences between phases segregate radially — useful when it puts the hard phase on the wear surface, a defect when it does not.

See the [surface finish chart](/charts/surface-finish-chart) for what boring and turning will deliver on the machined surfaces, and [metal melting points](/charts/metal-melting-points) for pour temperature planning.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Geometry | Cylinder, sleeve, ring, pipe | Non-round needs centrifuge casting | Rotation only pressurizes bodies of revolution |
| Bore machining stock | 0.125–0.25 in (3–6 mm) on radius | More on large castings | Inclusions and dross segregate to the bore |
| Wall thickness, iron/steel | ≥ 0.375 in (10 mm) | 0.25 in (6 mm) | Thin walls freeze before the pour completes |
| Wall uniformity | Constant along length | Tapered walls cost more | Uniform wall needs only a steady pour traverse |
| Cast features | OD steps and flanges | None on the bore | The bore is a free liquid surface |
| Risers and gates | None required | — | Centrifugal pressure feeds the freezing front |

## Cost drivers

The distinguishing economics are yield and tooling reuse. There is no gating system and no riser, so metal yield is very high compared with static casting, where risers and runners routinely consume a third or more of the poured weight. Permanent metal molds are reused for thousands of castings, so per-part tooling cost falls quickly. Against that, machining is not optional — every bore is bored — and machine time on a large ring is significant.

Volume breakpoints: sand-lined centrifugal molds make one-off large rings economic; a permanent steel die pays back over hundreds to thousands of parts. Very high volumes of small bushings are usually better served by press-and-sinter powder metallurgy, which produces them net shape with no machining at all.

1. **Size the casting to minimize bore stock.** Every extra thousandth of radius removed is chips and spindle time.
2. **Cast the longest tube you can, then part off.** Several short bushings from one cast tube spread the setup across many parts.
3. **Use bi-metal casting where a solid part of expensive alloy is not needed.** A thin bronze bearing layer on a steel backing is a large material saving.
4. **Standardize on a few mold diameters.** Because wall thickness comes from pour volume, one die can produce a family of wall thicknesses at the same OD.
5. **Do not specify tight wall-thickness tolerance.** Wall is a pour variable, not a tooling one; tighten the OD instead.

## FAQ

### Why is centrifugal casting metal denser than sand casting?

Rotation applies a continuous pressure gradient that feeds the solidification front from the outside inward, so shrinkage porosity never forms in the wall. Lighter inclusions, oxide, and gas migrate toward the bore, which is machined away. The result is a very clean outer wall with no risers at all.

### How much stock has to be machined out of the bore?

Typically 0.125–0.25 in (3–6 mm) on radius, and more on large castings or less clean melts. That layer is exactly where the process concentrates dross and inclusions, so leaving it in place forfeits the main quality advantage.

### What shapes can be centrifugally cast?

Bodies of revolution — pipe, tube, sleeves, bushings, rings, cylinder liners, and mill rolls. Outside-diameter steps and flanges can be cast into a stepped die. The bore is formed by a free liquid surface, so it cannot carry any cast feature.

### Does centrifugal casting need cores or risers?

No. The bore forms itself from the free surface of the spinning metal, so there is no core, and centrifugal pressure feeds solidification, so there are no risers. That is why metal yield is far higher than in static sand casting.

### Can two different alloys be cast into one part?

Yes. Pouring a second alloy after the first has partially solidified produces a metallurgically bonded bi-metal wall. This is how bronze-lined steel bushings and hard-faced mill rolls are made, and it saves a substantial amount of expensive alloy.

### Centrifugal casting or machining a bushing from bar stock?

Bar stock wins for small quantities of small bushings in common alloys, since no foundry is involved. Centrifugal casting wins on large diameters where bar stock is expensive or unavailable, on bronze bearing alloys, and wherever wall soundness through the full section matters.

## Alternative processes

- [Forging](https://manufacturingprocesses.org/processes/forming/forging.md): Forging deforms solid metal under compressive load in a hammer, press or dies, so the grain flow follows the part contour and fatigue strength rises.
- [CNC Turning](https://manufacturingprocesses.org/processes/cutting/cnc-turning.md): CNC turning rotates the workpiece against a stationary tool, producing round parts with excellent concentricity and short cycle times.
- [Powder Metallurgy (Press and Sinter)](https://manufacturingprocesses.org/processes/forming/powder-metallurgy-press-and-sinter.md): Press-and-sinter powder metallurgy compacts metal powder in a rigid die and sinters it below its melting point into a finished net-shape part.
- [Metal Extrusion](https://manufacturingprocesses.org/processes/forming/metal-extrusion.md): Metal extrusion pushes heated billet through a shaped die, producing continuous aluminum or copper profile of constant cross-section.

## Related processes

- [Sand Casting](https://manufacturingprocesses.org/processes/forming/sand-casting.md): Sand casting pours molten metal into a cavity formed in bonded sand around a pattern, then breaks the mold away to release the part.
- [Investment Casting](https://manufacturingprocesses.org/processes/forming/investment-casting.md): Investment casting builds a ceramic shell around a wax pattern, melts the wax out and pours metal into the cavity, giving fine detail in almost any alloy.
- [Die Casting](https://manufacturingprocesses.org/processes/forming/die-casting.md): Die casting injects molten aluminum, zinc or magnesium into a hardened steel die under high pressure, producing thin-walled net-shape metal parts.
- [Heat Treatment](https://manufacturingprocesses.org/processes/forming/heat-treatment.md): Heat treatment cycles metal through controlled heating and cooling to change its hardness, strength, ductility or internal stress state.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/forming/centrifugal-casting)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Ceramic Injection Molding (CIM)"
category: "Forming"
subcategory: "Glass and Ceramics"
materials: ["Ceramic"]
tolerances: "±0.3–0.5% of dimension as fired (about ±0.006–0.010 in on a 2 in feature); diamond-ground features hold ±0.0002 in (±0.005 mm) or better."
volumes: "10,000–1,000,000+ parts per year"
lead_time: "12–20 weeks for tooling and shrink-factor development, then 2–4 weeks per production batch, most of it furnace scheduling."
url: https://manufacturingprocesses.org/processes/forming/ceramic-injection-molding-cim
---

# Ceramic Injection Molding (CIM)

Ceramic injection molding molds a ceramic powder and binder feedstock, then debinds and sinters it into a dense technical ceramic part.

- **Category**: [Forming](https://manufacturingprocesses.org/processes/forming.md)
- **Family**: Glass and Ceramics
- **Materials**: Ceramic
- **Typical tolerances**: ±0.3–0.5% of dimension as fired (about ±0.006–0.010 in on a 2 in feature); diamond-ground features hold ±0.0002 in (±0.005 mm) or better.
- **Surface finish**: Ra 8–32 µin (0.2–0.8 µm) as fired on fine-grained zirconia, coarser on large-grain alumina. Lapping and polishing reach below Ra 2 µin (0.05 µm) for sealing and optical surfaces.
- **Typical volumes**: 10,000–1,000,000+ parts per year
- **Lead time**: 12–20 weeks for tooling and shrink-factor development, then 2–4 weeks per production batch, most of it furnace scheduling.

## Overview

Ceramic injection molding (CIM) injects a feedstock of fine ceramic powder — typically 50–65% by volume — bound in a thermoplastic and wax binder system, then removes the binder and sinters the part to full density. It is the only route to complex, near-net-shape technical ceramic components at production volume, and it works because the molding step behaves exactly like [injection molding](/processes/forming/injection-molding) while the material ends up as dense alumina, zirconia, or silicon nitride.

The defining number is shrinkage: parts contract 15–25% linearly during sintering, uniformly in all directions if — and only if — wall sections are uniform. Tooling is cut oversize by that factor, so the shrink rate must be characterized before the mold is built.

Typical materials are 96–99.8% alumina, 3Y-TZP zirconia, zirconia-toughened alumina, silicon nitride, and aluminum nitride. Parts are usually under 3.5 oz (100 g) and 0.02–0.24 in (0.5–6 mm) in wall thickness. Volumes start around 10,000 pieces a year, where tooling amortizes.

## How it works

1. **Compound the feedstock.** Ceramic powder — often sub-micron — is mixed with a multi-component binder of thermoplastic, wax, and surfactant to a solids loading of 50–65% by volume. Loading is critical: too low and the part slumps during debinding, too high and the feedstock will not flow.
2. **Mold.** The feedstock is injected on a standard or lightly modified injection molding machine at barrel temperatures around 265–375 °F (130–190 °C) into a tool held near 85–140 °F (30–60 °C). Because the feedstock is highly abrasive, tooling is made from hardened or carbide-faced steel and gates wear measurably faster than in plastic molding.
3. **Debind.** The binder is removed in stages. Solvent or catalytic debinding dissolves or depolymerizes the primary binder component, leaving an open pore network; a residual backbone binder holds the part together. Thermal debinding then burns out that backbone during furnace ramp-up. Debinding time scales roughly with the square of wall thickness, which is what caps section size.
4. **Sinter.** The brown part is fired to close porosity — roughly 2,730–3,000 °F (1,500–1,650 °C) for alumina and 2,460–2,730 °F (1,350–1,500 °C) for zirconia. Densification drives the 15–25% linear shrinkage.
5. **Finish.** Sintered technical ceramics can only be cut by diamond. Lapping, diamond grinding, and laser machining are used for sealing faces, bores, and any tolerance the as-fired process cannot hold.

## Design guidelines

### Uniform wall thickness
This is the whole of ceramic injection molding design. Non-uniform sections debind at different rates and shrink at different rates, and the part warps or cracks. Hold walls within a 2:1 ratio across the part, and blend transitions with generous radii rather than steps.

### Wall thickness limits
0.02–0.24 in (0.5–6 mm) is the practical range. Above roughly 0.4 in (10 mm) the debinding cycle becomes impractically long — debinding time rises with the square of thickness — and the risk of a trapped-binder blister rises with it. Core out heavy sections.

### Corner radii
Fillet every internal corner at 0.008 in (0.2 mm) or more. Sintered ceramics have essentially no ductility, so a sharp internal corner is a crack initiator both during firing and in service.

### Shrinkage and tolerance
Design for 15–25% linear shrinkage and expect ±0.3–0.5% of dimension as-fired. That is proportional: a 2 in (50 mm) feature carries roughly ±0.008 in (±0.2 mm). Anything tighter — a bearing bore, a sealing land, a fiber-optic ferrule — is diamond ground after sintering, so allow grinding stock.

### Draft
1–2° of draft eases ejection of the green part, which has low strength and is easily damaged. Zero-draft walls are possible with polished tooling but raise scrap.

### Gate location and knit lines
Weld lines in CIM survive into the fired part as strength defects, exactly as in [metal injection molding](/processes/forming/metal-injection-molding). Place gates so flow fronts meet away from loaded regions, and expect to iterate the gate on the first tool.

### Features to avoid
Threads, undercuts, and sharp edges belong in the machining operation, not the mold. Ceramics chip at edges — break them with a 0.004–0.008 in (0.1–0.2 mm) chamfer.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Wall thickness | 0.04–0.12 in (1–3 mm) | 0.02–0.24 in (0.5–6 mm) | Debind time scales with thickness squared |
| Section uniformity | Within 2:1 | 3:1 | Differential shrinkage warps the part |
| Internal radius | 0.02 in (0.5 mm) | 0.008 in (0.2 mm) | No ductility; corners initiate cracks |
| Draft | 1–2° | 0.5° | Green parts are fragile at ejection |
| Edge break | 0.008 in (0.2 mm) | 0.004 in (0.1 mm) | Sharp ceramic edges chip |
| Grinding stock | 0.008 in (0.2 mm) | 0.004 in (0.1 mm) | Tight features must be diamond ground |

## Cost drivers

Tooling dominates the entry cost. A CIM mold is built like an injection mold but from harder steel because the feedstock is abrasive, and it must be cut oversize to a shrink factor that is itself established by trial. Expect tool development to include at least one shrink correction.

Per-part cost splits between feedstock, molding cycle, and furnace time. Ceramic powder of the fineness CIM requires is expensive, and furnace cycles at 2,700 °F (1,500 °C) and above consume real energy per load. Because parts are fired in batches, throughput is set by furnace capacity rather than by molding cycle time.

Post-sintering diamond grinding is the cost that separates a well-designed CIM part from a poorly designed one. Every ground feature is a separate operation on hard, brittle material.

Volume breakpoints: below roughly 10,000 pieces a year, dry pressing or machining a sintered blank is usually cheaper. Above that, CIM's ability to mold complex geometry in one shot wins, and its advantage grows with part complexity.

1. Hold wall sections uniform — it reduces scrap more than any other change.
2. Design so no feature needs post-sinter grinding, or accept the grinding cost explicitly.
3. Choose alumina over zirconia where the mechanical requirement allows; the powder is substantially cheaper.
4. Consolidate assemblies into one molded part; ceramic joining is difficult and expensive.
5. Keep part mass low — furnace capacity, not molding time, sets throughput.

## FAQ

### How much do ceramic injection molded parts shrink?

15–25% linearly during sintering, depending on the ceramic and solids loading of the feedstock. The mold is cut oversize by that factor. Shrinkage is uniform only if wall sections are uniform, which is why constant wall thickness is the first design rule of the process.

### What tolerance can CIM hold?

±0.3–0.5% of the dimension as fired, so about ±0.006–0.010 in on a 2 in (50 mm) feature. Anything tighter must be diamond ground after sintering, which reaches ±0.0002 in (±0.005 mm) but adds an operation on very hard material.

### What is the maximum wall thickness for CIM?

About 0.24 in (6 mm) in routine production, with 0.4 in (10 mm) as a practical ceiling. Debinding time rises roughly with the square of wall thickness, so a thick section either takes an uneconomic cycle or traps binder and blisters during firing.

### Which ceramics can be injection molded?

The common production materials are 96–99.8% alumina, 3Y-TZP zirconia, zirconia-toughened alumina, silicon nitride, and aluminum nitride. Selection is driven by the property needed — alumina for insulation and wear at low cost, zirconia for toughness, silicon nitride for thermal shock and high-temperature strength.

### How is CIM different from metal injection molding?

The molding, debinding, and sintering steps are essentially the same, and so are the design rules. The differences are in the material: ceramics sinter at higher temperatures, shrink slightly more, cannot be machined with conventional tooling afterward, and have no ductility, so corner radii and edge breaks matter far more.

### At what volume does CIM make sense?

Around 10,000 parts a year and upward. Below that, dry pressing a simpler shape or machining a sintered blank is normally cheaper, because CIM tooling must be hardened against an abrasive feedstock and developed through at least one shrink-factor correction.

## Alternative processes

- [Ceramic Slip Casting](https://manufacturingprocesses.org/processes/forming/ceramic-slip-casting.md): Ceramic slip casting pours liquid clay slip into a porous plaster mold that draws off water, leaving a solid layer that becomes the part.
- [Press Molding Ceramics](https://manufacturingprocesses.org/processes/forming/press-molding-ceramics.md): Press molding ceramics forms clay between a die and a rotating profile tool or matched mold, producing flatware and repeat shapes quickly.
- [Metal Injection Molding](https://manufacturingprocesses.org/processes/forming/metal-injection-molding.md): Metal injection molding molds a powder-and-binder feedstock exactly like a plastic, then debinds and sinters it into a dense small metal part.
- [CNC Machining](https://manufacturingprocesses.org/processes/cutting/cnc-machining.md): CNC machining removes material from solid stock with a programmed cutting tool, producing accurate parts directly from a CAD model with no tooling.
- [Binder Jetting](https://manufacturingprocesses.org/processes/forming/binder-jetting.md): Binder jetting prints liquid binder into a powder bed to hold each layer, producing green parts that are later sintered or infiltrated.

## Related processes

- [Metal Injection Molding](https://manufacturingprocesses.org/processes/forming/metal-injection-molding.md): Metal injection molding molds a powder-and-binder feedstock exactly like a plastic, then debinds and sinters it into a dense small metal part.
- [Injection Molding](https://manufacturingprocesses.org/processes/forming/injection-molding.md): Injection molding forces molten thermoplastic into a steel mold under high pressure, producing high-volume identical plastic parts.
- [Press Molding Ceramics](https://manufacturingprocesses.org/processes/forming/press-molding-ceramics.md): Press molding ceramics forms clay between a die and a rotating profile tool or matched mold, producing flatware and repeat shapes quickly.
- [Ceramic Slip Casting](https://manufacturingprocesses.org/processes/forming/ceramic-slip-casting.md): Ceramic slip casting pours liquid clay slip into a porous plaster mold that draws off water, leaving a solid layer that becomes the part.
- [Powder Metallurgy (Press and Sinter)](https://manufacturingprocesses.org/processes/forming/powder-metallurgy-press-and-sinter.md): Press-and-sinter powder metallurgy compacts metal powder in a rigid die and sinters it below its melting point into a finished net-shape part.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/forming/ceramic-injection-molding-cim)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Ceramic Slip Casting"
category: "Forming"
subcategory: "Glass and Ceramics"
materials: ["Ceramic"]
tolerances: "Roughly ±1–2% of dimension on the fired part, dominated by variation in drying and firing shrinkage rather than by the mold."
volumes: "100–50,000 pieces per year with plaster molds; higher with pressure casting"
lead_time: "3–8 weeks for model and mold making. In production, plaster molds typically yield one or two castings per mold per day, so output is set by mold count."
url: https://manufacturingprocesses.org/processes/forming/ceramic-slip-casting
---

# Ceramic Slip Casting

Ceramic slip casting pours liquid clay slip into a porous plaster mold that draws off water, leaving a solid layer that becomes the part.

- **Category**: [Forming](https://manufacturingprocesses.org/processes/forming.md)
- **Family**: Glass and Ceramics
- **Materials**: Ceramic
- **Typical tolerances**: Roughly ±1–2% of dimension on the fired part, dominated by variation in drying and firing shrinkage rather than by the mold.
- **Typical volumes**: 100–50,000 pieces per year with plaster molds; higher with pressure casting
- **Lead time**: 3–8 weeks for model and mold making. In production, plaster molds typically yield one or two castings per mold per day, so output is set by mold count.

## Overview

Ceramic slip casting pours a fluid clay suspension into a porous plaster mold, which draws water out of the slip by capillary action and leaves a solid layer of consolidated clay against the mold face. Wall thickness is controlled by dwell time — the cast layer builds roughly with the square root of time, so a sanitaryware mold might be filled for an hour to reach 0.3–0.5 in (8–12 mm) while thin tableware casts in ten or twenty minutes.

The slip is not simply clay and water. A deflocculant, typically sodium silicate or sodium carbonate at a fraction of a percent of the dry weight, disperses the clay particles so the slip pours freely at 25–35% water instead of the 50%+ it would otherwise need. That reduction in water directly reduces drying shrinkage and cracking.

Slip casting is the only economical route to hollow, complex, thin-walled ceramic bodies — toilets and basins, teapots and vases, technical crucibles and thermocouple sheaths. Total linear shrinkage from wet cast to fired part is roughly 10–18%, which the mold is sized to compensate.

## How it works

1. **Prepare the slip.** The clay body is blunged with water and deflocculant to a controlled specific gravity and viscosity. Both are checked every shift; a slip that thickens or thins changes cast rate and therefore wall thickness.
2. **Fill the mold.** The plaster mold, assembled from two or more parts, is filled and topped up as the level drops — the mold absorbs water and the slip level falls as material is consumed building the wall.
3. **Dwell.** Water migrates into the plaster and clay particles pack against the mold face. Cast thickness grows approximately with the square root of dwell time, so doubling the wall takes roughly four times as long.
4. **Drain (hollow casting) or fill out (solid casting).** For hollow ware the mold is inverted and the surplus slip poured off, leaving a shell. For solid casting, the gap between two mold faces is narrow enough that the two advancing walls meet and the whole section becomes solid.
5. **Firm up and release.** As drying continues the casting shrinks away from the plaster and reaches leather-hard, at which point the mold is opened. Timing matters: too early and the piece slumps, too late and it cracks as it shrinks against fixed mold features.
6. **Fettle and dry.** Mold seams and the spare above the pouring gate are trimmed and sponged. The piece then dries slowly and evenly — uneven drying is the single largest source of scrap.
7. **Fire.** A bisque firing around 1,650–1,830 °F (900–1,000 °C) gives a porous, glaze-absorbent body. After glazing, the glost firing runs roughly 1,830–2,100 °F (1,000–1,150 °C) for earthenware and 2,190–2,550 °F (1,200–1,400 °C) for stoneware and porcelain.

## Design guidelines

### Uniform wall thickness
Wall thickness follows dwell time and is therefore naturally uniform — but only if the mold's plaster is uniformly absorbent and the geometry does not trap slip. Design so the mold drains cleanly, without pockets that hold residual slip and cast a locally thick, slow-drying section that will crack.

### Wall thickness range
0.12–0.20 in (3–5 mm) for tableware and decorative ware, 0.3–0.5 in (8–12 mm) for sanitaryware. Below about 0.08 in (2 mm) the casting is too weak to survive demolding.

### Undercuts and mold splits
Plaster molds are rigid, so undercuts require additional mold parts, and every split leaves a seam to be fettled by hand. Each additional mold part adds labor to every single casting for the life of the tool — mold-part count is a direct unit-cost decision.

### Draft and shrinkage
The casting shrinks away from the mold as it dries, so slip casting tolerates less draft than injection molding. Even so, 1–2° eases release on deep forms and reduces scrap.

### Radii and section changes
Fillet internal corners generously. Abrupt section changes dry at different rates and crack, and a sharp interior corner in the mold is also where plaster erodes first.

### Design for shrinkage
Total linear shrinkage from wet to fired is roughly 10–18% depending on the body — porcelain at the high end, coarse stoneware lower. The mold is made oversize by the measured figure for that body, so a body change is a tooling change.

### Attachments
Handles, spouts, and feet are cast separately and luted on with slip at leather-hard. Design the joint with enough contact area, and keep the attached part's drying rate similar to the body's or the joint cracks.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Wall thickness | 0.16 in (4 mm) tableware | 0.08 in (2 mm) | Below this, demolding breaks the piece |
| Mold parts | 2 | More adds fettling labor forever | Every seam is hand-finished |
| Internal radius | 0.12 in (3 mm) | Sharp corners crack | Differential drying |
| Draft | 1–2° | 0° possible | Casting shrinks away from plaster |
| Linear shrinkage allowance | 10–18% | Body-specific | Mold must be sized to the body |

## Cost drivers

Labor and cycle time dominate. A plaster mold absorbs a finite amount of water before it must dry out, so a mold typically produces one or two castings per shift unless the plant runs a drying regime between casts. Plants therefore hold large numbers of molds, and mold storage and handling is a real cost. Plaster molds also wear: 50–150 castings is a typical working life before absorbency and surface definition degrade.

Pressure casting changes this equation. Using porous polymer molds and slip injected at pressure, cycle times drop from hours to minutes and mold life extends by orders of magnitude, which is why high-volume sanitaryware moved to it. The tooling costs far more.

Firing is the other major cost: two firings, each consuming kiln time and energy, with kiln capacity setting plant throughput. Scrap discovered after the glost firing has absorbed all of that cost.

Volume breakpoints: plaster slip casting suits roughly 100–50,000 pieces a year. Above that, pressure casting or [press molding](/processes/forming/press-molding-ceramics) is more economical, and for small technical parts at high volume, [ceramic injection molding](/processes/forming/ceramic-injection-molding-cim) takes over.

1. Minimize mold parts — every seam is fettled on every piece, forever.
2. Design so the mold drains completely; trapped slip is scrap.
3. Keep sections uniform to reduce drying and firing losses.
4. Consider whether the design can be cast in one piece rather than assembled from luted components.
5. Design the ware to stack efficiently in the kiln; firing capacity is the plant bottleneck.

## FAQ

### How is wall thickness controlled in slip casting?

By dwell time. The plaster mold draws water out of the slip and clay consolidates against the mold face, with thickness building roughly in proportion to the square root of time. Doubling the wall thickness therefore takes about four times as long, and slip specific gravity and viscosity must be held constant for the relationship to repeat.

### Why does slip need a deflocculant?

A deflocculant such as sodium silicate or sodium carbonate, added at a fraction of a percent of dry weight, disperses the clay particles so the slip pours freely at 25–35% water instead of the much higher water content it would otherwise need. Less water means less drying shrinkage and far less cracking.

### How much does slip cast ceramic shrink?

Roughly 10–18% linearly from wet cast to fired part, combining drying and firing shrinkage. Porcelain sits at the high end and coarse stoneware bodies lower. Molds are made oversize by the measured figure for the specific body, so changing body means new tooling.

### How many castings does a plaster mold produce?

Typically 50–150 before absorbency and surface definition degrade enough to force replacement. A mold also needs to dry out between casts, so plants generally get one or two castings per mold per day and hold large mold inventories to make throughput.

### Can slip casting produce undercuts?

Yes, but each undercut needs an additional mold part, and every mold split leaves a seam that is fettled by hand on every casting. Mold-part count is therefore a permanent unit-cost decision rather than a one-time tooling decision.

### What is the difference between drain casting and solid casting?

In drain (hollow) casting, surplus slip is poured out after the wall reaches thickness, leaving a shell — this is how vases, teapots, and sanitaryware are made. In solid casting, the gap between two mold faces is narrow enough that the walls advancing from both sides meet, producing a solid section such as a handle or a knob.

## Alternative processes

- [Press Molding Ceramics](https://manufacturingprocesses.org/processes/forming/press-molding-ceramics.md): Press molding ceramics forms clay between a die and a rotating profile tool or matched mold, producing flatware and repeat shapes quickly.
- [Ceramic Injection Molding (CIM)](https://manufacturingprocesses.org/processes/forming/ceramic-injection-molding-cim.md): Ceramic injection molding molds a ceramic powder and binder feedstock, then debinds and sinters it into a dense technical ceramic part.
- [Clay Throwing](https://manufacturingprocesses.org/processes/forming/clay-throwing.md): Clay throwing forms a rotationally symmetric vessel by hand from a spinning ball of plastic clay on a potter's wheel.
- [Rotation Molding](https://manufacturingprocesses.org/processes/forming/rotation-molding.md): Rotational molding tumbles powdered polymer inside a heated mold on two axes so it fuses into a seamless, stress-free hollow part with a uniform wall.
- [Binder Jetting](https://manufacturingprocesses.org/processes/forming/binder-jetting.md): Binder jetting prints liquid binder into a powder bed to hold each layer, producing green parts that are later sintered or infiltrated.

## Related processes

- [Press Molding Ceramics](https://manufacturingprocesses.org/processes/forming/press-molding-ceramics.md): Press molding ceramics forms clay between a die and a rotating profile tool or matched mold, producing flatware and repeat shapes quickly.
- [Clay Throwing](https://manufacturingprocesses.org/processes/forming/clay-throwing.md): Clay throwing forms a rotationally symmetric vessel by hand from a spinning ball of plastic clay on a potter's wheel.
- [Ceramic Injection Molding (CIM)](https://manufacturingprocesses.org/processes/forming/ceramic-injection-molding-cim.md): Ceramic injection molding molds a ceramic powder and binder feedstock, then debinds and sinters it into a dense technical ceramic part.
- [Vacuum Casting (Urethane Casting)](https://manufacturingprocesses.org/processes/forming/vacuum-casting.md): Vacuum casting, known in the US as urethane casting, pours polyurethane resin into a silicone mold under vacuum to copy a master pattern for short runs.
- [Rotation Molding](https://manufacturingprocesses.org/processes/forming/rotation-molding.md): Rotational molding tumbles powdered polymer inside a heated mold on two axes so it fuses into a seamless, stress-free hollow part with a uniform wall.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/forming/ceramic-slip-casting)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Clay Throwing"
category: "Forming"
subcategory: "Glass and Ceramics"
materials: ["Ceramic"]
volumes: "1–500 pieces"
lead_time: "Days to weeks per batch. Throwing is minutes per piece, but drying takes days and pieces are held until a full kiln load is ready for each of two firings."
url: https://manufacturingprocesses.org/processes/forming/clay-throwing
---

# Clay Throwing

Clay throwing forms a rotationally symmetric vessel by hand from a spinning ball of plastic clay on a potter's wheel.

- **Category**: [Forming](https://manufacturingprocesses.org/processes/forming.md)
- **Family**: Glass and Ceramics
- **Materials**: Ceramic
- **Typical volumes**: 1–500 pieces
- **Lead time**: Days to weeks per batch. Throwing is minutes per piece, but drying takes days and pieces are held until a full kiln load is ready for each of two firings.

## Overview

Clay throwing forms a rotationally symmetric vessel by hand from a ball of plastic clay centered on a spinning wheel. The potter uses the wheel's rotation and the clay's plasticity — around 20–25% water content — to raise a wall in a few controlled pulls, then trims, dries, and fires the piece.

It is a skill process, not a tooling process. There is no mold, no die, and no fixed cycle: a small bowl takes a few minutes to throw, a large form considerably longer, and repeat accuracy comes from the maker rather than from the equipment. Total linear shrinkage from wet to fired is roughly 10–15%, and the piece passes through two firings — a bisque around 1,650–1,830 °F (900–1,000 °C) and a glost firing at 1,830–2,550 °F (1,000–1,400 °C) depending on body.

Throwing remains commercially relevant for studio and small-batch tableware, prototype development ahead of [slip casting](/processes/forming/ceramic-slip-casting) or [press molding](/processes/forming/press-molding-ceramics), and one-off work. For anything repeatable at volume, one of those tooled processes replaces it.

## How it works

1. **Wedge the clay.** The clay is kneaded to remove air pockets and homogenize moisture. An air bubble in the wall becomes a blister or a blown piece in the kiln, so this step is not optional.
2. **Center.** The ball is thrown onto the wheel head and forced into true rotation with braced hands, typically at the higher end of the wheel's range. Nothing after this works if the clay is not centered — an off-center ball produces a wall of varying thickness that collapses as it is raised.
3. **Open.** Thumbs or fingers press down through the center to establish the floor, leaving the base thickness the piece needs — enough to survive trimming later, since the foot is cut from it.
4. **Pull the wall.** Pressure between a hand inside and a hand outside draws clay upward, thinning the wall and raising height. Three to five pulls is normal; each pull moves clay from the base region upward and the wheel is slowed as the wall gets taller and less stable.
5. **Shape and finish.** The profile is formed with ribs and fingers, the rim is compressed to reduce cracking, and the piece is cut off the wheel head with a wire.
6. **Dry to leather-hard and trim.** At leather-hard the piece is inverted on the wheel and the foot ring is cut. Handles and spouts are attached now, with slip.
7. **Dry, bisque, glaze, fire.** Slow, even drying prevents cracks. The bisque firing produces a porous body that absorbs glaze; the glost firing vitrifies body and glaze together.

## Design guidelines

### Rotational symmetry only
The process makes surfaces of revolution. Non-round forms are made by altering a thrown piece while soft, or by another process entirely.

### Uniform wall thickness
0.12–0.3 in (3–8 mm) is the normal range, and the goal is uniformity rather than any particular number. A thick base under a thin wall dries and shrinks at a different rate and cracks at the junction — the most common failure in thrown ware.

### Base and foot
Leave enough thickness in the floor to trim a foot ring later, but no more. An untrimmed thick base is both wasted clay and a drying-crack risk.

### Rim treatment
Compress the rim with a rib or finger before cutting off. An uncompressed rim carries aligned particles and micro-tension and is where drying and firing cracks start.

### Section transitions
Blend every change of thickness. Ceramics have no ductility at any stage of the process, so differential shrinkage between adjacent sections has nowhere to go but into a crack.

### Attachments
Attach handles and spouts at leather-hard using slip and a scored joint, and match the moisture content of the attachment to the body. A drier handle on a wetter body shrinks less and pulls the joint apart.

### Design for shrinkage
Plan for 10–15% total linear shrinkage. A mug thrown to a 3.5 in (89 mm) rim finishes near 3 in (76 mm). Volume shrinks by roughly a third, which matters if the piece has to hold a specified quantity.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Wall thickness | 0.16–0.24 in (4–6 mm) | 0.12 in (3 mm) | Thinner walls collapse while throwing |
| Wall uniformity | Constant top to bottom | 2:1 variation | Differential drying cracks |
| Base thickness | Wall + trimming stock | Thick base cracks | Foot ring is cut from it |
| Linear shrinkage | 10–15% | Body-specific | Sizing must anticipate it |
| Form | Rotationally symmetric | — | Set by the process |

## Cost drivers

The cost of a thrown piece is the maker's time plus two firings. Throwing itself is often the shortest step: trimming, handle attachment, drying, glazing, kiln loading, and firing typically consume more hours per piece than the wheel does.

Kiln economics drive small studios more than most people expect. A firing costs roughly the same whether the kiln is full or half empty, so pieces are held back until a load is ready, and that queuing sets the effective lead time.

Loss rate is the third factor. Cracks from uneven drying, glaze faults, and thermal shock write off pieces that have already absorbed all the labor and one firing. A well-run studio keeps this in the single-digit percentages; new bodies and new glazes push it much higher.

Volume breakpoints: throwing is economical from 1 to a few hundred pieces. Beyond that, [slip casting](/processes/forming/ceramic-slip-casting) or jiggering under [press molding](/processes/forming/press-molding-ceramics) produces repeatable ware at a fraction of the labor, at the cost of model and mold making.

1. Batch by form — throwing twenty of one shape is far faster per piece than twenty different ones.
2. Standardize clay ball weight so wall thickness and finished size repeat.
3. Fill the kiln; a partial firing costs nearly the same as a full one.
4. Dry slowly and evenly under plastic to cut the loss rate.
5. Move to a tooled process as soon as the design is settled and the quantity justifies a model.

## FAQ

### How much does thrown clay shrink?

Roughly 10–15% linearly from wet to fired, combining drying and firing shrinkage. That means a rim thrown at 3.5 in (89 mm) finishes near 3 in (76 mm), and internal volume shrinks by around a third — worth calculating if the piece must hold a specified quantity.

### Why does a thrown pot crack at the base?

Almost always because the floor is much thicker than the wall. The thin wall dries and shrinks first while the thick base is still wet, and the resulting differential has nowhere to go in a material with no ductility. Trim the base to a thickness close to the wall and dry the piece slowly and evenly.

### How thick should a thrown wall be?

0.16–0.24 in (4–6 mm) for most functional ware, and uniform is more important than any specific figure. Below about 0.12 in (3 mm) the wall loses the stiffness it needs to stay up during throwing, and the rim will fold.

### Why is centering the clay so important?

An off-center ball produces a wall of varying thickness as it is pulled, and the thin side gives way while the thick side is still climbing. Nothing later in the process can recover from a poorly centered start, which is why it is the skill that takes longest to learn.

### When should throwing be replaced by a tooled process?

Once the design is settled and the quantity exceeds a few hundred pieces a year. Slip casting or jiggering produces repeatable ware with far less skilled labor per piece; the trade is model and mold making up front, and a loss of the variation that gives hand-thrown work its value.

### Why are ceramics fired twice?

The bisque firing at roughly 1,650–1,830 °F (900–1,000 °C) burns out organics and leaves a porous body strong enough to handle but still absorbent, so it takes up glaze evenly. The glost firing then vitrifies body and glaze together at 1,830–2,550 °F (1,000–1,400 °C) depending on the clay body.

## Alternative processes

- [Ceramic Slip Casting](https://manufacturingprocesses.org/processes/forming/ceramic-slip-casting.md): Ceramic slip casting pours liquid clay slip into a porous plaster mold that draws off water, leaving a solid layer that becomes the part.
- [Press Molding Ceramics](https://manufacturingprocesses.org/processes/forming/press-molding-ceramics.md): Press molding ceramics forms clay between a die and a rotating profile tool or matched mold, producing flatware and repeat shapes quickly.
- [Ceramic Injection Molding (CIM)](https://manufacturingprocesses.org/processes/forming/ceramic-injection-molding-cim.md): Ceramic injection molding molds a ceramic powder and binder feedstock, then debinds and sinters it into a dense technical ceramic part.
- [Metal Spinning](https://manufacturingprocesses.org/processes/forming/metal-spinning.md): Metal spinning presses a rotating metal disc against a mandrel with a roller or hand tool, forming it into an axially symmetric hollow shape.
- [Rotation Molding](https://manufacturingprocesses.org/processes/forming/rotation-molding.md): Rotational molding tumbles powdered polymer inside a heated mold on two axes so it fuses into a seamless, stress-free hollow part with a uniform wall.

## Related processes

- [Ceramic Slip Casting](https://manufacturingprocesses.org/processes/forming/ceramic-slip-casting.md): Ceramic slip casting pours liquid clay slip into a porous plaster mold that draws off water, leaving a solid layer that becomes the part.
- [Press Molding Ceramics](https://manufacturingprocesses.org/processes/forming/press-molding-ceramics.md): Press molding ceramics forms clay between a die and a rotating profile tool or matched mold, producing flatware and repeat shapes quickly.
- [Glassblowing](https://manufacturingprocesses.org/processes/forming/glassblowing.md): Glassblowing gathers molten glass on a blowpipe and inflates it, by hand or by machine, into hollow ware such as bottles, jars and vessels.
- [Metal Spinning](https://manufacturingprocesses.org/processes/forming/metal-spinning.md): Metal spinning presses a rotating metal disc against a mandrel with a roller or hand tool, forming it into an axially symmetric hollow shape.
- [Ceramic Injection Molding (CIM)](https://manufacturingprocesses.org/processes/forming/ceramic-injection-molding-cim.md): Ceramic injection molding molds a ceramic powder and binder feedstock, then debinds and sinters it into a dense technical ceramic part.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/forming/clay-throwing)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Composite Laminating"
category: "Forming"
subcategory: "Composites"
materials: ["Composite"]
tolerances: "±0.010 in (±0.25 mm) on tool-side surfaces. Laminate thickness varies ±10–20% on open (bag-side) surfaces; matched tooling or RTM is required to control both faces."
volumes: "1–5,000 parts per year, depending on route: wet layup at the low end, RTM at the high end"
lead_time: "2–6 weeks including tooling for prototype work; 1–5 days per part once tooling exists, dominated by layup labor and cure cycle."
url: https://manufacturingprocesses.org/processes/forming/composite-laminating
---

# Composite Laminating

Composite laminating lays reinforcing fiber into a mold and consolidates it with resin, curing into a stiff, light monocoque part.

- **Category**: [Forming](https://manufacturingprocesses.org/processes/forming.md)
- **Family**: Composites
- **Materials**: Composite
- **Typical tolerances**: ±0.010 in (±0.25 mm) on tool-side surfaces. Laminate thickness varies ±10–20% on open (bag-side) surfaces; matched tooling or RTM is required to control both faces.
- **Surface finish**: Tool-side surfaces replicate the mold and can be near-optical with a polished tool or gel coat. Bag-side surfaces carry peel-ply or breather texture and normally require filling and finishing if cosmetic.
- **Typical volumes**: 1–5,000 parts per year, depending on route: wet layup at the low end, RTM at the high end
- **Lead time**: 2–6 weeks including tooling for prototype work; 1–5 days per part once tooling exists, dominated by layup labor and cure cycle.

## Overview

Composite laminating builds a part by placing layers of reinforcing fiber into or onto a mold and consolidating them with a thermosetting resin, so the finished laminate is a single monocoque structure whose stiffness and strength follow the fiber directions the designer chose. It is the only process in which the engineer specifies the material and the part simultaneously.

Three routes dominate. Wet lay-up brushes or rolls liquid resin into dry fabric and reaches 30–40% fiber volume fraction by hand, 45–55% under vacuum. Pre-preg lay-up uses fabric pre-impregnated with B-staged resin, cured under vacuum and typically 90 psi (6 bar) of autoclave pressure at 250 °F or 350 °F (120 °C or 177 °C), reaching 55–65% fiber volume. Resin transfer molding (RTM) injects resin into dry fiber in a closed tool, giving two finished faces and better repeatability.

Fiber volume fraction is the number that matters: laminate stiffness scales almost linearly with it, which is why an autoclaved pre-preg part can be roughly 50% stiffer than the same layup done wet by hand.

## How it works

1. **Prepare the tool.** The mold defines the finished surface. Composite tooling is used for prototypes and low volume; machined aluminum or steel for production; invar where the tool's thermal expansion must match the part at cure temperature. Release agent or release film is applied, and gel coat where a finished cosmetic surface is required.
2. **Cut and kit the plies.** Plies are cut to a nested pattern, usually on a CNC ply cutter, and kitted in layup sequence with orientations marked. Typical cured ply thickness is about 0.005 in (0.13 mm) for a 150 gsm unidirectional pre-preg and 0.010 in (0.25 mm) for a 200 gsm woven fabric.
3. **Lay up.** Plies are placed to the ply book, orientation by orientation. Compaction — debulking under vacuum every three to five plies for pre-preg — removes trapped air before it becomes porosity.
4. **Bag.** Peel ply, release film, breather, and vacuum bag are applied and the bag is drawn to 25–29 inHg (0.85–1.0 bar). A leak here is the most common cause of a porous part.
5. **Cure.** Wet lay-up epoxies cure at room temperature and are usually post-cured at 120–180 °F (50–80 °C). Pre-pregs cure at 250 °F or 350 °F (120 °C or 177 °C) for roughly 2 hours, under 90 psi (6 bar) in an autoclave or vacuum-only for out-of-autoclave systems. RTM injects resin at 30–150 psi (2–10 bar) into a heated closed tool, with cycle times from 10 to 60 minutes.
6. **Demold and trim.** Edges are trimmed with diamond or carbide tooling, or [water jet cut](/processes/cutting/water-jet-cutting), and holes are drilled with specialized bits to avoid delamination on breakout.
7. **Inspect.** Void content, ply orientation, and thickness are verified; structural parts are ultrasonically inspected for delamination and porosity.

## Design guidelines

### Balance and symmetry
Lay up symmetric about the mid-plane and balanced in ±θ pairs. An unsymmetric laminate warps as it cools from cure temperature because the resin shrinks against an unbalanced fiber arrangement — the part comes out of the tool the wrong shape and no amount of tooling accuracy fixes it.

### Minimum ply count and the 10% rule
Use at least four plies in any structural laminate, and include a minimum of about 10% of fibers in each of the 0°, ±45°, and 90° directions unless the load case is genuinely uniaxial. All-0° laminates split under any off-axis or bearing load.

### Corner radii and spring-in
Inside radii should be 0.12 in (3 mm) minimum and 0.2 in (5 mm) preferred; outside radii at least twice the laminate thickness. Cured corners spring in by roughly 0.5–2° because through-thickness resin shrinkage exceeds in-plane shrinkage — production tooling is compensated for this, and prototypes should expect it.

### Thickness tolerance
The tool side is accurate to about ±0.010 in (±0.25 mm); the bag side is not, and laminate thickness typically varies ±10–20% unless matched tooling or RTM is used. Never design a fit against a bag-side surface.

### Ply drops
Taper thickness changes gradually: drop no more than one ply at a time and stagger drops by at least 0.4 in (10 mm). An abrupt ply drop is a delamination initiator.

### Joints and inserts
Bonded joints outperform bolted ones in composites, because a bolt hole cuts fibers and concentrates load in a material with no yield mechanism to redistribute it. Where fasteners are unavoidable, use a bonded metallic insert or a locally thickened bearing pad, and follow standard edge distance of at least 3× hole diameter.

### Galvanic compatibility
Carbon fiber is cathodic to aluminum and steel. Any carbon-to-aluminum joint needs an insulating glass ply or sealant between them, or the aluminum corrodes rapidly.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Laminate symmetry | Symmetric and balanced | — | Unsymmetric laminates warp after cure |
| Ply count | 6+ | 4 | Fewer plies cannot balance |
| Inside radius | 0.2 in (5 mm) | 0.12 in (3 mm) | Fiber bridging and resin-rich corners |
| Outside radius | 2× thickness | 1× thickness | Fibers cannot turn a sharp corner |
| Ply drop stagger | 0.8 in (20 mm) | 0.4 in (10 mm) | Abrupt drops delaminate |
| Fastener edge distance | 3× hole diameter | 2.5× | Bearing failure at the free edge |
| Fiber volume fraction | 55–60% (pre-preg) | 30–40% (hand wet) | Stiffness scales with fiber content |

## Variants

- Wet Lay-up
- Pre-preg Lay-up
- Resin Transfer Molding

## Cost drivers

Labor dominates almost every composite laminating operation. Hand layup is measured in plies placed per hour, so ply count, part size, and geometric complexity map almost directly onto cost. Automating with a ply cutter and kitting the layup sequence is the first and largest saving available.

Material cost is next and varies widely: E-glass fabric is inexpensive, standard-modulus carbon considerably more, and intermediate- or high-modulus carbon and aramid more again. Pre-preg carries a cost premium plus frozen storage and a finite out-life, so a shop paying for freezer capacity and tracking out-life has overheads a wet layup shop does not.

Cure equipment is the third. An autoclave is a large capital item with a per-cycle cost, which is why out-of-autoclave pre-pregs and RTM have taken so much of the market. RTM adds matched tooling cost but converts labor into cycle time, which is what makes it viable at volume.

Volume breakpoints: wet layup for 1–50 parts, vacuum-bagged pre-preg for 10–500, RTM from roughly 500 to several thousand a year, and [compression molding](/processes/forming/compression-molding) of [SMC](/processes/forming/dmc-and-smc-molding) above that where the mechanical requirement allows chopped fiber.

1. Reduce ply count and part count before optimizing anything else.
2. Design for a single-piece layup rather than bonded sub-assemblies.
3. Use glass where carbon's stiffness is not needed — much of a typical laminate is there for handling, not stiffness.
4. Accept a bag-side surface wherever cosmetics allow; matched tooling doubles tool cost.
5. Design bonded joints instead of bolted ones; every hole is machined labor plus a stress concentration.

## FAQ

### What fiber volume fraction should I expect?

30–40% for hand wet layup, 45–55% for vacuum-bagged wet layup, and 55–65% for autoclaved pre-preg. Laminate stiffness scales close to linearly with fiber volume fraction, so the consolidation method matters as much as the fiber choice.

### Why must a composite laminate be symmetric?

Resin shrinks as it cures and the laminate cools from cure temperature. If the ply stack is not symmetric about its mid-plane, that shrinkage is unbalanced through the thickness and the part warps off the tool. Symmetric and balanced layups are a hard rule, not a preference.

### What is spring-in and how much should I allow?

Corners of a cured laminate close slightly because through-thickness resin shrinkage exceeds in-plane shrinkage. Typical spring-in is 0.5–2° per corner. Production tooling is machined with the angle compensated; on a prototype tool, expect the part to come off the mold slightly tighter than nominal.

### Should composite parts be bolted or bonded?

Bonded, wherever possible. Drilling a hole cuts load-carrying fibers and concentrates stress in a material with no yield mechanism to redistribute it. Where fasteners are unavoidable, use a locally thickened bearing pad or a bonded metallic insert and keep edge distance at 3× hole diameter or more.

### What is the minimum inside radius for a laminated corner?

0.12 in (3 mm) as an absolute minimum, 0.2 in (5 mm) preferred. Tighter radii cause fiber bridging, where plies span the corner instead of conforming, leaving a resin-rich pocket that is the weakest point of the part.

### Can carbon fiber be bonded directly to aluminum?

Not without isolation. Carbon is strongly cathodic to aluminum, and a direct joint drives galvanic corrosion of the aluminum in any damp environment. Interleave a glass ply or an insulating adhesive layer, and seal the joint edge.

### When does RTM beat hand layup?

Roughly from a few hundred parts a year upward, and sooner if both surfaces must be finished. RTM trades tooling cost and cycle discipline for a large reduction in layup labor, more consistent fiber volume fraction, and two molded faces instead of one.

## Alternative processes

- [DMC and SMC Molding](https://manufacturingprocesses.org/processes/forming/dmc-and-smc-molding.md): DMC and SMC molding compression-molds a pre-mixed dough or sheet of chopped fiber and thermoset resin into a stiff, dimensionally stable part.
- [Filament Winding](https://manufacturingprocesses.org/processes/forming/filament-winding.md): Filament winding wraps resin-impregnated continuous fiber onto a rotating mandrel in a controlled pattern, producing tubes and pressure vessels.
- [Pultrusion](https://manufacturingprocesses.org/processes/forming/pultrusion.md): Pultrusion pulls continuous fiber through a resin bath and a heated die, curing it into a constant-section composite profile of unlimited length.
- [Injection Molding](https://manufacturingprocesses.org/processes/forming/injection-molding.md): Injection molding forces molten thermoplastic into a steel mold under high pressure, producing high-volume identical plastic parts.
- [Thermoforming](https://manufacturingprocesses.org/processes/forming/thermoforming.md): Thermoforming heats a plastic sheet until pliable and pulls or presses it over a single-sided mold, producing large thin-walled parts with low tooling cost.

## Related processes

- [Filament Winding](https://manufacturingprocesses.org/processes/forming/filament-winding.md): Filament winding wraps resin-impregnated continuous fiber onto a rotating mandrel in a controlled pattern, producing tubes and pressure vessels.
- [Pultrusion](https://manufacturingprocesses.org/processes/forming/pultrusion.md): Pultrusion pulls continuous fiber through a resin bath and a heated die, curing it into a constant-section composite profile of unlimited length.
- [DMC and SMC Molding](https://manufacturingprocesses.org/processes/forming/dmc-and-smc-molding.md): DMC and SMC molding compression-molds a pre-mixed dough or sheet of chopped fiber and thermoset resin into a stiff, dimensionally stable part.
- [Compression Molding](https://manufacturingprocesses.org/processes/forming/compression-molding.md): Compression molding presses a measured charge of rubber or thermoset into an open heated mold that closes to cure it into its final shape.
- [Vacuum Casting (Urethane Casting)](https://manufacturingprocesses.org/processes/forming/vacuum-casting.md): Vacuum casting, known in the US as urethane casting, pours polyurethane resin into a silicone mold under vacuum to copy a master pattern for short runs.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/forming/composite-laminating)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Compression Molding"
category: "Forming"
subcategory: "Plastics and Rubber"
materials: ["Plastic"]
tolerances: "About ±0.005 in (±0.13 mm) on features contained within one mold half; ±0.010 in (±0.25 mm) or more on dimensions spanning the parting line, since flash thickness varies with charge weight. Rubber tolerance classes follow ISO 3302-1."
volumes: "100–50,000 parts per year; above that, injection or transfer molding usually takes over"
lead_time: "3–8 weeks for tooling; cycles of 1–10+ minutes depending on section thickness, plus deflash and any post-cure"
url: https://manufacturingprocesses.org/processes/forming/compression-molding
---

# Compression Molding

Compression molding presses a measured charge of rubber or thermoset into an open heated mold that closes to cure it into its final shape.

- **Category**: [Forming](https://manufacturingprocesses.org/processes/forming.md)
- **Family**: Plastics and Rubber
- **Materials**: Plastic
- **Typical tolerances**: About ±0.005 in (±0.13 mm) on features contained within one mold half; ±0.010 in (±0.25 mm) or more on dimensions spanning the parting line, since flash thickness varies with charge weight. Rubber tolerance classes follow ISO 3302-1.
- **Surface finish**: Reproduces the mold finish, from polished gloss to blasted matte; a flash witness line at the parting line is unavoidable
- **Typical volumes**: 100–50,000 parts per year; above that, injection or transfer molding usually takes over
- **Lead time**: 3–8 weeks for tooling; cycles of 1–10+ minutes depending on section thickness, plus deflash and any post-cure

## Overview

Compression molding places a measured charge of uncured rubber or thermoset compound into an open, heated mold and closes it under pressure, so the material flows to fill the cavity and cures in place. It is the oldest production molding process still in wide industrial use and remains the default for rubber seals, gaskets, bushings, and mounts, and for thermoset parts in phenolic, melamine, urea, and epoxy — electrical insulators, cookware handles, brake pads, and closures.

The two branches diverge in what they optimize. Compression molding rubber handles high-viscosity elastomer stocks that would scorch in a long injection runner. Compression molding plastic covers glass-reinforced thermosets such as SMC and BMC, where the reinforcement must not be broken up by shearing through a gate.

Tooling is simpler and cheaper than an injection mold because there is no runner, gate, or high-pressure injection unit — but cycles are measured in minutes.

## How it works

1. **Prepare the charge.** Rubber stock is calendered and cut, or extruded and slug-cut, to a preform of controlled weight. Thermoset molding compound arrives as pellets, granules, or a preformed puck; SMC is cut from sheet. Charge weight tolerance directly becomes flash thickness and part height variation, so it is weighed rather than estimated.

2. **Load.** The charge is placed in the open, heated cavity. Placement matters: the compound flows outward from where it sits, so a poorly placed charge produces knit lines, entrapped air, and uneven fiber orientation in reinforced compounds. Charge coverage of roughly 30–70% of the cavity area is typical for SMC.

3. **Close and pressurize.** The press closes and applies pressure on the projected area, commonly in the range of 300–2,000 psi (2–14 MPa) depending on compound viscosity and part complexity. Bulky phenolic charges sit at the high end; low-viscosity rubber compounds at the low end.

4. **Cure under heat and pressure.** Mold temperatures typically run 275–320 °F (135–160 °C) for SMC, 300–360 °F (150–180 °C) for phenolics, and 320–380 °F (160–195 °C) for most rubber compounds. Curing is heat-transfer limited, so cure time rises steeply with section thickness: thin rubber gaskets cure in 1–3 minutes while thick sections take 10 minutes or more. Bumping (briefly opening the press early in the cycle) vents volatiles and trapped air.

5. **Demold and deflash.** Thermosets are demolded hot, since they do not soften on reheating. Flash at the parting line is trimmed, tumbled, or cryogenically deflashed for rubber. Rubber parts for critical sealing applications are then post-cured in an oven to complete crosslinking and drive off residual volatiles.

### Where does transfer molding fit?

Transfer molding is the intermediate step: the charge is loaded into a separate pot and forced through a sprue into a closed cavity. It gives better dimensional control and allows encapsulating delicate inserts, at the cost of sprue scrap and a more complex tool.

## Design guidelines

### Wall thickness

Practical range is 0.060–0.250 in (1.5–6 mm) for most thermoset parts, and thicker for rubber. Uniformity matters less than in injection molding — there is no gate to freeze off — but cure time scales steeply with the thickest section, so a single heavy boss can double the cycle for the whole part.

### Draft

1–2° per side is the norm for thermosets; rubber can go to 0.5° or effectively zero, since a cured elastomer strips off tooling that a rigid part could never leave. Add draft on textured or deep-drawn features.

### Radii

Internal radii of at least 0.030 in (0.8 mm), and more on structural corners. Sharp internal corners in a filled thermoset are both stress risers and points where the flow front has to turn hard, which locally reorients fibers and creates a weak line.

### The parting line and flash

Every compression molded part has flash, because the mold closes on an excess charge and squeezes it out. Put the parting line where flash removal is easy and where a witness line is acceptable. Dimensions that span the parting line vary with charge weight and press closure, so keep tight tolerances inside a single mold half wherever you can.

### Undercuts, inserts, and ribs

Metal inserts, threaded bushings, and reinforcing frames are loaded into the cavity and encapsulated. Undercuts in rubber can often be stripped off directly because of the material's elongation; in rigid thermosets they need a split cavity or a loose piece. Ribs behave much as they do in thermoplastics — the rib and boss ratios tabulated on [/charts/injection-molding-design-guidelines](/charts/injection-molding-design-guidelines) transfer directly, because the sink-mark mechanism is the same shrinkage differential.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Wall thickness | 0.090–0.190 in (2.3–4.8 mm) | 0.060–0.250 in (1.5–6 mm) | Cure time is set by the thickest section |
| Draft, thermoset | 1–2° per side | 0.5° | Rigid part must release from a hot mold |
| Draft, rubber | 0.5–1° per side | Effectively 0° | Elastomer elongation lets parts strip off |
| Internal radius | 0.060 in (1.5 mm) | 0.030 in (0.8 mm) | Sharp corners reorient fibers and concentrate stress |
| Rib base | 0.5–0.6 × wall | — | Same shrinkage-sink mechanism as thermoplastics |
| Tolerance, in one half | ±0.005 in (±0.13 mm) | — | Not affected by charge weight or press closure |
| Tolerance, across parting line | ±0.010 in (±0.25 mm) or more | — | Flash thickness varies with charge weight |

## Variants

- Compression Molding Rubber
- Compression Molding Plastic

## Cost drivers

Tooling is meaningfully cheaper than an injection mold: no runner system, no gates, no hot half, no injection unit, and lower clamp requirements per unit of projected area. Against that, the cycle is measured in minutes rather than seconds, charge preparation is a real labor step, and every part needs deflashing.

Volume breakpoints:

- Under 500 parts: single-cavity tooling, hand-loaded charges, hand deflash. Common for rubber prototypes and short-run seals.
- 500–50,000: multi-cavity tooling in a hydraulic press; charge preforms cut on a dedicated line.
- Above roughly 50,000 parts a year for a small rubber part, injection or transfer molding usually wins on cycle time and consistency.

Cost reduction:

1. **Reduce the thickest section.** Cure time follows the heaviest wall, and cure time is the cycle. Coring out one thick boss can cut minutes from every shot.
2. **Add cavities.** The press is heating and holding pressure regardless; a second, fourth, or sixteenth cavity spreads that fixed cycle across more parts.
3. **Control charge weight tightly.** Excess charge is both wasted material and thicker flash to trim; underweight charge is a short shot. Preform tooling pays for itself quickly.
4. **Design for automated deflash.** Parts that can be cryogenically tumbled avoid hand trimming entirely, which is often the largest labor line on a rubber part.
5. **Keep tight tolerances within one mold half.** Dimensions spanning the parting line carry the charge-weight variation and are far more expensive to hold.

## FAQ

### What is the difference between compression molding and injection molding?

Compression molding loads a measured charge into an open heated cavity and closes the press on it, so there is no runner, gate, or injection unit. That makes tooling cheaper and avoids shearing glass reinforcement through a gate, but the cycle runs minutes instead of seconds and every part carries parting-line flash.

### What pressure does compression molding use?

Commonly 300–2,000 psi (2–14 MPa) on the projected area, with bulky high-viscosity phenolic charges at the top of the range and low-viscosity rubber compounds near the bottom. That is far below injection molding's 5,000–20,000 psi, which is why compression tooling and presses are less expensive per unit of projected area.

### How long does a compression molding cycle take?

Cure is heat-transfer limited, so cycle time is set by the thickest section. Thin rubber gaskets cure in 1–3 minutes; thick sections take 10 minutes or more. Reducing the heaviest wall in the part is the single most effective way to shorten the cycle.

### What tolerance can compression molding hold?

About ±0.005 in (±0.13 mm) on features contained within one mold half. Dimensions spanning the parting line are looser — ±0.010 in (±0.25 mm) or more — because flash thickness varies with charge weight and press closure. Keep critical dimensions inside a single half wherever the design allows.

### Why does compression molding always produce flash?

The mold closes on a deliberately excess charge to guarantee the cavity fills, and the surplus escapes at the parting line. Charge weight is controlled to minimize it, but some flash is inherent. Place the parting line where a witness line is cosmetically acceptable and where trimming — ideally cryogenic tumbling — is easy.

### When should I use transfer molding instead?

When you need better dimensional control or you are encapsulating delicate inserts such as electrical terminals. Transfer molding loads the charge into a separate pot and forces it through a sprue into an already-closed cavity, which avoids the insert being displaced by the closing press. The trade is sprue scrap and a more complex tool.

## Alternative processes

- [Injection Molding](https://manufacturingprocesses.org/processes/forming/injection-molding.md): Injection molding forces molten thermoplastic into a steel mold under high pressure, producing high-volume identical plastic parts.
- [Liquid Silicone Rubber (LSR) Molding](https://manufacturingprocesses.org/processes/forming/liquid-silicone-rubber-lsr-molding.md): LSR molding injects two-part liquid silicone into a heated mold where it cures, producing flexible parts with a very wide service temperature range.
- [DMC and SMC Molding](https://manufacturingprocesses.org/processes/forming/dmc-and-smc-molding.md): DMC and SMC molding compression-molds a pre-mixed dough or sheet of chopped fiber and thermoset resin into a stiff, dimensionally stable part.
- [Reaction Injection Molding](https://manufacturingprocesses.org/processes/forming/reaction-injection-molding.md): Reaction injection molding mixes two liquid reactants that polymerize inside a low-pressure mold, producing large tough polyurethane parts on light tooling.

## Related processes

- [DMC and SMC Molding](https://manufacturingprocesses.org/processes/forming/dmc-and-smc-molding.md): DMC and SMC molding compression-molds a pre-mixed dough or sheet of chopped fiber and thermoset resin into a stiff, dimensionally stable part.
- [Liquid Silicone Rubber (LSR) Molding](https://manufacturingprocesses.org/processes/forming/liquid-silicone-rubber-lsr-molding.md): LSR molding injects two-part liquid silicone into a heated mold where it cures, producing flexible parts with a very wide service temperature range.
- [Injection Molding](https://manufacturingprocesses.org/processes/forming/injection-molding.md): Injection molding forces molten thermoplastic into a steel mold under high pressure, producing high-volume identical plastic parts.
- [Overmolding](https://manufacturingprocesses.org/processes/forming/overmolding.md): Overmolding molds a second material, usually a soft elastomer, directly onto an already-molded rigid substrate so the two bond permanently.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/forming/compression-molding)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "DMC and SMC Molding"
category: "Forming"
subcategory: "Composites"
materials: ["Composite", "Plastic"]
tolerances: "About ±0.1–0.3% of dimension, roughly ±0.005 in/in, thanks to the near-zero mold shrinkage of low-profile compounds. Tighter than most thermoplastic molding, and dimensionally stable with temperature."
volumes: "1,000–100,000 parts per year"
lead_time: "10–20 weeks for steel tooling; cycle times of 60–180 seconds per part in production."
url: https://manufacturingprocesses.org/processes/forming/dmc-and-smc-molding
---

# DMC and SMC Molding

DMC and SMC molding compression-molds a pre-mixed dough or sheet of chopped fiber and thermoset resin into a stiff, dimensionally stable part.

- **Category**: [Forming](https://manufacturingprocesses.org/processes/forming.md)
- **Family**: Composites
- **Materials**: Composite, Plastic
- **Typical tolerances**: About ±0.1–0.3% of dimension, roughly ±0.005 in/in, thanks to the near-zero mold shrinkage of low-profile compounds. Tighter than most thermoplastic molding, and dimensionally stable with temperature.
- **Surface finish**: Molded surfaces replicate the tool. Low-profile SMC with a polished tool and a coating step reaches Class A automotive paintability; standard compounds show a fine fiber pattern and are normally primed and painted.
- **Typical volumes**: 1,000–100,000 parts per year
- **Lead time**: 10–20 weeks for steel tooling; cycle times of 60–180 seconds per part in production.

## Overview

SMC (sheet molding compound) and DMC (dough molding compound, also called BMC) are compression molding compounds of chopped glass fiber, thermosetting polyester or vinyl ester resin, mineral filler, and a thickening system. SMC arrives as a tacky sheet with roughly 1 in (25 mm) fibers at 25–30% glass by weight; DMC arrives as a bulk dough with shorter 0.12–0.5 in (3–12 mm) fibers at 15–20% glass and more filler.

Both are compression molded in matched heated steel tools at roughly 285–320 °F (140–160 °C) under 700–1,500 psi (50–100 bar), curing in 60–180 seconds depending on wall thickness. Low-profile additives bring mold shrinkage to 0–0.2%, which is what makes Class A automotive body panels possible in a thermoset.

The result is a stiff, dimensionally stable, thermally and electrically stable part at 1.8–2.0 g/cm³ — heavier than a thermoplastic but far stiffer at temperature. Truck hoods, electrical switchgear and breaker components, headlamp reflectors, battery trays, and shower trays are typical. Volumes run 1,000 to 100,000 a year against steel tooling.

## How it works

1. **Compound.** For SMC, resin paste containing filler, catalyst, low-profile additive, and a thickener is doctored onto carrier film, chopped roving is rained onto it, a second filmed layer is laid on top, and the sandwich is compacted. For DMC, all constituents including the short fiber are mixed in a bulk mixer.
2. **Mature.** SMC is held for roughly 2–7 days at controlled temperature while the magnesium-oxide thickening system raises viscosity by orders of magnitude, turning a sticky paste into a leathery, handleable sheet. Under-matured SMC runs resin-rich; over-matured SMC will not flow.
3. **Cut the charge.** Sheet is cut into blanks and stacked to a mass that matches the part, and the carrier film is stripped. Charge mass is controlled tightly — the process has no runner to absorb error.
4. **Place the charge.** The stack covers roughly 30–70% of the cavity area. Placement determines flow direction, which in turn determines fiber orientation, knit line position, and where the part will be weakest. This is the single most important process variable.
5. **Close and cure.** The press closes and holds at roughly 700–1,500 psi (50–100 bar) with tool surfaces at 285–320 °F (140–160 °C). Material flows to fill the cavity and cures in place; cure time runs roughly 30–40 seconds per millimeter of wall thickness.
6. **Eject and finish.** The part is ejected hot and dimensionally stable — thermosets do not need cooling in the tool. Flash is removed, and painted parts go through a wash and prime cycle. Metal inserts can be molded in place, which is common in electrical hardware.

## Design guidelines

### Wall thickness
0.060–0.250 in (1.5–6 mm) is the working range, with 0.100 in (2.5 mm) typical for panels. Cure time scales with thickness, so a thick section costs cycle time everywhere in the part.

### Uniform sections and ribs
Keep the nominal wall uniform and add stiffness with ribs rather than thickness. Rib base thickness should be about 60–75% of the adjoining wall to limit sink, though thermoset shrinkage is low enough that SMC tolerates thicker ribs than a thermoplastic would.

### Draft
1–3° on all vertical faces, more on textured surfaces. Thermoset parts are ejected hot and rigid, so they do not shrink onto cores the way a thermoplastic does, but glass fiber makes the surface abrasive and low draft galls tooling.

### Radii
0.060 in (1.5 mm) minimum inside radius, 0.125 in (3 mm) preferred. Sharp corners create resin-rich, fiber-poor regions where the compound cannot carry fiber around the turn.

### Charge placement and knit lines
Where two flow fronts meet, fibers do not cross the boundary and the knit line is a strength defect. Design so the knit line falls in a low-stress region, and expect the tool trials to move it.

### Molded-in inserts
Threaded inserts, brackets, and grounding hardware can be placed in the tool and molded in. This is one of SMC and DMC's strongest advantages over sheet metal fabrication for electrical enclosures.

### Class A surfaces
Achievable with low-profile SMC and a polished tool, but it requires shrinkage control at 0–0.2%, careful charge placement, and usually an in-mold or post-mold coating step. Do not assume Class A comes free with the material.

### Electrical performance
DMC/BMC is a standard material for arc-resistant and tracking-resistant electrical components, which is a functional reason to choose it over a filled thermoplastic rather than a cost one.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Wall thickness | 0.100 in (2.5 mm) | 0.060–0.250 in (1.5–6 mm) | Cure time scales with thickness |
| Rib base thickness | 60–75% of wall | 100% of wall | Sink and cure differential |
| Draft | 2° | 1° | Abrasive glass galls tooling |
| Inside radius | 0.125 in (3 mm) | 0.060 in (1.5 mm) | Fiber cannot turn a sharp corner |
| Charge coverage | 50% of cavity | 30–70% | Sets flow, fiber orientation, knit lines |
| Molded-in insert | Knurled, with resin shoulder | — | Prevents pull-out and reduces stress |

## Cost drivers

Steel compression tooling is the entry cost, comparable to an injection mold of similar size, and it is the reason the process needs 1,000 parts a year or more to make sense. Compression tools are somewhat simpler than injection tools — no runner system, lower clamp requirements per unit area than injection — but they are still matched steel.

Cycle time is set by cure, which is set by wall thickness: roughly 30–40 seconds per millimeter. A 3 mm panel cures in around 90–120 seconds, so a single press produces on the order of 20–30 parts an hour. Press tonnage requirements are substantial, since molding pressure applies over the full projected area.

Material is inexpensive relative to other composites — filler is a large fraction of the compound by weight — but SMC has a finite shelf life after maturation and must be stored at controlled temperature, which is an operational cost.

Volume breakpoints: below roughly 1,000 parts a year, hand [composite laminating](/processes/forming/composite-laminating) or RTM is usually cheaper. From 1,000 to 100,000, SMC and DMC compression molding is the sweet spot. Above that, and where the material allows, injection-molded BMC or a filled thermoplastic can be faster.

1. Hold the nominal wall thin and uniform — cure time is the cycle.
2. Mold in inserts and hardware rather than assembling them afterward.
3. Consolidate a sheet metal assembly into a single molded part; that is where the process pays.
4. Place knit lines deliberately by controlling charge placement.
5. Specify Class A only on the surfaces that are actually visible.

## FAQ

### What is the difference between SMC and DMC?

SMC is supplied as a matured sheet with roughly 1 in (25 mm) chopped fibers at 25–30% glass by weight, giving higher strength and suiting large panels. DMC (also called BMC) is a bulk dough with 0.12–0.5 in (3–12 mm) fibers at 15–20% glass and more filler, which flows into intricate geometry and can also be injection molded.

### Why does SMC need to mature before molding?

A magnesium-oxide thickening system raises the compound's viscosity by orders of magnitude over roughly 2–7 days, turning a sticky paste into a handleable leathery sheet that will carry fiber with it as it flows. Under-matured compound runs resin-rich; over-matured compound will not fill the tool.

### Can SMC produce a Class A automotive surface?

Yes, and that is one of its historical justifications. It requires a low-profile additive package that holds mold shrinkage to 0–0.2%, a polished tool, controlled charge placement, and normally an in-mold or post-mold coating step. Class A is an engineered outcome, not a property of the material.

### What molding pressure and temperature does SMC need?

Roughly 700–1,500 psi (50–100 bar) over the projected area, with tool surfaces at 285–320 °F (140–160 °C). Cure time runs about 30–40 seconds per millimeter of wall thickness, so a 3 mm panel is in the tool for approximately 90–120 seconds.

### Why is BMC used for electrical components?

Thermoset polyester compounds have high comparative tracking index and arc resistance, retain stiffness at elevated temperature, and do not soften or drip. That combination — plus the ability to mold in metal contacts and inserts — makes DMC/BMC a standard material for breakers, switchgear, and terminal blocks.

### How does SMC compare with hand-laid composite?

SMC uses short chopped fiber and reaches lower specific strength than a continuous-fiber laminate, but it molds in minutes rather than hours, holds tighter tolerance, and gives two finished faces. Choose continuous-fiber laminating for structural efficiency and SMC for repeatable production geometry.

## Alternative processes

- [Compression Molding](https://manufacturingprocesses.org/processes/forming/compression-molding.md): Compression molding presses a measured charge of rubber or thermoset into an open heated mold that closes to cure it into its final shape.
- [Injection Molding](https://manufacturingprocesses.org/processes/forming/injection-molding.md): Injection molding forces molten thermoplastic into a steel mold under high pressure, producing high-volume identical plastic parts.
- [Composite Laminating](https://manufacturingprocesses.org/processes/forming/composite-laminating.md): Composite laminating lays reinforcing fiber into a mold and consolidates it with resin, curing into a stiff, light monocoque part.
- [Sheet Metal Fabrication](https://manufacturingprocesses.org/processes/forming/sheet-metal-fabrication.md): Sheet metal fabrication cuts, bends and joins flat stock into a finished assembly, typically laser cutting followed by press braking, hardware and welding.
- [Thermoforming](https://manufacturingprocesses.org/processes/forming/thermoforming.md): Thermoforming heats a plastic sheet until pliable and pulls or presses it over a single-sided mold, producing large thin-walled parts with low tooling cost.

## Related processes

- [Compression Molding](https://manufacturingprocesses.org/processes/forming/compression-molding.md): Compression molding presses a measured charge of rubber or thermoset into an open heated mold that closes to cure it into its final shape.
- [Composite Laminating](https://manufacturingprocesses.org/processes/forming/composite-laminating.md): Composite laminating lays reinforcing fiber into a mold and consolidates it with resin, curing into a stiff, light monocoque part.
- [Injection Molding](https://manufacturingprocesses.org/processes/forming/injection-molding.md): Injection molding forces molten thermoplastic into a steel mold under high pressure, producing high-volume identical plastic parts.
- [Pultrusion](https://manufacturingprocesses.org/processes/forming/pultrusion.md): Pultrusion pulls continuous fiber through a resin bath and a heated die, curing it into a constant-section composite profile of unlimited length.
- [Filament Winding](https://manufacturingprocesses.org/processes/forming/filament-winding.md): Filament winding wraps resin-impregnated continuous fiber onto a rotating mandrel in a controlled pattern, producing tubes and pressure vessels.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/forming/dmc-and-smc-molding)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Deep Drawing"
category: "Forming"
subcategory: "Metal"
materials: ["Metal"]
tolerances: "Roughly ±0.010 in (±0.25 mm) on drawn diameters and depth; ironed walls hold tighter, near ±0.002 in (±0.05 mm). Wall thickness varies 10–25% from the blank gauge by design"
volumes: "25,000–10,000,000+ parts; below about 5,000 a single-sided process is usually cheaper"
lead_time: "8–16 weeks for a multi-station draw die; 1–3 weeks per production run thereafter"
url: https://manufacturingprocesses.org/processes/forming/deep-drawing
---

# Deep Drawing

Deep drawing pushes a sheet metal blank through a die with a punch so the metal flows into a seamless cup or box deeper than its own diameter.

- **Category**: [Forming](https://manufacturingprocesses.org/processes/forming.md)
- **Family**: Metal
- **Materials**: Metal
- **Typical tolerances**: Roughly ±0.010 in (±0.25 mm) on drawn diameters and depth; ironed walls hold tighter, near ±0.002 in (±0.05 mm). Wall thickness varies 10–25% from the blank gauge by design
- **Surface finish**: Reflects the incoming coil finish, typically 16–63 µin Ra (0.4–1.6 µm), with draw lines possible on the wall
- **Typical volumes**: 25,000–10,000,000+ parts; below about 5,000 a single-sided process is usually cheaper
- **Lead time**: 8–16 weeks for a multi-station draw die; 1–3 weeks per production run thereafter

## Overview

Deep drawing pushes a sheet metal blank into a die cavity with a punch, so the flange material flows radially inward and becomes the wall of a seamless cup, can, shell, or box. The defining test is depth relative to diameter: a part drawn deeper than its own diameter is deep drawn, while shallower forming is stretch forming or simple stamping.

It is the process behind beverage cans, cartridge cases, kitchen sinks, oil filter shells, and automotive fuel tanks. Materials must be ductile and have good through-thickness anisotropy — drawing-quality low-carbon steel, 304 stainless, 1100/3003/5052 aluminum, and cartridge brass are the workhorses.

A well-designed first draw reduces blank diameter to roughly half, giving a limiting draw ratio of about 2.0; deeper cups need redraws, each taking a further 20–25% off the diameter. Walls thin by roughly 10–25% during the draw, so a drawn part is not a constant-thickness part.

## How it works

1. **Blank.** A circular or profiled blank is cut from coil, sized by surface-area development plus a trim allowance. Blank diameter versus punch diameter sets the draw ratio, which is the single most important number in the design.
2. **Load and clamp.** The blank sits on the draw ring, and a blank holder (binder) presses it down. Blank holder force is the process's balance point: too little and the flange buckles into wrinkles as it is compressed circumferentially, too much and the wall tears because the flange cannot feed inward.
3. **Draw.** The punch descends and the flange metal flows over the die radius into the cavity. Punch-to-die clearance is set at roughly 1.1–1.3× the sheet thickness per side so the wall is not ironed unintentionally.
4. **Redraws.** A cup deeper than roughly 75% of its diameter usually cannot be made in one hit. Each redraw pulls the cup through a smaller die, taking off another 20–25% of diameter. Work hardening accumulates; interstage annealing is required for stainless and heavily worked aluminum.
5. **Ironing (optional).** A pass with clearance deliberately less than the wall thickness thins and lengthens the wall to a controlled dimension. This is how a beverage can gets a wall far thinner than its base.
6. **Trim and pierce.** Earing — the scalloped top edge caused by planar anisotropy in the rolled sheet — is trimmed off, and holes are pierced after drawing rather than before.

Material selection is metallurgy, not just strength. The plastic strain ratio *r* measures a sheet's willingness to thin from the width rather than the thickness; drawing-quality steels reach r values around 1.4–2.0, which is why they outdraw aluminum, whose r is typically below 1. The strain hardening exponent *n* governs how much stretch the part can take before necking.

## Design guidelines

### Depth-to-diameter and the draw ratio

Aim for a first draw that reduces blank diameter by no more than about half — a limiting draw ratio near 2.0 for a good drawing steel, less for aluminum and stainless. In practice a cup up to about 75% as deep as it is wide is a single-operation part. Anything deeper is a multi-station tool, and each station is tooling money.

### Die and punch radii

Die profile radius should be about 4–8× sheet thickness. Too small and the sheet tears as it bends over the entry; too large and the flange is unsupported and wrinkles. Punch nose radius should be at least 3× thickness — a sharp punch nose concentrates strain at the cup bottom, which is where fractures start.

### Punch-to-die clearance

Use roughly 1.1–1.3× material thickness per side for a plain draw. Clearance below material thickness irons the wall, which is a deliberate operation with its own force and lubrication requirements, not something to fall into by accident.

### Corner radii on rectangular draws

Corners of a box draw are effectively small-diameter deep draws and are always the failure site. Keep corner radius at a minimum of 4–6× material thickness, and larger is much better. The bottom radius at the corner should be at least as generous as the side radius.

### Expect thinning, not uniform wall

The wall thins 10–25% relative to the blank, most severely just above the punch nose radius, while the flange actually thickens. Do not dimension a drawn wall as though it were the original gauge, and do not put a sealing or bearing surface at the thinnest region.

### Blank development and material

Blank diameter comes from surface-area equivalence plus 10–15% trim allowance for earing. Specify a drawing-quality grade explicitly — DDQ or EDDQ steel, 3003-O or 5052-O aluminum, 304 with a controlled ferrite level. A structural grade with the same nominal strength will not draw. Start from the [sheet metal gauge chart](/charts/sheet-metal-gauge-chart) when choosing blank thickness, and the [bend radius and K-factor chart](/charts/sheet-metal-bend-radius-k-factor) when developing flat patterns for any flanges added after the draw.

### Pierce after drawing

Holes punched in the blank distort into ovals as the metal flows. Pierce after the final draw unless the hole is in the flat bottom and well away from the punch radius.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| First draw ratio (blank ÷ punch dia) | 1.6–1.8 | ≈2.0 in drawing-quality steel | Above the limit the wall tears before the flange feeds |
| Cup depth, one operation | ≤ 0.75× diameter | Deeper needs redraws | Single-hit strain limit for most ductile sheet |
| Reduction per redraw | 20% | 25% | Work hardening reduces remaining ductility each pass |
| Die profile radius | 6× thickness | 4× thickness | Small radii tear the sheet at the die entry |
| Punch nose radius | 4–6× thickness | 3× thickness | Sharp noses concentrate strain at the cup bottom |
| Punch-to-die clearance | 1.1–1.2× thickness/side | 1.3× thickness/side | Less than 1× thickness becomes an ironing operation |
| Box corner radius | ≥ 6× thickness | 4× thickness | Corners are the highest-strain region of the draw |
| Wall thinning | Design for 10–20% | 25% | Metal comes from the wall as the flange feeds in |

## Cost drivers

Cost is dominated by the die set and the number of stations in it. A single-draw tool is a modest investment; a five-station transfer or progressive tool that draws, redraws, irons, trims, and pierces is a major one, and each added station multiplies both tooling cost and press requirements. Press time per part is short — a progressive draw line runs continuously — so at volume the material and the amortized tool dominate, not machine hours.

Volume breakpoints: below roughly 5,000 parts, metal spinning or hydroforming with a single-sided tool almost always beats a matched draw die. From 25,000 to a few hundred thousand, a conventional draw die pays back. Above a million, transfer presses with progressive tooling and optimized blank nesting take over, and blank utilization becomes the biggest single lever on cost.

1. **Cut a station.** Relaxing depth or opening a corner radius so the part draws in two hits instead of three removes a whole tool section.
2. **Nest the blank.** Round blanks waste 10–20% of the strip even when nested well. A profiled or squared blank, where the part allows, recovers much of it.
3. **Specify drawing-quality material.** Scrap from tearing dwarfs the price premium on a proper DDQ or EDDQ grade.
4. **Avoid interstage annealing.** Every anneal is a furnace pass and a handling cycle. Splitting the reduction across more stations is often cheaper than annealing.
5. **Let the wall thickness float.** Holding a drawn wall to a tight thickness forces an ironing station and much higher press tonnage.

## FAQ

### How deep can you deep draw in one operation?

About 75% of the cup diameter for a good drawing-quality steel, corresponding to a limiting draw ratio near 2.0 between blank and punch diameter. Deeper cups need redraws, each removing a further 20–25% of the diameter, and each redraw adds a die station.

### What punch and die radii should a draw tool use?

Die profile radius of roughly 4–8× sheet thickness and punch nose radius of at least 3× thickness. A die radius that is too small tears the sheet at the entry; one that is too large leaves the flange unsupported and it wrinkles.

### How much does the wall thin during deep drawing?

Typically 10–25% relative to the blank thickness, with the thinnest point just above the punch nose radius. The flange, in contrast, thickens. Never place a sealing or bearing surface at the thinnest region, and do not dimension the wall as though it were the original gauge.

### What causes wrinkles in a deep drawn part?

Insufficient blank holder force. As the flange is pulled inward its circumference must shrink, putting it in compression; without enough hold-down it buckles. Too much blank holder force causes the opposite failure — the flange cannot feed and the wall tears.

### What is the best material for deep drawing?

Drawing-quality low-carbon steel (DDQ or EDDQ), which reaches plastic strain ratios of roughly 1.4–2.0, plus annealed 3003 and 5052 aluminum, 304 stainless, and cartridge brass. The r-value matters more than tensile strength — it measures whether the sheet thins from its width or its thickness.

### Deep drawing or metal spinning?

Spinning needs only a mandrel, so it wins below roughly 5,000 pieces and for large diameters. Deep drawing wins at volume — cycle times are seconds rather than minutes — and it can produce non-round shapes such as rectangular pans, which spinning cannot.

## Alternative processes

- [Metal Spinning](https://manufacturingprocesses.org/processes/forming/metal-spinning.md): Metal spinning presses a rotating metal disc against a mandrel with a roller or hand tool, forming it into an axially symmetric hollow shape.
- [Hydroforming](https://manufacturingprocesses.org/processes/forming/hydroforming.md): Hydroforming presses sheet or tube into a single-sided die with high-pressure fluid, producing smooth seamless shapes that would need several stampings.
- [Die Casting](https://manufacturingprocesses.org/processes/forming/die-casting.md): Die casting injects molten aluminum, zinc or magnesium into a hardened steel die under high pressure, producing thin-walled net-shape metal parts.
- [CNC Turning](https://manufacturingprocesses.org/processes/cutting/cnc-turning.md): CNC turning rotates the workpiece against a stationary tool, producing round parts with excellent concentricity and short cycle times.

## Related processes

- [Metal Stamping](https://manufacturingprocesses.org/processes/forming/metal-stamping.md): Metal stamping shears and forms sheet metal between matched dies in a press, producing flat and shallow-formed parts at very high rates.
- [Press Braking](https://manufacturingprocesses.org/processes/forming/press-braking.md): Press braking bends sheet metal by driving a punch into a V-die, forming one straight-line bend at a time.
- [Hydroforming](https://manufacturingprocesses.org/processes/forming/hydroforming.md): Hydroforming presses sheet or tube into a single-sided die with high-pressure fluid, producing smooth seamless shapes that would need several stampings.
- [Metal Spinning](https://manufacturingprocesses.org/processes/forming/metal-spinning.md): Metal spinning presses a rotating metal disc against a mandrel with a roller or hand tool, forming it into an axially symmetric hollow shape.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/forming/deep-drawing)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Die Casting"
category: "Forming"
subcategory: "Metal"
materials: ["Metal"]
tolerances: "±0.010 in (±0.25 mm) on the first inch, then ±0.002 in per additional inch (NADCA standard linear tolerance, aluminum); dimensions crossing the parting line need added allowance"
volumes: "10,000–1,000,000+ parts per die; below about 5,000/year the tooling rarely pays back"
lead_time: "Roughly 10–16 weeks for die build and sampling; 2–4 weeks per production release thereafter"
url: https://manufacturingprocesses.org/processes/forming/die-casting
---

# Die Casting

Die casting injects molten aluminum, zinc or magnesium into a hardened steel die under high pressure, producing thin-walled net-shape metal parts.

- **Category**: [Forming](https://manufacturingprocesses.org/processes/forming.md)
- **Family**: Metal
- **Materials**: Metal
- **Typical tolerances**: ±0.010 in (±0.25 mm) on the first inch, then ±0.002 in per additional inch (NADCA standard linear tolerance, aluminum); dimensions crossing the parting line need added allowance
- **Surface finish**: Roughly 32–63 µin Ra (0.8–1.6 µm) as cast from a new die, coarsening as the die heat-checks
- **Typical volumes**: 10,000–1,000,000+ parts per die; below about 5,000/year the tooling rarely pays back
- **Lead time**: Roughly 10–16 weeks for die build and sampling; 2–4 weeks per production release thereafter

## Overview

Die casting injects molten aluminum, zinc, or magnesium into a hardened steel die under high pressure and holds it there while the part solidifies, producing net-shape metal parts with thin walls, cast-in bosses and ribs, and an as-cast skin good enough to paint. It is the highest-rate metal casting process: cycle times run 15–60 seconds, and gates fill a cavity in 20–100 milliseconds.

Typical parts are housings, gearbox cases, brackets, connector shells, and hardware from a few grams up to about 20 lb (9 kg). Aluminum A380 dominates by volume; zinc alloys (Zamak 3 and 5) allow thinner walls, finer detail, and far longer die life; magnesium AZ91D buys weight savings at higher metal cost. Standard linear tolerance is ±0.010 in (±0.25 mm) on the first inch and ±0.002 in per inch after that.

The constraint is tooling. A production die is a hardened H13 assembly that takes months to build, so die casting only makes economic sense above roughly 5,000–10,000 parts per year.

## How it works

Die casting splits into two machine architectures, chosen by the melting point of the alloy.

**Hot chamber** machines keep the injection cylinder submerged in the melt. They run zinc, magnesium, and lead alloys, whose melting points (Zamak 3 melts near 730°F / 390°C) are low enough not to attack the steel plunger. Shot-to-shot cycles are fast because there is no ladling step.

**Cold chamber** machines keep the furnace separate and ladle a measured shot into an unheated sleeve each cycle. Aluminum, poured at roughly 1,150–1,250°F (620–675°C), would dissolve a submerged hot-chamber cylinder, so it is always cast cold chamber. See [metal melting points](/charts/metal-melting-points) for the alloy temperatures that drive this split.

The cycle itself:

1. **Die prep.** The two die halves close under clamp force — machines are rated roughly 100–4,000 tons. The cavity has already been sprayed with a water-based die lubricant and blown dry; die surface temperature is held around 350–600°F (175–315°C) for aluminum.
2. **Shot.** The plunger drives metal through the runner and gate at roughly 1,500–15,000 psi (10–100 MPa). Gate velocity is high — commonly 100–200 ft/s (30–60 m/s) — which is what lets a 0.050 in wall fill before the metal freezes.
3. **Intensification.** Once the cavity is full, pressure is boosted and held while the casting solidifies, squeezing shrinkage porosity smaller.
4. **Solidification and open.** Dwell is a few seconds for thin zinc parts and 15–20 seconds for thick aluminum sections. The die opens and ejector pins push the casting out.
5. **Trim.** The shot — casting plus runner, biscuit, and overflows — goes into a trim die that shears off everything but the part. Runner metal is remelted.
6. **Secondary.** Machining of critical bores, deburring, vibratory finishing, resin impregnation for pressure-tight parts, and coating.

Air trapped during the high-velocity fill is the process's defining defect. Overflows and vents at the parting line give it somewhere to go; vacuum die casting evacuates the cavity before the shot when porosity must be low enough to heat treat or weld.

## Design guidelines

### Keep walls thin and uniform

Aluminum die castings run a nominal wall of 0.080–0.120 in (2–3 mm), with 0.050 in (1.2 mm) practical on small parts. Zinc goes thinner: 0.040–0.060 in (1–1.5 mm) nominal, 0.030 in (0.75 mm) achievable. Thick is worse than thin here — heavy sections solidify last, so they hold the shrinkage porosity. Core out any section more than about 1.5× the nominal wall.

### Draft every surface

Aluminum needs roughly 1° on outside walls and 2° on cored internal surfaces; zinc, which grips the core less aggressively, gets by with about half that. Deep cores need more, not less. Draft is not optional — without it the casting shrinks onto the die steel and galls on ejection.

### Radius the corners

Minimum fillet radius is about 0.060 in (1.5 mm), and a fillet equal to the adjacent wall thickness is better. Sharp internal corners are stress risers in the casting and heat-check initiation sites in the die, which makes them a tooling cost problem as much as a part problem.

### Ribs instead of thick walls

Ribs should be about 0.6× the nominal wall thickness so they do not sink, with generous root fillets and 1–2° of draft per side. Use several shallow ribs rather than a few tall ones.

### Cored holes

Minimum cored diameter is around 0.10 in (2.5 mm) in aluminum and 0.060 in (1.5 mm) in zinc, with core depth kept under about 4× diameter. Core pins are cantilevers loaded by a fluid moving at highway speed — long thin ones bend, then break. Small or deep holes are cheaper drilled after casting than cored.

### Plan for the parting line and ejectors

Flash at the parting line is normal and gets trimmed; dimensions that cross the parting line carry a larger tolerance than dimensions contained within one die half. Ejector pin marks land within about ±0.005 in of the surface, so keep them off cosmetic faces.

### Leave machining stock, but not too much

Allow 0.010–0.030 in (0.25–0.75 mm) on machined surfaces. The chill skin is the densest metal in the casting; cutting deeply through it exposes subsurface porosity, which is the usual cause of leaks in machined sealing faces.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Nominal wall, aluminum | 0.080–0.120 in (2–3 mm) | 0.050 in (1.2 mm) | Thinner sections freeze before the cavity fills |
| Nominal wall, zinc | 0.040–0.060 in (1–1.5 mm) | 0.030 in (0.75 mm) | Lower melting point tolerates thinner sections |
| Draft, outside wall | 1–2° aluminum | 0.5° zinc | Casting shrinks onto die steel and must release |
| Fillet radius | 1× adjacent wall | 0.060 in (1.5 mm) | Sharp corners crack castings and heat-check dies |
| Rib thickness | 0.6× nominal wall | 0.8× nominal wall | Thick ribs sink and trap porosity |
| Cored hole diameter | ≥ 0.10 in (2.5 mm) aluminum | 0.060 in (1.5 mm) zinc | Slender core pins deflect and break |
| Machining allowance | 0.015 in (0.4 mm) | 0.030 in (0.75 mm) | Cutting past the chill skin exposes porosity |

## Variants

- High Pressure Die Casting
- Low Pressure Die Casting

## Cost drivers

Cost splits into tooling, machine time, and metal. The die is the dominant fixed cost — a multi-cavity hardened steel die with slides costs substantially more than a plastic injection mold of similar size, because it runs hotter and dies of thermal fatigue rather than wear. Per-part cost is machine seconds plus alloy weight plus secondary operations, and the secondaries are routinely underestimated: trimming, machining, deburring, and impregnation frequently add more than the casting cycle itself.

Volume breakpoints: below roughly 5,000 parts per year the tooling rarely amortizes and investment casting, sand casting, or machining from billet wins. From 10,000 to 100,000, a single- or two-cavity die is normal. Above 250,000, multi-cavity dies and higher-tonnage machines cut per-part machine time proportionally.

1. **Cut the number of slides.** Every side-action mechanism adds tooling cost, cycle time, and a maintenance point. Reorient the part or move the feature onto the parting plane if you can.
2. **Design for as-cast surfaces.** Each machined feature adds a fixturing operation. Tolerance the casting to the ±0.010 in standard wherever function does not require better.
3. **Pick zinc when the part is small.** Zinc dies outlast aluminum dies by roughly an order of magnitude, cycle faster, and hold finer detail — often offsetting the higher metal cost.

## FAQ

### What tolerance can die casting hold?

The NADCA standard linear tolerance for aluminum is ±0.010 in (±0.25 mm) on the first inch, plus ±0.002 in for each additional inch. Precision tooling and tighter process control roughly halve that. Dimensions that span the parting line carry an extra allowance because they depend on how tightly the die halves close.

### What is the minimum wall thickness for die casting?

About 0.050 in (1.2 mm) for aluminum on small parts and 0.030 in (0.75 mm) for zinc. Nominal design walls are thicker — 0.080–0.120 in (2–3 mm) in aluminum — and uniformity matters more than absolute thinness, since heavy sections are where shrinkage porosity collects.

### What draft angle does a die casting need?

Roughly 1° on outside walls and 2° on cored internal surfaces for aluminum, and about half that for zinc. Deep cores need more. Zero-draft surfaces gall on ejection and shorten die life.

### Can die castings be welded or heat treated?

Conventional high-pressure die castings generally cannot, because entrapped gas expands at temperature and blisters the part. Vacuum and pore-free variants reduce entrapped gas enough to allow both, at a process premium. If welding or T6 heat treatment is required, say so before the die is designed.

### How many parts will a die casting die produce?

On the order of 100,000 shots in aluminum and over a million in zinc before major refurbishment. Aluminum's higher casting temperature drives thermal fatigue in the H13 die steel, which shows up as heat-check crazing transferred onto the part surface.

### Die casting or investment casting?

Die casting wins above roughly 10,000 parts a year on aluminum, zinc, and magnesium geometries with uniform thin walls. Investment casting wins for steel, stainless, and superalloys, for lower volumes, and for geometries with undercuts or zero draft that a steel die cannot release.

## Alternative processes

- [Investment Casting](https://manufacturingprocesses.org/processes/forming/investment-casting.md): Investment casting builds a ceramic shell around a wax pattern, melts the wax out and pours metal into the cavity, giving fine detail in almost any alloy.
- [Sand Casting](https://manufacturingprocesses.org/processes/forming/sand-casting.md): Sand casting pours molten metal into a cavity formed in bonded sand around a pattern, then breaks the mold away to release the part.
- [CNC Machining](https://manufacturingprocesses.org/processes/cutting/cnc-machining.md): CNC machining removes material from solid stock with a programmed cutting tool, producing accurate parts directly from a CAD model with no tooling.
- [Metal Extrusion](https://manufacturingprocesses.org/processes/forming/metal-extrusion.md): Metal extrusion pushes heated billet through a shaped die, producing continuous aluminum or copper profile of constant cross-section.

## Related processes

- [Investment Casting](https://manufacturingprocesses.org/processes/forming/investment-casting.md): Investment casting builds a ceramic shell around a wax pattern, melts the wax out and pours metal into the cavity, giving fine detail in almost any alloy.
- [Sand Casting](https://manufacturingprocesses.org/processes/forming/sand-casting.md): Sand casting pours molten metal into a cavity formed in bonded sand around a pattern, then breaks the mold away to release the part.
- [Metal Injection Molding](https://manufacturingprocesses.org/processes/forming/metal-injection-molding.md): Metal injection molding molds a powder-and-binder feedstock exactly like a plastic, then debinds and sinters it into a dense small metal part.
- [Heat Treatment](https://manufacturingprocesses.org/processes/forming/heat-treatment.md): Heat treatment cycles metal through controlled heating and cooling to change its hardness, strength, ductility or internal stress state.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/forming/die-casting)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Dip Molding"
category: "Forming"
subcategory: "Plastics and Rubber"
materials: ["Plastic"]
tolerances: "Inside diameter follows the former closely, about ±0.005 in (±0.13 mm); wall thickness is much looser at roughly ±0.010–0.030 in (±0.25–0.75 mm) and varies along the dip axis."
volumes: "100 to several million parts; economical at both ends because tooling is so inexpensive"
lead_time: "1–3 weeks for formers; hours per rack thereafter, with cycle time driven by fusion oven residence"
url: https://manufacturingprocesses.org/processes/forming/dip-molding
---

# Dip Molding

Dip molding withdraws a heated former from liquid plastisol or latex, leaving a coating that cures into a flexible open-ended part such as a grip or cap.

- **Category**: [Forming](https://manufacturingprocesses.org/processes/forming.md)
- **Family**: Plastics and Rubber
- **Materials**: Plastic
- **Typical tolerances**: Inside diameter follows the former closely, about ±0.005 in (±0.13 mm); wall thickness is much looser at roughly ±0.010–0.030 in (±0.25–0.75 mm) and varies along the dip axis.
- **Surface finish**: The inside surface reproduces the former's machined finish; the outside is a smooth, self-leveled gloss from drainage
- **Typical volumes**: 100 to several million parts; economical at both ends because tooling is so inexpensive
- **Lead time**: 1–3 weeks for formers; hours per rack thereafter, with cycle time driven by fusion oven residence

## Overview

Dip molding immerses a heated metal former in liquid PVC plastisol, latex, or polyurethane dispersion, holds it long enough for a layer to gel against the hot surface, then withdraws and fuses the coating in an oven. The cured part is stripped off the former, so every dip molded part is open at one end — grips, handle covers, end caps, plugs, protective boots, gloves, and dip-coated tool handles.

Tooling is the cheapest in plastics: a former is a machined or cast metal shape with no cavity, no parting line, and no clamp requirement. Add more formers to a rack and you have added cavitation. That combination makes dip molding economical from a few hundred parts, and it stays economical into the millions.

Wall thickness is set by former temperature and dwell time rather than by a mold, and typically lands between 0.020 and 0.125 in (0.5–3 mm).

## How it works

1. **Heat the former.** Formers are preheated in an oven, typically to 300–400 °F (150–205 °C) for PVC plastisol. Former temperature and its thermal mass are the primary controls on how much material gels, which is why formers are made from solid aluminum or steel rather than thin shells.

2. **Dip.** The hot former is lowered into a tank of plastisol — a suspension of PVC resin particles in liquid plasticizer — at a controlled rate. Dipping too fast entrains air; too slow leaves witness rings at each pause.

3. **Dwell.** Heat conducts from the former into the surrounding liquid and the PVC particles swell and gel into a soft solid layer. The layer builds quickly at first and then more slowly, because the gelled material insulates the former and the former is cooling. Dwell time is the main operator control on wall thickness.

4. **Withdraw and drain.** The former is lifted out and the excess drains back to the tank. Drainage during withdrawal is why dip molded parts are not perfectly uniform in wall thickness along the dip axis. Parts are often rotated or inverted during this phase to even out the distribution.

5. **Fuse.** The coated former passes through an oven at roughly 350–400 °F (175–205 °C) until the plastisol fully fuses, going from a soft gel to a tough, clear-to-opaque elastomer. Under-fusing leaves a weak, chalky part that tears easily; over-fusing discolors it.

6. **Cool and strip.** After a water quench or air cool, the part is peeled off the former, usually with compressed air introduced between part and former. Draft and the material's elongation are what make this possible.

### How does latex dipping differ?

Natural rubber latex is coagulant-dipped instead of heat-gelled: the former is first dipped in a coagulant salt solution, then in latex, where the coagulant destabilizes the emulsion at the surface. It runs near room temperature and is the route for gloves and thin-wall medical parts.

## Design guidelines

### Shape

The former has to come out, so the part must be open at one end and free of internal undercuts that would trap it. Draft of 1–2° per side makes stripping cleaner, though the elastomer's elongation lets you get away with less on soft compounds. Closed hollow shapes are impossible.

### Wall thickness

0.020–0.125 in (0.5–3 mm) is the practical range, with 0.040–0.080 in (1.0–2.0 mm) the most common. Thicker builds need longer dwell or repeated dips, which stretches the cycle. Expect real variation along the dip axis from drainage — design so a nominal-plus-or-minus wall is acceptable, and put any sealing or press-fit function on the inside diameter.

### Which dimension can you actually control?

The inside surface is formed against the machined former, so it reproduces it closely — this is where interference fits, retention beads, and location features belong. The outside surface is free-formed by drainage and surface tension, so it is the loose dimension. Getting this the right way round is the single most important decision in a dip molded part.

### Corners and edges

Radius everything. Sharp external corners on the former cause the gelled layer to thin as it drains and can leave a weak spot; sharp internal corners collect material and cure unevenly. A radius of at least one wall thickness is the floor.

### Features you can and cannot have

Retention beads, internal ribs, and grip textures on the inside surface all come free, because they are machined into the former. Text and logos can be engraved into the former and read through as raised detail. What you cannot have is a controlled outside profile, a closed end with a controlled internal void, or a sharp lip — the open end always finishes with a slightly rounded, drained edge unless it is trimmed.

### Former material

Aluminum heats and cools quickly and shortens the cycle; steel holds heat longer, which favors thicker builds and larger parts. Both machine easily and neither sees any pressure, which is why formers cost so little compared with any cavity tool — see [/charts/material-properties](/charts/material-properties) for the density and conductivity trade behind that choice.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Wall thickness | 0.040–0.080 in (1.0–2.0 mm) | 0.020–0.125 in (0.5–3 mm) | Thicker builds need repeated dips |
| Draft | 1–2° per side | About 0° on soft compounds | The former has to strip out |
| Controlled surface | Inside diameter | — | Formed directly against the machined former |
| Free surface | Outside profile | — | Set by drainage and surface tension |
| Radius | ≥ 1 × wall thickness | — | Sharp corners drain thin and cure unevenly |
| Geometry | Open at one end | No closed hollows | The former must be withdrawn |
| Hardness (PVC plastisol) | 40–90 Shore A | — | Set by plasticizer content in the compound |

## Cost drivers

Dip molding has the lowest tooling cost of any plastics process that produces a three-dimensional part. A former is a solid machined or cast shape with no cavity, no parting line, no ejection, and no pressure containment. Adding capacity means adding formers to the rack, so cavitation scales linearly and cheaply rather than in expensive multiples of a mold base.

The recurring costs are material, oven time, and labor at the strip station. Cycle time is dominated by heating the formers and fusing the coating, and it grows with wall thickness.

Volume breakpoints:

- Under 100 parts: a single machined former, hand-dipped. Viable for prototypes at very low tooling cost.
- 100–10,000: a rack of formers on a manual or semi-automatic line.
- Over 10,000: automated dip lines with multi-former racks, in-line fusion ovens, and automated stripping — the configuration behind commodity grips and caps.

Cost reduction:

1. **Thin the wall.** Material and dwell time both scale with it, and dwell time drives the cycle.
2. **Add formers, not shifts.** Formers are cheap; oven and tank capacity are the constraint, so fill the rack.
3. **Put functional features on the inside.** They are machined into the former once and reproduce free on every part, replacing secondary operations.
4. **Choose a color in the plastisol.** Compound is pigmented in the tank, so color is nearly free, whereas painting a flexible part is expensive and prone to cracking.
5. **Accept a drained open edge.** Specifying a trimmed, square lip adds a secondary operation to every part for what is often a cosmetic preference.

## FAQ

### What wall thickness can dip molding achieve?

Typically 0.020–0.125 in (0.5–3 mm), with 0.040–0.080 in (1.0–2.0 mm) most common. Thickness is controlled by former temperature and dwell time rather than by a mold cavity, and thicker builds need longer dwell or repeated dips, which lengthens the cycle.

### Which surface of a dip molded part is dimensionally accurate?

The inside. It forms directly against the machined former and holds roughly ±0.005 in (±0.13 mm), so interference fits, retention beads, and locating features belong there. The outside profile is free-formed by drainage and surface tension and is much looser — typically ±0.010–0.030 in (±0.25–0.75 mm) on wall.

### Why must dip molded parts be open at one end?

Because the former has to be withdrawn from the finished part. There is no split mold, so any geometry that would trap the former — a closed hollow, an internal undercut — is impossible. Draft of 1–2° makes stripping cleaner, though soft compounds tolerate very little.

### Why is dip molding tooling so cheap?

A former is a solid machined or cast shape with no cavity, no parting line, no ejection system, and no pressure to contain. Capacity scales by adding more formers to a rack rather than by building a multi-cavity mold, so both the entry cost and the cost of increasing output are far below any cavity-molding process.

### What is the difference between dip molding and dip coating?

Mechanically nothing — the same tank, heat, and dwell. The distinction is what comes off the former. In dip molding the coating is stripped and becomes the part; in dip coating the substrate stays inside and the coating is a permanent covering, as on a plier handle or a wire-formed rack.

### How is latex dipping different from plastisol dipping?

Latex is coagulant-dipped rather than heat-gelled. The former is dipped first in a coagulant salt solution and then in the latex, where the coagulant destabilizes the emulsion at the surface. It runs near room temperature instead of 300–400 °F and is the route used for gloves and thin-walled medical parts.

## Alternative processes

- [Injection Molding](https://manufacturingprocesses.org/processes/forming/injection-molding.md): Injection molding forces molten thermoplastic into a steel mold under high pressure, producing high-volume identical plastic parts.
- [Liquid Silicone Rubber (LSR) Molding](https://manufacturingprocesses.org/processes/forming/liquid-silicone-rubber-lsr-molding.md): LSR molding injects two-part liquid silicone into a heated mold where it cures, producing flexible parts with a very wide service temperature range.
- [Compression Molding](https://manufacturingprocesses.org/processes/forming/compression-molding.md): Compression molding presses a measured charge of rubber or thermoset into an open heated mold that closes to cure it into its final shape.
- [Rotation Molding](https://manufacturingprocesses.org/processes/forming/rotation-molding.md): Rotational molding tumbles powdered polymer inside a heated mold on two axes so it fuses into a seamless, stress-free hollow part with a uniform wall.

## Related processes

- [Rotation Molding](https://manufacturingprocesses.org/processes/forming/rotation-molding.md): Rotational molding tumbles powdered polymer inside a heated mold on two axes so it fuses into a seamless, stress-free hollow part with a uniform wall.
- [Liquid Silicone Rubber (LSR) Molding](https://manufacturingprocesses.org/processes/forming/liquid-silicone-rubber-lsr-molding.md): LSR molding injects two-part liquid silicone into a heated mold where it cures, producing flexible parts with a very wide service temperature range.
- [Overmolding](https://manufacturingprocesses.org/processes/forming/overmolding.md): Overmolding molds a second material, usually a soft elastomer, directly onto an already-molded rigid substrate so the two bond permanently.
- [Compression Molding](https://manufacturingprocesses.org/processes/forming/compression-molding.md): Compression molding presses a measured charge of rubber or thermoset into an open heated mold that closes to cure it into its final shape.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/forming/dip-molding)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Direct Metal Laser Sintering (DMLS)"
category: "Forming"
subcategory: "Additive Manufacturing"
materials: ["Metal"]
tolerances: "About ±0.004 in (±0.1 mm) on features up to 1 in (25 mm), then roughly ±0.2% of nominal. Suppliers commonly quote ±0.003 in (±0.076 mm) plus ±0.001 in/in for well-controlled geometry; machined features are far tighter."
volumes: "1–500 parts; economical wherever the geometry cannot be machined or cast"
lead_time: "5–15 business days. The print itself is typically 1–4 days, with stress relief, plate removal, support removal, and any machining or HIP adding most of the balance."
url: https://manufacturingprocesses.org/processes/forming/direct-metal-laser-sintering-dmls
---

# Direct Metal Laser Sintering (DMLS)

Direct metal laser sintering fuses metal powder layer by layer with a laser, building dense metal parts anchored to a supported build plate.

- **Category**: [Forming](https://manufacturingprocesses.org/processes/forming.md)
- **Family**: Additive Manufacturing
- **Materials**: Metal
- **Typical tolerances**: About ±0.004 in (±0.1 mm) on features up to 1 in (25 mm), then roughly ±0.2% of nominal. Suppliers commonly quote ±0.003 in (±0.076 mm) plus ±0.001 in/in for well-controlled geometry; machined features are far tighter.
- **Surface finish**: Ra 200–500 µin (5–13 µm) as built on vertical walls; down-facing supported surfaces reach Ra 800–1,200 µin (20–30 µm). Bead blasting, tumbling, machining, or polishing are used where finish matters, particularly for fatigue-critical surfaces.
- **Typical volumes**: 1–500 parts; economical wherever the geometry cannot be machined or cast
- **Lead time**: 5–15 business days. The print itself is typically 1–4 days, with stress relief, plate removal, support removal, and any machining or HIP adding most of the balance.

## Overview

Direct metal laser sintering (DMLS) is laser powder bed fusion of metal: a fiber laser of 200–1,000 W fully melts 0.0008–0.0024 in (20–60 µm) layers of 15–45 µm alloy powder under argon or nitrogen, building parts anchored to a solid substrate plate. Finished density is typically above 99.5%, with mechanical properties at or above cast values and, after hot isostatic pressing, approaching wrought.

The production alloy set is small but useful: AlSi10Mg, Ti-6Al-4V and Ti-6Al-4V ELI, 316L and 17-4 PH stainless, maraging steel, Inconel 625 and 718, and CoCrMo. Typical build envelopes are around 10 × 10 × 13 in (250 × 250 × 325 mm), with large-frame machines reaching 400–800 mm in each axis.

Accuracy runs roughly ±0.004 in (±0.1 mm) on the first inch. DMLS earns its cost where geometry cannot be machined or cast — conformal cooling channels, topology-optimized brackets, internal lattices, and consolidated assemblies — at volumes of 1 to a few hundred parts.

## How it works

1. **Prepare the build.** Parts are oriented, supported, and sliced. Supports here are structural and thermal: they anchor the part against the enormous residual stresses of solidification and conduct heat away from downward-facing surfaces.
2. **Inert the chamber.** The chamber is purged with argon (reactive alloys such as titanium and aluminum) or nitrogen (steels), typically to below 0.1% residual oxygen. Oxygen pickup embrittles titanium and causes spatter and porosity in any alloy.
3. **Recoat.** A blade or roller spreads a 0.0008–0.0024 in (20–60 µm) layer of powder, with 0.0012–0.0016 in (30–40 µm) the common production setting.
4. **Melt.** The laser scans the cross-section with a 0.003–0.004 in (70–100 µm) spot at scan speeds of roughly 0.5–3 m/s and hatch spacing near 0.003–0.006 in (0.08–0.14 mm). Scan strategies split the layer into stripes or islands and rotate the hatch direction each layer to break up residual stress. Typical build rate is 5–20 cm³/h.
5. **Repeat to full height.** Build time is set almost entirely by Z height and cross-sectional area, not by part count.
6. **Stress relieve on the plate.** The part is heat-treated *before* it leaves the substrate, or it will spring and distort the moment it is cut free. Typical cycles are around 800 °C (1,470 °F) for 2 hours in argon or vacuum for Ti-6Al-4V, roughly 300 °C (570 °F) for 2 hours for AlSi10Mg, and 550–650 °C (1,020–1,200 °F) for austenitic stainless.
7. **Separate and finish.** The part is cut from the plate by wire EDM or bandsaw, supports are removed by hand or by machining, and critical features are machined. Fatigue-critical aerospace and medical parts are hot isostatically pressed — for Ti-6Al-4V, commonly around 900 °C (1,650 °F) at 15 ksi (100 MPa) for 2 hours — to close internal porosity.

## Design guidelines

### Overhang angle
Surfaces below about 45° from vertical need support. Unsupported downskins do not merely sag — they sit on loose powder, which conducts heat an order of magnitude worse than solid metal, so the melt pool runs hot and the surface comes out rough and dross-covered. Design self-supporting angles wherever you can; every support you delete is money saved twice, at build time and at removal.

### Holes and internal channels
Round horizontal holes above about 0.3 in (8 mm) diameter sag at the crown. Redraw them as teardrops or diamonds and they self-support to much larger sizes. Vertical holes are unrestricted. Every internal channel needs at least one powder evacuation port of 0.08–0.12 in (2–3 mm) or larger; trapped powder in a closed cavity is permanent.

### Wall thickness
0.016 in (0.4 mm) is the process minimum; use 0.040 in (1 mm) for structure. Keep sections as uniform as practical — abrupt changes in cross-section concentrate residual stress and are the usual cause of a build peeling off its supports mid-print.

### Machining allowance
Add 0.020–0.040 in (0.5–1.0 mm) of stock to any surface that must be flat, round, or held to a fit. Bearing bores, sealing faces, and datum surfaces should always be machined after stress relief — see the [ISO 286 fits and tolerances chart](/charts/iso-286-fits-tolerances) for what those fits require.

### Threads
Do not build threads. Print a pilot hole with machining stock and tap it; the [tap drill chart](/charts/tap-drill-chart) gives pilot diameters. As-built threads are dimensionally poor and their roots concentrate the stress that additive metal is least good at resisting.

### Orientation for properties
Fatigue and tensile properties are direction-dependent: Z-direction ductility is usually the lowest, and as-built surface roughness is itself the dominant fatigue initiator. Orient critical loads in the build plane and machine or polish highly stressed surfaces.

### Fillets and stress
Fillet every internal corner at 0.04 in (1 mm) or more. Sharp re-entrant corners in a process that solidifies under steep thermal gradients are crack initiators.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Overhang from vertical | 40° | 45° | Downskin sits on powder, not metal |
| Wall thickness | 0.040 in (1.0 mm) | 0.016 in (0.4 mm) | Thin walls distort and warp |
| Horizontal hole | Teardrop profile | 0.3 in (8 mm) round | Crown of a round hole sags |
| Powder escape port | 0.16 in (4 mm) | 0.08 in (2 mm) | Trapped powder cannot be removed |
| Machining stock | 0.040 in (1.0 mm) | 0.020 in (0.5 mm) | Distortion after cut-off must be cleaned up |
| Internal fillet | 0.08 in (2 mm) | 0.04 in (1 mm) | Sharp corners crack under residual stress |
| Layer thickness | 0.0012 in (30 µm) | 0.0008 in (20 µm) | Halving the layer doubles build time |

## Cost drivers

Build height is the first cost driver: machine time scales with layer count, so a part standing 8 in (200 mm) tall costs far more than the same volume lying flat, even though it uses identical powder. Cross-sectional area matters second, since it sets scan time per layer.

Powder is expensive and alloy-dependent — titanium and nickel superalloy powders cost several times what stainless does per kilogram — and only powder that is sieved and requalified can be reused, so contamination or a cross-alloy mistake writes off a full charge.

Post-processing is the cost most people underestimate. Stress relief, plate cut-off by wire EDM, manual support removal, HIP where required, and CNC finishing of critical features routinely total more than the print itself.

Volume breakpoints: DMLS is competitive from 1 part upward for geometry that cannot be made otherwise. Against machining, it wins below roughly 50–100 parts for complex shapes and loses immediately on simple prismatic ones. Against [investment casting](/processes/forming/investment-casting), the crossover is typically a few hundred parts, where pattern tooling amortizes.

1. Reduce build height before anything else — reorient rather than redesign if you can.
2. Design for self-supporting angles above 45° and teardrop holes; support removal is manual labor.
3. Hollow and lattice bulk sections; you pay for every cubic centimeter melted.
4. Nest several parts on one plate — the stress relief cycle and plate cut-off are shared.
5. Machine only what must be machined, and say so explicitly on the drawing rather than applying a blanket tolerance.

## FAQ

### What tolerance can DMLS hold?

Roughly ±0.004 in (±0.1 mm) up to 1 in (25 mm), then about ±0.2% of the dimension. Some suppliers quote ±0.003 in plus ±0.001 in/in on well-supported geometry. Any feature that needs a fit, a seal, or a datum should be printed with 0.020–0.040 in (0.5–1.0 mm) of stock and machined.

### What is the minimum overhang angle for DMLS?

About 45° from vertical. Below that, the melt pool sits on loose powder rather than solid metal, which conducts heat far more poorly, so the surface comes out rough and dross-covered and the part risks curling off its supports. Teardrop and diamond hole profiles avoid supports entirely.

### Why do DMLS parts need stress relief before removal from the build plate?

Each melt track solidifies and contracts against already-solid material below it, locking in large residual stresses. Cutting the part free before heat treatment releases those stresses all at once and the part springs out of tolerance. Stress relief is performed with the part still bolted to the plate.

### How dense are DMLS parts?

Above 99.5% of theoretical density with correctly developed parameters, and mechanical properties at or above cast values. Hot isostatic pressing — around 900 °C (1,650 °F) at 15 ksi (100 MPa) for Ti-6Al-4V — closes residual porosity and is standard practice for fatigue-critical aerospace and medical parts.

### Is DMLS the same as SLM?

In practice, yes. Both are laser powder bed fusion, both fully melt the powder, and both build under inert gas. DMLS began as an EOS trade name and SLM as a competing one; the ISO/ASTM term for the family is laser powder bed fusion (PBF-LB/M).

### When does DMLS beat machining or casting?

When the geometry cannot be made another way — conformal cooling channels, internal lattices, topology-optimized load paths, or an assembly consolidated into one piece — and at quantities below roughly 50–100 for complex parts. For simple prismatic geometry, machining is cheaper at any quantity.

## Alternative processes

- [Investment Casting](https://manufacturingprocesses.org/processes/forming/investment-casting.md): Investment casting builds a ceramic shell around a wax pattern, melts the wax out and pours metal into the cavity, giving fine detail in almost any alloy.
- [CNC Machining](https://manufacturingprocesses.org/processes/cutting/cnc-machining.md): CNC machining removes material from solid stock with a programmed cutting tool, producing accurate parts directly from a CAD model with no tooling.
- [Metal Injection Molding](https://manufacturingprocesses.org/processes/forming/metal-injection-molding.md): Metal injection molding molds a powder-and-binder feedstock exactly like a plastic, then debinds and sinters it into a dense small metal part.
- [Electron Beam Melting (EBM)](https://manufacturingprocesses.org/processes/forming/electron-beam-melting-ebm.md): Electron beam melting fuses metal powder with an electron beam in vacuum at high preheat temperature, which cuts residual stress in titanium parts.
- [Powder Metallurgy (Press and Sinter)](https://manufacturingprocesses.org/processes/forming/powder-metallurgy-press-and-sinter.md): Press-and-sinter powder metallurgy compacts metal powder in a rigid die and sinters it below its melting point into a finished net-shape part.

## Related processes

- [Selective Laser Melting (SLM)](https://manufacturingprocesses.org/processes/forming/selective-laser-melting-slm.md): Selective laser melting fully melts metal powder with a laser under inert gas, producing parts at close to wrought density.
- [Electron Beam Melting (EBM)](https://manufacturingprocesses.org/processes/forming/electron-beam-melting-ebm.md): Electron beam melting fuses metal powder with an electron beam in vacuum at high preheat temperature, which cuts residual stress in titanium parts.
- [Binder Jetting](https://manufacturingprocesses.org/processes/forming/binder-jetting.md): Binder jetting prints liquid binder into a powder bed to hold each layer, producing green parts that are later sintered or infiltrated.
- [Directed Energy Deposition (DED)](https://manufacturingprocesses.org/processes/forming/directed-energy-deposition-ded.md): Directed energy deposition feeds metal wire or powder into a moving melt pool, building up large features or repairing existing parts.
- [Metal Injection Molding](https://manufacturingprocesses.org/processes/forming/metal-injection-molding.md): Metal injection molding molds a powder-and-binder feedstock exactly like a plastic, then debinds and sinters it into a dense small metal part.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/forming/direct-metal-laser-sintering-dmls)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Directed Energy Deposition (DED)"
category: "Forming"
subcategory: "Additive Manufacturing"
materials: ["Metal"]
tolerances: "As-deposited surfaces are held to roughly ±0.02–0.04 in (±0.5–1.0 mm) at best, and large WAAM structures to considerably less. Final tolerances come from the machining operation that follows, not from the deposition."
volumes: "1–100 parts; repairs and one-off large structures are the core of the market"
lead_time: "1–4 weeks depending on size, heat treatment, and the machining that follows. Repairs of existing components are often turned around in days."
url: https://manufacturingprocesses.org/processes/forming/directed-energy-deposition-ded
---

# Directed Energy Deposition (DED)

Directed energy deposition feeds metal wire or powder into a moving melt pool, building up large features or repairing existing parts.

- **Category**: [Forming](https://manufacturingprocesses.org/processes/forming.md)
- **Family**: Additive Manufacturing
- **Materials**: Metal
- **Typical tolerances**: As-deposited surfaces are held to roughly ±0.02–0.04 in (±0.5–1.0 mm) at best, and large WAAM structures to considerably less. Final tolerances come from the machining operation that follows, not from the deposition.
- **Surface finish**: Ra 500–1,600 µin (12–40 µm) as deposited, with visible bead waviness on top of that. Every functional surface is machined, after which normal machining finishes apply.
- **Typical volumes**: 1–100 parts; repairs and one-off large structures are the core of the market
- **Lead time**: 1–4 weeks depending on size, heat treatment, and the machining that follows. Repairs of existing components are often turned around in days.

## Overview

Directed energy deposition (DED) feeds metal powder or wire into a melt pool created by a laser, electron beam, or electric arc, and moves that melt pool along a path with a robot, gantry, or CNC machine. Unlike powder bed processes, nothing constrains the build envelope but the motion system, so DED is used on parts measured in meters — and it is the only additive family that routinely adds material to an *existing* component.

Deposition rates are one to two orders of magnitude above powder bed fusion: roughly 0.5–2 kg/h for blown-powder laser systems and 1–10 kg/h for wire arc additive manufacturing (WAAM). The price of that rate is resolution. Beads are 0.04–0.16 in (1–4 mm) wide, layers are 0.010–0.040 in (0.25–1.0 mm) thick or more, and as-deposited surfaces need 0.04–0.12 in (1–3 mm) of machining stock on every functional face.

The three dominant uses are near-net-shape preforms for large aerospace structures, repair of high-value components such as turbine blades and forming dies, and hardfacing or corrosion cladding of steel substrates.

## How it works

1. **Fixture the substrate.** DED always builds onto something — a plate, a forging, or the worn component being repaired. The substrate is fixtured and, for repairs, machined back to sound material first.
2. **Establish the melt pool.** A laser (typically 1–10 kW), an electron beam under vacuum, or a welding arc creates a localized melt pool on the substrate.
3. **Feed material into it.** Blown-powder heads deliver powder coaxially through nozzles with a shielding gas shroud, capturing perhaps 40–90% of what is blown depending on geometry. Wire feed captures essentially all of it, which is why WAAM is cheaper per kilogram but coarser.
4. **Move.** The head or the part traverses along a planned path, laying down a bead 0.04–0.16 in (1–4 mm) wide. Adjacent beads overlap by roughly 30–50% to give a continuous layer.
5. **Manage heat.** Heat input is enormous compared with powder bed fusion, so interpass temperature control, dwell times, and sometimes active cooling are used to control distortion and microstructure. Large WAAM structures may be built on a heated or actively cooled table.
6. **Post-process.** Stress relief or full heat treatment follows, then machining of every functional surface. Some systems are hybrid — DED head and milling spindle in the same enclosure — so deposition and machining alternate and internal features can be cut while still accessible.

## Design guidelines

### Design a preform, not a finished part
Treat DED output the way you would treat a forging or a casting: it is a near-net shape carrying stock for machining. Add 0.04–0.12 in (1–3 mm) per surface, more on large or thermally distorted structures. Nothing usable comes off a DED machine as-built.

### Wall thickness and features
Minimum wall is roughly one bead width — 0.08–0.16 in (2–4 mm) for most systems. Fine detail, small holes, and thin ribs are outside the process entirely; those get machined in.

### Overhangs
Most DED systems are 3-axis or 5-axis but deposit without a support powder bed, so overhangs beyond about 45° require either a support structure that must later be machined off, or a 5-axis motion plan that keeps the deposition head normal to the growing surface. The second option is the reason multi-axis DED exists.

### Distortion and build sequence
Heat input is the design constraint. Long straight walls bow, and asymmetric features pull the substrate. Plan symmetric build sequences, alternate deposition direction, and design in enough stock that predicted distortion can be machined away rather than avoided.

### Material transitions
DED can change alloy mid-build, since composition is set by what is fed into the melt pool. That enables functionally graded parts and repairs where a wear-resistant alloy is deposited onto a tough substrate — but the transition must be metallurgically compatible. Check dilution and the risk of brittle intermetallics before specifying a bimetallic joint.

### Repair geometry
For repair work, machine the damaged region to a smooth open groove with no sharp corners or blind pockets before depositing. The melt pool cannot reach into a re-entrant feature, and lack-of-fusion defects at the substrate interface are the usual failure mode.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Machining stock | 0.08 in (2 mm) | 0.04 in (1 mm) | As-deposited surfaces are rough and wavy |
| Wall thickness | 0.16 in (4 mm) | 0.08 in (2 mm) | One to two bead widths |
| Layer thickness | 0.020 in (0.5 mm) | 0.010 in (0.25 mm) | Set by bead geometry and feed rate |
| Overhang | 45° or 5-axis path | Support required | No powder bed to hold the bead |
| Repair groove | Open, filleted | No blind pockets | Melt pool needs line of sight |

## Cost drivers

DED is priced on deposition time and feedstock. Deposition rate is the whole argument for the process — 0.5–2 kg/h for blown powder and 1–10 kg/h for wire arc, against a small fraction of that for powder bed fusion — so large parts that would take weeks on a laser powder bed machine become viable.

Feedstock form drives cost sharply. Welding wire is far cheaper per kilogram than atomized powder and is deposited with essentially complete capture efficiency, while blown powder loses 10–60% to overspray. That is why WAAM dominates large structural work and blown powder dominates cladding and precision repair, where the finer bead matters.

Machining is the other half of the bill. Every functional surface is cut after deposition, and on a large preform that can exceed the deposition cost. Hybrid machines reduce fixturing and handling but not the metal removal itself.

The strongest economic case is repair. Rebuilding a turbine blade tip or a worn forming die costs a fraction of a new component, and DED is often the only way to add material to a finished part at all.

1. Deposit only where material is needed — use a substrate, forging, or plate for the bulk.
2. Choose wire over powder wherever the bead resolution allows it.
3. Design symmetric build sequences to cut distortion, and therefore machining stock.
4. Consider hybrid deposition-plus-milling when internal features become inaccessible later.
5. Evaluate repair before replacement on any high-value component.

## FAQ

### How fast is DED compared with powder bed fusion?

Roughly 0.5–2 kg/h for blown-powder laser DED and 1–10 kg/h for wire arc additive manufacturing, which is one to two orders of magnitude above laser powder bed fusion. The trade is resolution: beads are 0.04–0.16 in (1–4 mm) wide and every functional surface must be machined.

### How much machining stock should I leave on a DED part?

0.04–0.12 in (1–3 mm) per surface for most work, and more on large structures where thermal distortion is significant. Treat a DED part like a forging or casting — it is a near-net preform, not a finished component.

### Can DED repair existing parts?

Yes, and this is its most common industrial use. Worn turbine blade tips, forming dies, shafts, and seal surfaces are machined back to sound metal and rebuilt by deposition, then heat treated and re-machined. It is usually a fraction of the cost of replacement.

### What is the difference between WAAM and laser DED?

WAAM uses a welding arc and wire feedstock, giving very high deposition rates at low feedstock cost but coarse resolution and high heat input. Laser DED with blown powder deposits more slowly with a finer bead and better control, which suits cladding and precision repair.

### Does DED have a build volume limit?

Only the motion system's reach. Robot-mounted and gantry systems build structures measured in meters, which is why DED is used for large aerospace preforms and pressure vessel components that no powder bed machine could accommodate.

### Can DED deposit one alloy onto another?

Yes. Composition is set by what is fed into the melt pool, so functionally graded structures and dissimilar-metal cladding — hardfacing or corrosion-resistant layers on a steel substrate — are routine. Check dilution at the interface and the potential for brittle intermetallics before specifying the pairing.

## Alternative processes

- [Forging](https://manufacturingprocesses.org/processes/forming/forging.md): Forging deforms solid metal under compressive load in a hammer, press or dies, so the grain flow follows the part contour and fatigue strength rises.
- [CNC Machining](https://manufacturingprocesses.org/processes/cutting/cnc-machining.md): CNC machining removes material from solid stock with a programmed cutting tool, producing accurate parts directly from a CAD model with no tooling.
- [Arc Welding](https://manufacturingprocesses.org/processes/joining/arc-welding.md): Arc welding melts and fuses metal with an electric arc between an electrode and the work, shielded from the atmosphere by gas or flux.
- [Selective Laser Melting (SLM)](https://manufacturingprocesses.org/processes/forming/selective-laser-melting-slm.md): Selective laser melting fully melts metal powder with a laser under inert gas, producing parts at close to wrought density.
- [Sand Casting](https://manufacturingprocesses.org/processes/forming/sand-casting.md): Sand casting pours molten metal into a cavity formed in bonded sand around a pattern, then breaks the mold away to release the part.

## Related processes

- [Direct Metal Laser Sintering (DMLS)](https://manufacturingprocesses.org/processes/forming/direct-metal-laser-sintering-dmls.md): Direct metal laser sintering fuses metal powder layer by layer with a laser, building dense metal parts anchored to a supported build plate.
- [Selective Laser Melting (SLM)](https://manufacturingprocesses.org/processes/forming/selective-laser-melting-slm.md): Selective laser melting fully melts metal powder with a laser under inert gas, producing parts at close to wrought density.
- [Arc Welding](https://manufacturingprocesses.org/processes/joining/arc-welding.md): Arc welding melts and fuses metal with an electric arc between an electrode and the work, shielded from the atmosphere by gas or flux.
- [Power Beam Welding](https://manufacturingprocesses.org/processes/joining/power-beam-welding.md): Power beam welding fuses metal with a laser or electron beam focused to a very small spot, producing deep, narrow welds with little distortion.
- [CNC Machining](https://manufacturingprocesses.org/processes/cutting/cnc-machining.md): CNC machining removes material from solid stock with a programmed cutting tool, producing accurate parts directly from a CAD model with no tooling.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/forming/directed-energy-deposition-ded)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Electroforming"
category: "Forming"
subcategory: "Metal"
materials: ["Metal"]
tolerances: "The mandrel-contacting surface replicates the mandrel closely, so its accuracy is the mandrel's accuracy; wall thickness typically varies ±10–25% across a part unless shields and current thieves are used"
volumes: "1–10,000 parts; single pieces are entirely practical, and permanent mandrels amortize over hundreds"
lead_time: "4–10 weeks including mandrel manufacture and process development; plating alone runs days per part for thick walls"
url: https://manufacturingprocesses.org/processes/forming/electroforming
---

# Electroforming

Electroforming grows a metal shell by electrodeposition onto a mandrel that is afterwards removed, producing thin parts with sub-micron detail.

- **Category**: [Forming](https://manufacturingprocesses.org/processes/forming.md)
- **Family**: Metal
- **Materials**: Metal
- **Typical tolerances**: The mandrel-contacting surface replicates the mandrel closely, so its accuracy is the mandrel's accuracy; wall thickness typically varies ±10–25% across a part unless shields and current thieves are used
- **Surface finish**: The mandrel side reproduces the mandrel finish, including mirror and sub-micron textures; the solution side is rough and is machined where it matters
- **Typical volumes**: 1–10,000 parts; single pieces are entirely practical, and permanent mandrels amortize over hundreds
- **Lead time**: 4–10 weeks including mandrel manufacture and process development; plating alone runs days per part for thick walls

## Overview

Electroforming grows a metal part atom by atom, electrodepositing it onto a conductive mandrel until the wall reaches the required thickness, then separating the two. It is electroplating taken to the point where the deposit becomes the part rather than a coating.

Because the part's inner surface is a negative of the mandrel, replication fidelity is limited only by how well the mandrel was made — sub-micron features and mirror finishes transfer faithfully, which is why electroforming produces optical disc stampers, textured mold inserts, precision meshes and screens, waveguides, bellows, foil, and nozzles.

Nickel and copper are the workhorse metals, deposited from sulfamate or acid baths at roughly 0.001–0.002 in (25–50 µm) per hour. That rate is the process's defining constraint: a 0.020 in wall is a day or more of continuous plating. Typical wall thicknesses run 0.001–0.030 in (25 µm–0.75 mm), and volumes from one part to a few thousand.

## How it works

1. **Mandrel.** The mandrel carries the geometry, so it is made to the tolerance and finish the part requires — diamond turned, polished, photolithographically patterned, or machined. It must be electrically conductive; non-conductive mandrels (wax, plastic, glass) are metallized first with a sputtered or electroless conductive layer.
2. **Mandrel type.** A **permanent** mandrel is drafted, passivated stainless steel or chrome-plated steel, designed so the electroform can be pulled off and the mandrel reused hundreds of times. An **expendable** mandrel — aluminum dissolved in caustic, low-melting alloy, or wax — is destroyed to release the part and is what makes undercut and fully enclosed geometry possible.
3. **Passivation.** For permanent mandrels, a controlled thin passive film is deliberately created so the deposit adheres well enough to grow but releases cleanly at the end. Getting this wrong is the classic electroforming failure: the shell either falls off mid-build or refuses to come off at all.
4. **Deposition.** The mandrel is made the cathode in an electrolyte — nickel sulfamate is the standard for structural electroforms, typically operated around 120–140°F (50–60°C) in the pH 3.5–4.5 region — and current is applied. Metal deposits at a rate proportional to current density.
5. **Managing thickness distribution.** Current density is not uniform: it concentrates at edges, points, and protruding features, and starves in recesses. Left alone, a deposit is thick on the corners and thin in the grooves. Conforming anodes, current thieves, and non-conductive shields are used to even it out, and thickness variation of ±10–25% is normal without careful control.
6. **Separation.** The electroform is pulled from a permanent mandrel or the expendable mandrel is dissolved or melted away.
7. **Finishing.** The outer (solution-side) surface is rough and uncontrolled and is machined or ground where it matters; the mandrel side is finished already.

Internal stress in the deposit is a real design variable. Sulfamate nickel is chosen over other nickel baths largely because it deposits at low internal stress, and stress additives are used to tune it — a highly stressed thick electroform will distort or crack when it comes off the mandrel.

## Design guidelines

### The mandrel side is the good side

Every dimension, finish, and detail you care about must be on the mandrel-contacting surface. The solution side grows freely, follows the current distribution, and ends up rough and roughly parallel at best. Design the part so its functional surface is the one against the mandrel.

### Avoid deep, narrow recesses

Throwing power is poor. A deep narrow groove in the mandrel receives much less current density at its bottom than at its mouth, so the deposit there is thin — sometimes dramatically so. Keep aspect ratios of recessed features low, or accept a wall thickness that varies with depth. See the [surface finish chart](/charts/surface-finish-chart) for context on the finishes the mandrel needs to carry.

### Break every sharp external edge

A sharp external corner on the part corresponds to a sharp internal corner on the mandrel, where current crowds, deposits build fast, and nodules or treeing form. Radius the mandrel geometry wherever possible.

### Release geometry decides the mandrel type

If the part can be pulled straight off, use a permanent mandrel with a degree or two of draft and get hundreds of parts from one tool. If it has undercuts, re-entrant features, or is fully enclosed, an expendable mandrel is required — and it has to be remade for every single part, which changes the economics completely.

### Wall thickness and time

Design for the thinnest wall that carries the load. At roughly 0.001–0.002 in per hour, wall thickness translates directly into machine hours: 0.005 in is a shift, 0.030 in is most of a week. Typical production electroforms fall in the 0.001–0.030 in (25 µm–0.75 mm) band.

### Expect thickness variation

Without extensive shielding and thieving, plan on ±10–25% variation in wall thickness across a part, biased thick at edges and thin in recesses. If a uniform wall is critical, say so — it is achievable, but it costs anode tooling and development time.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Functional surface | On the mandrel side | Solution side is uncontrolled | The deposit replicates the mandrel exactly |
| Wall thickness | 0.002–0.010 in (50–250 µm) | 0.030 in (0.75 mm) practical | Deposition runs 0.001–0.002 in per hour |
| Recess aspect ratio | Shallow and open | Deep grooves plate thin at the bottom | Current density falls inside recesses |
| External corners | Radiused | Sharp corners grow nodules | Current crowds at internal mandrel corners |
| Draft, permanent mandrel | 1–2° | 0° needs an expendable mandrel | The shell must pull off cleanly |
| Undercuts | Expendable mandrel only | New mandrel per part | Nothing can be pulled over an undercut |
| Wall uniformity | Accept ±10–25% | Tighter needs shields and thieves | Current distribution is inherently uneven |

## Cost drivers

Two things drive cost: mandrel work and plating hours. The mandrel is a precision part in its own right, made to the finish and tolerance the final component requires, and on a permanent-mandrel job it is the dominant fixed cost — amortized over hundreds of parts, it becomes small. On an expendable-mandrel job it is a per-part cost, and the process becomes expensive fast. Plating time is directly proportional to wall thickness at roughly 0.001–0.002 in per hour, and tank time is not compressible: doubling the current to speed it up degrades the deposit.

Volume breakpoints: single parts are entirely practical, which is unusual for a process capable of this precision. Permanent-mandrel work amortizes across hundreds to a few thousand parts. There is no high-volume regime — electroforming is never the cheap answer at scale, it is the only answer for certain geometries and surface requirements.

1. **Use a permanent mandrel wherever the geometry allows.** Adding a degree of draft to make a mandrel reusable is the single biggest cost lever in the process.
2. **Thin the wall.** Wall thickness is tank hours, one for one.
3. **Nest multiple parts per mandrel.** Plating several cavities on one mandrel in one tank cycle spreads the fixed time.

## FAQ

### How thick can an electroformed part be?

Typical production walls run 0.001–0.030 in (25 µm–0.75 mm). Thicker is possible but rarely economic, because deposition proceeds at only about 0.001–0.002 in (25–50 µm) per hour — a 0.030 in wall represents most of a week of continuous plating, and increasing the current to speed it up degrades the deposit.

### Which surface of an electroform is accurate?

The one that was against the mandrel. It is a faithful negative of the mandrel surface, reproducing sub-micron detail and mirror finishes. The solution side grows freely according to the current distribution and comes out rough and only roughly parallel, so it is machined afterward if it matters.

### What is the difference between a permanent and an expendable mandrel?

A permanent mandrel is passivated stainless or chrome-plated steel with a degree or two of draft, so the electroform pulls off and the mandrel is reused hundreds of times. An expendable mandrel — aluminum dissolved in caustic, low-melting alloy, or wax — is destroyed to release the part, which is the only way to make undercut or fully enclosed geometry.

### Why is electroformed wall thickness uneven?

Because current density is uneven. It concentrates at edges, points, and protrusions and starves in recesses, so the deposit builds fast on corners and slowly at the bottom of grooves. Expect ±10–25% variation without control measures; conforming anodes, current thieves, and non-conductive shields even it out at additional development cost.

### What metals can be electroformed?

Nickel and copper are the workhorses, along with nickel-cobalt for higher hardness and gold and silver for specialty work. Nickel sulfamate is the standard structural bath because it deposits at low internal stress, which matters because a highly stressed thick deposit distorts or cracks when it is separated from the mandrel.

### Electroforming or CNC machining?

Machining wins for anything thick, structural, or geometrically prismatic. Electroforming wins where the requirement is a thin shell with a precisely replicated surface — mold texture inserts, meshes, waveguide interiors, and optical stampers — because it copies a mandrel exactly rather than cutting a surface, and it can make walls far thinner than a cutter can reach.

## Alternative processes

- [Photochemical Machining](https://manufacturingprocesses.org/processes/cutting/photochemical-machining.md): Photochemical machining masks sheet metal photographically and etches away the unmasked areas, cutting burr-free flat parts with no hard tooling.
- [Metal Injection Molding](https://manufacturingprocesses.org/processes/forming/metal-injection-molding.md): Metal injection molding molds a powder-and-binder feedstock exactly like a plastic, then debinds and sinters it into a dense small metal part.
- [CNC Machining](https://manufacturingprocesses.org/processes/cutting/cnc-machining.md): CNC machining removes material from solid stock with a programmed cutting tool, producing accurate parts directly from a CAD model with no tooling.
- [Direct Metal Laser Sintering (DMLS)](https://manufacturingprocesses.org/processes/forming/direct-metal-laser-sintering-dmls.md): Direct metal laser sintering fuses metal powder layer by layer with a laser, building dense metal parts anchored to a supported build plate.

## Related processes

- [Electroplating](https://manufacturingprocesses.org/processes/finishing/electroplating.md): Electroplating deposits a metal coating from solution onto a conductive part by passing current through an electrolyte bath.
- [Photochemical Machining](https://manufacturingprocesses.org/processes/cutting/photochemical-machining.md): Photochemical machining masks sheet metal photographically and etches away the unmasked areas, cutting burr-free flat parts with no hard tooling.
- [Photo Etching](https://manufacturingprocesses.org/processes/finishing/photo-etching.md): Photo etching masks a surface photographically and etches shallow detail into it for decoration, marking or texture.
- [Vacuum Metalizing](https://manufacturingprocesses.org/processes/finishing/vacuum-metalizing.md): Vacuum metalizing evaporates aluminum inside a vacuum chamber so it condenses as a mirror-bright film on the part surface.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/forming/electroforming)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Electron Beam Melting (EBM)"
category: "Forming"
subcategory: "Additive Manufacturing"
materials: ["Metal"]
tolerances: "About ±0.012–0.016 in (±0.3–0.4 mm), looser than laser powder bed fusion because of the thicker layers and coarser powder. Machined interfaces hold normal machining tolerances."
volumes: "1–10,000 parts per year; unusually well suited to serial production of a repeating titanium part"
lead_time: "7–20 business days. Build time is competitive, but the controlled cooldown from process temperature adds many hours to every cycle."
url: https://manufacturingprocesses.org/processes/forming/electron-beam-melting-ebm
---

# Electron Beam Melting (EBM)

Electron beam melting fuses metal powder with an electron beam in vacuum at high preheat temperature, which cuts residual stress in titanium parts.

- **Category**: [Forming](https://manufacturingprocesses.org/processes/forming.md)
- **Family**: Additive Manufacturing
- **Materials**: Metal
- **Typical tolerances**: About ±0.012–0.016 in (±0.3–0.4 mm), looser than laser powder bed fusion because of the thicker layers and coarser powder. Machined interfaces hold normal machining tolerances.
- **Surface finish**: Ra 800–1,400 µin (20–35 µm) as built — the roughest of the powder bed processes. Advantageous for osseointegration on implant surfaces, unacceptable for seals or bearings, which must be machined.
- **Typical volumes**: 1–10,000 parts per year; unusually well suited to serial production of a repeating titanium part
- **Lead time**: 7–20 business days. Build time is competitive, but the controlled cooldown from process temperature adds many hours to every cycle.

## Overview

Electron beam melting (EBM) fuses metal powder with a focused electron beam inside a vacuum chamber, at a bed preheat temperature high enough — around 650–750 °C (1,200–1,380 °F) for Ti-6Al-4V — that residual stress is largely relieved as the part is built. That single difference from laser powder bed fusion drives everything else: EBM parts come off the plate essentially stress-free, need no post-build stress relief, and need far less support.

Layers are thicker at 0.002–0.004 in (50–100 µm), powder is coarser at 45–106 µm, and beam power reaches several kilowatts with electromagnetic deflection that steers far faster than any galvanometer. The result is high deposition rate but a rougher surface, typically Ra 800–1,400 µin (20–35 µm), and looser accuracy near ±0.012–0.016 in (±0.3–0.4 mm).

The material set is narrow and deliberate: Ti-6Al-4V and Ti-6Al-4V ELI, titanium aluminides, CoCr, and some nickel alloys. Orthopedic implants with porous ingrowth lattices and turbine hardware are the dominant applications.

## How it works

1. **Evacuate.** The chamber is pumped to roughly 10⁻⁴–10⁻⁵ mbar, often with a small helium bleed to stabilize the beam and help cooling. Vacuum is what makes EBM viable for titanium and titanium aluminide — there is no atmosphere to pick up oxygen from.
2. **Preheat the layer.** Before melting, the beam is defocused and swept rapidly across the whole bed to lightly sinter the powder and bring it to the process temperature — roughly 650–750 °C (1,200–1,380 °F) for Ti-6Al-4V, and higher still for titanium aluminides. This preheat both prevents powder from being blown out of the bed by electrostatic charge and eliminates the steep gradients that cause residual stress.
3. **Melt.** The focused beam melts the cross-section. Deflection is electromagnetic and effectively inertia-free, so the beam can be time-shared across multiple melt pools at once.
4. **Index and repeat** at 0.002–0.004 in (50–100 µm) per layer.
5. **Cool under vacuum or inert gas.** A full build cools slowly from process temperature, which can take many hours and is a substantial share of total cycle time.
6. **Recover the part.** The build sits in a lightly sintered powder cake rather than loose powder. It is freed in a blasting cabinet that uses the same alloy powder as the blast media, so the recovered material can be sieved and reused.
7. **Finish.** Because the part is already stress-relieved, it can go straight to HIP (for fatigue-critical work), machining of interfaces, and surface treatment. Implant lattices are typically left as built, since the rough surface is the point.

## Design guidelines

### Supports
EBM needs far fewer supports than laser powder bed fusion because the sintered cake carries load and the low thermal gradient removes most of the curl. Supports are still used to conduct heat out of large down-facing solid areas, but their function is thermal rather than mechanical.

### Wall thickness and resolution
The coarser powder and thicker layers set a higher floor than laser processes: plan on 0.024–0.040 in (0.6–1.0 mm) minimum wall, and do not expect crisp detail below about 0.04 in (1 mm).

### Powder removal from lattices
Porous ingrowth structures are EBM's signature capability, but the semi-sintered cake inside them has to be blasted out. Keep pore sizes and strut spacing large enough that the blast media can reach — this is a manufacturing constraint on the lattice design, not an afterthought.

### Machining allowance
Add 0.020–0.040 in (0.5–1.0 mm) to any interface surface. Given the as-built roughness, taper locks, bearing seats, and thread features are always machined.

### Surface roughness as a design feature
Ra 800–1,400 µin (20–35 µm) is poor for a sealing face and excellent for bone ingrowth. Decide surface by surface which one you want, and machine the ones that need it. The [surface finish chart](/charts/surface-finish-chart) puts these values in context against machined and ground finishes.

### Alloy selection
EBM's vacuum and high preheat suit alloys that are hard to process any other way — titanium aluminides in particular, which crack readily under the thermal gradients of laser powder bed fusion.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Wall thickness | 0.040 in (1.0 mm) | 0.024 in (0.6 mm) | Coarse powder, 50–100 µm layers |
| Overhang from vertical | 45° | Cake supports much steeper | Sintered cake carries the surface |
| Detail size | 0.06 in (1.5 mm) | 0.04 in (1 mm) | Beam spot and powder size |
| Machining stock | 0.040 in (1.0 mm) | 0.020 in (0.5 mm) | As-built roughness must be removed |
| Lattice strut | 0.03 in (0.7 mm) | 0.02 in (0.5 mm) | Powder must be blasted out |

## Cost drivers

EBM's economics differ from laser powder bed fusion in a way that matters at quantity. Because the process runs hot and supports are minimal, parts can be stacked in the powder cake rather than each being anchored to the plate, so a build can be packed far more densely. Combined with high beam deposition rates, that makes EBM comparatively strong for serial production of a repeating part — which is exactly why acetabular cups and similar implants are made this way in volume.

Against that: cycle time includes a long controlled cooldown, the machine population is small, and the qualified alloy list is short. Post-processing is lighter than laser PBF (no stress relief step) but blasting parts out of a sintered cake is manual work.

Volume breakpoints: EBM makes sense from a handful of parts for titanium geometry that a laser process would struggle with, and scales well into thousands per year for a repeating implant or blade. For a one-off prototype in a common alloy, laser powder bed fusion is usually faster to source.

1. Stack parts vertically in the cake — the process supports it and machine hours are shared.
2. Design lattices that can actually be blasted clean.
3. Accept the as-built surface wherever it is functionally acceptable, and machine only interfaces.
4. Choose EBM specifically for titanium and titanium aluminide; for steels and aluminum, laser PBF has more mature parameters.

## FAQ

### Why do EBM parts not need stress relief?

The powder bed is preheated to roughly 650–750 °C (1,200–1,380 °F) for Ti-6Al-4V before each layer is melted, so the thermal gradient between the melt pool and its surroundings is small. Residual stress relaxes continuously during the build rather than accumulating, and parts come off the plate essentially stress-free.

### Why does EBM run in a vacuum?

An electron beam requires vacuum to propagate — gas molecules would scatter it. The chamber is pumped to about 10⁻⁴–10⁻⁵ mbar. The side benefit is that reactive alloys such as titanium and titanium aluminide pick up no oxygen or nitrogen during processing.

### Why are EBM surfaces so rough?

Powder is coarser (45–106 µm versus 15–45 µm for laser processes), layers are thicker at 0.002–0.004 in (50–100 µm), and the preheat lightly sinters powder onto every surface. As-built Ra of 800–1,400 µin (20–35 µm) is normal, and any functional surface must be machined.

### Which materials can EBM process?

The qualified set is narrow: Ti-6Al-4V, Ti-6Al-4V ELI, titanium aluminides, CoCr alloys, and some nickel alloys. Aluminum is impractical, and the alloy range is much smaller than for laser powder bed fusion — EBM is a specialist process, not a general one.

### When should I choose EBM over laser powder bed fusion?

For titanium and titanium aluminide parts where residual stress or cracking is the limiting problem, for porous implant lattices where a rough surface is functionally desirable, and for serial production of a repeating part that can be stacked densely in the powder cake.

### Does EBM need supports?

Far fewer than laser processes. The semi-sintered powder cake carries the part and the low thermal gradient removes most of the curling force. Supports that remain are there to conduct heat out of large down-facing solid regions rather than to hold the part down.

## Alternative processes

- [Selective Laser Melting (SLM)](https://manufacturingprocesses.org/processes/forming/selective-laser-melting-slm.md): Selective laser melting fully melts metal powder with a laser under inert gas, producing parts at close to wrought density.
- [Direct Metal Laser Sintering (DMLS)](https://manufacturingprocesses.org/processes/forming/direct-metal-laser-sintering-dmls.md): Direct metal laser sintering fuses metal powder layer by layer with a laser, building dense metal parts anchored to a supported build plate.
- [Investment Casting](https://manufacturingprocesses.org/processes/forming/investment-casting.md): Investment casting builds a ceramic shell around a wax pattern, melts the wax out and pours metal into the cavity, giving fine detail in almost any alloy.
- [CNC Machining](https://manufacturingprocesses.org/processes/cutting/cnc-machining.md): CNC machining removes material from solid stock with a programmed cutting tool, producing accurate parts directly from a CAD model with no tooling.
- [Metal Injection Molding](https://manufacturingprocesses.org/processes/forming/metal-injection-molding.md): Metal injection molding molds a powder-and-binder feedstock exactly like a plastic, then debinds and sinters it into a dense small metal part.

## Related processes

- [Direct Metal Laser Sintering (DMLS)](https://manufacturingprocesses.org/processes/forming/direct-metal-laser-sintering-dmls.md): Direct metal laser sintering fuses metal powder layer by layer with a laser, building dense metal parts anchored to a supported build plate.
- [Selective Laser Melting (SLM)](https://manufacturingprocesses.org/processes/forming/selective-laser-melting-slm.md): Selective laser melting fully melts metal powder with a laser under inert gas, producing parts at close to wrought density.
- [Binder Jetting](https://manufacturingprocesses.org/processes/forming/binder-jetting.md): Binder jetting prints liquid binder into a powder bed to hold each layer, producing green parts that are later sintered or infiltrated.
- [Directed Energy Deposition (DED)](https://manufacturingprocesses.org/processes/forming/directed-energy-deposition-ded.md): Directed energy deposition feeds metal wire or powder into a moving melt pool, building up large features or repairing existing parts.
- [Investment Casting](https://manufacturingprocesses.org/processes/forming/investment-casting.md): Investment casting builds a ceramic shell around a wax pattern, melts the wax out and pours metal into the cavity, giving fine detail in almost any alloy.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/forming/electron-beam-melting-ebm)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Filament Winding"
category: "Forming"
subcategory: "Composites"
materials: ["Composite"]
tolerances: "Inside diameter ±0.005–0.010 in (±0.13–0.25 mm), set by the mandrel. Wall thickness and outside diameter carry roughly ±0.5–1% because thickness builds in whole layers."
volumes: "100–500,000 parts per year"
lead_time: "6–16 weeks for mandrel and pattern development on a new part; hours per part in production once the winding program exists."
url: https://manufacturingprocesses.org/processes/forming/filament-winding
---

# Filament Winding

Filament winding wraps resin-impregnated continuous fiber onto a rotating mandrel in a controlled pattern, producing tubes and pressure vessels.

- **Category**: [Forming](https://manufacturingprocesses.org/processes/forming.md)
- **Family**: Composites
- **Materials**: Composite
- **Typical tolerances**: Inside diameter ±0.005–0.010 in (±0.13–0.25 mm), set by the mandrel. Wall thickness and outside diameter carry roughly ±0.5–1% because thickness builds in whole layers.
- **Surface finish**: The bore replicates the mandrel and is smooth. The outer surface shows the winding pattern as a visible fiber texture; a peel ply, shrink tape, or surfacing veil is used where a smoother or paintable exterior is needed.
- **Typical volumes**: 100–500,000 parts per year
- **Lead time**: 6–16 weeks for mandrel and pattern development on a new part; hours per part in production once the winding program exists.

## Overview

Filament winding lays resin-impregnated continuous fiber onto a rotating mandrel along a computer-controlled path, so every fiber follows the load line the designer chose. It produces the highest fiber volume fractions of any laminating process — typically 60–70% wet wound — because winding tension consolidates each layer as it is laid.

The signature result is netting analysis's optimum helical angle of 54.7° for a closed-end cylindrical pressure vessel, the angle at which fiber stresses balance the 2:1 ratio of hoop to axial stress in a thin-walled cylinder. Real vessels combine hoop windings near 85–90° with helical and polar passes to reach that balance across domes and cylinder alike.

Because fiber must be laid in tension over a convex surface, filament winding makes only surfaces of revolution and gentle variations on them: CNG and hydrogen cylinders, pipe and pressure vessels, rocket motor cases, drive shafts, rollers, and poles. It is highly automated, so labor per part is low and volumes reach hundreds of thousands.

## How it works

1. **Prepare the mandrel.** The mandrel defines the inside surface. It may be a reusable steel or aluminum shaft withdrawn afterward, a collapsible or segmented mandrel for closed shapes, a soluble sand or plaster core washed out after cure, or a permanent liner that stays in the part — an aluminum liner for a Type III cylinder or a blow-molded HDPE or polyamide liner for Type IV.
2. **Impregnate.** In wet winding, tow passes through a resin bath and a set of wiper dies or rollers that meter resin content. Towpreg and prepreg tape winding skip the bath and give tighter control of resin fraction at higher material cost.
3. **Wind.** The mandrel rotates while a delivery eye traverses. The ratio of traverse speed to rotation sets the winding angle: roughly 85–90° for hoop windings that carry hoop stress, 15–80° for helical passes that carry combined loads, and 5–15° for polar windings that run over the domes of a closed vessel. Fiber tension — commonly a few newtons up to tens of newtons per tow — consolidates the laminate and squeezes out excess resin.
4. **Build the pattern.** Layers are wound in a repeating pattern that closes on itself after a whole number of circuits, so coverage is uniform. Modern machines plan these patterns automatically along with the dome turn-around geometry.
5. **Cure.** Polyester and vinyl ester systems cure at ambient to about 180 °F (80 °C); epoxies at 250–350 °F (120–180 °C). Curing is done with the mandrel still rotating so resin cannot drain to the bottom of the part before gelation.
6. **Extract and finish.** The mandrel is withdrawn, collapsed, or dissolved. Ends are trimmed, bosses and fittings installed, and pressure vessels are proof tested and often autofrettaged — pressurized beyond the liner's yield point so the metal liner is left in beneficial compression.

## Design guidelines

### Convex surfaces of revolution only
Fiber under tension takes the shortest geodesic path over the mandrel. It will not stay in a concave region and will slip off a surface that turns away from it. Any concavity, sharp shoulder, or re-entrant feature must be added afterward as a bonded or laid-up detail.

### Winding angle follows the load
Hoop stress in a thin-walled cylinder is twice the axial stress, so a vessel wound entirely at one angle is inefficient. Use hoop windings near 90° for the cylindrical section and helical windings at 54.7° or lower where axial and dome loads must be carried. For a drive shaft loaded in torsion, ±45° is the efficient answer; for a bending-loaded pole, low-angle windings near 0° carry the load.

### Dome geometry
The dome profile of a pressure vessel is set by the fiber path, not chosen freely. Geodesic and near-geodesic dome contours exist precisely so fiber does not slip; specifying an arbitrary dome shape forces non-geodesic winding with friction-dependent stability.

### Boss and port design
Openings must be reinforced and the fiber turned around a boss of adequate diameter. A polar boss too small in relation to the vessel diameter causes fiber build-up and a resin-rich, weak turnaround region.

### Wall thickness and tolerance
Inside diameter is set by the mandrel and is the accurate dimension — typically ±0.005–0.010 in (±0.13–0.25 mm). Wall thickness builds up in whole layers and varies with pattern and tension, so outside diameter carries roughly ±0.5–1% of the wall.

### Liner compatibility
For Type IV vessels, the thermoplastic liner must survive the resin cure temperature. This constrains resin selection to systems that cure below the liner's softening point, or requires a liner material chosen for the cure cycle.

### Design to code
Composite pressure vessels are governed by design codes that require burst pressures of roughly 2.25–3.5 times service pressure depending on cylinder type and jurisdiction, plus cycle and environmental qualification. Establish the applicable code before choosing a layup.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Surface geometry | Convex, surface of revolution | No concavity | Fiber under tension will not follow it |
| Helical angle, closed vessel | 54.7° | Set by netting analysis | Balances 2:1 hoop-to-axial stress |
| Hoop winding angle | 88–90° | 85° | Efficient hoop stress carriage |
| Fiber volume fraction | 60–65% | 70% | Above this, resin starvation |
| ID tolerance | ±0.010 in (0.25 mm) | Mandrel-limited | Mandrel defines the bore |
| Wall build-up | Whole layer increments | — | Thickness is quantized by pattern |

## Cost drivers

Filament winding is a machine-hour process with very little touch labor, which is what separates it economically from hand layup. Cost per part is driven by winding time — a function of part surface area, layer count, and how fast the machine can traverse without disturbing fiber placement — plus material.

Mandrels are the tooling cost, and the type chosen changes the economics completely. A reusable steel mandrel amortizes across thousands of parts. A soluble sand or plaster mandrel is consumed every cycle and adds both material and washout labor. A permanent liner is part of the product, so it appears as material rather than tooling.

Material is a large share of a wound part because fiber volume fraction is high by design. Wet winding with bulk roving and resin is the cheapest feedstock route; towpreg costs more but reduces variability and eliminates the resin bath.

Volume breakpoints: filament winding becomes attractive at roughly 100 parts a year where a reusable mandrel can be used, and scales to hundreds of thousands for cylinders and pipe. Below that, hand [composite laminating](/processes/forming/composite-laminating) over a male plug is usually cheaper.

1. Design for a reusable mandrel — extraction geometry is worth more than shape freedom.
2. Keep the shape a true surface of revolution; every non-axisymmetric feature is added labor.
3. Use hoop windings where hoop stress dominates instead of thickening the whole laminate.
4. Specify glass where stiffness allows and reserve carbon for the layers that need it — hybrid layups are routine.
5. Batch parts on one mandrel where geometry permits, then cut them apart after cure.

## FAQ

### Why is 54.7 degrees the optimum winding angle?

In a thin-walled closed cylinder under internal pressure, hoop stress is twice the axial stress. Netting analysis — which assumes the fibers carry all the load and the resin none — gives an angle whose tangent is the square root of two, or about 54.7°, at which fiber tension balances that 2:1 ratio exactly. Real vessels combine this with hoop and polar windings.

### What shapes can filament winding produce?

Convex surfaces of revolution: cylinders, cones, spheres, and closed pressure vessels with domed ends. Fiber laid under tension follows a geodesic path and will not stay in a concave region, so concavities, flanges, and re-entrant features must be added afterward as bonded or hand-laid details.

### What fiber volume fraction does filament winding achieve?

Typically 60–70% wet wound, higher than hand layup or vacuum-bagged laminating, because winding tension consolidates each layer as it is placed and squeezes out excess resin. Since laminate stiffness scales nearly linearly with fiber volume fraction, this is the process's main structural advantage.

### What is the difference between Type III and Type IV pressure vessels?

Both are fully wrapped composite cylinders. A Type III uses a metallic liner, usually aluminum, which carries some load and provides the gas barrier. A Type IV uses a thermoplastic liner that provides only the barrier, making the vessel lighter but requiring a resin system that cures below the liner's softening temperature.

### How is the mandrel removed from a closed vessel?

Four ways: withdraw a tapered or straight mandrel through an open end, collapse a segmented mandrel and remove it in pieces, wash out a soluble sand or plaster core, or leave a permanent liner in place as part of the finished product — which is how pressure vessels are made.

### What tolerance can filament winding hold?

The bore is accurate because the mandrel defines it, typically ±0.005–0.010 in (±0.13–0.25 mm). Wall thickness builds up in whole layers and varies with pattern and tension, so the outside diameter carries about ±0.5–1% of the wall thickness.

## Alternative processes

- [Composite Laminating](https://manufacturingprocesses.org/processes/forming/composite-laminating.md): Composite laminating lays reinforcing fiber into a mold and consolidates it with resin, curing into a stiff, light monocoque part.
- [Pultrusion](https://manufacturingprocesses.org/processes/forming/pultrusion.md): Pultrusion pulls continuous fiber through a resin bath and a heated die, curing it into a constant-section composite profile of unlimited length.
- [Metal Spinning](https://manufacturingprocesses.org/processes/forming/metal-spinning.md): Metal spinning presses a rotating metal disc against a mandrel with a roller or hand tool, forming it into an axially symmetric hollow shape.
- [Deep Drawing](https://manufacturingprocesses.org/processes/forming/deep-drawing.md): Deep drawing pushes a sheet metal blank through a die with a punch so the metal flows into a seamless cup or box deeper than its own diameter.
- [Rotation Molding](https://manufacturingprocesses.org/processes/forming/rotation-molding.md): Rotational molding tumbles powdered polymer inside a heated mold on two axes so it fuses into a seamless, stress-free hollow part with a uniform wall.

## Related processes

- [Composite Laminating](https://manufacturingprocesses.org/processes/forming/composite-laminating.md): Composite laminating lays reinforcing fiber into a mold and consolidates it with resin, curing into a stiff, light monocoque part.
- [Pultrusion](https://manufacturingprocesses.org/processes/forming/pultrusion.md): Pultrusion pulls continuous fiber through a resin bath and a heated die, curing it into a constant-section composite profile of unlimited length.
- [DMC and SMC Molding](https://manufacturingprocesses.org/processes/forming/dmc-and-smc-molding.md): DMC and SMC molding compression-molds a pre-mixed dough or sheet of chopped fiber and thermoset resin into a stiff, dimensionally stable part.
- [Rotation Molding](https://manufacturingprocesses.org/processes/forming/rotation-molding.md): Rotational molding tumbles powdered polymer inside a heated mold on two axes so it fuses into a seamless, stress-free hollow part with a uniform wall.
- [Blow Molding](https://manufacturingprocesses.org/processes/forming/blow-molding.md): Blow molding inflates a heated plastic tube or preform against the walls of a mold with compressed air, producing hollow one-piece containers and tanks.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/forming/filament-winding)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Forging"
category: "Forming"
subcategory: "Metal"
materials: ["Metal"]
tolerances: "Commercial impression-die forgings hold roughly ±0.030 in (±0.8 mm) on small parts, widening with size, plus separate allowances for die wear, mismatch, and die closure; machine anything needing better than about ±0.010 in"
volumes: "1–100 pieces open die; 1,000–500,000+ per year impression die"
lead_time: "8–16 weeks to sink an impression die and produce first articles; 3–8 weeks per production run. Open-die work in 2–6 weeks with no tooling"
url: https://manufacturingprocesses.org/processes/forming/forging
---

# Forging

Forging deforms solid metal under compressive load in a hammer, press or dies, so the grain flow follows the part contour and fatigue strength rises.

- **Category**: [Forming](https://manufacturingprocesses.org/processes/forming.md)
- **Family**: Metal
- **Materials**: Metal
- **Typical tolerances**: Commercial impression-die forgings hold roughly ±0.030 in (±0.8 mm) on small parts, widening with size, plus separate allowances for die wear, mismatch, and die closure; machine anything needing better than about ±0.010 in
- **Surface finish**: Roughly 125–500 µin Ra (3.2–12.5 µm) as forged after descaling; critical surfaces are machined
- **Typical volumes**: 1–100 pieces open die; 1,000–500,000+ per year impression die
- **Lead time**: 8–16 weeks to sink an impression die and produce first articles; 3–8 weeks per production run. Open-die work in 2–6 weeks with no tooling

## Overview

Forging shapes solid metal under compressive load — in a hammer, a press, or between rotating dies — rather than by melting or cutting it. The metal is worked while hot enough to flow plastically, typically 2,100–2,300°F (1,150–1,260°C) for carbon steel and 700–850°F (370–455°C) for aluminum, and the grain structure deforms with the shape instead of being cut through.

That grain flow is the reason forging exists. A forged crankshaft, connecting rod, hook, or landing gear component has continuous fiber following its contour, giving fatigue and impact performance that a casting or a part machined from bar cannot match at the same weight. Porosity is closed rather than created.

Forging covers a huge range: open-die work on one-off shafts and rings weighing tons, impression-die production of automotive and hand-tool components at hundreds of thousands per year, and cold heading of fasteners at hundreds of parts per minute.

## How it works

Three families cover most production.

**Open-die forging** works a heated billet between flat or simply contoured dies, with the operator repositioning it between blows. It produces shafts, blocks, discs, and — with a mandrel — rolled rings, from a few pounds to hundreds of tons. There is no part-specific tooling, so it suits one-offs and very large parts.

**Impression-die (closed-die) forging** squeezes the billet into a cavity cut into matched die halves. Excess metal escapes as flash around the parting line, which is what generates the back pressure that fills the cavity corners. Flash is trimmed afterward and recycled.

**Cold forging and heading** work the metal below its recrystallization temperature, gaining excellent surface finish and tight tolerance at the cost of much higher forces. This is how bolts, rivets, and small precision parts are made at rate.

The hot impression-die cycle:

1. **Cut and heat.** Bar or billet is sheared to weight and heated — induction for speed, gas furnace for large sections. Getting the temperature right matters at both ends: too cold and the die loads spike and the part cracks, too hot and the grain coarsens or the metal burns.
2. **Preform.** Fullering and edging operations redistribute the mass along the billet's length so the finish impression only has to do local work.
3. **Blocker.** An intermediate cavity brings the part near shape with generous radii.
4. **Finisher.** The final impression forms the part and squeezes flash out around the parting line.
5. **Trim.** Flash is sheared off in a trim die while the part is still hot.
6. **Heat treatment.** Normalize, quench and temper, or solution and age to reach the specified properties. See [metal melting points](/charts/metal-melting-points) for how forging temperatures sit relative to melting.
7. **Finishing.** Shot blast to remove scale, then straighten and machine the critical features.

A forging is almost never a finished part: it is a near-net shape with the metallurgy already right, delivered to machining with stock only where it is needed.

## Design guidelines

### Draft

Impression-die forgings need substantial draft — commonly 3–7° for steel and 1–3° for aluminum, which flows more readily and releases better. Internal surfaces (pockets shrinking onto a die projection) need more draft than external ones. Precision and cold forgings can approach zero draft at higher tooling cost.

### Radii, generously

Corner radius should be at least 0.125 in (3 mm) and fillet radii larger still, growing with the depth of the cavity being filled. Sharp corners in a die cavity are unfillable — metal simply will not flow into them — and they crack the die. Where a forging keeps rejecting for underfill, an undersized fillet is the usual cause.

### Parting line

Put the parting line on a single plane through the largest cross-section wherever possible. A flat parting line is the cheapest die to sink and the easiest to trim. Expect a small mismatch across it and never place a tight dimension so it depends on die-half alignment.

### Webs and ribs

Thin webs chill against the die and resist filling. Keep web thickness at a workable minimum — thicker than a casting would need in the same place — and keep rib height-to-width ratios modest. Deep, narrow ribs are the most expensive feature to fill and the fastest way to wear out a die.

### Machining stock

Add 0.030–0.125 in (0.75–3 mm) on surfaces that will be machined, and more on large forgings and on surfaces far from the parting line. Enough stock must remain to clean up scale, decarburized skin, and die wear variation. For achievable machined finishes, see the [surface finish chart](/charts/surface-finish-chart).

### Orient the grain flow deliberately

The whole point of forging is that the fiber follows the shape. Orient the part so the principal load path runs along the grain flow rather than across it, and avoid machining operations that cut through the flow lines at a highly stressed section — doing so throws away the advantage that justified forging.

### Tolerances

Commercial impression-die tolerances are on the order of ±0.030 in (±0.8 mm) on small parts, widening with size, plus separate allowances for die wear, mismatch, and die closure. Precision forging holds substantially tighter but costs more per part; machine any feature that needs better than about ±0.010 in.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Draft, steel | 5–7° | 3° | Hot part grips the die and must release |
| Draft, aluminum | 2–3° | 1° | Aluminum flows and releases more readily |
| Corner radius | 0.19–0.25 in (5–6 mm) | 0.125 in (3 mm) | Metal will not flow into a sharp die corner |
| Fillet radius | Increase with cavity depth | — | Deep cavities need generous fillets to fill |
| Parting line | Flat, through the largest section | Contoured costs more | Cheapest die to sink and trim |
| Machining stock | 0.060 in (1.5 mm) typical | 0.030 in (0.75 mm) | Must clean up scale, decarb, and die wear |
| Ribs | Low height-to-width ratio | — | Deep narrow ribs underfill and wear dies fast |

## Variants

- Drop Forging
- Roll Forging

## Cost drivers

Die cost and press time dominate impression-die forging, and both scale with part size and complexity. Dies are sunk from hot-work tool steel and wear out — die life is a real per-part cost, not a fixed one, and it falls sharply with thin ribs, sharp radii, and high forging temperatures. Material cost is billet weight including flash, so flash allowance is money burned every stroke. Heat treatment and machining of the forged part are frequently the largest cost lines of all.

Volume breakpoints: open-die forging is economic at one piece and stays competitive to a few hundred. Impression-die forging typically needs 1,000 or more to justify sinking a die, works well from 5,000 to several hundred thousand, and at the top of that range is usually running on automated hot formers. Below those volumes, machining from bar or investment casting is normally cheaper.

1. **Open up the radii.** Larger fillets fill better, forge with lower loads, and extend die life — three cost reductions from one change.
2. **Keep the parting line flat.** It cuts die sinking, trim die cost, and mismatch scrap.
3. **Design the machining stock, don't default it.** Stock only where a machined surface is needed, at the minimum that reliably cleans up.
4. **Consider a preform.** Better mass distribution before the finisher reduces forging load, flash weight, and die wear at once.
5. **Match the alloy to forgeability.** Plain carbon and low-alloy steels and 6000-series aluminum forge easily; stainless, nickel alloys, and titanium need narrower temperature windows, more force, and cost far more per pound to shape.

## FAQ

### Why is a forged part stronger than a casting?

Forging deforms the existing grain structure so the fiber follows the part contour, and it closes porosity rather than creating it. The result is better fatigue and impact performance in the direction of the flow lines. A casting solidifies from liquid with no directional fiber and typically some residual porosity.

### What draft angle does a forging need?

Commonly 3–7° for steel and 1–3° for aluminum, with more on internal surfaces that shrink onto a die projection than on external ones. Precision and cold forging can approach zero draft, at higher tooling cost and force.

### What temperature is steel forged at?

Typically 2,100–2,300°F (1,150–1,260°C) for carbon and low-alloy steels. Aluminum forges at 700–850°F (370–455°C) and copper alloys around 1,300–1,450°F (700–790°C). Forging below the window spikes die loads and cracks the part; above it, the grain coarsens or the metal burns.

### What tolerance can forging hold?

Commercial impression-die forgings run about ±0.030 in (±0.8 mm) on small parts, widening with size, plus allowances for die wear, mismatch across the parting line, and die closure. Anything tighter than roughly ±0.010 in should be machined into the forging afterward.

### What volume justifies impression-die forging?

Roughly 1,000 pieces as a lower bound for sinking a die, with the process at its best from 5,000 to several hundred thousand per year. Below that, open-die forging, machining from bar, or investment casting is normally cheaper because they carry no die cost.

### Forging or machining from billet?

Machining wins for one-offs, complex prismatic geometry, and tight tolerances everywhere. Forging wins once fatigue life matters, once volumes reach the thousands, and on expensive alloys, because it puts material only where the part needs it instead of cutting most of the billet into chips.

## Alternative processes

- [Investment Casting](https://manufacturingprocesses.org/processes/forming/investment-casting.md): Investment casting builds a ceramic shell around a wax pattern, melts the wax out and pours metal into the cavity, giving fine detail in almost any alloy.
- [CNC Machining](https://manufacturingprocesses.org/processes/cutting/cnc-machining.md): CNC machining removes material from solid stock with a programmed cutting tool, producing accurate parts directly from a CAD model with no tooling.
- [Sand Casting](https://manufacturingprocesses.org/processes/forming/sand-casting.md): Sand casting pours molten metal into a cavity formed in bonded sand around a pattern, then breaks the mold away to release the part.
- [Powder Metallurgy (Press and Sinter)](https://manufacturingprocesses.org/processes/forming/powder-metallurgy-press-and-sinter.md): Press-and-sinter powder metallurgy compacts metal powder in a rigid die and sinters it below its melting point into a finished net-shape part.

## Related processes

- [Heat Treatment](https://manufacturingprocesses.org/processes/forming/heat-treatment.md): Heat treatment cycles metal through controlled heating and cooling to change its hardness, strength, ductility or internal stress state.
- [Metal Extrusion](https://manufacturingprocesses.org/processes/forming/metal-extrusion.md): Metal extrusion pushes heated billet through a shaped die, producing continuous aluminum or copper profile of constant cross-section.
- [Swaging](https://manufacturingprocesses.org/processes/forming/swaging.md): Swaging hammers or presses the end of a tube or bar inward against a die to reduce or taper its diameter without removing any material.
- [Powder Metallurgy (Press and Sinter)](https://manufacturingprocesses.org/processes/forming/powder-metallurgy-press-and-sinter.md): Press-and-sinter powder metallurgy compacts metal powder in a rigid die and sinters it below its melting point into a finished net-shape part.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/forming/forging)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Fused Deposition Modeling (FDM)"
category: "Forming"
subcategory: "Additive Manufacturing"
materials: ["Plastic", "Composite"]
tolerances: "±0.020 in (±0.5 mm) or ±0.5% of nominal, whichever is greater, on desktop-class machines. Industrial systems hold ±0.008 in (±0.2 mm) or ±0.002 in/in. Drilled and reamed features are far tighter."
volumes: "1–500 parts"
lead_time: "1–5 business days; small parts print in 1–4 hours and a full plate in 10–30 hours. No tooling, so the first part ships as fast as the last."
url: https://manufacturingprocesses.org/processes/forming/fused-deposition-modeling-fdm
---

# Fused Deposition Modeling (FDM)

Fused deposition modeling extrudes molten thermoplastic filament along programmed paths so each layer welds to the one beneath it.

- **Category**: [Forming](https://manufacturingprocesses.org/processes/forming.md)
- **Family**: Additive Manufacturing
- **Materials**: Plastic, Composite
- **Typical tolerances**: ±0.020 in (±0.5 mm) or ±0.5% of nominal, whichever is greater, on desktop-class machines. Industrial systems hold ±0.008 in (±0.2 mm) or ±0.002 in/in. Drilled and reamed features are far tighter.
- **Surface finish**: Ra 200–1,000 µin (5–25 µm), with visible layer lines whose spacing equals the layer height. Sanding, vapor smoothing (ABS in acetone), or filler primer are needed for a cosmetic finish.
- **Typical volumes**: 1–500 parts
- **Lead time**: 1–5 business days; small parts print in 1–4 hours and a full plate in 10–30 hours. No tooling, so the first part ships as fast as the last.

## Overview

Fused deposition modeling (FDM), also called fused filament fabrication, extrudes molten thermoplastic filament through a heated nozzle — 0.4 mm is the standard size — and lays it down path by path so each 0.002–0.016 in (0.05–0.4 mm) layer welds to the one beneath. It is the cheapest and most widely available additive process, and the only common one that uses genuine engineering thermoplastics: PLA, PETG, ABS, ASA, nylon, polycarbonate, and on industrial machines PEI (ULTEM) and PEEK.

The trade-off is anisotropy. Bonds between deposited roads are weaker than the filament itself, so Z-axis tensile strength typically runs 30–70% of the XY value. Accuracy is the loosest of the mainstream processes at roughly ±0.020 in (±0.5 mm) or ±0.5% of nominal, whichever is greater, on desktop machines; industrial systems reach ±0.008 in (±0.2 mm).

FDM suits jigs, fixtures, housings, ducting, and functional prototypes from 1 to a few hundred parts, with build volumes from 200 mm cubes to 36 × 24 × 36 in (914 × 610 × 914 mm).

## How it works

1. **Slice and plan tool paths.** The model is sliced at 0.002–0.016 in (0.05–0.4 mm) — 0.008 in (0.2 mm) is the usual default — and each layer is decomposed into perimeters (typically 2–4), solid top and bottom skins, and a sparse infill of 15–40%.
2. **Heat and prime.** The nozzle reaches material temperature and the bed is heated to control first-layer adhesion and warp: roughly 190–220 °C (375–430 °F) nozzle with a 50–60 °C bed for PLA, 230–250 °C (445–480 °F) with a 100–110 °C bed for ABS and PETG, and 260–310 °C for polycarbonate. High-temperature materials such as PEI 9085 need a heated chamber above 160 °C (320 °F).
3. **Extrude the first layer.** The first layer is deliberately squashed against the bed. Get this wrong and nothing else matters — first-layer squish also produces the "elephant foot" flare of 0.1–0.2 mm that shows up on close-fitting bases.
4. **Build up layers.** The head traces perimeters and infill; the platform or gantry indexes one layer height. Extrusion width is normally 100–120% of nozzle diameter, so a 0.4 mm nozzle lays roads about 0.42–0.48 mm wide.
5. **Print supports.** Overhangs beyond about 45° from vertical need support, either the same material (breakaway) or a dissolvable one such as HIPS or PVA on dual-extruder machines.
6. **Cool and remove.** Amorphous materials such as ABS and PC shrink 0.4–0.8% on cooling and warp when cooled unevenly, which is why they need an enclosed chamber. Semi-crystalline nylons need dry filament — nylon absorbs moisture from air within hours and prints with visible bubbles.
7. **Post-process.** Remove supports, drill and ream critical holes, tap threads or install heat-set inserts, and sand or vapor-smooth cosmetic surfaces.

## Design guidelines

### Wall thickness
Design walls as a multiple of extrusion width. With a 0.4 mm nozzle, 0.032 in (0.8 mm) equals two perimeters and is the practical minimum; 0.048–0.063 in (1.2–1.6 mm) gives three or four perimeters and is where structural walls should sit. A wall specified at 1.0 mm forces the slicer to fill an awkward 0.2 mm gap with a weak thin extrusion.

### Overhangs and bridges
Anything overhanging more than about 45° from vertical needs support. Unsupported bridges span up to roughly 0.4 in (10 mm) reliably if both ends are anchored. Chamfer or fillet overhanging edges to 45° and you can often delete supports entirely.

### Holes
Printed holes come out 0.004–0.016 in (0.1–0.4 mm) undersize because of extrusion width compensation and corner cutting. Drill and ream any hole that locates something. Horizontal holes should be drawn as teardrops or hexagons so the top does not need support.

### Orientation and anisotropy
Orient the part so the principal tensile and bending loads act in the XY plane. Layer-to-layer bonds carry 30–70% of the in-plane strength, so a printed hook loaded along Z will fail at a layer boundary. Where load direction cannot be controlled, increase perimeter count rather than infill.

### Threads and inserts
Do not print threads below M6. Use heat-set brass inserts (the cleanest option in FDM), or drill and tap — see the [tap drill chart](/charts/tap-drill-chart) for pilot sizes and the [thread size chart](/charts/thread-size-chart) for the corresponding major diameters.

### Fits and clearances
Allow 0.008–0.012 in (0.2–0.3 mm) per side for a slip fit and 0.016 in (0.4 mm) for a free-running fit. Add a 0.02 in (0.5 mm) chamfer at the base of mating features to clear the elephant-foot flare.

### Minimum features
The nozzle diameter sets the floor: 0.016 in (0.4 mm) for a standalone rib or pin, embossed text at 0.02 in (0.5 mm) line width and 0.12 in (3 mm) cap height. Engraved text reads better than embossed at small sizes.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Wall thickness | 0.048 in (1.2 mm) | 0.032 in (0.8 mm) | Multiples of a 0.4 mm extrusion width |
| Overhang angle | 45° from vertical | 60° with cooling | Each road needs the one below for support |
| Unsupported bridge | 0.2 in (5 mm) | 0.4 in (10 mm) | Molten road sags between anchors |
| Hole diameter | 0.12 in (3 mm), then ream | 0.08 in (2 mm) | Printed holes run 0.1–0.4 mm undersize |
| Layer height | 0.008 in (0.2 mm) | 0.004 in (0.1 mm) cosmetic | Time scales inversely with layer height |
| Clearance, moving | 0.016 in (0.4 mm) | 0.008 in (0.2 mm) | Extrusion overshoot closes tight gaps |
| Pin or rib | 0.08 in (2 mm) | 0.032 in (0.8 mm) | Two-perimeter minimum for stiffness |

## Cost drivers

FDM cost is almost entirely machine hours. Print time scales with the number of layers times path length, so height and solid volume both matter, but layer height is the strongest single lever: doubling from 0.004 in (0.1 mm) to 0.008 in (0.2 mm) roughly halves the build.

Material is secondary on desktop-class work — PLA, PETG, and ABS filament are commodity-priced by the kilogram — but rises sharply for polycarbonate, nylon composites, and especially PEI and PEEK, which also demand a high-temperature machine. Support material and the labor to remove it are a real line item on parts with deep internal overhangs; dissolvable support trades that labor for tank time.

Volume breakpoints: FDM is normally the cheapest route for 1–50 parts. Between 50 and 500, look at [SLS](/processes/forming/selective-laser-sintering-sls) or [MJF](/processes/forming/multi-jet-fusion-mjf), which produce better parts at similar unit cost once the build is packed. Above about 1,000 parts, [injection molding](/processes/forming/injection-molding) wins.

1. Increase layer height on non-cosmetic parts — the single biggest time saver.
2. Reduce infill to 15–20% and add perimeters instead; perimeters carry most of the bending stiffness.
3. Orient the part to eliminate supports before adding them.
4. Split large parts at natural seams and bond them, rather than paying for a tall build.
5. Stay on commodity materials unless a specific thermal or chemical requirement forces PC, nylon, or PEI.

## FAQ

### What tolerance can FDM hold?

Desktop-class FDM holds about ±0.020 in (±0.5 mm) or ±0.5% of the dimension, whichever is greater. Industrial systems with heated chambers reach ±0.008 in (±0.2 mm) or ±0.002 in/in. If a hole or shaft needs to be tighter, print it undersize and drill or ream it.

### Why are FDM parts weaker in the Z direction?

Each deposited road bonds to the layer below by partial remelting, and that weld is smaller than the road's own cross-section. Z-axis tensile strength typically lands at 30–70% of the XY value, so parts should be oriented with the main load in the print plane.

### What is the minimum wall thickness for FDM?

0.032 in (0.8 mm), which is two perimeters at a standard 0.4 mm nozzle. Use 0.048–0.063 in (1.2–1.6 mm) for structural walls. Always specify wall thickness as a multiple of extrusion width so the slicer does not leave a weak partial gap.

### What overhang angle can FDM print without support?

About 45° from vertical is the reliable limit; good part cooling stretches it toward 60° on small features. Horizontal bridges span up to roughly 0.4 in (10 mm) when both ends are anchored. Chamfering overhanging edges to 45° often removes the need for support entirely.

### How do you get threads into an FDM part?

Heat-set brass inserts are the most reliable option: print a straight pilot hole and melt the insert in with a soldering iron. Tapping directly into the plastic works down to about M6 but strips more easily. Printed threads below M6 are not worth attempting.

### Why does my ABS part warp or crack at the corners?

ABS shrinks roughly 0.4–0.8% as it cools, and uneven cooling puts the lower layers in tension. Print it in an enclosed, heated chamber with a bed at 100–110 °C (212–230 °F), add generous corner fillets, and avoid large flat first layers. PETG or ASA warp less if the chamber is not available.

## Alternative processes

- [Stereolithography (SLA)](https://manufacturingprocesses.org/processes/forming/stereolithography-sla.md): Stereolithography cures liquid photopolymer layer by layer with a scanning UV laser, giving the smoothest surfaces in additive manufacturing.
- [Selective Laser Sintering (SLS)](https://manufacturingprocesses.org/processes/forming/selective-laser-sintering-sls.md): Selective laser sintering fuses nylon powder layer by layer with a laser, so the surrounding loose powder supports the part and no supports are needed.
- [CNC Machining](https://manufacturingprocesses.org/processes/cutting/cnc-machining.md): CNC machining removes material from solid stock with a programmed cutting tool, producing accurate parts directly from a CAD model with no tooling.
- [Injection Molding](https://manufacturingprocesses.org/processes/forming/injection-molding.md): Injection molding forces molten thermoplastic into a steel mold under high pressure, producing high-volume identical plastic parts.
- [Thermoforming](https://manufacturingprocesses.org/processes/forming/thermoforming.md): Thermoforming heats a plastic sheet until pliable and pulls or presses it over a single-sided mold, producing large thin-walled parts with low tooling cost.

## Related processes

- [Rapid Prototyping](https://manufacturingprocesses.org/processes/forming/rapid-prototyping.md): Rapid prototyping builds a part layer by layer directly from a CAD model, with no tooling and no minimum order quantity.
- [Selective Laser Sintering (SLS)](https://manufacturingprocesses.org/processes/forming/selective-laser-sintering-sls.md): Selective laser sintering fuses nylon powder layer by layer with a laser, so the surrounding loose powder supports the part and no supports are needed.
- [Multi Jet Fusion (MJF)](https://manufacturingprocesses.org/processes/forming/multi-jet-fusion-mjf.md): Multi Jet Fusion prints a fusing agent onto nylon powder and drives it with infrared energy, fusing whole layers at once for high throughput.
- [Stereolithography (SLA)](https://manufacturingprocesses.org/processes/forming/stereolithography-sla.md): Stereolithography cures liquid photopolymer layer by layer with a scanning UV laser, giving the smoothest surfaces in additive manufacturing.
- [Directed Energy Deposition (DED)](https://manufacturingprocesses.org/processes/forming/directed-energy-deposition-ded.md): Directed energy deposition feeds metal wire or powder into a moving melt pool, building up large features or repairing existing parts.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/forming/fused-deposition-modeling-fdm)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Glassblowing"
category: "Forming"
subcategory: "Glass and Ceramics"
materials: ["Glass"]
volumes: "1–1,000 pieces hand blown; hundreds of thousands to hundreds of millions machine formed"
lead_time: "Hours to days per piece in a studio, plus annealing time that scales with section thickness. Container mold sets take weeks to months; machine production is continuous."
url: https://manufacturingprocesses.org/processes/forming/glassblowing
---

# Glassblowing

Glassblowing gathers molten glass on a blowpipe and inflates it, by hand or by machine, into hollow ware such as bottles, jars and vessels.

- **Category**: [Forming](https://manufacturingprocesses.org/processes/forming.md)
- **Family**: Glass and Ceramics
- **Materials**: Glass
- **Typical volumes**: 1–1,000 pieces hand blown; hundreds of thousands to hundreds of millions machine formed
- **Lead time**: Hours to days per piece in a studio, plus annealing time that scales with section thickness. Container mold sets take weeks to months; machine production is continuous.

## Overview

Glassblowing gathers molten glass on the end of a blowpipe and inflates it into a hollow form, either by hand at the bench and glory hole or automatically in a machine. Glass is melted in the furnace at roughly 2,000–2,400 °F (1,100–1,300 °C) and worked in a narrower window of about 1,600–2,000 °F (870–1,100 °C), where it is viscous enough to hold shape but soft enough to move. There is no sharp melting point — glass simply gets stiffer as it cools, and the whole craft is the management of that viscosity gradient against time.

The industrial branch is container glass. Individual section (IS) machines form bottles and jars by blow-and-blow for narrow-neck ware or press-and-blow for wide-mouth jars and lightweight narrow-neck bottles, with multi-section double- or triple-gob machines producing several hundred containers a minute.

Both branches share one non-negotiable step: annealing. A blown piece cooled freely locks in stresses large enough to shatter it days later, so every piece is held near the annealing point and cooled slowly through it — roughly 900–1,000 °F (480–540 °C) for soda-lime glass.

## How it works

1. **Melt.** Soda-lime batch — silica sand with soda ash and lime, plus cullet — is melted at roughly 2,400 °F (1,300 °C) and refined so bubbles rise out. Studio furnaces hold glass at a working temperature continuously.
2. **Gather.** The blowpipe is dipped and rotated in the melt to pick up a gather of glass. Rotation is constant from this point on: glass at working temperature flows under gravity, and the only thing keeping the form symmetric is spinning it.
3. **Form the parison.** The gather is shaped on a steel marver and given a first puff of air to establish a small bubble. In machine forming, a gob of controlled weight drops into a blank mold and is either blown or pressed into this preliminary shape.
4. **Inflate and shape.** Repeated cycles of reheating in the glory hole and blowing, with jacks, paddles, and wet newspaper shaping the outside, bring the piece to form. Machine forming transfers the parison to the blow mold and inflates it against the cavity.
5. **Add and finish.** Handles, feet, and stems are added from separate gathers. Containers get their finish (the threaded or sealing neck) from the blank mold end, which is why the neck is formed first and to tighter control than the body.
6. **Detach and anneal.** The piece is knocked off the pipe or punty and goes immediately into an annealing lehr, held near the annealing point — roughly 900–1,000 °F (480–540 °C) for soda-lime — then cooled slowly through the strain point so no residual stress remains.
7. **Inspect.** Container plants inspect every unit automatically for dimensional and stress faults; studio work is checked with polarized light for residual stress.

## Design guidelines

### Design for rotation
Blown glass is a body of revolution unless it is blown into a mold. Non-round forms come from a mold, from working the piece off-axis after inflation, or from assembly of separate elements.

### Wall thickness
Container glass typically runs 0.04–0.12 in (1–3 mm). Wall distribution is controlled by gob weight, parison shape, and reheat timing rather than by any direct means, so thin spots are a process outcome — this is why container specifications are written around gob weight and minimum wall rather than a nominal dimension.

### Uniform thickness and thermal stress
Thick and thin sections in the same piece cool at different rates and set up permanent stress. Abrupt junctions — a heavy base meeting a thin wall — are where thermal shock failures start in service.

### Draft and mold release
Blown-into-mold forms need draft and a parting line. Glass shrinks on cooling, which assists release, but sharp corners in the mold produce thin, chilled, weak regions in the glass because the glass touches cold steel there first.

### Annealing is a design constraint
Section thickness determines annealing time — thicker pieces need much longer holds and slower ramps. A design with a solid heavy element in it commits the piece to a long lehr cycle, and skipping it produces a piece that may fail without warning.

### Assembly while hot
Handles, stems, and applied decoration must be attached at working temperature and with compatible glass. Mismatched thermal expansion between two glasses joined in one piece cracks the joint on cooling — the reason art glass is specified by coefficient of expansion.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Form | Body of revolution | Mold or hand work otherwise | Glass is worked while rotating |
| Wall thickness (container) | 0.06 in (1.5 mm) | 0.04 in (1 mm) | Thin spots fail under internal pressure |
| Section uniformity | Uniform | Avoid abrupt junctions | Differential cooling locks in stress |
| Applied elements | Matched expansion glass | — | Mismatch cracks the joint on cooling |
| Annealing | Required on every piece | — | Residual stress causes delayed failure |

## Variants

- Studio Glassblowing
- Machine Blow and Blow
- Machine Press and Blow

## Cost drivers

The two branches have almost nothing in common economically.

Hand glassblowing is priced as skilled labor plus furnace energy. A glass furnace runs continuously — it cannot be cycled without damaging the refractory — so energy is a standing cost whether the shop is working or not, and studios generally work in teams because larger pieces need more hands. Loss rate is significant and every loss carries full labor cost.

Machine container forming is a capital and volume business. An IS machine line, its mold sets, and the annealing lehr represent large fixed investment, amortized across production measured in hundreds of containers per minute. Mold sets are the per-design cost, and they are replicated across every section of the machine, so a design change is a substantial tooling commitment. Lightweighting — reducing gob weight while holding strength — is the industry's central cost lever, since glass is sold by the container but bought by the ton.

Volume breakpoints: hand blowing suits 1 to a few thousand pieces. Machine forming needs hundreds of thousands to millions to justify mold sets and line time.

1. In hand work, design for the fewest separate gathers and applied elements.
2. Keep sections uniform to shorten annealing cycles.
3. In container design, reduce weight rather than dimensions; glass is bought by mass.
4. Reuse existing finish (neck) geometry — it is the most tightly controlled and most expensive part of a container mold set.
5. Maximize cullet content; remelting scrap glass uses less energy than melting fresh batch.

## FAQ

### Why must blown glass be annealed?

Glass cooling freely develops large differential stresses between its surface and interior, and those stresses stay locked in. Annealing holds the piece near its annealing point — roughly 900–1,000 °F (480–540 °C) for soda-lime glass — and cools it slowly through the strain point so the stress relaxes. Unannealed glass can shatter days or weeks later without being touched.

### What temperature is glass worked at?

Soda-lime glass is melted around 2,000–2,400 °F (1,100–1,300 °C) and worked in a window of roughly 1,600–2,000 °F (870–1,100 °C). Glass has no melting point, only a viscosity that falls continuously with temperature, so the whole process is a race against the piece stiffening.

### What is the difference between blow-and-blow and press-and-blow?

Both are machine container processes. Blow-and-blow forms the parison in the blank mold with compressed air and suits narrow-neck bottles. Press-and-blow presses the parison with a plunger, which gives better control of glass distribution and is used for wide-mouth jars and for lightweight narrow-neck containers.

### How fast can a container glass machine run?

A multi-section IS machine running double or triple gob produces several hundred containers a minute. Each section operates independently on the same gob feed, so the line continues running while an individual section is serviced.

### Can two different glasses be joined in one piece?

Only if their coefficients of thermal expansion match. Two glasses with different expansion joined while hot will pull against each other as they cool and crack the joint, sometimes immediately and sometimes weeks later. This is why art glass is specified and sold by COE.

### Why is glass rotated constantly while being worked?

At working temperature glass flows under its own weight. Continuous rotation of the blowpipe is what keeps the gather symmetric about the axis; stop rotating for a moment and the glass sags to one side. It is also why blown forms are naturally bodies of revolution.

## Alternative processes

- [Blow Molding](https://manufacturingprocesses.org/processes/forming/blow-molding.md): Blow molding inflates a heated plastic tube or preform against the walls of a mold with compressed air, producing hollow one-piece containers and tanks.
- [Lampworking](https://manufacturingprocesses.org/processes/forming/lampworking.md): Lampworking softens glass rod and tube in a bench torch and forms it directly, producing scientific glassware, neon tube and small artwork.
- [Rotation Molding](https://manufacturingprocesses.org/processes/forming/rotation-molding.md): Rotational molding tumbles powdered polymer inside a heated mold on two axes so it fuses into a seamless, stress-free hollow part with a uniform wall.
- [Ceramic Slip Casting](https://manufacturingprocesses.org/processes/forming/ceramic-slip-casting.md): Ceramic slip casting pours liquid clay slip into a porous plaster mold that draws off water, leaving a solid layer that becomes the part.
- [Injection Molding](https://manufacturingprocesses.org/processes/forming/injection-molding.md): Injection molding forces molten thermoplastic into a steel mold under high pressure, producing high-volume identical plastic parts.

## Related processes

- [Lampworking](https://manufacturingprocesses.org/processes/forming/lampworking.md): Lampworking softens glass rod and tube in a bench torch and forms it directly, producing scientific glassware, neon tube and small artwork.
- [Clay Throwing](https://manufacturingprocesses.org/processes/forming/clay-throwing.md): Clay throwing forms a rotationally symmetric vessel by hand from a spinning ball of plastic clay on a potter's wheel.
- [Blow Molding](https://manufacturingprocesses.org/processes/forming/blow-molding.md): Blow molding inflates a heated plastic tube or preform against the walls of a mold with compressed air, producing hollow one-piece containers and tanks.
- [Glass Scoring](https://manufacturingprocesses.org/processes/cutting/glass-scoring.md): Glass scoring runs a hardened wheel across the surface to create a controlled fracture line, then breaks the sheet cleanly along it.
- [Press Molding Ceramics](https://manufacturingprocesses.org/processes/forming/press-molding-ceramics.md): Press molding ceramics forms clay between a die and a rotating profile tool or matched mold, producing flatware and repeat shapes quickly.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/forming/glassblowing)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Heat Treatment"
category: "Forming"
subcategory: "Metal"
materials: ["Metal"]
volumes: "1 piece to millions; furnace loads are priced by weight, so batching drives the unit cost"
lead_time: "3–10 business days for standard atmosphere hardening and tempering; longer for carburizing, and 1–3 weeks for long nitriding cycles"
url: https://manufacturingprocesses.org/processes/forming/heat-treatment
---

# Heat Treatment

Heat treatment cycles metal through controlled heating and cooling to change its hardness, strength, ductility or internal stress state.

- **Category**: [Forming](https://manufacturingprocesses.org/processes/forming.md)
- **Family**: Metal
- **Materials**: Metal
- **Typical volumes**: 1 piece to millions; furnace loads are priced by weight, so batching drives the unit cost
- **Lead time**: 3–10 business days for standard atmosphere hardening and tempering; longer for carburizing, and 1–3 weeks for long nitriding cycles

## Overview

Heat treatment changes a metal's properties without changing its shape, by cycling it through controlled heating, holding, and cooling. In steel it works by manipulating phase transformations: heating above the austenitizing temperature dissolves carbon into solution, and how fast the part is then cooled determines whether it ends up soft pearlite or hard martensite.

The lever is enormous. The same 4140 bar can be delivered soft enough to machine at roughly 20 HRC or hardened and tempered to over 50 HRC, from identical chemistry. Aluminum alloys gain most of their strength from solution treatment and aging: 6061 is solution treated near 985°F (530°C), quenched, and aged around 350°F (175°C) to reach T6.

It is a service operation, not a shaping process, and it sits between machining steps. Everything about designing for heat treatment is about the two side effects: distortion, and dimensional change.

## How it works

### The main cycles

**Annealing** heats steel above its transformation range and cools it slowly in the furnace, producing the softest, most machinable, most ductile condition.

**Normalizing** austenitizes and then air cools, giving a uniform fine grain structure and relieving the effects of prior forging or casting.

**Stress relieving** heats below the transformation range — commonly 1,100–1,250°F (595–675°C) for steel — and cools slowly, relaxing residual stress from machining, welding, or forming without changing the microstructure. It is what you do between roughing and finishing on a part that must stay flat.

**Hardening and tempering** is the two-step production sequence. Medium-carbon and alloy steels are austenitized around 1,500–1,600°F (815–870°C), quenched in water, oil, polymer, or gas, and then immediately tempered. As-quenched martensite is hard and brittle; tempering trades hardness for toughness in a predictable way. Tempering 4140 near 400°F leaves it in the mid-50s HRC, while tempering near 1,000–1,200°F brings it down toward the low 30s.

**Case hardening** puts a hard skin on a tough core. Carburizing diffuses carbon into low-carbon steel at roughly 1,650–1,750°F (900–955°C), then quenches, producing a surface around 58–62 HRC over a case typically 0.010–0.060 in (0.25–1.5 mm) deep. Nitriding diffuses nitrogen at 950–1,050°F (510–565°C) — below the transformation temperature, so there is no quench and very little distortion — giving a thinner case, typically 0.005–0.025 in (0.13–0.6 mm), at very high surface hardness on nitriding-grade alloys. Induction hardening heats only the surface with an induction coil and quenches immediately, which is how gear teeth and shaft journals are hardened locally.

**Solution treatment and aging** is the aluminum, 17-4PH stainless, and superalloy route: dissolve the strengthening elements at high temperature, quench to trap them in solution, then hold at a lower temperature so fine precipitates form.

### Why hardenability matters

Only the metal that cools fast enough transforms to martensite. In a plain carbon steel, that is a thin skin on anything but a small section; alloying elements such as chromium, molybdenum, and nickel slow the transformation and let thicker sections harden through. This is why 4140 and 4340 exist, and why section size has to be part of the specification.

## Design guidelines

### Specify hardness properly

A drawing note reading "harden" is not a specification. State the hardness range (a band, not a single number), where it is measured, the test scale, and for case hardening the case depth and whether it is total or effective. Convert scales carefully — hardness taken on a thin case with a Rockwell C indenter is not valid.

### Leave grinding stock

Parts move during heat treatment. Rough machine, heat treat, then finish grind, leaving 0.010–0.020 in (0.25–0.5 mm) of stock on precision surfaces — more on long, thin, or asymmetric parts. Designing a part that must be finish machined before hardening is designing in a tolerance problem. See the [surface finish chart](/charts/surface-finish-chart) for what grinding delivers and [ISO 286 fits and tolerances](/charts/iso-286-fits-tolerances) for the fits those finish operations have to hit.

### Design for symmetry

Distortion is driven by uneven cooling and by residual stress released at temperature. Symmetric sections, uniform wall thickness, and generous radii all reduce it. Sharp internal corners are quench-crack initiation sites, and a part with a heavy hub and a thin web will distort no matter how carefully it is processed.

### Choose the quench for the geometry

Water and brine are fastest and cause the most distortion and the most quench cracks; oil is the general-purpose middle ground; gas quenching in a vacuum furnace is the gentlest and is standard for tool steels and precision parts. A part that will not survive an oil quench may need a more hardenable alloy so it can be quenched more slowly.

### Watch for decarburization and scale

Heating steel in air pulls carbon out of the surface, leaving a soft skin, and forms scale. Controlled-atmosphere, vacuum, and salt bath furnaces avoid it. If a part is processed in air, enough stock has to remain to machine the decarburized layer off.

### Account for size change

Through hardening changes dimensions — martensite occupies more volume than the structure it replaced — and carburizing adds carbon to the surface. Size change is repeatable for a given part and process and is normally absorbed in the finishing allowance, but it must be anticipated when heat treating a finished-size feature such as a bore.

| Consideration | Recommended practice | Why |
| --- | --- | --- |
| Hardness callout | Range, location, scale, and case depth | A single number is unverifiable |
| Grinding stock | 0.010–0.020 in (0.25–0.5 mm) | Parts distort; finishing follows hardening |
| Internal corners | Radiused, no sharp changes of section | Sharp corners initiate quench cracks |
| Section uniformity | Symmetric, uniform walls | Uneven cooling drives distortion |
| Alloy vs section size | Match hardenability to thickness | Only metal that cools fast enough hardens |
| Quench medium | Oil or gas for precision parts | Water quenching maximizes distortion risk |
| Furnace atmosphere | Controlled atmosphere or vacuum | Air heating decarburizes the surface |
| Tempering | Immediately after quench | As-quenched martensite is brittle and cracks |

## Cost drivers

Heat treatment is priced by furnace time and load weight, so it is a batch economy: a full basket costs little more to run than a half-empty one. What drives cost is the process, not the part. Atmosphere-controlled batch and belt furnace work is the commodity end; vacuum and salt bath processing, long carburizing cycles, and nitriding runs measured in tens of hours cost substantially more per pound.

The larger cost is usually indirect. Distortion means extra grinding stock and an extra finishing operation. Scrap from quench cracking is total — a cracked part cannot be reworked. And heat treatment is almost always an outside service, so it inserts shipping time and a queue into the middle of the manufacturing route.

1. **Batch the work.** Furnace charges are priced by load; combining part numbers with the same cycle spreads the setup.
2. **Pick the least aggressive process that meets the spec.** Induction hardening a journal is cheaper and lower-distortion than through hardening the whole shaft; nitriding avoids the quench entirely.
3. **Design out the distortion instead of grinding it out.** Symmetry and uniform sections cost nothing at design time and save a finishing operation on every part.

## FAQ

### What is the difference between hardening and tempering?

Hardening austenitizes the steel — typically 1,500–1,600°F (815–870°C) for medium-carbon and alloy grades — and quenches it to form martensite, which is hard but brittle. Tempering then reheats below the transformation range to trade some hardness for toughness. The two are always done together; as-quenched parts crack.

### How much distortion should I allow for heat treatment?

Enough to grind out. Leave 0.010–0.020 in (0.25–0.5 mm) of stock on precision surfaces, and more on long, thin, or asymmetric parts. Rough machine, heat treat, then finish grind. A part that must be finish machined before hardening will not hold tolerance.

### What is the difference between carburizing and nitriding?

Carburizing diffuses carbon at 1,650–1,750°F (900–955°C) and then quenches, giving a case typically 0.010–0.060 in deep at roughly 58–62 HRC, with real distortion from the quench. Nitriding diffuses nitrogen at 950–1,050°F (510–565°C) below the transformation temperature, so there is no quench, distortion is minimal, and the case is thinner at 0.005–0.025 in.

### Why does alloy steel harden deeper than plain carbon steel?

Because alloying elements such as chromium, molybdenum, and nickel slow the transformation from austenite, so martensite can form at slower cooling rates. Plain carbon steel only hardens where it cools very fast, which on any substantial section is a thin skin. This is why section thickness has to be considered when selecting the grade.

### How should hardness be specified on a drawing?

As a range rather than a single value, with the test scale, the location of measurement, and — for case hardening — the case depth and whether it is total or effective. Hardness measured on a thin case with a Rockwell C indenter is invalid because the indenter punches through into the soft core.

### What causes quench cracking?

Sharp internal corners, abrupt section changes, an overly aggressive quench medium, and delaying the temper after quenching. Water and brine quench fastest and crack most; oil is the general-purpose choice; vacuum gas quenching is gentlest. A cracked part cannot be reworked, so the design changes are worth making up front.

## Alternative processes

- [PVD Coating](https://manufacturingprocesses.org/processes/finishing/pvd-coating.md): PVD coating vaporises a solid target inside a vacuum chamber so it condenses on the part as a very thin, extremely hard film.
- [Electroplating](https://manufacturingprocesses.org/processes/finishing/electroplating.md): Electroplating deposits a metal coating from solution onto a conductive part by passing current through an electrolyte bath.
- [Black Oxide](https://manufacturingprocesses.org/processes/finishing/black-oxide.md): Black oxide converts the steel surface to magnetite in a hot alkaline bath, giving a black finish that adds essentially no thickness.
- [Passivation](https://manufacturingprocesses.org/processes/finishing/passivation.md): Passivation removes free iron from a stainless steel surface with an acid bath so the chromium oxide layer can reform unbroken.

## Related processes

- [Forging](https://manufacturingprocesses.org/processes/forming/forging.md): Forging deforms solid metal under compressive load in a hammer, press or dies, so the grain flow follows the part contour and fatigue strength rises.
- [Metal Injection Molding](https://manufacturingprocesses.org/processes/forming/metal-injection-molding.md): Metal injection molding molds a powder-and-binder feedstock exactly like a plastic, then debinds and sinters it into a dense small metal part.
- [Powder Metallurgy (Press and Sinter)](https://manufacturingprocesses.org/processes/forming/powder-metallurgy-press-and-sinter.md): Press-and-sinter powder metallurgy compacts metal powder in a rigid die and sinters it below its melting point into a finished net-shape part.
- [Investment Casting](https://manufacturingprocesses.org/processes/forming/investment-casting.md): Investment casting builds a ceramic shell around a wax pattern, melts the wax out and pours metal into the cavity, giving fine detail in almost any alloy.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/forming/heat-treatment)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Hydroforming"
category: "Forming"
subcategory: "Metal"
materials: ["Metal"]
tolerances: "Roughly ±0.010–0.030 in (±0.25–0.75 mm) on formed features; sections calibrated hard against the die hold better than regions that are only partly expanded"
volumes: "1,000–250,000 parts per year; above roughly 500,000, stamping's cycle time wins"
lead_time: "8–16 weeks for die design, build, and process development; 2–6 weeks per production run thereafter"
url: https://manufacturingprocesses.org/processes/forming/hydroforming
---

# Hydroforming

Hydroforming presses sheet or tube into a single-sided die with high-pressure fluid, producing smooth seamless shapes that would need several stampings.

- **Category**: [Forming](https://manufacturingprocesses.org/processes/forming.md)
- **Family**: Metal
- **Materials**: Metal
- **Typical tolerances**: Roughly ±0.010–0.030 in (±0.25–0.75 mm) on formed features; sections calibrated hard against the die hold better than regions that are only partly expanded
- **Surface finish**: Retains the incoming tube or sheet finish, since fluid pressure produces no sliding contact against a die half
- **Typical volumes**: 1,000–250,000 parts per year; above roughly 500,000, stamping's cycle time wins
- **Lead time**: 8–16 weeks for die design, build, and process development; 2–6 weeks per production run thereafter

## Overview

Hydroforming replaces one half of a matched forming die with pressurized fluid. Sheet or tube is placed against a single-sided tool and hydraulic pressure — commonly 10,000–20,000 psi (700–1,400 bar) in tube work — expands the metal into the die until it takes the tool's shape exactly.

Because only one die half is cut, tooling is cheaper and faster than matched stamping dies, and because the fluid applies pressure uniformly rather than through a rigid punch, the metal thins more evenly and the surface stays unmarked. A single hydroformed part often replaces a two-piece stamped-and-welded assembly, removing the weld flange, the weld, and the tolerance stack across it.

Typical work is exhaust manifolds, engine cradles, radiator supports, bicycle and motorcycle frame members, structural nodes, and — on the sheet side — kitchen sinks and appliance panels. Cycle times of 30 seconds to a few minutes make it a mid-volume process: roughly 1,000 to 250,000 parts a year.

## How it works

### Tube hydroforming

1. **Pre-bend.** The tube is bent to approximately the part's centerline path on a rotary draw bender, since hydroforming expands a section but cannot create large bends.
2. **Load and seal.** The tube is placed in the closed die and axial punches seal both ends and inject fluid.
3. **Pressurize with axial feed.** Internal pressure rises while the end punches push material inward along the tube axis. This feed is the critical variable: without it, expansion comes entirely from thinning the wall and the tube bursts. With it, material is fed into the expanding region and thinning stays manageable.
4. **Calibration.** A final pressure spike forces the metal into the sharpest corners of the die. Required pressure rises steeply as the target corner radius falls, so sharp corners drive the machine tonnage and the die cost.
5. **Depressurize, open, and pierce.** Holes can be punched from inside the part with the fluid acting as the die, which eliminates separate piercing operations.

### Sheet hydroforming

A blank is clamped over a single-sided punch or cavity, and a fluid-filled bladder or chamber presses it into shape. Fluid-cell presses form shallow panels against a simple form block; deep-draw hydroforming uses a pressurized chamber as the "die" against a conventional punch, allowing draw ratios beyond what a matched die achieves because the fluid pressure holds the blank against the punch and reduces the tendency to thin at the nose.

Because pressure acts normal to the surface everywhere at once, there is no sliding contact against a hard die half, so pre-finished and coated sheet survives forming better than in conventional drawing.

## Design guidelines

### Design to the perimeter, not the shape

The governing constraint in tube hydroforming is that the tube's circumference must reach the perimeter of every section along the part. Expansion of roughly 20–30% in circumference is achievable in low-carbon steel, more in annealed stainless and some aluminum tempers, less in high-strength grades. Sections that need more than that will burst regardless of pressure. Start from the largest required perimeter and work backward to the starting tube.

### Corner radii cost pressure

The pressure required to fill a corner scales inversely with corner radius. Generous corner radii — several times the wall thickness — form at moderate pressure; sharp corners force a calibration spike that drives press tonnage, die strength, and cost. Never specify a sharp corner for cosmetic reasons on a hydroformed part.

### Feed the expansion

Design so material can be fed axially from the tube ends. Expansion in the middle of a long part far from either end has no feed path and must come entirely from wall thinning.

### Expect variable wall thickness

Wall thins in the expanded regions and can thicken slightly where feed accumulates. A hydroformed part is not a constant-wall part; do not dimension it as one, and place welds and highly loaded features away from the thinnest sections.

### Pre-bend before you form

Large bends come from a bending operation before hydroforming. Keep bend radii within normal rotary draw practice, and remember that pre-bending already thins the outside of each bend — that thinning is then compounded by the expansion.

### Sheet hydroforming rules follow deep drawing

Draft, corner radii, and depth limits track conventional draw practice, with the fluid allowing somewhat deeper draws. Blank thickness from the [sheet metal gauge chart](/charts/sheet-metal-gauge-chart); flange and secondary bend development from the [bend radius and K-factor chart](/charts/sheet-metal-bend-radius-k-factor).

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Circumferential expansion | 10–20% | 20–30% in mild steel | Beyond the limit the tube bursts at any pressure |
| Corner radius | ≥ 4× wall thickness | 2× wall thickness | Calibration pressure rises steeply as radius falls |
| Axial feed | Design a feed path from both ends | No feed means pure thinning | Feed is what limits wall loss during expansion |
| Wall thinning | Design for 10–20% | 25% | Expansion draws material from the wall |
| Pre-bend radius | Standard rotary draw practice | — | Pre-bend thinning compounds with expansion thinning |
| Part length | Moderate | Long parts starve mid-span feed | Feed comes from the ends only |
| Pierced holes | Pierce in the die | — | Internal pressure acts as the piercing die |

## Cost drivers

Hydroforming's economic argument is tooling: one die half instead of two, and one part instead of a stamped-and-welded assembly. Against that, cycle time is long by pressing standards — 30 seconds to a few minutes versus a fraction of a second on a progressive die — and the press itself, with its high-pressure intensifiers and heavy clamping, is expensive capital that shows up in the hourly rate.

Volume breakpoints: below roughly 1,000 parts, conventional bending and welding usually still wins because setup dominates. From 5,000 to 250,000 a year, hydroforming is at its best, particularly where it eliminates welded joints. Above roughly 500,000, high-speed stamping's cycle time advantage overtakes the tooling saving.

1. **Consolidate assemblies.** The real saving is removing weld flanges, welding labor, and the fixture that held the assembly, not the forming step itself.
2. **Open the corner radii.** Corner radius drives calibration pressure, which drives press size, die cost, and cycle time simultaneously.
3. **Keep expansion modest.** A part designed for 15% expansion runs reliably; one at 28% runs at the edge of the material's limit with a scrap rate to match.
4. **Pierce in the die.** In-die piercing removes an entire downstream operation and its fixture.
5. **Start from a standard tube size.** Custom-drawn starting tube carries a mill minimum and a lead time that can exceed the tooling.

## FAQ

### How much can a tube expand during hydroforming?

Roughly 20–30% in circumference for low-carbon steel, more for annealed stainless and some aluminum tempers, and less for high-strength grades. Beyond that limit the tube bursts no matter how much pressure is applied, so the part's largest section perimeter has to be planned against the starting tube circumference.

### What pressure does hydroforming use?

Tube hydroforming commonly runs 10,000–20,000 psi (700–1,400 bar), with a higher calibration spike at the end of the cycle to force metal into the sharpest die corners. Sheet hydroforming generally works at lower pressures. Required pressure rises steeply as the target corner radius falls.

### Why is axial feed important in tube hydroforming?

Because expansion has to get its material from somewhere. Punches at both tube ends push material inward as pressure rises, feeding the expanding region. Without feed, all the extra circumference comes out of wall thickness, and the tube thins until it splits.

### Is hydroforming tooling cheaper than stamping tooling?

Generally yes, because only one die half is cut instead of a matched pair, and because a single hydroformed part often replaces a stamped-and-welded assembly along with its weld fixture. The offset is a long cycle time and an expensive high-pressure press, which raises the hourly rate.

### What volume suits hydroforming?

Roughly 1,000 to 250,000 parts a year. Below that, bending and welding usually wins on setup cost. Above roughly 500,000, high-speed stamping's fraction-of-a-second cycle overtakes hydroforming's tooling advantage.

### Does a hydroformed part have uniform wall thickness?

No. The wall thins in expanded regions — typically 10–25% — and can thicken slightly where axial feed accumulates. Pre-bending before hydroforming adds its own thinning on the outside of each bend. Place welds and highly loaded features away from the thinnest sections.

## Alternative processes

- [Deep Drawing](https://manufacturingprocesses.org/processes/forming/deep-drawing.md): Deep drawing pushes a sheet metal blank through a die with a punch so the metal flows into a seamless cup or box deeper than its own diameter.
- [Metal Stamping](https://manufacturingprocesses.org/processes/forming/metal-stamping.md): Metal stamping shears and forms sheet metal between matched dies in a press, producing flat and shallow-formed parts at very high rates.
- [Superforming](https://manufacturingprocesses.org/processes/forming/superforming.md): Superforming heats a superplastic aluminum alloy sheet and forms it with gas pressure over a single-sided tool, producing deep, complex panels.
- [Metal Spinning](https://manufacturingprocesses.org/processes/forming/metal-spinning.md): Metal spinning presses a rotating metal disc against a mandrel with a roller or hand tool, forming it into an axially symmetric hollow shape.

## Related processes

- [Deep Drawing](https://manufacturingprocesses.org/processes/forming/deep-drawing.md): Deep drawing pushes a sheet metal blank through a die with a punch so the metal flows into a seamless cup or box deeper than its own diameter.
- [Metal Stamping](https://manufacturingprocesses.org/processes/forming/metal-stamping.md): Metal stamping shears and forms sheet metal between matched dies in a press, producing flat and shallow-formed parts at very high rates.
- [Tube and Section Bending](https://manufacturingprocesses.org/processes/forming/tube-and-section-bending.md): Tube and section bending draws or presses metal tube and profile around a former, with an internal mandrel wherever the wall would otherwise collapse.
- [Press Braking](https://manufacturingprocesses.org/processes/forming/press-braking.md): Press braking bends sheet metal by driving a punch into a V-die, forming one straight-line bend at a time.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/forming/hydroforming)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Injection Molding"
category: "Forming"
subcategory: "Plastics and Rubber"
materials: ["Plastic"]
tolerances: "±0.005 in (±0.127 mm) commercial on features under 1 in; ±0.002 in (±0.05 mm) precision; ±0.1–0.2% of dimension on larger parts. Add tolerance for anything spanning the parting line or a side action."
volumes: "1,000–1,000,000+ parts; economics are strongest above 10,000"
lead_time: "2–4 weeks for an aluminum bridge tool, 6–12 weeks for a hardened steel production tool; parts in hours to days once the tool is sampled"
url: https://manufacturingprocesses.org/processes/forming/injection-molding
---

# Injection Molding

Injection molding forces molten thermoplastic into a steel mold under high pressure, producing high-volume identical plastic parts.

- **Category**: [Forming](https://manufacturingprocesses.org/processes/forming.md)
- **Family**: Plastics and Rubber
- **Materials**: Plastic
- **Typical tolerances**: ±0.005 in (±0.127 mm) commercial on features under 1 in; ±0.002 in (±0.05 mm) precision; ±0.1–0.2% of dimension on larger parts. Add tolerance for anything spanning the parting line or a side action.
- **Surface finish**: SPI A-1 mirror (about 0.5 µin / 0.012 µm Ra) through SPI D-3 dry blast (about 125 µin / 3.2 µm Ra), plus MT-series molded textures
- **Typical volumes**: 1,000–1,000,000+ parts; economics are strongest above 10,000
- **Lead time**: 2–4 weeks for an aluminum bridge tool, 6–12 weeks for a hardened steel production tool; parts in hours to days once the tool is sampled

## Overview

Injection molding forces molten thermoplastic into a closed steel mold under high pressure, holds it while it cools, and ejects a finished part — typically on a 15–60 second cycle. It is the default process for plastic parts made in the thousands to the millions: enclosures, connectors, closures, gears, medical disposables, and most molded consumer goods. Nearly any thermoplastic runs on it — ABS, PP, PC, nylon, acetal, PEEK — plus glass- and mineral-filled compounds, and thermosets on modified machines.

The economics reduce to one number: the mold. Tooling is a multi-week capital item; after it exists, the marginal cost of a part is measured in cents. Below roughly 1,000 parts that trade rarely pays, and above 10,000 nothing else competes on unit cost. Commercial tolerances run about ±0.005 in (±0.127 mm) on molded features, with ±0.002 in (±0.05 mm) achievable on well-controlled dimensions in a hardened precision tool.

## How it works

1. **Dry the resin.** Hygroscopic polymers — nylon, PC, PET, ABS, PBT — are dried 2–4 hours at 175–250 °F (80–120 °C) to well under 0.02% moisture. Wet resin hydrolyzes in the barrel, showing up as silver streaking and a loss of impact strength that no process adjustment recovers.

2. **Plasticize.** A reciprocating screw melts pellets through barrel heat and shear, then retracts as it meters a shot. Typical melt temperatures: PP 400–480 °F (200–250 °C), ABS 410–480 °F (210–250 °C), nylon 500–550 °F (260–290 °C), PC 540–610 °F (280–320 °C).

3. **Inject.** The screw drives forward as a ram and fills the cavity in roughly 0.5–4 seconds. Injection pressure at the nozzle commonly runs 5,000–20,000 psi (35–140 MPa). Clamp tonnage is sized from projected area — a working rule is 2–5 tons per square inch, at the higher end for stiff, low-flow resins and thin walls.

4. **Pack and hold.** At about 95–98% fill the machine switches from velocity to pressure control and packs additional melt in to compensate for cooling shrinkage, holding until the gate freezes off. This phase, not the fill, determines sink marks, dimensional repeatability, and residual stress.

5. **Cool.** Coolant circulates through the mold, held at 60–140 °F (15–60 °C) for commodity resins and 180–250 °F (80–120 °C) for PC, PPS, and PEEK.

### Why is cooling most of the cycle?

Cooling time scales with the square of the wall thickness, because heat has to conduct out through that wall. It is normally 50–80% of total cycle time. Going from a 2.5 mm wall to 2.0 mm cuts the cooling term by roughly a third — far more leverage than any barrel-temperature tweak.

6. **Eject.** The mold opens and ejector pins, sleeves, or a stripper plate push the part off the core. Draft is what makes this possible; parts with no draft gall and scuff on the way out.

7. **Degate and regrind.** Cold sprues and runners are trimmed and can usually be reground and blended back at 10–25%. Hot runner systems eliminate that scrap entirely at higher tooling cost.

### Which variants of the process matter?

**Moldflow analysis** simulates fill, pack, cooling, and warpage before steel is cut, and is standard practice on any tool with a long flow path or a tight flatness requirement. **Gas-assisted injection molding** injects nitrogen into a partly filled cavity to hollow out thick sections, which removes sink marks from handles and structural ribs and cuts material and cycle. **Multishot (two-shot) molding** runs two injection units into a rotating tool to combine two resins or colors in one part. **In-mold decoration** places a printed film in the cavity so graphics are molded in, permanently, rather than printed on afterward.

## Design guidelines

### Wall thickness

Keep it uniform and thin. Most thermoplastics mold well between 0.040 and 0.140 in (1.0–3.5 mm): ABS 0.045–0.140 in (1.14–3.56 mm), PC 0.040–0.150 in (1.02–3.81 mm), PP 0.025–0.150 in (0.64–3.81 mm). Where thickness must change, blend over at least 3× the step rather than jumping. Per-polymer minimums and maximums are tabulated on [/charts/injection-molding-design-guidelines](/charts/injection-molding-design-guidelines).

### Draft

Use 1–2° per side as the untextured default. Shutoffs and features under 1 mm tall can go down to 0.25–0.5°. Ribs and internal bosses need 0.5–1.5°. Textured surfaces need much more: 3° for light MT-11010/11020 textures, 3–5° for medium texture, and 5–7° for heavy leather grain. The shop rule is to add roughly 1.5° per 0.001 in (0.025 mm) of texture depth on top of the base draft.

### Corners and radii

Internal corner radius ≥ 0.5 × nominal wall; external radius = internal radius + wall, which keeps the wall uniform through the corner. Sharp internal corners are the most common cause of field cracking: at R/T = 0.5 the stress concentration factor sits near 1.5, but at R/T = 0.1 it climbs past 3.

### Ribs

Rib base thickness 0.5–0.6 × wall, dropping to 0.4–0.5 × wall behind a cosmetic or textured face. Height ≤ 3 × wall. Center-to-center spacing ≥ 2 × wall. Base fillet 0.25–0.5 × wall. Ribs stiffen far more efficiently than thick walls, because bending stiffness goes with the cube of section depth.

### Bosses

Boss outside diameter 2–2.5 × the screw's major diameter; boss wall 0.5–0.6 × nominal wall. Never blend a boss straight into a side wall — connect it with a rib or gusset so you do not create a heavy junction that sinks. Core the boss all the way to the base.

### Holes and cores

Keep hole-to-hole and hole-to-edge spacing ≥ 2 × hole diameter. Blind core pins should stay under about 3:1 length-to-diameter; a core shut off at both ends can run to about 5:1. Beyond that the pin deflects and the hole comes out tapered or out of position.

### Undercuts and threads

Every side action, lifter, or unscrewing core adds tooling cost and a tolerance stack. Shallow undercuts in flexible resins can often be bumped off with no mechanism at all. External threads split cleanly across the parting line; internal threads need an unscrewing or collapsible core, or a molded-in insert.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Nominal wall | 0.060–0.120 in (1.5–3.0 mm) | 0.020–0.200 in (0.5–5 mm) | Thin fills fast and cools fast; thick sinks and voids |
| Wall transition | Taper over 3× the step | Abrupt step | Sudden changes freeze flow and warp the part |
| Draft, untextured | 1.5° per side | 0.25° at shutoffs | Below this the part drags and scuffs on ejection |
| Draft, textured | 3° light, 5–7° heavy grain | — | About 1.5° per 0.001 in of texture depth |
| Internal radius | 0.5 × wall | 0.25 × wall | Stress concentration climbs sharply below 0.5 T |
| Rib thickness | 0.5 × wall | 0.6 × wall | Above 0.6 T a sink mark shows on the opposite face |
| Rib height | 2.5 × wall | 3 × wall | Taller ribs are hard to fill and to eject |
| Boss OD | 2 × screw major dia. | 2.5 × screw major dia. | Resists hoop stress from a thread-forming screw |
| Hole spacing | 2 × hole diameter | — | Thin webs between holes are weak and hard to fill |

## Variants

- Moldflow analysis
- Gas-assisted Injection Molding
- Multishot Injection Molding
- In-Mold Decoration

## Cost drivers

Tooling dominates everything else. A single-cavity aluminum bridge tool, a single-cavity P20 steel tool, and a 16-cavity hardened H13 tool with a hot runner differ by more than an order of magnitude in both price and break-even volume. Per-part cost is then machine rate × cycle time ÷ cavity count, plus material, plus scrap.

Practical breakpoints:

- Under 100 parts: urethane casting or additive — no mold pays back.
- 100–1,000: aluminum bridge tooling, single cavity.
- 1,000–100,000: P20 or NAK80 steel, one to four cavities.
- Over 100,000: hardened steel, high cavitation, hot runner, automated part removal.

Ways to take cost out:

1. **Thin the wall.** Cooling time goes with wall squared. Dropping a 2.5 mm wall to 2.0 mm cuts cooling roughly a third and takes material out at the same time.
2. **Kill the side actions.** Redesigning an undercut into a through-hole formed by a shutoff between core and cavity removes a slide, its maintenance, and its tolerance stack.
3. **Add cavities, not machines.** Once annual volume passes a few tens of thousands, a second or fourth cavity is almost always cheaper per part than more machine hours.
4. **Loosen non-functional tolerances.** Call out ±0.002 in only where it does something; ±0.010 in elsewhere lets the molder run a faster, cooler cycle.
5. **Texture instead of polish.** SPI A-series finishes are hand-polished with diamond compound and are among the most expensive line items on a mold quote. An MT texture hides sink and knit lines and costs far less.

## FAQ

### What draft angle do I need for injection molding?

Use 1–2° per side on untextured surfaces as a default, and 0.5–1.5° on ribs and internal bosses. Textured surfaces need substantially more: about 3° for a light MT texture and 5–7° for heavy leather grain. The working rule is to add roughly 1.5° of draft for every 0.001 in (0.025 mm) of texture depth.

### What is the minimum wall thickness for injection molding?

For most thermoplastics the practical range is 0.040–0.140 in (1.0–3.5 mm). High-flow resins such as polypropylene, acetal, and LCP fill down to about 0.025–0.030 in (0.64–0.76 mm), while low-flow polycarbonate wants at least 0.040 in (1.02 mm). Uniformity matters more than the absolute number.

### What tolerance can injection molding hold?

About ±0.005 in (±0.127 mm) is standard commercial practice on features under an inch, and ±0.002 in (±0.05 mm) is achievable on well-controlled dimensions in a hardened tool. On larger parts, think in percentage terms — ±0.1 to ±0.2% of the dimension — because polymer shrinkage, not machining, sets the variation.

### How thick can a rib be before it causes a sink mark?

Keep the rib base at 0.5–0.6 times the nominal wall thickness. Behind a cosmetic or gloss face, drop to 0.4–0.5 times the wall. Above 0.6 T the rib holds enough heat that the opposite surface pulls in as it cools, which reads as a visible sink line.

### How many parts do I need before injection molding makes sense?

Roughly 1,000 parts is the crossover against urethane casting and additive manufacturing, and above about 10,000 parts nothing else competes on unit cost. Below 1,000, the mold cost per part swamps the savings on the part itself.

### Why is cooling most of the injection molding cycle?

Heat has to conduct out through the part wall, so cooling time scales with the square of wall thickness. It typically accounts for 50–80% of a 15–60 second cycle. Reducing a 2.5 mm wall to 2.0 mm cuts the cooling term by roughly a third.

### Can injection molding produce threads?

Yes. External threads split across the parting line at no extra tooling complexity. Internal threads need an unscrewing core or a collapsible core, both of which add cost and maintenance — for low volumes a molded-in or heat-staked brass insert is usually cheaper and stronger.

## Alternative processes

- [Thermoforming](https://manufacturingprocesses.org/processes/forming/thermoforming.md): Thermoforming heats a plastic sheet until pliable and pulls or presses it over a single-sided mold, producing large thin-walled parts with low tooling cost.
- [Vacuum Casting (Urethane Casting)](https://manufacturingprocesses.org/processes/forming/vacuum-casting.md): Vacuum casting, known in the US as urethane casting, pours polyurethane resin into a silicone mold under vacuum to copy a master pattern for short runs.
- [CNC Machining](https://manufacturingprocesses.org/processes/cutting/cnc-machining.md): CNC machining removes material from solid stock with a programmed cutting tool, producing accurate parts directly from a CAD model with no tooling.
- [Die Casting](https://manufacturingprocesses.org/processes/forming/die-casting.md): Die casting injects molten aluminum, zinc or magnesium into a hardened steel die under high pressure, producing thin-walled net-shape metal parts.
- [Reaction Injection Molding](https://manufacturingprocesses.org/processes/forming/reaction-injection-molding.md): Reaction injection molding mixes two liquid reactants that polymerize inside a low-pressure mold, producing large tough polyurethane parts on light tooling.

## Related processes

- [Insert Molding](https://manufacturingprocesses.org/processes/forming/insert-molding.md): Insert molding loads metal inserts into the cavity and molds polymer around them, capturing threads, terminals or stampings in a single shot.
- [Overmolding](https://manufacturingprocesses.org/processes/forming/overmolding.md): Overmolding molds a second material, usually a soft elastomer, directly onto an already-molded rigid substrate so the two bond permanently.
- [Liquid Silicone Rubber (LSR) Molding](https://manufacturingprocesses.org/processes/forming/liquid-silicone-rubber-lsr-molding.md): LSR molding injects two-part liquid silicone into a heated mold where it cures, producing flexible parts with a very wide service temperature range.
- [Metal Injection Molding](https://manufacturingprocesses.org/processes/forming/metal-injection-molding.md): Metal injection molding molds a powder-and-binder feedstock exactly like a plastic, then debinds and sinters it into a dense small metal part.
- [Blow Molding](https://manufacturingprocesses.org/processes/forming/blow-molding.md): Blow molding inflates a heated plastic tube or preform against the walls of a mold with compressed air, producing hollow one-piece containers and tanks.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/forming/injection-molding)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Insert Molding"
category: "Forming"
subcategory: "Plastics and Rubber"
materials: ["Plastic", "Metal"]
tolerances: "Molded features hold normal injection molding tolerances, about ±0.005 in (±0.127 mm). Insert position relative to molded features is looser and depends on how positively the insert is located — budget the insert's own tolerance plus the locating clearance."
volumes: "1,000–1,000,000+ parts; below 1,000, post-mold insert installation is usually cheaper"
lead_time: "6–12 weeks for tooling including insert locating features and presence sensing; cycles run several seconds longer than the equivalent plain molding"
url: https://manufacturingprocesses.org/processes/forming/insert-molding
---

# Insert Molding

Insert molding loads metal inserts into the cavity and molds polymer around them, capturing threads, terminals or stampings in a single shot.

- **Category**: [Forming](https://manufacturingprocesses.org/processes/forming.md)
- **Family**: Plastics and Rubber
- **Materials**: Plastic, Metal
- **Typical tolerances**: Molded features hold normal injection molding tolerances, about ±0.005 in (±0.127 mm). Insert position relative to molded features is looser and depends on how positively the insert is located — budget the insert's own tolerance plus the locating clearance.
- **Surface finish**: SPI A-1 through D-3 on molded surfaces, as for any injection molded part; exposed insert faces retain their own finish
- **Typical volumes**: 1,000–1,000,000+ parts; below 1,000, post-mold insert installation is usually cheaper
- **Lead time**: 6–12 weeks for tooling including insert locating features and presence sensing; cycles run several seconds longer than the equivalent plain molding

## Overview

Insert molding loads a metal component into the mold cavity before the shot and injects plastic around it, so the finished part comes out of the press already assembled. Threaded brass inserts, stamped terminals, contact pins, shafts, bearings, and reinforcing plates are all routinely captured this way.

The value is assembly elimination. One molding operation replaces a molded part, a separate insert, a press or heat-stake station, and the labor and quality checks around them. It is the standard construction for electrical connectors, encapsulated sensors, gear-and-shaft assemblies, and any plastic housing that needs a metal thread capable of repeated disassembly.

The costs are equally concrete: an operator or robot has to load inserts every cycle, which lengthens the cycle and makes the mold vulnerable to a mis-set insert, and the mismatch in thermal expansion between metal and polymer must be designed around rather than ignored.

## How it works

1. **Prepare the inserts.** Inserts arrive clean and, for critical applications, degreased — mold release or drawing oil on an insert becomes a bond-line failure. Preheating inserts, typically to somewhere near the mold temperature, is common practice: it reduces the thermal shock as melt hits cold metal, which lowers residual stress in the surrounding plastic and improves encapsulation.

2. **Load and locate.** Inserts are placed onto core pins, magnets, vacuum cups, or shouldered pockets in the cavity. Location has to be positive, because injection at 5,000–20,000 psi (35–140 MPa) will move anything that is not properly restrained. This step is the reason insert molding cycles are longer than plain injection molding, and it is where robotic loading pays back.

3. **Close and verify.** Production tools normally carry insert-presence sensing. A missing insert leaves a core pin unsupported; a proud or tilted insert gets crushed when the mold closes, which can damage the cavity — an expensive failure mode compared with a short shot.

4. **Inject.** The melt flows around the insert and welds to itself on the far side, forming a knit line downstream of every insert. That knit line is a genuine weak point, so gate placement should push it away from any loaded region.

5. **Cool and eject.** The polymer shrinks onto the insert, which is exactly what generates the grip. Encapsulated knurls and undercut grooves convert that shrinkage into torque and pull-out resistance.

### What holds the insert in?

Three mechanisms, in order of importance. Shrinkage grip from the polymer contracting onto the metal, mechanical interlock from knurls and undercut grooves, and only rarely chemical adhesion. Straight knurls resist torque, annular undercut grooves resist pull-out, and a diamond knurl combined with a groove resists both — which is why standard molded-in inserts carry that geometry.

## Design guidelines

### Boss geometry around the insert

The boss outside diameter should be at least 2× the insert outside diameter. Too little plastic around the insert and the hoop stress from cooling shrinkage — plus the hoop stress from any screw driven into it — cracks the boss, often not immediately but weeks later in the field. Keep the boss wall at 0.5–0.6 × the nominal part wall, and connect it to adjacent walls with ribs or gussets rather than a thick blend, exactly as tabulated on [/charts/injection-molding-design-guidelines](/charts/injection-molding-design-guidelines).

### Thermal expansion mismatch

This is the failure mode most often missed. Steel expands at about 6.5 µin/in·°F (11.7 µm/m·K) and brass at about 10 µin/in·°F (19 µm/m·K), while unfilled thermoplastics run several times higher. Over a thermal cycle the polymer moves and the metal does not, which loads the interface every time. Keep encapsulating walls thick enough to carry that stress, avoid long metal inserts fully bonded along their length, and prefer glass-filled resins where the part sees wide temperature swings — the filler pulls the polymer's expansion much closer to the metal's.

### Insert support

Support every insert against the injection pressure on at least two features, and shut off cleanly on any surface that must stay free of plastic — a thread, a contact face, a sealing land. Flash on a molded-in thread is a scrap part; there is no rework.

### Keep inserts off the parting line

An insert that straddles the parting line will flash and will resist ejection unevenly. Set it fully within one half.

### Thread selection

Molded-in inserts are specified by the screw they accept, so start from the fastener. Thread sizes and their pilot dimensions are on [/charts/thread-size-chart](/charts/thread-size-chart) and [/charts/tap-drill-chart](/charts/tap-drill-chart); size the boss from the screw's major diameter, not from the insert catalog number.

### When not to insert mold

If the insert count is high, the parts are small, or the program is short, post-mold installation — heat staking or ultrasonic insertion — is often cheaper and lower risk. It keeps the molding cycle fast, keeps the tool simple, and lets the same tool run parts with and without inserts.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Boss OD | 2.5 × insert OD | 2 × insert OD | Below this the boss cracks from hoop stress |
| Boss wall | 0.5–0.6 × nominal wall | — | Same sink-mark rule as any molded boss |
| Insert location | Fully within one mold half | Never across the parting line | Straddling inserts flash and eject unevenly |
| Insert retention | Diamond knurl plus undercut groove | Knurl alone | Knurl resists torque, groove resists pull-out |
| Resin for thermal cycling | Glass-filled | Unfilled | Filler brings polymer expansion nearer the metal's |
| Knit line | Placed away from loaded regions | — | Melt rejoins downstream of every insert |

## Cost drivers

Insert molding trades assembly cost for cycle cost. The saving is real — one operation instead of three or four, no separate insertion press, no assembly inspection — but every second spent loading inserts is a second added to a cycle that would otherwise be 20–40 seconds. On a high-volume part, that arithmetic decides the process.

Volume breakpoints:

- Under 1,000 parts: mold plain and install inserts afterward by heat staking or ultrasonics. Tooling stays simple and there is no loading fixture to build.
- 1,000–100,000: manual insert loading with a shuttle or rotary table so the operator loads one half while the other runs.
- Over 100,000: robotic loading with insert-presence verification, which removes both the cycle penalty and the crushed-insert risk.

Cost reduction:

1. **Cut the insert count.** Every insert is loading time on every cycle. Two well-placed inserts often carry a joint that four were specified for.
2. **Use a shuttle or rotary tool.** Loading time overlaps the molding cycle instead of adding to it.
3. **Standardize on one insert size.** A single insert across a product family simplifies loading fixtures, reduces mis-set risk, and improves purchasing leverage.
4. **Add insert-presence sensing.** A crushed insert can damage a cavity, and the repair plus lost production dwarfs the cost of the sensor.
5. **Reconsider post-mold installation for short programs.** Heat-staked inserts reach most of the pull-out strength of molded-in ones without the cycle penalty or the tool risk.

## FAQ

### How much plastic do I need around a molded-in insert?

Design the boss outside diameter to at least twice the insert outside diameter, and keep the boss wall at 0.5–0.6 times the nominal part wall. Less material than that and hoop stress from cooling shrinkage plus screw installation cracks the boss, frequently weeks later in service rather than at assembly.

### What holds a metal insert in a molded part?

Mainly the polymer shrinking onto the metal as it cools, reinforced by mechanical interlock. Straight knurls resist torque, annular undercut grooves resist pull-out, and standard molded-in inserts combine a diamond knurl with a groove to resist both. Chemical adhesion between metal and thermoplastic contributes very little.

### Should I mold inserts in or install them afterward?

Mold them in above roughly 1,000 parts, where eliminating a separate insertion operation pays for the longer cycle. Below that, or when insert counts are high, heat staking or ultrasonic insertion after molding is usually cheaper: the tool stays simple, the cycle stays fast, and a mis-set insert cannot damage a cavity.

### Why does thermal expansion matter in insert molding?

Steel expands at about 6.5 µin/in·°F (11.7 µm/m·K) and brass around 10 µin/in·°F (19 µm/m·K), while unfilled thermoplastics expand several times faster. Every thermal cycle loads the plastic-to-metal interface. Glass-filled resins bring the polymer's expansion much closer to the metal's and are the standard answer for parts that see wide temperature swings.

### Where do knit lines form in an insert molded part?

Directly downstream of every insert, where the melt splits around the metal and rejoins on the far side. Knit lines carry a fraction of the base material's strength, so gate placement should push them away from loaded regions, mounting points, and anywhere the part sees impact.

### What happens if an insert is loaded incorrectly?

A proud or tilted insert gets crushed when the mold closes, which can damage the cavity steel — a repair far more expensive than the scrapped part. A missing insert leaves the core pin unsupported. Production tools therefore carry insert-presence sensing, which pays for itself the first time it prevents a cavity strike.

## Alternative processes

- [Staking](https://manufacturingprocesses.org/processes/joining/staking.md): Heat staking softens a molded plastic boss with hot air or ultrasonic energy and reforms it into a head that traps the mating part.
- [Mechanical Fastening](https://manufacturingprocesses.org/processes/joining/mechanical-fastening.md): Mechanical fastening joins parts with screws, inserts, clinch hardware or snap fits so the joint can be assembled and taken apart again.
- [Overmolding](https://manufacturingprocesses.org/processes/forming/overmolding.md): Overmolding molds a second material, usually a soft elastomer, directly onto an already-molded rigid substrate so the two bond permanently.
- [Injection Molding](https://manufacturingprocesses.org/processes/forming/injection-molding.md): Injection molding forces molten thermoplastic into a steel mold under high pressure, producing high-volume identical plastic parts.

## Related processes

- [Overmolding](https://manufacturingprocesses.org/processes/forming/overmolding.md): Overmolding molds a second material, usually a soft elastomer, directly onto an already-molded rigid substrate so the two bond permanently.
- [Injection Molding](https://manufacturingprocesses.org/processes/forming/injection-molding.md): Injection molding forces molten thermoplastic into a steel mold under high pressure, producing high-volume identical plastic parts.
- [Staking](https://manufacturingprocesses.org/processes/joining/staking.md): Heat staking softens a molded plastic boss with hot air or ultrasonic energy and reforms it into a head that traps the mating part.
- [Liquid Silicone Rubber (LSR) Molding](https://manufacturingprocesses.org/processes/forming/liquid-silicone-rubber-lsr-molding.md): LSR molding injects two-part liquid silicone into a heated mold where it cures, producing flexible parts with a very wide service temperature range.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/forming/insert-molding)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Investment Casting"
category: "Forming"
subcategory: "Metal"
materials: ["Metal"]
tolerances: "About ±0.005 in (±0.13 mm) on the first inch plus roughly ±0.003 in per additional inch; flatness and straightness are usually the limiting characteristics on long or thin parts"
volumes: "100–50,000 parts per year; single prototypes possible with printed wax patterns"
lead_time: "6–12 weeks for wax tooling and first article, then 4–8 weeks per production run; printed-wax prototypes in 2–4 weeks with no tool"
url: https://manufacturingprocesses.org/processes/forming/investment-casting
---

# Investment Casting

Investment casting builds a ceramic shell around a wax pattern, melts the wax out and pours metal into the cavity, giving fine detail in almost any alloy.

- **Category**: [Forming](https://manufacturingprocesses.org/processes/forming.md)
- **Family**: Metal
- **Materials**: Metal
- **Typical tolerances**: About ±0.005 in (±0.13 mm) on the first inch plus roughly ±0.003 in per additional inch; flatness and straightness are usually the limiting characteristics on long or thin parts
- **Surface finish**: Roughly 63–125 µin Ra (1.6–3.2 µm) as cast
- **Typical volumes**: 100–50,000 parts per year; single prototypes possible with printed wax patterns
- **Lead time**: 6–12 weeks for wax tooling and first article, then 4–8 weeks per production run; printed-wax prototypes in 2–4 weeks with no tool

## Overview

Investment casting — lost wax casting — builds a ceramic shell around an injection-molded wax pattern, melts the wax out, and pours metal into the resulting cavity. Because the pattern is destroyed rather than withdrawn, the process needs no draft and tolerates undercuts, so it produces detail no steel die could release.

It runs the full alloy range: carbon and stainless steels, tool steels, cobalt-chrome and nickel superalloys, titanium, aluminum, and bronze. Typical parts are turbine blades, orthopedic implants, firearm and valve components, impellers, and structural fittings from a few grams to about 100 lb (45 kg).

Standard linear tolerance is roughly ±0.005 in (±0.13 mm) on the first inch, and as-cast surfaces run 63–125 µin Ra — good enough that many features never see a cutting tool. The cost is a multi-step, labor-intensive shell build plus a wax injection tool, which keeps per-part cost comparatively flat as volume rises.

## How it works

1. **Wax tooling.** An aluminum wax injection die is cut, sized up to account for both wax and metal shrinkage — total allowance is commonly 1.5–2.5% depending on alloy and geometry.
2. **Pattern injection.** Wax is injected at low pressure and modest temperature, typically around 150–200°F (65–95°C). Internal passages come from soluble wax or preformed ceramic cores.
3. **Assembly.** Patterns are welded to a wax sprue with a hot knife to form a tree. Tree layout is a real engineering decision: it determines how metal feeds every part on it.
4. **Shell build.** The tree is dipped in ceramic slurry, drained, rained with refractory stucco, and dried. This repeats 5–8 times over roughly 24–48 hours. The first coat uses the finest flour — it is what reproduces surface detail — and later coats are coarser, for strength.
5. **Dewax.** The shell is flash-dewaxed in a steam autoclave, typically around 300°F (150°C) under pressure, so the wax melts and runs out faster than it can expand and crack the shell. Wax is reclaimed.
6. **Burnout and preheat.** The shell is fired at roughly 1,600–1,900°F (870–1,040°C), burning out residual wax, sintering the ceramic, and leaving the mold hot — which is exactly what lets sections down to 0.030 in (0.75 mm) fill.
7. **Pour.** Metal is poured into the hot shell, by gravity in air for most steels, or under vacuum for superalloys and titanium that would otherwise oxidize.
8. **Knockout and cutoff.** The shell is vibrated and blasted off, parts are cut from the tree with an abrasive wheel, and gate stubs are ground flush.
9. **Finishing.** Heat treatment, straightening, machining of critical features, and inspection — commonly radiographic or fluorescent penetrant on aerospace and medical work.

The hot-shell pour is the whole trick. Because the mold is near red heat when metal enters, fluidity stays high all the way to the last thin web, which is why investment casting fills sections that would misrun in sand.

## Design guidelines

### Walls

Minimum wall is roughly 0.030–0.060 in (0.75–1.5 mm) over short spans, with 0.090–0.125 in (2.3–3 mm) a comfortable nominal in steel. Large flat thin panels are harder than the numbers suggest — they distort during solidification — so break them up with ribs or a crown.

### Draft is optional, not mandatory

Because the wax pattern is melted out rather than pulled, zero draft is castable. Draft still helps in one place: pulling wax from the aluminum injection die. Give 0.5–1° where it is free, and accept that zero-draft or undercut features may need a split or soluble-core wax tool, which costs more.

### Fillets and corners

Use a minimum internal fillet of 0.030 in (0.75 mm), and 0.060 in (1.5 mm) or more at load-bearing junctions. Sharp internal corners crack the ceramic shell during dewax; sharp external corners on the wax pattern break off in handling.

### Holes and passages

Cast holes down to roughly 0.060 in (1.5 mm) diameter, with blind-hole depth under about 3–4× diameter. Anything smaller or deeper should be drilled. Long internal passages require ceramic cores that must be leached out afterward — a real cost and lead-time adder.

### Gates and witness marks

Every part carries at least one gate stub, ground flush and usually leaving a visible witness. Put the gate on a non-critical, non-cosmetic surface and agree it with the foundry early; moving a gate after tooling is expensive.

### Tolerances and datums

Standard tolerance is about ±0.005 in (±0.13 mm) on the first inch plus ±0.003 in per additional inch. Flatness and straightness are the weak axes: long or thin parts move during solidification and are often straightened afterward. Machine critical bores and mating faces rather than casting them to a tight tolerance, and locate on cast datum pads. For achievable machined finishes see the [surface finish chart](/charts/surface-finish-chart); for pour temperature planning see [metal melting points](/charts/metal-melting-points).

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Nominal wall, steel | 0.090–0.125 in (2.3–3 mm) | 0.030 in (0.75 mm) short spans | Sections fill because the shell is poured hot |
| Draft | 0.5–1° where free | 0° is castable | Only needed to pull wax from the injection die |
| Internal fillet | 0.060 in (1.5 mm) | 0.030 in (0.75 mm) | Sharp corners crack the ceramic shell at dewax |
| Cast hole diameter | ≥ 0.090 in (2.3 mm) | 0.060 in (1.5 mm) | Fine ceramic detail has to survive shell handling |
| Blind hole depth | ≤ 3× diameter | 4× diameter | Ceramic in a deep pocket is unsupported |
| Linear tolerance | ±0.005 in first inch | ±0.003 in per added inch | Wax and metal shrinkage stack across length |
| Thin flat panels | Add ribs or machine flat | — | Unsupported flats distort on solidification |

## Cost drivers

Three cost centers: the wax tool, the per-part shell build, and finishing. The wax injection die is a fraction of a die-casting die and can be soft aluminum, so entry cost is moderate. But unlike a die or permanent mold, the shell is consumed every cycle and takes 5–8 dip-and-dry passes across a day or two. That makes investment casting's per-part cost high and relatively volume-insensitive — going from 1,000 to 100,000 parts does not collapse unit cost the way it does in die casting.

Volume breakpoints: below roughly 100 parts, printed wax or printed shells skip the tool entirely and prototypes arrive in weeks. From 500 to 50,000 a year is the sweet spot for a hard wax tool. Above roughly 100,000 in an aluminum or zinc geometry, die casting almost always wins on cost.

1. **Get more parts per tree.** Tree density drives shell, metal, and labor cost per part. Compact parts that nest tightly cost less than sprawling ones.
2. **Avoid ceramic cores.** Leachable cores add tooling, cycle time, and a leaching operation. Redesign internal passages to be castable from outside, or drill them.
3. **Reduce gate count.** Each gate is a cutoff and a grind. One well-placed gate on a non-critical face beats three convenient ones.
4. **Cast near-net and machine only what matters.** The 63–125 µin Ra as-cast surface is acceptable for most non-sealing faces.
5. **Choose an air-melt alloy where you can.** Vacuum-melted superalloys and titanium carry a substantial premium over air-melted stainless and carbon steel.

## FAQ

### What tolerance can investment casting hold?

Roughly ±0.005 in (±0.13 mm) on the first inch, plus about ±0.003 in per additional inch — around six times tighter than green sand casting. Flatness and straightness on long thin parts are usually the harder characteristics and often need post-cast straightening.

### Does investment casting need draft?

No. The wax pattern is melted out of the ceramic shell rather than withdrawn, so zero-draft walls and undercuts are castable. Draft only helps when pulling the wax pattern from the aluminum injection die, and even that can be handled with a split or soluble wax tool at extra cost.

### What is the minimum wall thickness for investment casting?

About 0.030–0.060 in (0.75–1.5 mm) over short spans, with 0.090–0.125 in (2.3–3 mm) a comfortable nominal in steel. Thin sections fill because the ceramic shell is poured while still near 1,600–1,900°F, keeping the metal fluid to the end of the flow path.

### What metals can be investment cast?

Nearly all of them — carbon and stainless steels, tool steels, cobalt-chrome, nickel superalloys, titanium, aluminum, and copper alloys. Reactive alloys such as titanium and many superalloys are poured under vacuum, which adds cost relative to air-melted stainless.

### Why doesn't investment casting get much cheaper at high volume?

The ceramic shell is consumed on every cycle and takes 5–8 dip-and-dry coats over 24–48 hours to build. That per-part labor does not amortize with volume, so unit cost stays comparatively flat while die casting's falls. Above roughly 100,000 aluminum or zinc parts a year, die casting normally wins.

### Investment casting or machining from billet?

Machining wins for one-offs, simple prismatic shapes, and anything needing tolerances tighter than about ±0.003 in on every feature. Investment casting wins once the geometry has internal cavities, thin curved webs, or undercuts, and once volumes reach a few hundred, because it removes most of the material before the first cut.

## Alternative processes

- [Die Casting](https://manufacturingprocesses.org/processes/forming/die-casting.md): Die casting injects molten aluminum, zinc or magnesium into a hardened steel die under high pressure, producing thin-walled net-shape metal parts.
- [Sand Casting](https://manufacturingprocesses.org/processes/forming/sand-casting.md): Sand casting pours molten metal into a cavity formed in bonded sand around a pattern, then breaks the mold away to release the part.
- [CNC Machining](https://manufacturingprocesses.org/processes/cutting/cnc-machining.md): CNC machining removes material from solid stock with a programmed cutting tool, producing accurate parts directly from a CAD model with no tooling.
- [Direct Metal Laser Sintering (DMLS)](https://manufacturingprocesses.org/processes/forming/direct-metal-laser-sintering-dmls.md): Direct metal laser sintering fuses metal powder layer by layer with a laser, building dense metal parts anchored to a supported build plate.

## Related processes

- [Sand Casting](https://manufacturingprocesses.org/processes/forming/sand-casting.md): Sand casting pours molten metal into a cavity formed in bonded sand around a pattern, then breaks the mold away to release the part.
- [Die Casting](https://manufacturingprocesses.org/processes/forming/die-casting.md): Die casting injects molten aluminum, zinc or magnesium into a hardened steel die under high pressure, producing thin-walled net-shape metal parts.
- [Centrifugal Casting](https://manufacturingprocesses.org/processes/forming/centrifugal-casting.md): Centrifugal casting pours molten metal into a spinning mold so centrifugal force packs it against the wall and drives inclusions toward the bore.
- [Metal Injection Molding](https://manufacturingprocesses.org/processes/forming/metal-injection-molding.md): Metal injection molding molds a powder-and-binder feedstock exactly like a plastic, then debinds and sinters it into a dense small metal part.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/forming/investment-casting)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Lampworking"
category: "Forming"
subcategory: "Glass and Ceramics"
materials: ["Glass"]
volumes: "1–1,000 pieces; unit cost is flat with quantity because there is no tooling"
lead_time: "Hours to days per piece, plus a kiln annealing cycle sized to the thickest section. Complex laboratory assemblies are quoted in bench hours rather than lead weeks."
url: https://manufacturingprocesses.org/processes/forming/lampworking
---

# Lampworking

Lampworking softens glass rod and tube in a bench torch and forms it directly, producing scientific glassware, neon tube and small artwork.

- **Category**: [Forming](https://manufacturingprocesses.org/processes/forming.md)
- **Family**: Glass and Ceramics
- **Materials**: Glass
- **Typical volumes**: 1–1,000 pieces; unit cost is flat with quantity because there is no tooling
- **Lead time**: Hours to days per piece, plus a kiln annealing cycle sized to the thickest section. Complex laboratory assemblies are quoted in bench hours rather than lead weeks.

## Overview

Lampworking — also called flameworking or torchwork — softens glass rod and tube in a bench torch and forms it directly by hand, without a furnace. A surface-mix oxygen-propane or oxygen-natural-gas torch supplies a flame hot enough to bring a local region of the glass into its working range while the rest of the piece stays rigid, which is what makes precise, localized work possible.

The two commercial branches are scientific glassware and neon. Laboratory glass is worked almost exclusively in borosilicate, whose low thermal expansion lets it survive the thermal shock of the flame and of laboratory service; it softens around 1,510 °F (820 °C) and reaches a comfortable working viscosity near 2,280 °F (1,250 °C). Neon tube is bent from soda-lime or lead glass tubing of roughly 8–15 mm outside diameter, then electroded, evacuated, and backfilled.

The third branch is artwork — beads, sculpture, and borosilicate art — which shares the same tools. All of it requires kiln annealing: roughly 1,050 °F (565 °C) for borosilicate and 900–960 °F (480–515 °C) for soft glass.

## How it works

1. **Select stock.** Work begins from manufactured rod or tube, so wall thickness and diameter are inherited from the supplier rather than created. Tube is specified by outside diameter and wall, and the glass type — borosilicate or soft glass — is chosen for both working temperature and thermal expansion.
2. **Preheat.** Glass is introduced to the flame gradually. Borosilicate tolerates thermal shock far better than soft glass, but any cold glass pushed straight into a hot flame cracks.
3. **Work in the flame.** Only the heated region moves. Tubes are joined by heating both ends to working temperature, pressing them together, and blowing gently to restore the bore. Bends are made by heating a length evenly, removing it from the flame, and bending while supporting the wall with internal air pressure so it does not collapse.
4. **Control the bore.** The recurring problem in tube work is keeping the inside diameter open. Blowing into the tube while bending or joining keeps the wall from folding inward; a partly collapsed bore is the most common defect in both scientific and neon work.
5. **Add features.** Side arms, joints, stopcock seats, and thermometer wells are added by heating a spot, blowing out a bulge, and opening it to receive the next component.
6. **Anneal.** The finished piece goes into a kiln and is held near the annealing point — roughly 1,050 °F (565 °C) for borosilicate — then cooled slowly. Small beads may be annealed in batches; large scientific assemblies are annealed as complete units.
7. **Process, for neon.** Electrodes are sealed into the tube ends, the tube is evacuated and bombarded with high current to drive out contaminants, then backfilled with neon at roughly 10–20 torr for red, or argon with a trace of mercury for blue, and sealed.

## Design guidelines

### Design in standard tube sizes
Everything starts from commercially available rod and tube. Specify from standard outside diameters and wall thicknesses; asking for a non-standard section means either drawing custom tube or working the glass down from a larger size, which changes the wall in ways that are hard to control.

### Match coefficients of expansion
Any two glasses joined in one piece must have matched expansion, or the joint cracks as it cools. Borosilicate and soft glass cannot be joined directly. Where a transition is genuinely needed — glass to metal in an electrode seal, for instance — it is made through a graded seal of intermediate glasses.

### Keep joints away from stress
A flame-worked joint is as strong as the parent glass when properly made, but it is also where the wall thickness is least predictable. Place joints away from regions that will be clamped, loaded, or thermally cycled.

### Support the bore
Bends and joints in tubing collapse inward unless supported by internal pressure while hot. Design bend radii generously — a tight bend in thin-wall tube is difficult to make without a flat or a thin spot on the outside of the curve.

### Anneal by section thickness
Annealing time scales with the thickest section in the piece. A design that combines thin tubing with a heavy solid element commits the whole assembly to the long cycle the heavy part requires.

### Glass-to-metal seals
Electrode seals and feedthroughs require a metal whose thermal expansion matches the glass across the whole cooling range, not merely at one temperature. This is a materials selection problem, not a technique problem.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Stock | Standard tube OD and wall | Custom drawing required otherwise | Wall is inherited from supplier |
| Glass compatibility | Matched expansion only | Graded seal for transitions | Mismatch cracks the joint |
| Bend radius | Generous | Tight bends flatten and thin | Bore collapses without support |
| Joint location | Away from clamps and loads | — | Wall thickness least predictable there |
| Annealing | Required, sized to thickest section | — | Residual stress causes delayed failure |

## Variants

- Blowing
- Hole Boring
- Bending
- Mandrel Forming

## Cost drivers

Lampworking is priced as skilled labor. There is no tooling of any kind — a torch, hand tools, and a kiln — so cost is time at the bench plus material plus annealing energy. Complexity translates almost directly into hours: a laboratory manifold with a dozen joints costs roughly what a dozen joints cost, because each is made individually.

Material is a modest share for scientific work and a larger one for specialty colored art glass. Neon adds gas fill, electrodes, and processing equipment, plus the transformer and installation that come with the finished sign.

Loss is meaningful and comes late. A crack during a final joint destroys everything already invested in the piece, and annealing failures show up after the piece is otherwise complete.

Volume breakpoints: lampworking has no volume story — cost per piece is essentially flat because there is no tooling to amortize. Repeat scientific components in quantity move to semi-automatic tube-working lathes, and simple hollow ware at volume moves to [glassblowing](/processes/forming/glassblowing) into a mold.

1. Reduce the number of joints; each one is separately paid for.
2. Design around standard tube sizes to avoid custom stock.
3. Standardize on one glass type throughout a piece to avoid graded seals.
4. Batch pieces for annealing; kiln cycles cost the same part-loaded.
5. For repeat production of a simple form, evaluate a glass lathe or mold-blown alternative before committing to hand work.

## FAQ

### What is the difference between lampworking and glassblowing?

Glassblowing gathers molten glass from a furnace and works the whole mass at once. Lampworking starts from manufactured rod or tube and heats only a local region in a bench torch, leaving the rest rigid. That local control is what makes it the process for scientific glassware, neon tube, and small detailed work.

### Why is borosilicate used for scientific glassware?

Its thermal expansion is roughly a third that of soda-lime glass, so it survives both the localized heating of the torch and the thermal shock of laboratory service. It softens around 1,510 °F (820 °C) and works comfortably near 2,280 °F (1,250 °C), a higher and wider working range than soft glass.

### Can borosilicate and soft glass be joined?

Not directly. Their thermal expansions are very different, so a direct joint develops stress as it cools and cracks. A transition requires a graded seal — a series of intermediate glasses whose expansions step gradually from one to the other.

### How is neon tube made?

Soda-lime or lead glass tubing of roughly 8–15 mm outside diameter is bent in the flame to the letter or figure, electrodes are sealed into the ends, the tube is evacuated and bombarded with high current to drive out contaminants, and it is then backfilled — neon at roughly 10–20 torr for red, or argon with a trace of mercury for blue and its phosphor-coated variants.

### Does lampworked glass need annealing?

Yes, always. Localized heating puts large stresses into the glass around every joint and bend. The piece must be held near its annealing point — around 1,050 °F (565 °C) for borosilicate, 900–960 °F (480–515 °C) for soft glass — and cooled slowly, or it may fail spontaneously later.

### Why do bends in glass tube collapse?

The wall on the inside of the bend goes into compression and folds inward unless supported. Lampworkers blow gently into the tube while bending to hold the bore open with internal pressure. Generous bend radii make this far easier, particularly in thin-wall tube.

## Alternative processes

- [Glassblowing](https://manufacturingprocesses.org/processes/forming/glassblowing.md): Glassblowing gathers molten glass on a blowpipe and inflates it, by hand or by machine, into hollow ware such as bottles, jars and vessels.
- [Glass Scoring](https://manufacturingprocesses.org/processes/cutting/glass-scoring.md): Glass scoring runs a hardened wheel across the surface to create a controlled fracture line, then breaks the sheet cleanly along it.
- [Tube and Section Bending](https://manufacturingprocesses.org/processes/forming/tube-and-section-bending.md): Tube and section bending draws or presses metal tube and profile around a former, with an internal mandrel wherever the wall would otherwise collapse.
- [Injection Molding](https://manufacturingprocesses.org/processes/forming/injection-molding.md): Injection molding forces molten thermoplastic into a steel mold under high pressure, producing high-volume identical plastic parts.
- [Blow Molding](https://manufacturingprocesses.org/processes/forming/blow-molding.md): Blow molding inflates a heated plastic tube or preform against the walls of a mold with compressed air, producing hollow one-piece containers and tanks.

## Related processes

- [Glassblowing](https://manufacturingprocesses.org/processes/forming/glassblowing.md): Glassblowing gathers molten glass on a blowpipe and inflates it, by hand or by machine, into hollow ware such as bottles, jars and vessels.
- [Clay Throwing](https://manufacturingprocesses.org/processes/forming/clay-throwing.md): Clay throwing forms a rotationally symmetric vessel by hand from a spinning ball of plastic clay on a potter's wheel.
- [Glass Scoring](https://manufacturingprocesses.org/processes/cutting/glass-scoring.md): Glass scoring runs a hardened wheel across the surface to create a controlled fracture line, then breaks the sheet cleanly along it.
- [Tube and Section Bending](https://manufacturingprocesses.org/processes/forming/tube-and-section-bending.md): Tube and section bending draws or presses metal tube and profile around a former, with an internal mandrel wherever the wall would otherwise collapse.
- [Soldering and Brazing](https://manufacturingprocesses.org/processes/joining/soldering-and-brazing.md): Soldering and brazing join metals with a molten filler that wets the joint by capillary action while the parent metal stays solid.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/forming/lampworking)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Liquid Silicone Rubber (LSR) Molding"
category: "Forming"
subcategory: "Plastics and Rubber"
materials: ["Plastic"]
tolerances: "About ±0.002–0.004 in (±0.05–0.10 mm) on small precision features; ISO 3302-1 class M1 is achievable on tightly controlled dimensions. Larger dimensions scale with the 2–3.5% shrinkage allowance."
volumes: "1,000–1,000,000+ parts; below about 500, compression molding avoids the tooling cost"
lead_time: "6–12 weeks for a production LSR tool with cold runner and vacuum sealing; cycles of 20–60 seconds thereafter, plus any post-cure"
url: https://manufacturingprocesses.org/processes/forming/liquid-silicone-rubber-lsr-molding
---

# Liquid Silicone Rubber (LSR) Molding

LSR molding injects two-part liquid silicone into a heated mold where it cures, producing flexible parts with a very wide service temperature range.

- **Category**: [Forming](https://manufacturingprocesses.org/processes/forming.md)
- **Family**: Plastics and Rubber
- **Materials**: Plastic
- **Typical tolerances**: About ±0.002–0.004 in (±0.05–0.10 mm) on small precision features; ISO 3302-1 class M1 is achievable on tightly controlled dimensions. Larger dimensions scale with the 2–3.5% shrinkage allowance.
- **Surface finish**: Reproduces the mold finish from polished gloss to matte; a parting-line flash witness is inherent to the material's low viscosity
- **Typical volumes**: 1,000–1,000,000+ parts; below about 500, compression molding avoids the tooling cost
- **Lead time**: 6–12 weeks for a production LSR tool with cold runner and vacuum sealing; cycles of 20–60 seconds thereafter, plus any post-cure

## Overview

Liquid silicone rubber molding injects a two-part, platinum-cured silicone into a heated mold, where it cures in seconds into a flexible elastomer. Unlike thermoplastic injection molding, the barrel runs cold and the mold runs hot: the material is a low-viscosity liquid at room temperature and crosslinks only once it reaches the cavity.

LSR is chosen for what silicone does that organic elastomers do not — continuous service from roughly -58 °F to 392 °F (-50 °C to 200 °C), excellent resistance to UV, ozone, and steam sterilization, and biocompatibility grades qualified to USP Class VI and ISO 10993. Baby bottle nipples, medical seals and valves, respirator masks, automotive connector seals, and LED optics are all LSR.

Hardnesses run 5–80 Shore A, most commonly 30–70. The material's very low viscosity means it will flash through any gap the tool leaves, so LSR tooling is built to tighter shutoffs than thermoplastic tooling.

## How it works

1. **Meter and mix.** Two components — one carrying a platinum catalyst, the other a hydride crosslinker — are pumped from drums at a 1:1 ratio through a static mixer, with pigment and any additive dosed in line. Ratio control matters: an off-ratio shot undercures and stays tacky.

2. **Feed cold.** The mixed material passes through a temperature-controlled barrel held near ambient, roughly 60–77 °F (15–25 °C). Crosslinking is heat-activated, so keeping the barrel cold is what preserves working life. Any heat leak from the mold back into the nozzle will cure material in the barrel and shut the machine down.

3. **Inject into a hot mold.** The mold runs at roughly 300–400 °F (150–205 °C). Because LSR has a viscosity closer to honey than to polymer melt, injection pressure is far lower than for thermoplastics and the cavity must be evacuated — production LSR tools pull vacuum before the shot to avoid trapping air the material is too thin to push out.

4. **Cure.** Crosslinking begins on contact with the hot cavity. Cure time depends on section thickness and catalyst level; typical cycles run 20–60 seconds for parts under about 0.2 in (5 mm) section. Cure is chemical, so unlike thermoplastic molding, cooling is not part of the cycle at all.

5. **Demold.** LSR elongates 300–700% before break, so parts are usually stripped, air-blown, or peeled off cores rather than pushed off by ejector pins — which is why LSR parts can carry undercuts that would require a side action in a rigid material.

6. **Post-cure if required.** Food-contact and medical parts are typically post-cured about 4 hours at 392 °F (200 °C) to drive off residual volatiles and stabilize properties. Post-cure is not needed for most industrial parts.

### Why a cold runner?

LSR is a thermoset — runners and sprues cannot be reground. A cold runner block keeps the feed system below cure temperature so no material is wasted between shots, and it is standard on any LSR tool running meaningful volume.

## Design guidelines

### Wall thickness

LSR handles a far wider range than thermoplastics: from about 0.010 in (0.25 mm) in membranes and diaphragms up to 0.400 in (10 mm) or more in solid seals. Thick and thin sections in one part are acceptable, because cure is chemical rather than heat-conduction limited, though very thick sections do lengthen the cycle. Uniformity, the governing rule for thermoplastics, is much less important here.

### Draft — often zero

Parts strip off tooling on their own elongation, so 0–0.5° is workable and many LSR parts carry no draft at all. Where a part must be pulled over a deep core, a small draft still eases automation.

### Undercuts

Moderate undercuts, snap features, and internal grooves release directly because the material stretches. This removes the side actions that the same geometry would require in a rigid molded part and is one of the strongest reasons to choose LSR over a molded thermoplastic elastomer.

### Shutoffs, flash, and vents

The design consequence of low viscosity: LSR flashes through clearances that would be perfectly sealed for a thermoplastic. Parting lines must land where a thin flash witness is acceptable and where deflashing is practical. Plan the parting line as a functional design decision, not a tooling afterthought.

### Shrinkage

Budget 2–3.5% linear shrinkage, higher than most thermoplastics and dependent on the specific grade, hardness, and whether the part is post-cured. Confirm the number for the actual compound before the tool is cut, because the entire cavity is scaled by it.

### Bonding to a substrate

Self-bonding LSR grades adhere directly to some thermoplastics and to primed metals during cure; standard grades do not. If you are overmolding LSR onto a rigid part, either specify a self-bonding grade, prime the substrate, or design a mechanical interlock — a through-hole or undercut groove — and do not rely on the chemistry alone.

### Material comparison

Where the requirement is only flexibility rather than temperature or biocompatibility, a thermoplastic elastomer molded conventionally is usually cheaper and faster. Compare against the production polymers on [/charts/material-properties](/charts/material-properties) before committing to silicone.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Wall thickness | 0.040–0.200 in (1–5 mm) | 0.010–0.400 in (0.25–10 mm) | Chemical cure tolerates thick and thin in one part |
| Draft | 0–0.5° per side | 0° | Elongation of 300–700% lets parts strip |
| Undercuts | Moderate, released by stretch | Deep sharp undercuts | No side action needed for most snap geometry |
| Shrinkage allowance | 2–3.5% linear | — | Grade and post-cure dependent; confirm before cutting steel |
| Hardness | 30–70 Shore A | 5–80 Shore A | Standard commercial compound range |
| Service temperature | -58 to 392 °F (-50 to 200 °C) | — | The main reason to choose silicone at all |
| Shutoff clearance | As tight as the tool can hold | — | Low viscosity flashes through any gap |

## Cost drivers

LSR tooling costs more than an equivalent thermoplastic tool. It is heated rather than cooled, it needs vacuum sealing, it needs a cold runner block to avoid wasting a thermoset that cannot be reground, and its shutoffs are cut to tighter clearances to control flash. The raw material is also several times the cost per pound of commodity thermoplastics.

Against that, cycles are short — 20–60 seconds for most parts — the process automates well, and the finished part often eliminates an assembly that would otherwise need a separate seal, gasket, or grip.

Volume breakpoints:

- Under 500 parts: compression molding of high-consistency rubber, or machining from silicone sheet, avoids the tooling investment.
- 1,000–100,000: single or low-cavity LSR tool with a cold runner.
- Over 100,000: multi-cavity tooling with automated demolding, valve-gated cold runners, and in-line vision inspection.

Cost reduction:

1. **Specify post-cure only where it is required.** Four hours at 392 °F is a real oven cost and is unnecessary for most industrial parts.
2. **Design the parting line for flash you can live with.** Deflashing is manual labor on every part unless the geometry allows cryogenic tumbling.
3. **Use the elongation.** Designing undercuts that strip out eliminates side actions and their maintenance entirely.
4. **Choose the softest durometer that works.** Softer compounds generally flow and fill more easily, which shortens the cycle.
5. **Check whether a TPE will do.** If the part does not need silicone's temperature range, chemical resistance, or biocompatibility, a thermoplastic elastomer molded conventionally is usually cheaper on both tooling and material.

## FAQ

### What temperature range can LSR parts survive?

Roughly -58 °F to 392 °F (-50 °C to 200 °C) in continuous service, with short excursions higher. That span, combined with UV, ozone, and steam-sterilization resistance, is the main reason to choose silicone over a thermoplastic elastomer, which typically tops out far lower.

### Why is the LSR mold hot and the barrel cold?

It is the inverse of thermoplastic molding. LSR crosslinks on heating, so the barrel is held near ambient — roughly 60–77 °F (15–25 °C) — to preserve working life, and the mold runs at 300–400 °F (150–205 °C) to trigger cure. Heat leaking back from the mold into the nozzle will cure material in the barrel and stop production.

### What draft angle does LSR molding need?

Often none. LSR elongates 300–700% before break, so parts strip off cores on their own flexibility; 0–0.5° per side is typical and many parts carry zero draft. The same elongation lets moderate undercuts and snap features release without any side action in the tool.

### How much does LSR shrink?

Budget 2–3.5% linear shrinkage, which is higher than most thermoplastics. The exact figure depends on the grade, the durometer, and whether the part is post-cured, so confirm it against the specific compound before the cavity is cut — the entire tool is scaled by that number.

### Does LSR need to be post-cured?

Only for food-contact and medical applications, where a post-cure of about 4 hours at 392 °F (200 °C) drives off residual volatiles and stabilizes properties. Most industrial parts ship as molded, and specifying an unnecessary post-cure adds real oven cost to every batch.

### Will LSR bond to a plastic substrate when overmolded?

Only with a self-bonding grade, or with a primed substrate. Standard LSR grades do not adhere to thermoplastics during cure. Where the joint is structural, design a mechanical interlock — a through-hole or undercut groove the silicone fills — rather than relying on chemistry alone.

### Why does LSR flash more than thermoplastic molding?

Its viscosity before cure is closer to honey than to polymer melt, so it penetrates clearances that a thermoplastic would seal against. LSR tooling is cut to tighter shutoffs and usually pulls vacuum on the cavity, but a thin parting-line witness is inherent. Place the parting line where that witness is cosmetically and functionally acceptable.

## Alternative processes

- [Compression Molding](https://manufacturingprocesses.org/processes/forming/compression-molding.md): Compression molding presses a measured charge of rubber or thermoset into an open heated mold that closes to cure it into its final shape.
- [Injection Molding](https://manufacturingprocesses.org/processes/forming/injection-molding.md): Injection molding forces molten thermoplastic into a steel mold under high pressure, producing high-volume identical plastic parts.
- [Dip Molding](https://manufacturingprocesses.org/processes/forming/dip-molding.md): Dip molding withdraws a heated former from liquid plastisol or latex, leaving a coating that cures into a flexible open-ended part such as a grip or cap.
- [Vacuum Casting (Urethane Casting)](https://manufacturingprocesses.org/processes/forming/vacuum-casting.md): Vacuum casting, known in the US as urethane casting, pours polyurethane resin into a silicone mold under vacuum to copy a master pattern for short runs.

## Related processes

- [Injection Molding](https://manufacturingprocesses.org/processes/forming/injection-molding.md): Injection molding forces molten thermoplastic into a steel mold under high pressure, producing high-volume identical plastic parts.
- [Overmolding](https://manufacturingprocesses.org/processes/forming/overmolding.md): Overmolding molds a second material, usually a soft elastomer, directly onto an already-molded rigid substrate so the two bond permanently.
- [Compression Molding](https://manufacturingprocesses.org/processes/forming/compression-molding.md): Compression molding presses a measured charge of rubber or thermoset into an open heated mold that closes to cure it into its final shape.
- [Insert Molding](https://manufacturingprocesses.org/processes/forming/insert-molding.md): Insert molding loads metal inserts into the cavity and molds polymer around them, capturing threads, terminals or stampings in a single shot.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/forming/liquid-silicone-rubber-lsr-molding)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Masked Stereolithography (MSLA / DLP)"
category: "Forming"
subcategory: "Additive Manufacturing"
materials: ["Plastic"]
tolerances: "About ±0.004–0.008 in (±0.1–0.2 mm) on general geometry; ±0.002 in (±0.05 mm) is achievable on small parts with fine-pitch panels and a compensated first article."
volumes: "1–5,000 small parts; cost per part falls in proportion to how densely the plate is filled"
lead_time: "Same day to 3 business days. A 4 in (100 mm) tall build runs roughly 3–8 hours plus wash and post-cure, regardless of how many parts are on the plate."
url: https://manufacturingprocesses.org/processes/forming/masked-stereolithography-msla-dlp
---

# Masked Stereolithography (MSLA / DLP)

Masked stereolithography cures a whole resin layer at once through an LCD or DLP image, so build time depends on height rather than part count.

- **Category**: [Forming](https://manufacturingprocesses.org/processes/forming.md)
- **Family**: Additive Manufacturing
- **Materials**: Plastic
- **Typical tolerances**: About ±0.004–0.008 in (±0.1–0.2 mm) on general geometry; ±0.002 in (±0.05 mm) is achievable on small parts with fine-pitch panels and a compensated first article.
- **Surface finish**: Ra 20–120 µin (0.5–3 µm) — close to laser SLA, but with a faint pixel texture on curved walls that laser scanning does not produce.
- **Typical volumes**: 1–5,000 small parts; cost per part falls in proportion to how densely the plate is filled
- **Lead time**: Same day to 3 business days. A 4 in (100 mm) tall build runs roughly 3–8 hours plus wash and post-cure, regardless of how many parts are on the plate.

## Overview

Masked stereolithography cures an entire resin layer in a single exposure by projecting the whole cross-section at once — through an LCD panel acting as a photomask (MSLA) or from a digital micromirror device (DLP) — rather than tracing it with a laser as [SLA](/processes/forming/stereolithography-sla) does. Exposure takes 1.5–4 seconds per layer regardless of how much of that layer is solid, so build time depends only on part height. One part or a full platform of them costs the same machine hours.

XY resolution equals the mask's pixel pitch: roughly 0.0008–0.002 in (20–50 µm) on current mono LCD panels, and 0.0012–0.004 in (30–100 µm) with DLP projection optics depending on the projected field size. Layers are 0.001–0.004 in (25–100 µm).

Build volumes are small — typically around 8 × 5 × 9 in (200 × 125 × 220 mm) — because enlarging the field either costs resolution or requires a bigger, more expensive panel. That makes the process ideal for high-volume small parts: dental models and surgical guides, aligner thermoforming molds, castable jewelry patterns, hearing aid shells, and miniatures.

## How it works

1. **Slice into images.** Each layer becomes a bitmap at the native pixel resolution of the mask. Anti-aliasing and gray-scale exposure are used to soften the pixel staircase on curved walls.
2. **Dip and settle.** The build platform lowers into the vat until the gap between the last cured layer and the transparent film at the vat floor equals one layer thickness. Resin must flow into that gap before exposure — viscous resins need a longer settle time.
3. **Expose the whole layer.** A 405 nm LED array shines through the LCD mask, or a DLP projector images the cross-section directly. Exposure is typically 1.5–4 seconds per layer on a mono LCD. Because the whole layer cures simultaneously, adding parts to the platform does not add time.
4. **Peel.** The platform lifts, separating the cured layer from the release film. This is the critical mechanical event of the process: peel force scales with the cured cross-sectional area, and large flat layers are the usual cause of a part tearing off its supports or delaminating.
5. **Repeat.** Vertical build rate is commonly 0.4–1.2 in/h (10–30 mm/h).
6. **Wash and post-cure.** Isopropyl alcohol wash of 5–15 minutes, support removal, then UV post-cure — 15–60 minutes, often warmed to 60 °C (140 °F). As with any photopolymer process, published mechanical properties assume a completed post-cure.

## Design guidelines

### Orient to minimize cross-sectional area
This is the dominant rule of the process. Peel force is proportional to the area cured against the release film, so a flat plate printed parallel to the platform will fail where the same plate tilted 20–30° succeeds. Tilting also breaks the layer up so no single exposure has a large continuous area.

### Hollow and vent
Hollow shells reduce both resin cost and peel area. Use a 0.060–0.080 in (1.5–2 mm) wall and at least two vent holes of 0.12 in (3 mm) or larger. A sealed hollow section also creates a suction cup against the vat film, which is a separate and equally destructive failure mode.

### Feature size versus pixel size
Nothing smaller than one pixel can be resolved, and features under about three pixels reproduce unreliably. On a 50 µm panel, that means 0.006 in (0.15 mm) as an absolute floor for a rib or engraved line. Fine dental and jewelry work is done specifically on panels with the smallest available pitch.

### Wall thickness
0.016 in (0.4 mm) supported and 0.032 in (0.8 mm) unsupported. Thin walls fail during peel, not during cure.

### Supports
Use many small contact points rather than a few large ones — typically 0.3–0.6 mm tips at a density that spreads peel load. Keep them off cosmetic and fitting surfaces, and orient so the first layers land on sacrificial geometry.

### Dimensional compensation
Photopolymer shrinks on cure and again on post-cure, and light bleed through the mask grows XY features slightly. Fitting features are usually compensated by 0.002–0.004 in (0.05–0.1 mm) per side after a first article; expect to iterate once on any press or slip fit. The [ISO 286 fits and tolerances chart](/charts/iso-286-fits-tolerances) is a useful reference for what fit class you actually need.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Orientation | 20–30° tilt | Flat only for thin parts | Peel force scales with layer area |
| Wall thickness | 0.032 in (0.8 mm) | 0.016 in (0.4 mm) | Peel loading, not cure resolution |
| Detail size | 3 pixels | 1 pixel | Sub-pixel features do not reproduce |
| Vent hole | 0.16 in (4 mm) | 0.12 in (3 mm) | Prevents suction against the vat film |
| Layer height | 0.002 in (50 µm) | 0.001 in (25 µm) | Time scales directly with layer count |
| Hollow wall | 0.080 in (2 mm) | 0.060 in (1.5 mm) | Shell must resist peel without support |

## Cost drivers

The defining economic feature of masked SLA is that build time is a function of height alone. A platform holding fifty dental models costs the same machine hours as one, so unit cost falls almost linearly with how many parts fit on the plate. This is why the process took over dental and jewelry production while laser SLA stayed in engineering prototyping.

Resin volume is the second cost, which is why hollowing pays twice — less resin, lower peel force. Consumables are the third: LCD panels degrade under constant UV and are replaced periodically, and release films are changed regularly.

Labor for support removal, washing, and post-curing does not scale down with quantity and eventually dominates. Production shops mitigate it with automated wash-and-cure stations and support layouts designed for fast breakoff.

Volume breakpoints: masked SLA is economical from 1 part and stays competitive to a few thousand small parts a year. Beyond that, or where a tougher material is needed, look at molding.

1. Fill the plate — it costs nothing extra in time.
2. Hollow everything with a wall over 0.16 in (4 mm) and vent it.
3. Use the largest layer height the surface finish tolerates.
4. Design supports for fast manual removal; labor is the cost that does not scale.
5. Match panel pixel pitch to the smallest feature you actually need rather than buying resolution you will not use.

## FAQ

### What is the difference between MSLA, DLP, and SLA?

All three cure photopolymer with UV light. SLA traces each layer with a scanning laser, so time scales with how much area is solid. MSLA masks an LED array with an LCD panel and DLP images the layer with a micromirror chip — both cure the whole layer in one exposure, so time depends only on part height.

### Why does build time not increase when I add more parts?

Because the entire layer is exposed at once. Whether one part or thirty occupy that layer, the exposure is the same 1.5–4 seconds. Only the number of layers — that is, the height of the tallest part — determines build time.

### What resolution can MSLA achieve?

XY resolution equals the mask pixel pitch, roughly 0.0008–0.002 in (20–50 µm) on current mono LCD panels. Features smaller than about three pixels reproduce unreliably, so on a 50 µm panel plan on 0.006 in (0.15 mm) as the practical minimum detail.

### Why do my large flat MSLA prints fail or delaminate?

Peel force is proportional to the cured area contacting the vat film, and a large flat layer generates enough of it to tear the part off its supports. Tilt the part 20–30°, hollow it, and add vents — tilting reduces the area cured in any single exposure.

### Do MSLA parts need vent holes?

Any hollow section does. Without vents, uncured resin is trapped inside and the sealed cavity acts as a suction cup against the vat film during peel. Use at least two holes of 0.12 in (3 mm) or larger, placed so resin can drain and air can enter.

### Is MSLA accurate enough for dental and jewelry work?

Yes — it is the dominant process in both. With a fine-pitch panel and a compensated first article, ±0.002 in (±0.05 mm) is achievable on small parts, which covers dental models, surgical guides, aligner molds, and castable patterns for [investment casting](/processes/forming/investment-casting).

## Alternative processes

- [Stereolithography (SLA)](https://manufacturingprocesses.org/processes/forming/stereolithography-sla.md): Stereolithography cures liquid photopolymer layer by layer with a scanning UV laser, giving the smoothest surfaces in additive manufacturing.
- [Material Jetting (PolyJet)](https://manufacturingprocesses.org/processes/forming/material-jetting-polyjet.md): Material jetting sprays droplets of photopolymer and cures them layer by layer, allowing several materials and colors in one build.
- [Fused Deposition Modeling (FDM)](https://manufacturingprocesses.org/processes/forming/fused-deposition-modeling-fdm.md): Fused deposition modeling extrudes molten thermoplastic filament along programmed paths so each layer welds to the one beneath it.
- [Vacuum Casting (Urethane Casting)](https://manufacturingprocesses.org/processes/forming/vacuum-casting.md): Vacuum casting, known in the US as urethane casting, pours polyurethane resin into a silicone mold under vacuum to copy a master pattern for short runs.
- [Injection Molding](https://manufacturingprocesses.org/processes/forming/injection-molding.md): Injection molding forces molten thermoplastic into a steel mold under high pressure, producing high-volume identical plastic parts.

## Related processes

- [Stereolithography (SLA)](https://manufacturingprocesses.org/processes/forming/stereolithography-sla.md): Stereolithography cures liquid photopolymer layer by layer with a scanning UV laser, giving the smoothest surfaces in additive manufacturing.
- [Material Jetting (PolyJet)](https://manufacturingprocesses.org/processes/forming/material-jetting-polyjet.md): Material jetting sprays droplets of photopolymer and cures them layer by layer, allowing several materials and colors in one build.
- [Rapid Prototyping](https://manufacturingprocesses.org/processes/forming/rapid-prototyping.md): Rapid prototyping builds a part layer by layer directly from a CAD model, with no tooling and no minimum order quantity.
- [Vacuum Casting (Urethane Casting)](https://manufacturingprocesses.org/processes/forming/vacuum-casting.md): Vacuum casting, known in the US as urethane casting, pours polyurethane resin into a silicone mold under vacuum to copy a master pattern for short runs.
- [Investment Casting](https://manufacturingprocesses.org/processes/forming/investment-casting.md): Investment casting builds a ceramic shell around a wax pattern, melts the wax out and pours metal into the cavity, giving fine detail in almost any alloy.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/forming/masked-stereolithography-msla-dlp)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Material Jetting (PolyJet)"
category: "Forming"
subcategory: "Additive Manufacturing"
materials: ["Plastic"]
tolerances: "About ±0.004 in (±0.1 mm) on parts up to 4 in (100 mm), and roughly ±0.008 in (±0.2 mm) or ±0.1% of nominal on larger geometry."
volumes: "1–50 parts"
lead_time: "1–3 business days. Printing is typically 4–12 hours, with support removal adding anywhere from minutes to several hours depending on internal geometry."
url: https://manufacturingprocesses.org/processes/forming/material-jetting-polyjet
---

# Material Jetting (PolyJet)

Material jetting sprays droplets of photopolymer and cures them layer by layer, allowing several materials and colors in one build.

- **Category**: [Forming](https://manufacturingprocesses.org/processes/forming.md)
- **Family**: Additive Manufacturing
- **Materials**: Plastic
- **Typical tolerances**: About ±0.004 in (±0.1 mm) on parts up to 4 in (100 mm), and roughly ±0.008 in (±0.2 mm) or ±0.1% of nominal on larger geometry.
- **Surface finish**: Ra 20–40 µin (0.5–1 µm) on up-facing glossy surfaces — the smoothest available from an additive process. Surfaces built against support gel are matte and noticeably rougher.
- **Typical volumes**: 1–50 parts
- **Lead time**: 1–3 business days. Printing is typically 4–12 hours, with support removal adding anywhere from minutes to several hours depending on internal geometry.

## Overview

Material jetting sprays droplets of liquid photopolymer from a piezo inkjet head and cures each layer with a UV lamp on the same carriage, at layer heights of 0.0006–0.0013 in (14–32 µm) — the finest of any additive process. Because the print head can lay down several resins in the same pass, one build can combine rigid and rubber-like materials, transparent and opaque regions, and full-color texture maps in a single part.

That capability, not raw accuracy, is why the process exists. A PolyJet anatomical model can carry compliant vessels inside a clear rigid shell; an overmolded handle can be prototyped with the grip already in place at Shore A 30–95; a color-mapped concept model needs no painting.

Accuracy is around ±0.004 in (±0.1 mm) on parts under 4 in (100 mm), and as-built surfaces reach Ra 20–40 µin (0.5–1 µm) in glossy mode. Support is a separate gel resin removed by waterjet or caustic bath. The materials are brittle photopolymers with poor UV stability, so material jetting is a communication and validation process rather than a production one.

## How it works

1. **Slice and assign materials.** Layers are sliced at 0.0006–0.0013 in (14–32 µm) and each voxel is assigned a resin. Multi-material systems blend base resins on the fly to produce intermediate stiffnesses and colors, so a gradient between rigid and elastomeric can be printed rather than assembled.
2. **Jet.** The carriage traverses the build tray, and hundreds of piezo nozzles jet picoliter droplets of model and support resin simultaneously. Support is a soft gel-like photopolymer placed wherever the geometry overhangs and, unavoidably, on every down-facing surface.
3. **Level.** A roller passes over the freshly jetted layer to bring it to exact thickness and remove excess; that excess is scavenged to waste.
4. **Cure.** UV lamps mounted on the carriage cure the layer immediately behind the jetting heads, so each layer is fully solid before the next is deposited. No separate post-cure is required.
5. **Repeat.** The tray indexes down one layer. Build time scales with tray area covered and total height, not with part count in a given layer.
6. **Remove support.** The gel is stripped with a waterjet for accessible geometry, or dissolved in a heated caustic solution for internal channels and lattices. This is the process's main labor step, and delicate features are frequently damaged during it.

## Design guidelines

### Wall thickness
0.024 in (0.6 mm) minimum in rigid resin and 0.035 in (0.9 mm) in rubber-like grades. The limit is not the printer — layers are extremely fine — but the waterjet used to strip support, which will destroy anything thinner.

### Support removal access
Every down-facing surface is printed on support gel, and every internal cavity fills with it. Design access for either waterjet or solution: openings of 0.12 in (3 mm) or larger for waterjet, and remember that soluble support in a long narrow channel takes hours and may never fully clear.

### Multi-material design
Assign materials by body in the CAD assembly rather than trying to patch them at the slicer. Rigid-to-elastomer transitions bond chemically during printing and are far stronger than a mechanical joint, but a sharp stiffness discontinuity still concentrates strain — blend the transition over 0.04–0.08 in (1–2 mm) where the part will flex.

### Glossy versus matte
Up-facing surfaces printed without support come out glossy at Ra 20–40 µin (0.5–1 µm). Any surface that touches support prints matte and slightly rougher. If appearance matters, orient the show face upward.

### Color and texture
Full-color systems accept texture-mapped models. Color is printed as a thin shell over a white core, so engraved detail can cut through the color layer — keep engraving shallower than the shell or design the color into the geometry.

### Do not design for load
Material jetting resins are brittle, creep under sustained load, and yellow under UV within weeks of sun exposure. Use the process for fit, appearance, and communication; move functional testing to [SLS](/processes/forming/selective-laser-sintering-sls) or a molded thermoplastic.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Wall, rigid | 0.040 in (1.0 mm) | 0.024 in (0.6 mm) | Waterjet support removal |
| Wall, elastomeric | 0.060 in (1.5 mm) | 0.035 in (0.9 mm) | Soft material tears during cleaning |
| Support access opening | 0.20 in (5 mm) | 0.12 in (3 mm) | Waterjet needs line of sight |
| Detail size | 0.016 in (0.4 mm) | 0.008 in (0.2 mm) | Survivable through cleaning |
| Layer height | 0.0011 in (28 µm) | 0.0006 in (14 µm) | Finer layers roughly double time |
| Clearance, moving | 0.016 in (0.4 mm) | 0.008 in (0.2 mm) | Support gel must be removed from the gap |

## Cost drivers

Material jetting has the highest material cost per part of the common additive processes, for two reasons. The resins themselves are expensive, and support gel is consumed in large quantity — a part with substantial overhang can consume as much support resin as model resin, and all of it is thrown away. Roller scavenging adds further waste on every layer.

Machine time scales with the area of the tray covered and the total build height, so nesting several parts across the tray is efficient in a way that stacking them vertically is not.

Support removal labor is the third cost and the one that varies most. A simple exterior shell cleans in minutes; a lattice or a closed internal channel can take hours of soaking and picking.

Volume breakpoints: material jetting is a 1–50 part process. It has no volume story at all — nothing about it gets cheaper with quantity beyond tray packing — and any functional or production requirement should move to another process.

1. Orient so cosmetic faces point up and support lands on hidden geometry.
2. Reduce overhang volume; you pay full price for support resin.
3. Print only the region that needs multi-material or color, and machine or print the rest elsewhere.
4. Use a single material and standard finish for pure fit checks — multi-material builds cost substantially more.
5. Spread parts across the tray rather than stacking them tall.

## FAQ

### What layer height does PolyJet print at?

0.0006–0.0013 in (14–32 µm), the finest of any commercial additive process. High-quality mode is typically 16 µm and high-speed mode around 28–32 µm. Layer height affects build time roughly proportionally.

### Can material jetting print rigid and flexible material in one part?

Yes — that is its main reason for existing. Rigid resins and rubber-like grades from about Shore A 30 to 95 can be jetted in the same pass and bond chemically at the interface, so overmolds, gaskets, and living hinges can be prototyped as a single printed assembly.

### How accurate is PolyJet?

Roughly ±0.004 in (±0.1 mm) on parts up to 4 in (100 mm), and ±0.008 in (±0.2 mm) or about ±0.1% of nominal on larger ones. Accuracy is good but the material properties, not the dimensions, are what limit the process.

### How is PolyJet support removed?

A gel-like support photopolymer is jetted alongside the model material and stripped afterward with a waterjet, or dissolved in a heated caustic solution for internal geometry. Waterjet cleaning is what sets the 0.024 in (0.6 mm) minimum wall thickness — thinner walls do not survive it.

### Are PolyJet parts usable as functional parts?

Rarely. The resins are brittle, creep under sustained load, and yellow and embrittle under UV within weeks of sun exposure. Use material jetting for appearance, fit, and communication models, and move functional testing to sintered nylon or molded thermoplastic.

### Why is material jetting expensive?

Model resin is costly, support gel is consumed in comparable quantity and discarded, and the roller scavenges excess material from every layer. On top of that, support removal is a manual operation that can take hours on parts with internal geometry.

## Alternative processes

- [Stereolithography (SLA)](https://manufacturingprocesses.org/processes/forming/stereolithography-sla.md): Stereolithography cures liquid photopolymer layer by layer with a scanning UV laser, giving the smoothest surfaces in additive manufacturing.
- [Selective Laser Sintering (SLS)](https://manufacturingprocesses.org/processes/forming/selective-laser-sintering-sls.md): Selective laser sintering fuses nylon powder layer by layer with a laser, so the surrounding loose powder supports the part and no supports are needed.
- [Vacuum Casting (Urethane Casting)](https://manufacturingprocesses.org/processes/forming/vacuum-casting.md): Vacuum casting, known in the US as urethane casting, pours polyurethane resin into a silicone mold under vacuum to copy a master pattern for short runs.
- [Overmolding](https://manufacturingprocesses.org/processes/forming/overmolding.md): Overmolding molds a second material, usually a soft elastomer, directly onto an already-molded rigid substrate so the two bond permanently.
- [CNC Machining](https://manufacturingprocesses.org/processes/cutting/cnc-machining.md): CNC machining removes material from solid stock with a programmed cutting tool, producing accurate parts directly from a CAD model with no tooling.

## Related processes

- [Stereolithography (SLA)](https://manufacturingprocesses.org/processes/forming/stereolithography-sla.md): Stereolithography cures liquid photopolymer layer by layer with a scanning UV laser, giving the smoothest surfaces in additive manufacturing.
- [Masked Stereolithography (MSLA / DLP)](https://manufacturingprocesses.org/processes/forming/masked-stereolithography-msla-dlp.md): Masked stereolithography cures a whole resin layer at once through an LCD or DLP image, so build time depends on height rather than part count.
- [Rapid Prototyping](https://manufacturingprocesses.org/processes/forming/rapid-prototyping.md): Rapid prototyping builds a part layer by layer directly from a CAD model, with no tooling and no minimum order quantity.
- [Binder Jetting](https://manufacturingprocesses.org/processes/forming/binder-jetting.md): Binder jetting prints liquid binder into a powder bed to hold each layer, producing green parts that are later sintered or infiltrated.
- [Overmolding](https://manufacturingprocesses.org/processes/forming/overmolding.md): Overmolding molds a second material, usually a soft elastomer, directly onto an already-molded rigid substrate so the two bond permanently.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/forming/material-jetting-polyjet)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Metal Extrusion"
category: "Forming"
subcategory: "Metal"
materials: ["Metal"]
tolerances: "Standard mill tolerances per published aluminum extrusion practice — roughly ±0.010 in (±0.25 mm) on small cross-section dimensions, wall thickness held as a percentage, plus separate straightness, twist, and flatness allowances"
volumes: "Several hundred pounds minimum per die run, up to millions of pounds; unit cost is nearly flat across that range"
lead_time: "2–4 weeks for a new die and first run; 2–6 weeks per production order thereafter"
url: https://manufacturingprocesses.org/processes/forming/metal-extrusion
---

# Metal Extrusion

Metal extrusion pushes heated billet through a shaped die, producing continuous aluminum or copper profile of constant cross-section.

- **Category**: [Forming](https://manufacturingprocesses.org/processes/forming.md)
- **Family**: Metal
- **Materials**: Metal
- **Typical tolerances**: Standard mill tolerances per published aluminum extrusion practice — roughly ±0.010 in (±0.25 mm) on small cross-section dimensions, wall thickness held as a percentage, plus separate straightness, twist, and flatness allowances
- **Surface finish**: As-extruded aluminum takes anodizing and powder coat directly; die lines running along the extrusion direction are inherent and are removed by mechanical finishing where cosmetic
- **Typical volumes**: Several hundred pounds minimum per die run, up to millions of pounds; unit cost is nearly flat across that range
- **Lead time**: 2–4 weeks for a new die and first run; 2–6 weeks per production order thereafter

## Overview

Metal extrusion pushes a heated billet through a shaped die under thousands of tons of force, producing a continuous profile of constant cross-section — solid, hollow, or multi-void — in a single operation. Aluminum dominates: 6063 is extruded at roughly 800–930°F (425–500°C), well below its melting point, and comes out as a continuous section that is stretched straight, cut to length, and aged to temper.

It is how architectural framing, heat sinks, machine framing, window and curtain wall sections, structural rails, tubing, and enclosure profiles are made. Copper, brass, and magnesium extrude as well; steel extrusion is far less common because of the die temperatures involved.

The economics are unusual. Die cost is low compared to almost any other metal tooling, and a die makes profile by the mile — so extrusion is competitive from a few hundred pounds, and the material cost per pound is a fraction of what machining the same section from plate would cost.

## How it works

1. **Billet heating.** A cast aluminum log is cut to billet length and heated to roughly 800–930°F (425–500°C) for 6000-series alloys. The container and die are preheated too; a cold die is the fastest way to a scrapped run.
2. **Loading.** The billet is loaded into the container and the ram advances behind a dummy block.
3. **Extrusion.** The ram forces the metal through the die aperture. Reduction is described by the extrusion ratio — billet area divided by profile area — which for aluminum commonly runs from about 10:1 to 100:1. The metal does not melt; it flows plastically like toothpaste.
4. **Hollow profiles.** A hollow is made with a porthole or bridge die, which splits the flow around a mandrel and rewelds it downstream under pressure and temperature. Those longitudinal seam welds are why hollow aluminum extrusions are possible at all, and also why 6063 and 6061 — which reweld cleanly — dominate hollow work while 7000-series alloys do not.
5. **Quench.** The profile is quenched on the run-out table by air or water spray. For 6000-series alloys this is the solution treatment step, and it must be fast enough to hold the alloying elements in solution.
6. **Stretch.** The cooled profile is pulled in tension, typically a fraction of a percent, to straighten it and relieve residual stress. Every commercial extrusion is stretched; without it, profiles are bowed and twisted.
7. **Age.** Artificial aging — several hours at a few hundred degrees Fahrenheit — develops the T5 or T6 temper. A profile pulled straight off the press is soft.
8. **Finish.** Cut to length, then machine, anodize, or powder coat. Anodizing is common and imposes its own alloy and finish requirements.

Die design is the discipline here. Metal flows fastest where the section is thickest and where it is nearest the billet center, so the die is corrected — apertures choked, bearing lengths varied — until every part of the profile emerges at the same speed. A section with wildly different wall thicknesses fights this correction and runs slower, with more scrap.

## Design guidelines

### Keep wall thickness uniform

This is the dominant rule. Thick and thin regions in the same profile flow at different speeds, and the die correction needed to balance them costs development time, press speed, and scrap. Aim for a constant wall; where thickness must change, taper the transition rather than stepping it.

### Minimum wall

For small 6063 profiles, roughly 0.040–0.060 in (1–1.5 mm) is achievable on solid sections, with hollows needing more — typically 0.060–0.080 in (1.5–2 mm). Minimum wall scales up with the size of the profile, because a big section needs more pressure and the die is under more load.

### Circumscribing circle

Profiles are quoted by the diameter of the smallest circle enclosing the cross-section. Press capacity is described the same way — common presses handle profiles in the range of about 7–12 in circumscribing circle, with large presses going considerably beyond. A profile that just fits a smaller press is much cheaper to run than one that forces the next size up.

### Symmetry and tongue ratio

Symmetric profiles flow evenly and run fast. Deep, narrow channels create a fragile die tongue — the ratio of channel depth to opening width should be kept modest, commonly no more than about 3:1, because a slender tongue deflects under pressure and eventually breaks.

### Radius the corners

Sharp internal corners concentrate stress in the die and produce tear lines in the profile. Use generous internal fillets and break external corners.

### Screw bosses and functional features

Extrude the features in: screw bosses for self-tapping screws, T-slots for fasteners, snap channels, hinge knuckles, and heat sink fins are all free once they are in the die. This is the biggest design lever the process offers — features that would each be a machining operation cost nothing extra per foot.

### Tolerances

Standard mill tolerances follow published aluminum association practice: roughly ±0.010 in on small cross-section dimensions, wall thickness held to a percentage rather than an absolute, plus separate straightness, twist, and flatness allowances. Anything tighter than the standard needs machining or a precision-tolerance agreement with the mill. Material selection and property comparison from the [material properties chart](/charts/material-properties); machined-feature fits from [ISO 286 fits and tolerances](/charts/iso-286-fits-tolerances).

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Wall thickness | Uniform throughout | Taper any transition | Uneven walls flow at different speeds |
| Minimum wall, small solid 6063 | 0.060 in (1.5 mm) | 0.040 in (1 mm) | Die pressure and metal flow limits |
| Minimum wall, hollow | 0.080 in (2 mm) | 0.060 in (1.5 mm) | Mandrel support and seam weld quality |
| Tongue ratio (channel depth ÷ opening) | ≤ 2:1 | ≈3:1 | Slender die tongues deflect and break |
| Internal corners | Radiused | Sharp corners tear | Stress concentration in the die and profile |
| Symmetry | Symmetric about an axis | Asymmetric runs slower | Balanced flow across the die aperture |
| Functional features | Extrude them in | — | Bosses, slots, and fins are free in the die |

## Cost drivers

Extrusion has the most favorable tooling-to-output ratio in metal forming. A solid-profile die is cheap relative to any casting or stamping tool and can be cut in a couple of weeks; hollow porthole dies cost more but are still modest. Once running, cost per pound is press time plus metal, and scrap goes back into the melt. What actually drives price is press speed, which is governed by profile complexity, wall uniformity, and alloy: a well-balanced 6063 section runs fast, an asymmetric thin-walled 6061 hollow runs slowly and scraps more.

Volume breakpoints: mills impose a minimum order per die — commonly several hundred pounds — so tiny quantities are impractical unless you buy from a standard-profile catalog. From that minimum up to millions of pounds, unit cost is nearly flat, which makes extrusion unusually forgiving of volume uncertainty.

1. **Use a stock profile if one fits.** Catalog sections carry no die cost and no minimum-run risk.
2. **Balance the wall thickness.** It is the single biggest determinant of press speed, and press speed is the price.
3. **Extrude features instead of machining them.** Screw bosses, T-slots, and fins cost nothing per foot; each machined equivalent costs an operation.

## FAQ

### What is the minimum wall thickness for an aluminum extrusion?

Roughly 0.040–0.060 in (1–1.5 mm) for small solid 6063 profiles and 0.060–0.080 in (1.5–2 mm) for hollows. The minimum scales up with the size of the profile, because larger cross-sections need higher pressure and put more load on the die.

### Why does wall thickness need to be uniform in an extrusion?

Metal flows faster through thick sections than thin ones. An unbalanced profile forces the die maker to choke apertures and vary bearing lengths to equalize the exit speed, which costs development iterations, reduces press speed, and increases scrap — all of which show up in the price per pound.

### How are hollow aluminum extrusions made?

With a porthole or bridge die that splits the metal flow around a mandrel and rewelds it downstream under pressure and temperature. Those longitudinal seam welds are why 6063 and 6061, which reweld cleanly, dominate hollow work, and why high-strength 7000-series alloys are rarely extruded hollow.

### Why are extrusions stretched after they come off the press?

Straightening and stress relief. A profile leaves the press bowed and twisted from uneven cooling, so it is pulled in tension by a fraction of a percent on the stretcher. Every commercial extrusion goes through this step, and the resulting temper designation reflects it.

### 6063 or 6061 for an extruded profile?

6063 extrudes faster, holds finer detail and thinner walls, and anodizes to a better cosmetic finish — it is the standard for architectural and enclosure profiles. 6061 is significantly stronger and better for structural and machined parts, but it runs slower, needs thicker walls, and is harder on dies.

### Extrusion or machining from plate?

Machining wins for one-offs and for geometry that varies along the length. Extrusion wins as soon as the cross-section is constant and the quantity clears the mill's minimum run, because die cost is low and features like screw bosses, T-slots, and fins come out of the die at no extra cost per foot.

## Alternative processes

- [Roll Forming](https://manufacturingprocesses.org/processes/forming/roll-forming.md): Roll forming passes coil stock through a sequence of contoured roller stands that bend it progressively into a constant cross-section profile.
- [CNC Machining](https://manufacturingprocesses.org/processes/cutting/cnc-machining.md): CNC machining removes material from solid stock with a programmed cutting tool, producing accurate parts directly from a CAD model with no tooling.
- [Pultrusion](https://manufacturingprocesses.org/processes/forming/pultrusion.md): Pultrusion pulls continuous fiber through a resin bath and a heated die, curing it into a constant-section composite profile of unlimited length.
- [Die Casting](https://manufacturingprocesses.org/processes/forming/die-casting.md): Die casting injects molten aluminum, zinc or magnesium into a hardened steel die under high pressure, producing thin-walled net-shape metal parts.

## Related processes

- [Forging](https://manufacturingprocesses.org/processes/forming/forging.md): Forging deforms solid metal under compressive load in a hammer, press or dies, so the grain flow follows the part contour and fatigue strength rises.
- [Roll Forming](https://manufacturingprocesses.org/processes/forming/roll-forming.md): Roll forming passes coil stock through a sequence of contoured roller stands that bend it progressively into a constant cross-section profile.
- [Heat Treatment](https://manufacturingprocesses.org/processes/forming/heat-treatment.md): Heat treatment cycles metal through controlled heating and cooling to change its hardness, strength, ductility or internal stress state.
- [Metal Stamping](https://manufacturingprocesses.org/processes/forming/metal-stamping.md): Metal stamping shears and forms sheet metal between matched dies in a press, producing flat and shallow-formed parts at very high rates.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/forming/metal-extrusion)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Metal Injection Molding"
category: "Forming"
subcategory: "Metal"
materials: ["Metal"]
tolerances: "About ±0.3% of the dimension as standard (roughly ±0.003 in on a 1 in feature), with ±0.1–0.2% achievable on controlled features or after sizing; machine anything tighter"
volumes: "20,000–1,000,000+ parts per year; below about 5,000 machining is usually cheaper"
lead_time: "8–14 weeks for tooling and first articles; 4–8 weeks per production run thereafter, set largely by furnace scheduling"
url: https://manufacturingprocesses.org/processes/forming/metal-injection-molding
---

# Metal Injection Molding

Metal injection molding molds a powder-and-binder feedstock exactly like a plastic, then debinds and sinters it into a dense small metal part.

- **Category**: [Forming](https://manufacturingprocesses.org/processes/forming.md)
- **Family**: Metal
- **Materials**: Metal
- **Typical tolerances**: About ±0.3% of the dimension as standard (roughly ±0.003 in on a 1 in feature), with ±0.1–0.2% achievable on controlled features or after sizing; machine anything tighter
- **Surface finish**: Roughly 32 µin Ra (0.8 µm) as sintered, reflecting the mold finish
- **Typical volumes**: 20,000–1,000,000+ parts per year; below about 5,000 machining is usually cheaper
- **Lead time**: 8–14 weeks for tooling and first articles; 4–8 weeks per production run thereafter, set largely by furnace scheduling

## Overview

Metal injection molding (MIM) blends fine metal powder with a thermoplastic binder into a feedstock that is molded on a conventional injection molding machine, then chemically and thermally strips the binder away and sinters the remaining powder skeleton into a dense metal part. The molded "green" part shrinks 15–20% linearly during sintering — uniformly, and by an amount the toolmaker builds into the cavity.

The result is a small, geometrically complex metal part at injection-molding economics: sintered density typically 96–99% of theoretical, so mechanical properties approach those of wrought material, unlike conventional press-and-sinter powder metallurgy.

MIM's sweet spot is parts under about 100 g with sections under 0.25 in (6 mm) — surgical instrument jaws, firearm components, orthodontic brackets, watch cases, connector shells, and phone hinges. Standard dimensional tolerance is roughly ±0.3% of the dimension. Below about 5,000–20,000 parts a year, machining usually wins.

## How it works

1. **Feedstock.** Metal powder finer than about 20 µm is compounded with a multi-component binder — typically wax or polymer blends — to roughly 55–65% powder by volume. The powder must be spherical and fine, which is why MIM material costs far more per pound than bar stock.
2. **Molding.** The feedstock is injected into a steel tool on a standard injection molding machine. The green part looks like the finished part but oversized, and it is fragile — it is essentially powder held together by wax.
3. **Debinding.** The primary binder is removed by solvent immersion, catalytic decomposition, or thermal means, leaving an open pore network and a secondary backbone binder holding the shape. This is the slowest step and its rate is governed by section thickness, which is exactly why MIM has a wall thickness ceiling.
4. **Sintering.** The brown part is fired in a controlled atmosphere or vacuum furnace below the melting point — roughly 2,300–2,450°F (1,260–1,340°C) for stainless steels. The residual binder burns out and the powder particles fuse and densify, with the part shrinking 15–20% in every direction.
5. **Secondary.** Optional sizing, machining of critical features, heat treatment (17-4PH is commonly solution treated and aged), and finishing. See the [surface finish chart](/charts/surface-finish-chart) for achievable finishes.

The whole process depends on the shrinkage being isotropic and repeatable. Part-to-part variation comes from density gradients in the green part — the same molding defects that cause sink and warp in plastic — which is why MIM design rules track plastic injection molding rules closely. The [injection molding design guidelines](/charts/injection-molding-design-guidelines) apply almost directly, and the [metal melting points](/charts/metal-melting-points) chart shows why sintering happens well below the melt.

## Design guidelines

### Wall thickness: uniform and thin

Nominal walls of 0.040–0.160 in (1–4 mm) are ideal, with a practical ceiling around 0.25 in (6 mm). Thick sections are not a strength problem, they are a debinding problem — the binder has to diffuse out of the center, and time scales steeply with thickness. Core out heavy sections just as you would in plastic.

### Part size and mass

MIM is economic on small parts. Typical parts fall between about 0.1 g and 100 g; above that, feedstock cost and furnace capacity make it uncompetitive against machining or investment casting.

### Draft

0.5–2° of draft is normal and makes ejection reliable. Zero draft is achievable on short walls with a good tool finish, since sintering shrinkage pulls the part away from the steel, but it should be a deliberate decision rather than an oversight.

### Ribs, bosses, and corners

Follow injection molding practice: ribs at roughly 0.6× nominal wall to avoid sink, radiused internal corners, and cored bosses rather than solid ones. Sharp internal corners crack the green part during handling and debinding, before the metal ever exists.

### Undercuts, threads, and holes

Side actions and slides work the same as in plastic tooling. Cross holes, external threads, and internal detail are all moldable. Small holes down to about 0.010 in (0.25 mm) can be molded, but blind holes deeper than roughly 5× diameter risk pin damage. Threads are moldable but often finished by machining when the class of fit matters.

### Support during sintering

The part is soft at sintering temperature and sags under its own weight. Long unsupported spans, thin overhanging flanges, and tall thin walls distort. Flatness and straightness on a MIM part are set by the sintering setter, not the mold, so give the supplier flat surfaces to rest on.

### Tolerances

Standard tolerance is about ±0.3% of a dimension, with ±0.1–0.2% achievable on well-controlled features or with post-sinter sizing. On a 1.000 in dimension, that is roughly ±0.003 in standard. Anything tighter — a bearing bore or a sealing face — should be machined after sintering.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Nominal wall | 0.040–0.160 in (1–4 mm) | 0.25 in (6 mm) | Debinding time rises steeply with section |
| Part mass | 0.1–60 g | ~100 g | Feedstock cost and furnace capacity |
| Draft | 0.5–2° | 0° on short walls | Sintering shrinkage aids release |
| Rib thickness | 0.6× nominal wall | 0.8× wall | Thick ribs sink and create density gradients |
| Molded hole diameter | ≥ 0.020 in (0.5 mm) | 0.010 in (0.25 mm) | Fine core pins survive only short depths |
| Blind hole depth | ≤ 3× diameter | 5× diameter | Long core pins deflect in the feedstock |
| Linear tolerance | ±0.3% of dimension | ±0.1–0.2% with control | Shrinkage repeatability governs everything |
| Flatness | Provide a flat setter face | — | Parts sag on their supports during sintering |

## Cost drivers

Three cost centers: the tool, the feedstock, and the furnace. The mold is a conventional injection molding tool sized up for shrinkage, so its cost is comparable to a plastic tool of the same complexity — modest against a die-casting die. Feedstock is expensive because MIM requires fine spherical powder, so material cost per part is high relative to bar stock. Furnace time is a batch operation, and long debinding and sintering cycles mean throughput, not cycle time, sets capacity.

Volume breakpoints: below roughly 5,000 parts a year, machining or 3D printing beats MIM because the tool cannot amortize. From 20,000 to a few hundred thousand is the classic MIM window. Above a million, the process is fully in its element, with multi-cavity tooling driving molding cost toward negligible and furnace throughput becoming the constraint.

1. **Consolidate an assembly into one part.** MIM's biggest wins come from replacing three machined components and a welding step with one molded part.
2. **Thin the walls.** Thinner sections debind faster, use less expensive powder, and cost less per part in every dimension.
3. **Avoid post-machining.** Every machined feature adds a fixturing operation on a small part that is awkward to hold. Tolerance to ±0.3% wherever function allows.
4. **Use multi-cavity tooling.** Molding is the cheap step; more cavities spread it further.
5. **Pick a common alloy.** 316L and 17-4PH have the deepest supply base and the best-characterized shrinkage; exotic alloys carry both a material and a process-development premium.

## FAQ

### How much does a MIM part shrink during sintering?

15–20% linearly, in every direction. The mold cavity is cut oversize by exactly that factor, which is why shrinkage repeatability — not molding precision — is what sets MIM's dimensional tolerance of roughly ±0.3%.

### What is the maximum wall thickness for metal injection molding?

Around 0.25 in (6 mm), with 0.040–0.160 in (1–4 mm) ideal. The limit comes from debinding: the binder has to diffuse out of the section, and the time required rises steeply with thickness. Core out heavy sections exactly as you would in plastic injection molding.

### How dense are MIM parts?

Typically 96–99% of theoretical density, which is why mechanical properties approach wrought material. This is the main distinction from conventional press-and-sinter powder metallurgy, which usually leaves 10–15% porosity and correspondingly lower strength and fatigue performance.

### What materials can be metal injection molded?

316L and 17-4PH stainless are the highest-volume grades, along with low-alloy steels such as 4605, soft magnetic iron-nickel alloys, tool steels, titanium, and tungsten heavy alloys. Alloy availability follows powder availability, and fine spherical powder is what drives the material cost.

### What volume justifies MIM over machining?

Roughly 5,000–20,000 parts a year as a crossover, depending on how much machining the part would otherwise need. MIM's advantage grows with geometric complexity — a part needing five machining setups crosses over far sooner than a simple turned one.

### MIM or investment casting?

MIM wins below about 100 g, where its molded detail, 0.040 in walls, and 32 µin Ra surfaces beat what a ceramic shell can deliver. Investment casting wins above that mass and for larger sections, since MIM's debinding limit caps section thickness near 0.25 in regardless of part size.

## Alternative processes

- [Investment Casting](https://manufacturingprocesses.org/processes/forming/investment-casting.md): Investment casting builds a ceramic shell around a wax pattern, melts the wax out and pours metal into the cavity, giving fine detail in almost any alloy.
- [CNC Machining](https://manufacturingprocesses.org/processes/cutting/cnc-machining.md): CNC machining removes material from solid stock with a programmed cutting tool, producing accurate parts directly from a CAD model with no tooling.
- [Direct Metal Laser Sintering (DMLS)](https://manufacturingprocesses.org/processes/forming/direct-metal-laser-sintering-dmls.md): Direct metal laser sintering fuses metal powder layer by layer with a laser, building dense metal parts anchored to a supported build plate.
- [Swiss Screw Machining](https://manufacturingprocesses.org/processes/cutting/swiss-screw-machining.md): Swiss screw machining feeds bar stock through a guide bushing so cutting happens right next to the support, holding tight tolerances on slender parts.

## Related processes

- [Injection Molding](https://manufacturingprocesses.org/processes/forming/injection-molding.md): Injection molding forces molten thermoplastic into a steel mold under high pressure, producing high-volume identical plastic parts.
- [Powder Metallurgy (Press and Sinter)](https://manufacturingprocesses.org/processes/forming/powder-metallurgy-press-and-sinter.md): Press-and-sinter powder metallurgy compacts metal powder in a rigid die and sinters it below its melting point into a finished net-shape part.
- [Investment Casting](https://manufacturingprocesses.org/processes/forming/investment-casting.md): Investment casting builds a ceramic shell around a wax pattern, melts the wax out and pours metal into the cavity, giving fine detail in almost any alloy.
- [Die Casting](https://manufacturingprocesses.org/processes/forming/die-casting.md): Die casting injects molten aluminum, zinc or magnesium into a hardened steel die under high pressure, producing thin-walled net-shape metal parts.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/forming/metal-injection-molding)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Metal Spinning"
category: "Forming"
subcategory: "Metal"
materials: ["Metal"]
tolerances: "Roughly ±0.010–0.030 in (±0.25–0.75 mm) on diameters, tighter on CNC machines with hard mandrels; wall thickness varies through the part in conventional spinning and follows the sine law in shear spinning"
volumes: "1–5,000 parts; above about 5,000 deep drawing usually wins on cycle time"
lead_time: "1–3 weeks including mandrel; days for repeat work on an existing mandrel"
url: https://manufacturingprocesses.org/processes/forming/metal-spinning
---

# Metal Spinning

Metal spinning presses a rotating metal disc against a mandrel with a roller or hand tool, forming it into an axially symmetric hollow shape.

- **Category**: [Forming](https://manufacturingprocesses.org/processes/forming.md)
- **Family**: Metal
- **Materials**: Metal
- **Typical tolerances**: Roughly ±0.010–0.030 in (±0.25–0.75 mm) on diameters, tighter on CNC machines with hard mandrels; wall thickness varies through the part in conventional spinning and follows the sine law in shear spinning
- **Surface finish**: Spiral tool marks are inherent and are polished out where cosmetics matter; a burnished, work-hardened surface otherwise
- **Typical volumes**: 1–5,000 parts; above about 5,000 deep drawing usually wins on cycle time
- **Lead time**: 1–3 weeks including mandrel; days for repeat work on an existing mandrel

## Overview

Metal spinning presses a rotating sheet metal disc against a mandrel with a roller or hand tool, working it progressively until it takes the mandrel's shape. It is the cheapest way to make an axially symmetric hollow metal part: the tooling is a single mandrel, often turned from hardwood, MDF, or aluminum, and it costs a small fraction of a draw die.

Typical parts are cones, hemispheres, dished heads, reflectors, tank ends, cookware, lighting housings, nose cones, and ducting transitions. Blank thickness usually runs 0.020–0.250 in (0.5–6 mm), and diameters range from a few inches to well over ten feet on large machines. Aluminum, copper, brass, low-carbon steel, and 304 stainless all spin well.

Cycle times of minutes rather than seconds cap the practical volume near a few thousand pieces a year, but with no die to pay for, spinning is economic from a single part and is the standard choice for prototypes of parts that will eventually be deep drawn.

## How it works

1. **Blank and mandrel.** A circular blank is cut and clamped between the mandrel (mounted on the headstock) and a tailstock follower block. The mandrel is the part's internal shape.
2. **Spin up.** The lathe brings the assembly to speed, typically several hundred to a couple of thousand rpm depending on diameter — bigger blanks run slower to keep surface speed reasonable.
3. **Forming passes.** A roller or hand tool is worked against the blank in a sequence of passes, walking the metal down against the mandrel a little at a time. The metal is pushed, not stretched: in **conventional spinning** the wall thickness stays approximately constant while the blank diameter reduces, which means every pass moves material inward rather than thinning it.
4. **Lubrication and heat.** Spinning generates real heat, and lubricant is applied continuously. Thick blanks and hard materials are sometimes spun hot.
5. **Interstage annealing.** The metal work hardens as it deforms. Stainless steel and heavily worked aluminum usually need annealing between stages, or they will crack.
6. **Trim and finish.** The rim of the spun part is uneven and gets trimmed; the mandrel is withdrawn, and spiral tool marks are polished off if cosmetics matter.

### Shear spinning and flow forming

**Shear spinning** deliberately thins the wall instead of feeding material inward. Wall thickness follows the sine law — the formed wall equals the blank thickness multiplied by the sine of the half-angle between the mandrel surface and the axis — so a 30° cone from a 0.100 in blank yields a 0.050 in wall. The blank diameter stays the same as the finished part diameter, which makes shear spinning far more material-efficient for cones.

**Flow forming** applies the same principle to tubes, extending a preform axially over a mandrel with rollers to produce long, thin-walled, very concentric cylinders — pressure vessels, rocket motor cases, and drive shafts.

## Design guidelines

### Axial symmetry is mandatory

Everything about the part must be a body of revolution. Flats, lugs, and off-axis holes are secondary operations. Re-entrant shapes — where the diameter reduces and then increases again — require a segmented or collapsible mandrel, which is a substantially more expensive tool.

### Corner radii

Sharp corners at the transition between the base and the wall thin locally and are the usual failure point. Use a radius of at least 1–2× material thickness, and larger on hard materials. A radius equal to several times the thickness spins far more easily than the minimum.

### Depth and wall angle

Shallow, open shapes spin easily. Deep cylindrical shapes with near-vertical walls are the hardest cases in conventional spinning, because material must travel a long way with little support. Cones and dished shapes are the natural geometry.

### Wall thickness

In conventional spinning, wall thickness stays close to the blank thickness but is not constant — expect some thinning at the transitions and some thickening near the rim. If a specific finished wall is required, shear spinning gives a predictable result from the sine law; conventional spinning does not.

### Material selection

Annealed tempers spin best: 1100, 3003, and 5052-O aluminum, C11000 copper, cartridge brass, drawing-quality low-carbon steel, and 304 stainless. Hardened and precipitation-strengthened tempers crack. Specify the temper explicitly. Blank thickness selection from the [sheet metal gauge chart](/charts/sheet-metal-gauge-chart); flat-pattern development for any secondary flanges from the [bend radius and K-factor chart](/charts/sheet-metal-bend-radius-k-factor).

### Trim allowance

Leave 0.25–0.5 in (6–13 mm) of extra length at the open end for trimming. The rim of a spun part is always uneven.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Geometry | Body of revolution | Re-entrant shapes need a collapsible mandrel | The mandrel must withdraw from the finished part |
| Base-to-wall radius | ≥ 3× thickness | 1–2× thickness | Sharp transitions thin locally and split |
| Wall angle | Cones and dished forms | Deep vertical walls are hardest | Unsupported metal must travel far |
| Wall thickness control | Accept variation | Use shear spinning for a defined wall | Conventional spinning does not control thickness |
| Material temper | Annealed (O temper) | Hard tempers crack | Work hardening accumulates every pass |
| Trim allowance | 0.25–0.5 in (6–13 mm) | — | The formed rim is always uneven |
| Interstage anneal | Required for stainless | — | Work hardening exhausts ductility |

## Cost drivers

Spinning inverts the deep drawing cost structure: tooling is trivial and per-part labor is high. A mandrel in hardwood, MDF, or aluminum can be turned in a day or two, so entry cost is minimal. But every part takes minutes of skilled operator or CNC machine time, and multi-stage parts need annealing cycles between passes.

Volume breakpoints: 1 to a few hundred parts is where spinning is unbeatable — no other process gets a 40 in diameter dished head made this week. From a few hundred to a few thousand, CNC spinning with a hardened steel mandrel keeps it competitive. Above roughly 5,000 a year, deep drawing's seconds-per-part cycle overtakes spinning despite the die cost.

1. **Design a family of parts on one mandrel.** Different depths and trim diameters from the same mandrel reuse the whole tool.
2. **Open the corner radii.** Generous radii spin in fewer passes and crack less, which shows up directly in labor.
3. **Avoid re-entrant geometry.** A collapsible or segmented mandrel can cost more than the rest of the job.
4. **Use shear spinning for cones.** It starts from a blank the same diameter as the finished part, cutting material use substantially versus conventional spinning.
5. **Specify an annealed temper.** Cracked parts from hard-temper stock are the most common avoidable scrap in spinning.

## FAQ

### What shapes can be metal spun?

Bodies of revolution — cones, hemispheres, dished heads, cylinders, and flared or stepped profiles. Re-entrant shapes, where the diameter reduces and then grows again, require a segmented or collapsible mandrel, which is much more expensive than a solid one.

### Does metal spinning change the wall thickness?

In conventional spinning, thickness stays close to the blank thickness but varies through the part, thinning at transitions and thickening near the rim. Shear spinning deliberately thins the wall following the sine law: the wall equals blank thickness times the sine of the half-angle, so a 30° cone from a 0.100 in blank gives a 0.050 in wall.

### What volume justifies metal spinning over deep drawing?

Spinning wins from one part up to a few thousand a year, because the mandrel is inexpensive and can be turned in days. Deep drawing wins above roughly 5,000, where its seconds-per-part cycle time overcomes the cost of a matched draw die.

### What materials spin well?

Annealed tempers: 1100, 3003, and 5052-O aluminum, copper, cartridge brass, drawing-quality low-carbon steel, and 304 stainless. Hard and precipitation-strengthened tempers crack. Stainless and heavily worked aluminum usually need annealing between stages, since the metal work hardens on every pass.

### What tolerance can metal spinning hold?

Roughly ±0.010–0.030 in (±0.25–0.75 mm) on diameters, and tighter on CNC machines with hardened steel mandrels. Diameters formed hard against the mandrel hold better than free-formed rim regions, which is also why a trim allowance of 0.25–0.5 in is standard.

### What is flow forming?

Shear spinning applied to tube. A short thick preform is extended axially over a mandrel by rollers, producing a long, thin-walled, highly concentric cylinder. It is how pressure vessel bodies, rocket motor cases, and lightweight drive shafts are made.

## Alternative processes

- [Deep Drawing](https://manufacturingprocesses.org/processes/forming/deep-drawing.md): Deep drawing pushes a sheet metal blank through a die with a punch so the metal flows into a seamless cup or box deeper than its own diameter.
- [Hydroforming](https://manufacturingprocesses.org/processes/forming/hydroforming.md): Hydroforming presses sheet or tube into a single-sided die with high-pressure fluid, producing smooth seamless shapes that would need several stampings.
- [Sand Casting](https://manufacturingprocesses.org/processes/forming/sand-casting.md): Sand casting pours molten metal into a cavity formed in bonded sand around a pattern, then breaks the mold away to release the part.
- [CNC Turning](https://manufacturingprocesses.org/processes/cutting/cnc-turning.md): CNC turning rotates the workpiece against a stationary tool, producing round parts with excellent concentricity and short cycle times.

## Related processes

- [Deep Drawing](https://manufacturingprocesses.org/processes/forming/deep-drawing.md): Deep drawing pushes a sheet metal blank through a die with a punch so the metal flows into a seamless cup or box deeper than its own diameter.
- [Press Braking](https://manufacturingprocesses.org/processes/forming/press-braking.md): Press braking bends sheet metal by driving a punch into a V-die, forming one straight-line bend at a time.
- [Panel Beating](https://manufacturingprocesses.org/processes/forming/panel-beating.md): Panel beating shapes sheet metal by hand with hammers, dollies and an English wheel, producing one-off compound-curved panels without any dies.
- [Hydroforming](https://manufacturingprocesses.org/processes/forming/hydroforming.md): Hydroforming presses sheet or tube into a single-sided die with high-pressure fluid, producing smooth seamless shapes that would need several stampings.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/forming/metal-spinning)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Metal Stamping"
category: "Forming"
subcategory: "Metal"
materials: ["Metal"]
tolerances: "±0.005 in (±0.13 mm) typical on blanked and pierced features; ±0.001–0.002 in (±0.025–0.05 mm) achievable in precision dies. Formed dimensions spanning several stations hold looser than single-station features"
volumes: "25,000–10,000,000+ parts per year; below about 10,000 laser cutting and press braking usually win"
lead_time: "10–20 weeks for progressive die design, build, and tryout; 1–3 weeks per production release thereafter"
url: https://manufacturingprocesses.org/processes/forming/metal-stamping
---

# Metal Stamping

Metal stamping shears and forms sheet metal between matched dies in a press, producing flat and shallow-formed parts at very high rates.

- **Category**: [Forming](https://manufacturingprocesses.org/processes/forming.md)
- **Family**: Metal
- **Materials**: Metal
- **Typical tolerances**: ±0.005 in (±0.13 mm) typical on blanked and pierced features; ±0.001–0.002 in (±0.025–0.05 mm) achievable in precision dies. Formed dimensions spanning several stations hold looser than single-station features
- **Surface finish**: Inherits the coil finish, typically 16–63 µin Ra (0.4–1.6 µm); sheared edges carry a burr up to about 10% of thickness
- **Typical volumes**: 25,000–10,000,000+ parts per year; below about 10,000 laser cutting and press braking usually win
- **Lead time**: 10–20 weeks for progressive die design, build, and tryout; 1–3 weeks per production release thereafter

## Overview

Metal stamping shears and forms sheet metal between matched punch-and-die sets in a mechanical or servo press, producing flat blanks and shallow-formed parts at rates no other metal process approaches. A progressive die running at 200 strokes per minute makes 12,000 parts an hour from a single coil, and high-speed lines for terminals and lead frames exceed 1,000 strokes per minute.

Operations combine freely in one die: blanking, piercing, notching, bending, coining, embossing, lancing, and drawing. Typical parts are brackets, clips, contacts, shielding cans, hinges, lamination stacks, and structural body panels, in low-carbon steel, high-strength steel, stainless, aluminum, copper, and brass from about 0.005 to 0.250 in (0.13 to 6 mm).

The economics are tooling-driven. A progressive die is a substantial investment with a multi-month build, so stamping is a high-volume answer — generally above 25,000–50,000 parts a year.

## How it works

### How does the cut actually happen?

Shearing is fracture, not cutting. The punch first pushes the sheet down (rollover), then penetrates and burnishes a smooth band, and then a crack initiates from the punch and die edges and runs through the remaining thickness. A correctly clearanced cut has the two cracks meet, giving a clean break; wrong clearance makes them miss, leaving a secondary shear band and a heavy burr. Die clearance is roughly 5–10% of material thickness per side — toward the low end for soft aluminum, the high end for high-strength steel.

### The die types

- **Progressive dies** carry the strip through a series of stations on a carrier web, adding one operation per stroke, and cut the part free at the last station. This is the high-volume standard.
- **Transfer dies** cut the blank free first and move it mechanically between stations, which allows deeper forms than a carrier strip can support.
- **Compound dies** do several operations in a single stroke at one station, giving excellent flatness and concentricity for simple parts.

### The cycle

1. **Coil feed.** Stock is uncoiled, straightened through a leveler, and fed by a servo roll feed to a pitch accuracy typically within a few thousandths of an inch.
2. **Pilot.** Pilot pins enter pre-pierced holes to register the strip precisely before any station works it.
3. **Stroke.** All stations act simultaneously. Tonnage required is approximately cut perimeter × thickness × material shear strength; shear strength runs roughly 70–80% of tensile strength for common steels.
4. **Form.** Bending, drawing, and coining stations shape the part while it is still attached to the carrier.
5. **Cutoff.** The final station separates the part; the skeleton scrap is chopped and recycled.
6. **Secondary.** Deburring, tumbling, tapping, plating, or heat treatment as required.

Press selection follows tonnage and speed. Servo presses allow the slide velocity to be programmed within the stroke, which is what makes high-strength steels and deep forms viable at rate.

## Design guidelines

### Hole size and spacing

Minimum punched hole diameter is about 1× material thickness in mild steel and 1.5–2× in stainless and high-strength grades — below that, punches break. Keep holes at least 2× thickness from any edge and 2× thickness from each other, and make slots at least 1.5× thickness wide.

### Distance from a hole to a bend

Keep the edge of a hole at least 2.5× material thickness plus the bend radius away from the bend line. Closer and the hole distorts into an oval as the metal stretches around the bend.

### Bend radius and grain direction

Inside bend radius should be at least 1× material thickness for mild steel and more for high-strength or hardened tempers. Bend across the rolling direction where possible; bending parallel to the grain cracks the outer fiber, which is a routine failure in 5000- and 6000-series aluminum. Flat patterns come from the [bend radius and K-factor chart](/charts/sheet-metal-bend-radius-k-factor), and stock thickness from the [sheet metal gauge chart](/charts/sheet-metal-gauge-chart).

### Bend relief

Where a bend ends at a formed edge, add a relief notch at least as wide as the material thickness and deeper than the bend radius plus thickness. Without relief the material tears at the bend end.

### Edge condition and burr

Every sheared edge has a burr on the die side. Specify the burr side on the drawing if it matters; a common acceptance limit is 10% of material thickness. Call out deburring or tumbling explicitly rather than assuming it.

### Flatness

Blanking induces stress and parts come out with some bow and camber. If flatness is functional, specify it — restrike or a leveling operation adds cost but is the only way to get it.

### Tolerances

Features made within one die station hold much better than features spanning several stations, because station-to-station accuracy depends on strip pitch and pilot registration. Group critical dimensions into the same station wherever the part permits.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Punched hole diameter | ≥ 1.5× thickness | 1× thickness (mild steel) | Slender punches deflect and snap |
| Hole to edge | ≥ 2× thickness | 1.5× thickness | Thin webs bulge and tear during shearing |
| Slot width | ≥ 1.5× thickness | 1× thickness | Narrow punch sections lack column strength |
| Hole to bend line | ≥ 2.5× thickness + bend radius | — | Metal stretching around the bend ovals the hole |
| Inside bend radius | ≥ 1× thickness | Larger for high-strength grades | Outer fiber cracks below this |
| Bend relief | Width ≥ 1× thickness, depth > radius + thickness | — | Prevents tearing at the end of the bend |
| Die clearance | 5–8% of thickness per side | 10% for high-strength steel | Sets whether the shear cracks meet cleanly |
| Burr height | ≤ 10% of thickness | — | Sheared edges always carry a die-side burr |

## Variants

- Secondary Pressing

## Cost drivers

Tooling is the whole story at the front end. A progressive die's cost scales with station count, part size, tolerance, and die material, and it takes months to build and try out. Per-part cost is press seconds plus material, and material is usually the larger of the two: strip utilization on a typical part is 50–80%, meaning a fifth to a half of the purchased steel leaves as skeleton scrap.

Volume breakpoints: below roughly 10,000 parts, laser cutting plus press braking has no tooling cost and wins outright. From 25,000 to 250,000 a year, a simple progressive or compound die pays back. Above a million, high-speed dies, multi-out layouts, and material utilization improvements dominate the cost equation.

1. **Improve the nest.** Rotating the part on the strip or running a two-out layout can recover several percent of material on every stroke, forever.
2. **Reduce station count.** Each station is tooling, maintenance, and press length. Combining two bends into one station or dropping a non-functional feature is real money.
3. **Loosen non-critical tolerances.** Tolerances tighter than about ±0.002 in force precision die construction, harder die steel, and more frequent sharpening.
4. **Design for one grain direction.** Parts that require bends in two directions constrain the nest and can force a wider strip.
5. **Choose a common gauge and grade.** Standard coil widths and thicknesses avoid mill minimums and slitting charges.

## FAQ

### What is the minimum hole size for metal stamping?

About 1× material thickness in mild steel and 1.5–2× in stainless or high-strength grades. The limit is punch strength: a punch narrower than the sheet is thick lacks the column strength to survive repeated penetration and will break.

### How far should a hole be from a bend?

At least 2.5× material thickness plus the inside bend radius, measured from the edge of the hole to the bend line. Any closer and the metal stretching around the bend pulls the hole into an oval. If the hole must be closer, pierce it after forming.

### What die clearance should be used for stamping?

Roughly 5–10% of material thickness per side — nearer 5% for soft aluminum, 6–8% for mild steel, and up to 10% for high-strength steel. Correct clearance makes the cracks from the punch and die edges meet, which gives a clean fracture and a small burr.

### How much press tonnage does a stamping need?

Approximately the cut perimeter multiplied by material thickness and by material shear strength. Shear strength runs roughly 70–80% of tensile strength for common steels. Forming, coining, and drawing stations add their own force on top of the shearing requirement.

### Stamping or laser cutting and bending?

Laser cutting plus press braking has zero tooling cost and wins below roughly 10,000 parts a year or wherever the design is still changing. Stamping wins above 25,000–50,000, where the die amortizes and cycle time drops from minutes per part to a fraction of a second.

### Why are stamped parts not perfectly flat?

Shearing and forming leave residual stress that relaxes into bow and camber once the part is cut free. If flatness is functional, specify it on the drawing so the toolmaker can add a restrike station or a post-process leveling operation.

## Alternative processes

- [Laser Cutting](https://manufacturingprocesses.org/processes/cutting/laser-cutting.md): Laser cutting melts, burns or vaporises a narrow kerf through flat stock with a focused beam, producing profiles from a 2D file with no tooling.
- [Sheet Metal Fabrication](https://manufacturingprocesses.org/processes/forming/sheet-metal-fabrication.md): Sheet metal fabrication cuts, bends and joins flat stock into a finished assembly, typically laser cutting followed by press braking, hardware and welding.
- [Punching and Blanking](https://manufacturingprocesses.org/processes/cutting/punching-and-blanking.md): Punching and blanking shear sheet metal between a punch and a die, producing holes and flat outlines at very high speed.
- [Die Casting](https://manufacturingprocesses.org/processes/forming/die-casting.md): Die casting injects molten aluminum, zinc or magnesium into a hardened steel die under high pressure, producing thin-walled net-shape metal parts.

## Related processes

- [Press Braking](https://manufacturingprocesses.org/processes/forming/press-braking.md): Press braking bends sheet metal by driving a punch into a V-die, forming one straight-line bend at a time.
- [Deep Drawing](https://manufacturingprocesses.org/processes/forming/deep-drawing.md): Deep drawing pushes a sheet metal blank through a die with a punch so the metal flows into a seamless cup or box deeper than its own diameter.
- [Punching and Blanking](https://manufacturingprocesses.org/processes/cutting/punching-and-blanking.md): Punching and blanking shear sheet metal between a punch and a die, producing holes and flat outlines at very high speed.
- [Roll Forming](https://manufacturingprocesses.org/processes/forming/roll-forming.md): Roll forming passes coil stock through a sequence of contoured roller stands that bend it progressively into a constant cross-section profile.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/forming/metal-stamping)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Multi Jet Fusion (MJF)"
category: "Forming"
subcategory: "Additive Manufacturing"
materials: ["Plastic"]
tolerances: "±0.008 in (±0.2 mm) or ±0.3% of nominal, whichever is greater. Tighter features are machined or reamed after the build."
volumes: "10–10,000 parts per year; unit cost is close to flat across that range when builds are nested well"
lead_time: "3–7 business days. A full build runs roughly 10–15 hours of printing plus a comparable controlled cooling cycle before depowdering."
url: https://manufacturingprocesses.org/processes/forming/multi-jet-fusion-mjf
---

# Multi Jet Fusion (MJF)

Multi Jet Fusion prints a fusing agent onto nylon powder and drives it with infrared energy, fusing whole layers at once for high throughput.

- **Category**: [Forming](https://manufacturingprocesses.org/processes/forming.md)
- **Family**: Additive Manufacturing
- **Materials**: Plastic
- **Typical tolerances**: ±0.008 in (±0.2 mm) or ±0.3% of nominal, whichever is greater. Tighter features are machined or reamed after the build.
- **Surface finish**: Ra 250–400 µin (6–10 µm) as built, a fine matte grain slightly finer than SLS. Vapor smoothing reaches roughly Ra 40–120 µin (1–3 µm) and seals the surface.
- **Typical volumes**: 10–10,000 parts per year; unit cost is close to flat across that range when builds are nested well
- **Lead time**: 3–7 business days. A full build runs roughly 10–15 hours of printing plus a comparable controlled cooling cycle before depowdering.

## Overview

Multi Jet Fusion (MJF) is a powder bed fusion process that inkjets a carbon-black fusing agent onto nylon powder and then sweeps an infrared lamp across the whole layer, fusing every printed region at once rather than tracing them with a laser. A detailing agent is jetted at the boundaries to sharpen edges and suppress thermal bleed. Standard layer height is 0.0031 in (80 µm).

Fusing whole layers is what makes MJF fast: a full 15 × 11.2 × 15 in (380 × 284 × 380 mm) build takes on the order of 10–15 hours regardless of how many parts are packed into it, so unit cost falls sharply with nesting density. Parts come out gray to near-black, dense, and closer to isotropic than [SLS](/processes/forming/selective-laser-sintering-sls) — PA12 lands near 7,000 psi (48 MPa) tensile with 15–20% elongation in XY.

Accuracy is ±0.008 in (±0.2 mm) or ±0.3% of nominal, whichever is greater. MJF is the usual choice for functional nylon parts from tens to about 10,000 pieces a year.

## How it works

1. **Preheat the powder bed.** PA12 powder is spread and the build area is held just below the melting range, so the lamps need only supply the final energy to fuse.
2. **Recoat.** A 0.0031 in (80 µm) layer is spread across the bed by a recoater moving in one axis.
3. **Jet agents.** A carriage carrying thermal inkjet arrays passes over the layer, depositing fusing agent where the cross-section is solid and detailing agent immediately around its perimeter. The fusing agent is carbon-loaded so it absorbs infrared strongly; the detailing agent cools the boundary to keep it sharp.
4. **Fuse.** Infrared lamps on the same carriage sweep the full layer. Every printed region reaches fusion temperature simultaneously — this is the throughput advantage over point-by-point laser scanning.
5. **Repeat.** The bed indexes down and steps 2–4 repeat. Build time is a function of layer count, not part count.
6. **Cool.** The build unit is moved to a processing station and cooled in a controlled cycle. Cooling typically takes as long as the print or longer, and rushing it warps flat parts.
7. **Depowder and finish.** Loose powder is vacuumed and bead-blasted away and recycled at a refresh ratio around 20–30% virgin. Parts are naturally gray with a slightly mottled surface, so most production work is dyed black; vapor smoothing and vibratory finishing are also common.

## Design guidelines

### Wall thickness
0.020 in (0.5 mm) prints, but use 0.032 in (0.8 mm) for anything structural and 0.040 in (1.0 mm) for larger panels. As with any powder bed process, distortion happens during cooling rather than printing.

### Escape holes
Enclosed cavities fill with un-fused powder. Provide two or more escape holes of at least 0.20 in (5 mm) diameter, positioned so compressed air can be blown through in a straight line. Powder that stays in an internal channel is essentially permanent.

### Bulk sections
Keep solid sections under about 0.6 in (15 mm). Heat retained in a thick region continues to fuse surrounding powder, growing the part and causing sink and warp. Shell and rib instead.

### Detail and text
The detailing agent gives MJF crisper edges than SLS. Embossed detail works from 0.020 in (0.5 mm) wide and tall; engraved text is legible from about 0.08 in (2 mm) cap height, which is finer than SLS supports.

### Clearances
Allow 0.016 in (0.4 mm) between moving surfaces on small features and 0.032 in (0.8 mm) on larger assemblies. MJF's thermal bleed is tighter than laser sintering, so gaps can be smaller, but powder still has to be blown out from between the surfaces.

### Nesting and orientation
Because you are billed on the fraction of the build unit consumed, geometry that nests densely is materially cheaper. Orientation still matters for cosmetics: down-facing surfaces and steep walls pick up slightly more texture, and long flat parts should be angled to spread thermal load.

### Threads
Use heat-set or press-in inserts below M6. Direct-tapped threads in MJF PA12 hold better than in FDM but still strip under repeated assembly; see the [tap drill chart](/charts/tap-drill-chart) for pilot diameters.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Wall thickness | 0.032 in (0.8 mm) | 0.020 in (0.5 mm) | Thin walls distort during cooling |
| Escape hole | 0.30 in (8 mm) | 0.20 in (5 mm) | Powder must be blown clear |
| Hole diameter | 0.040 in (1.0 mm) | 0.020 in (0.5 mm) | Smaller holes retain fused powder |
| Embossed detail | 0.032 in (0.8 mm) | 0.020 in (0.5 mm) | Detailing agent limits edge sharpness |
| Clearance, moving | 0.032 in (0.8 mm) | 0.016 in (0.4 mm) | Powder must clear the gap |
| Solid section | 0.4 in (10 mm) | 0.6 in (15 mm) | Heat soak grows and warps the region |

## Cost drivers

MJF is priced on the fraction of the build unit a part occupies, and because a full build takes roughly the same time whether it holds one part or a thousand, packing density dominates everything else. Bureaus running MJF at high utilization can quote per-part prices that stay nearly flat from 10 pieces to several thousand — the reason MJF has displaced low-volume injection molding for many nylon housings and brackets.

Powder is the second cost. A refresh ratio near 20–30% virgin is typical, better than laser sintering, but you still pay for powder that leaves as waste. Dyeing, vapor smoothing, and inserts are line items on top.

Volume breakpoints: MJF competes with FDM and SLA below about 50 parts, dominates from roughly 50 to 10,000 parts a year for functional nylon, and gives way to [injection molding](/processes/forming/injection-molding) above that — though the crossover moves higher for large or geometrically complex parts where tooling would be expensive.

1. Design for nesting: compact, interlocking geometry buys real savings.
2. Hollow and rib bulk sections; you are charged on the volume you occupy.
3. Batch orders into one build rather than releasing them piecemeal.
4. Skip dyeing on internal parts — undyed gray is functionally identical.
5. Consolidate multi-part assemblies into one printed component to remove fasteners and assembly labor.

## FAQ

### What tolerance can MJF hold?

±0.008 in (±0.2 mm) or ±0.3% of the nominal dimension, whichever is larger. That is tighter than SLS, largely because the detailing agent controls thermal bleed at part boundaries. Bearing bores and dowel holes are still normally reamed.

### Why are MJF parts gray or black?

The fusing agent is carbon-loaded so it will absorb infrared energy, which leaves the finished part a mottled gray. Most production parts are dyed black to even out the appearance; the dye penetrates a few tenths of a millimeter and does not change mechanical properties.

### Is MJF stronger than SLS?

Slightly, and more consistently. MJF PA12 reaches roughly 7,000 psi (48 MPa) tensile with 15–20% elongation in XY and shows less Z-axis penalty than laser sintering because whole layers fuse at once. The practical difference matters most on thin-walled parts loaded across layers.

### What is the minimum wall thickness for MJF?

0.020 in (0.5 mm) will print, but 0.032 in (0.8 mm) is the sensible minimum for anything that carries load, and 0.040 in (1.0 mm) for larger flat panels. Distortion comes from the cooling cycle, not the print itself.

### Does MJF require support structures?

No. Un-fused powder supports every layer, so parts nest in all three axes and internal geometry prints freely. The consequence is that every enclosed cavity must have escape holes of at least 0.20 in (5 mm) to get the powder out.

### How does MJF cost scale with quantity?

Nearly flat. A build takes about the same time whether it contains one part or hundreds, so cost per part is set by how much of the build unit each one occupies. That makes MJF unusually predictable for quantities from tens to several thousand.

### When should I switch from MJF to injection molding?

Somewhere around 5,000–10,000 parts a year for small, simple geometry, where tooling amortizes quickly. For large parts or those with many undercuts and sliding actions, the tooling cost rises and MJF stays competitive well past that point.

## Alternative processes

- [Selective Laser Sintering (SLS)](https://manufacturingprocesses.org/processes/forming/selective-laser-sintering-sls.md): Selective laser sintering fuses nylon powder layer by layer with a laser, so the surrounding loose powder supports the part and no supports are needed.
- [Injection Molding](https://manufacturingprocesses.org/processes/forming/injection-molding.md): Injection molding forces molten thermoplastic into a steel mold under high pressure, producing high-volume identical plastic parts.
- [Fused Deposition Modeling (FDM)](https://manufacturingprocesses.org/processes/forming/fused-deposition-modeling-fdm.md): Fused deposition modeling extrudes molten thermoplastic filament along programmed paths so each layer welds to the one beneath it.
- [CNC Machining](https://manufacturingprocesses.org/processes/cutting/cnc-machining.md): CNC machining removes material from solid stock with a programmed cutting tool, producing accurate parts directly from a CAD model with no tooling.
- [Vacuum Casting (Urethane Casting)](https://manufacturingprocesses.org/processes/forming/vacuum-casting.md): Vacuum casting, known in the US as urethane casting, pours polyurethane resin into a silicone mold under vacuum to copy a master pattern for short runs.

## Related processes

- [Selective Laser Sintering (SLS)](https://manufacturingprocesses.org/processes/forming/selective-laser-sintering-sls.md): Selective laser sintering fuses nylon powder layer by layer with a laser, so the surrounding loose powder supports the part and no supports are needed.
- [Rapid Prototyping](https://manufacturingprocesses.org/processes/forming/rapid-prototyping.md): Rapid prototyping builds a part layer by layer directly from a CAD model, with no tooling and no minimum order quantity.
- [Fused Deposition Modeling (FDM)](https://manufacturingprocesses.org/processes/forming/fused-deposition-modeling-fdm.md): Fused deposition modeling extrudes molten thermoplastic filament along programmed paths so each layer welds to the one beneath it.
- [Binder Jetting](https://manufacturingprocesses.org/processes/forming/binder-jetting.md): Binder jetting prints liquid binder into a powder bed to hold each layer, producing green parts that are later sintered or infiltrated.
- [Stereolithography (SLA)](https://manufacturingprocesses.org/processes/forming/stereolithography-sla.md): Stereolithography cures liquid photopolymer layer by layer with a scanning UV laser, giving the smoothest surfaces in additive manufacturing.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/forming/multi-jet-fusion-mjf)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Overmolding"
category: "Forming"
subcategory: "Plastics and Rubber"
materials: ["Plastic", "Metal"]
tolerances: "Substrate features hold normal injection molding tolerances, about ±0.005 in (±0.127 mm). Dimensions taken over the elastomer are looser and compressible; in insert-transfer overmolding, add the substrate loading clearance to the position tolerance of the second shot."
volumes: "5,000–1,000,000+ parts; two-shot tooling pays back above roughly 100,000"
lead_time: "8–14 weeks for a two-shot rotating tool; 6–10 weeks for a pair of insert-transfer tools; cycles are one conventional molding cycle plus the elastomer shot"
url: https://manufacturingprocesses.org/processes/forming/overmolding
---

# Overmolding

Overmolding molds a second material, usually a soft elastomer, directly onto an already-molded rigid substrate so the two bond permanently.

- **Category**: [Forming](https://manufacturingprocesses.org/processes/forming.md)
- **Family**: Plastics and Rubber
- **Materials**: Plastic, Metal
- **Typical tolerances**: Substrate features hold normal injection molding tolerances, about ±0.005 in (±0.127 mm). Dimensions taken over the elastomer are looser and compressible; in insert-transfer overmolding, add the substrate loading clearance to the position tolerance of the second shot.
- **Surface finish**: Substrate takes any SPI finish; the elastomer is normally textured or matte, since a gloss soft-touch surface shows fingerprints and wear quickly
- **Typical volumes**: 5,000–1,000,000+ parts; two-shot tooling pays back above roughly 100,000
- **Lead time**: 8–14 weeks for a two-shot rotating tool; 6–10 weeks for a pair of insert-transfer tools; cycles are one conventional molding cycle plus the elastomer shot

## Overview

Overmolding injects a second material onto an already-molded rigid substrate so the two become one part — almost always a soft thermoplastic elastomer over a rigid engineering plastic. Power tool grips, toothbrush handles, sealing gaskets molded onto housings, cable strain reliefs, and soft-touch instrument bezels are all overmolded.

It runs two ways. In two-shot (multishot) molding, one machine with two injection units and a rotating or indexing tool molds the substrate and the overmold back to back without the part ever leaving the mold. In insert-transfer overmolding, the substrate is molded separately, then reloaded into a second tool — slower and more labor-intensive, but with far cheaper tooling and no requirement for a two-shot press.

The design question that decides everything is material compatibility: the elastomer must bond chemically to that specific substrate, or the joint has to be mechanical.

## How it works

1. **Mold the substrate.** The rigid first shot is molded conventionally, in ABS, PC, PC/ABS, PP, nylon, or PBT. Its surface must be clean and dry — mold release, fingerprints, and absorbed moisture all degrade the bond.

2. **Position for the second shot.** In two-shot molding the tool rotates the core carrying the substrate into a second cavity, so position is fixed by the tool and repeatable to molding tolerances. In insert-transfer the substrate is reloaded by hand or robot into a separate tool, and the joint now carries the substrate's tolerance plus the locating clearance.

3. **Inject the elastomer.** TPE melt temperatures typically run 355–445 °F (180–230 °C). The hot second shot momentarily softens the substrate skin, and the two polymer chains interdiffuse at the interface. That interdiffusion is the bond — it is not adhesion in the glue sense, and it only happens between compatible chemistries.

4. **Cool and eject.** The elastomer shrinks more than the substrate, which is why a thin or asymmetric substrate can bow after overmolding. A substrate stiff enough to resist that differential is a design requirement, not a nicety.

### Which elastomers bond to which substrates?

The compatibility rule follows chemical family. SEBS-based TPEs bond to polypropylene, polyethylene, and polystyrene. TPU bonds to ABS, PC, PC/ABS, nylon, and PBT. TPV bonds to polypropylene. Standard silicone bonds to essentially nothing without a self-bonding grade or a primer. Bond quality is verified with a 90° peel test, and a good result fails cohesively — tearing within the elastomer — rather than adhesively, cleanly separating at the interface.

### What if the materials do not bond?

Design a mechanical interlock: through-holes the elastomer flows into and rivets over, undercut grooves, dovetails, or a wrap-around edge. Any structural or sealing overmold should carry mechanical retention regardless, since chemical bonds degrade with heat aging, UV, and chemical exposure.

## Design guidelines

### Overmold wall thickness

0.020–0.080 in (0.5–2.0 mm) covers most soft-touch and grip layers, with 0.030–0.060 in (0.75–1.5 mm) the common target. Below about 0.020 in the elastomer will not fill reliably at typical grip lengths. Above about 0.120 in (3 mm) the soft layer feels spongy and its own shrinkage starts pulling on the substrate.

### Keep the elastomer layer uniform

The same rule as any molded part, but it matters more here because thick regions shrink more and drag the substrate. Where the grip must vary in thickness, put the variation in the substrate and keep the overmold constant.

### Substrate design

Design the substrate as a proper molded part first — the wall, rib, boss, corner, and draft rules on [/charts/injection-molding-design-guidelines](/charts/injection-molding-design-guidelines) all apply to it unchanged. Then add stiffness where the overmold will pull: a thin, flat substrate with a heavy elastomer on one face will bow.

### Terminate the edge properly

Never end an overmold on a flat, exposed edge — it becomes a lifting point that a fingernail or a cleaning cloth will start peeling. Terminate into a step, a groove, or a shallow recess in the substrate so the elastomer edge is captured and sits flush.

### Gate the second shot carefully

The elastomer is being injected against a plastic substrate rather than steel. Gate so the flow front does not impinge directly on a thin substrate wall, or it will deflect or wash. Gate location also determines where the elastomer knit lines land, and knit lines in a soft grip are visible as well as weak.

### Bond area and mechanical backup

Give the bond meaningful surface area — a narrow strip of TPE on a small footprint will peel regardless of chemistry. For any overmold that seals, carries load, or is exposed to heat, solvents, or UV, add through-holes or undercut grooves as mechanical retention rather than relying on interdiffusion that degrades over the product's life.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Overmold wall | 0.030–0.060 in (0.75–1.5 mm) | 0.020–0.120 in (0.5–3 mm) | Thin will not fill; thick feels spongy and pulls the substrate |
| Overmold uniformity | Constant thickness | — | Thick regions shrink more and bow the substrate |
| Edge termination | Into a step or groove | Never a flat exposed edge | An exposed edge is a peel initiation point |
| Substrate stiffness | Ribbed or thick enough to resist bow | — | Elastomer shrinkage loads the substrate asymmetrically |
| Bond verification | 90° peel, cohesive failure | Adhesive failure | Cohesive failure means the bond exceeds the material |
| Mechanical interlock | Through-holes or undercut grooves | — | Required where chemistry does not bond or will age |
| Substrate condition | Clean, dry, release-free | — | Contamination is the most common bond failure cause |

## Cost drivers

The route decides the cost structure. Two-shot molding needs a multi-material press and a rotating or indexing tool, so tooling runs well above a pair of single-shot molds — but the cycle is one operation, position is fixed by the tool, and there is no handling between shots. Insert-transfer needs only two conventional tools and a standard press, but adds a load-and-unload step to every part and stacks the substrate's tolerance onto the joint.

Volume breakpoints:

- Under 5,000 parts: insert-transfer with two simple tools, or a mechanically assembled elastomer part.
- 5,000–100,000: insert-transfer with a robot loader, or a two-shot tool if the program justifies it.
- Over 100,000: two-shot, essentially always. The eliminated handling step dominates.

Cost reduction:

1. **Pick a compatible material pair first.** Choosing a TPE that bonds to the chosen substrate avoids primers, plasma treatment, or a redesign around mechanical interlocks — all of which are more expensive than the material decision would have been.
2. **Reduce overmold coverage.** Soft material is usually the more expensive resin, and a grip that covers only the contact zones costs less than one that wraps the whole housing.
3. **Design a single, simple parting for the second shot.** Complex second-shot shutoffs against a molded substrate are a common source of flash and rework.
4. **Consider an assembled elastomer part at low volume.** A separately molded sleeve or a mechanically retained pad can beat overmolding entirely below a few thousand parts.
5. **Keep the overmold off tight-tolerance features.** The soft layer shrinks more and is compressible, so any dimension taken over it is inherently loose.

## FAQ

### Which TPE bonds to which plastic substrate?

Compatibility follows chemical family. SEBS-based TPEs bond to polypropylene, polyethylene, and polystyrene; TPU bonds to ABS, PC, PC/ABS, nylon, and PBT; TPV bonds to polypropylene. Standard silicone bonds to essentially nothing without a self-bonding grade or a primer, so pick the pair before finalizing either material.

### How thick should an overmolded grip be?

0.030–0.060 in (0.75–1.5 mm) is the usual target, within a practical range of 0.020–0.120 in (0.5–3 mm). Below about 0.020 in the elastomer will not fill reliably over a normal grip length; above about 0.120 in it feels spongy and its own shrinkage begins pulling the substrate out of shape.

### What is the difference between two-shot molding and insert-transfer overmolding?

Two-shot molds both materials in one machine with a rotating or indexing tool, so the part never leaves the mold and position is fixed by the tooling. Insert-transfer molds the substrate separately and reloads it into a second tool — much cheaper tooling and no multi-material press, but an added handling step and a looser positional tolerance.

### How do I test whether an overmold bond is good?

A 90° peel test. A good bond fails cohesively, tearing within the elastomer itself, which means the interface is stronger than the material. Adhesive failure — the elastomer separating cleanly from the substrate — indicates incompatible chemistry, a contaminated substrate, or too low a second-shot melt temperature.

### Do I need a mechanical interlock if the materials bond chemically?

For anything structural, sealing, or exposed to heat, solvents, or UV, yes. Chemical bonds degrade with age and environmental exposure, while a through-hole the elastomer rivets over or an undercut groove does not. Treat the interlock as insurance on a joint that has to last the product's life.

### Why does an overmolded part bow?

The elastomer shrinks more than the rigid substrate, and if it sits on one face of a thin, flat part that differential pulls the substrate into a curve. Stiffen the substrate with ribs or additional wall, keep the overmold thickness uniform, and where possible balance the soft material across the neutral axis.

### Where should an overmold edge terminate?

Into a step, groove, or shallow recess in the substrate, never on a flat exposed edge. An exposed elastomer edge is a peel initiation point that a fingernail, a cleaning cloth, or normal handling will eventually lift, and once peeling starts the bond unzips along the interface.

## Alternative processes

- [Insert Molding](https://manufacturingprocesses.org/processes/forming/insert-molding.md): Insert molding loads metal inserts into the cavity and molds polymer around them, capturing threads, terminals or stampings in a single shot.
- [Adhesive Bonding](https://manufacturingprocesses.org/processes/joining/adhesive-bonding.md): Adhesive bonding joins parts with a cured polymer layer that spreads load over the whole bond area instead of concentrating it at fastener points.
- [Dip Molding](https://manufacturingprocesses.org/processes/forming/dip-molding.md): Dip molding withdraws a heated former from liquid plastisol or latex, leaving a coating that cures into a flexible open-ended part such as a grip or cap.
- [Mechanical Fastening](https://manufacturingprocesses.org/processes/joining/mechanical-fastening.md): Mechanical fastening joins parts with screws, inserts, clinch hardware or snap fits so the joint can be assembled and taken apart again.

## Related processes

- [Insert Molding](https://manufacturingprocesses.org/processes/forming/insert-molding.md): Insert molding loads metal inserts into the cavity and molds polymer around them, capturing threads, terminals or stampings in a single shot.
- [Injection Molding](https://manufacturingprocesses.org/processes/forming/injection-molding.md): Injection molding forces molten thermoplastic into a steel mold under high pressure, producing high-volume identical plastic parts.
- [Liquid Silicone Rubber (LSR) Molding](https://manufacturingprocesses.org/processes/forming/liquid-silicone-rubber-lsr-molding.md): LSR molding injects two-part liquid silicone into a heated mold where it cures, producing flexible parts with a very wide service temperature range.
- [Dip Molding](https://manufacturingprocesses.org/processes/forming/dip-molding.md): Dip molding withdraws a heated former from liquid plastisol or latex, leaving a coating that cures into a flexible open-ended part such as a grip or cap.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/forming/overmolding)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Panel Beating"
category: "Forming"
subcategory: "Metal"
materials: ["Metal"]
volumes: "1–50 pieces; a form block for jig chasing becomes worthwhile above a handful of repeats"
lead_time: "Days to weeks per panel, driven entirely by shape difficulty and finish requirements"
url: https://manufacturingprocesses.org/processes/forming/panel-beating
---

# Panel Beating

Panel beating shapes sheet metal by hand with hammers, dollies and an English wheel, producing one-off compound-curved panels without any dies.

- **Category**: [Forming](https://manufacturingprocesses.org/processes/forming.md)
- **Family**: Metal
- **Materials**: Metal
- **Typical volumes**: 1–50 pieces; a form block for jig chasing becomes worthwhile above a handful of repeats
- **Lead time**: Days to weeks per panel, driven entirely by shape difficulty and finish requirements

## Overview

Panel beating shapes sheet metal into compound curves by hand, using hammers, dollies, a shot bag, an English wheel, and a planishing hammer. There is no die and no press — the panel is raised, stretched, shrunk, and smoothed a small area at a time until it matches a buck or a template.

It is how one-off and restoration body panels, custom coachwork, aircraft cowlings and fairings, and prototype body surfaces are made, in materials that stretch well: 1100 and 3003 aluminum around 0.050–0.080 in (1.2–2 mm), and 20 to 18 gauge (0.036–0.048 in / 0.9–1.2 mm) mild steel.

The economics are pure labor. A complex compound-curved panel is measured in hours or days of skilled work, so panel beating is a process for quantities of one to a few dozen — where any die-based alternative would cost more in tooling than the entire job. Below that quantity, nothing competes; above it, nothing else is slower.

## How it works

Every panel beating operation is one of two things: making the metal locally longer (stretching) or locally shorter (shrinking). Compound curvature — curvature in two directions at once, like a fender crown — cannot be made by bending alone, because a flat sheet has no way to become a doubly curved surface without changing the length of its fibers.

1. **Buck or template.** A wooden or tubular buck defines the target surface, and profile templates check the shape as work proceeds. The buck is the specification.
2. **Dishing.** The blank is hammered into a hollow — a shot bag, a wooden dishing block, or a hollowed stump — which stretches the struck area and raises a hollow form.
3. **Wheeling.** The panel is rolled between the two wheels of an English wheel, whose lower anvil wheel is crowned. Each pass thins the metal slightly along a track and adds curvature. Building crown with a wheel is slow, controlled, and leaves a smooth surface, which is why it does the bulk of the shaping on large panels.
4. **Shrinking.** Where the metal has too much length — typically at a flange or a reverse curve — it is gathered into small tucks and hammered down, or shrunk with a shrinking disc or localized heat. Shrinking is harder than stretching and is what separates experienced panel beaters from beginners.
5. **Planishing.** The panel is worked between a planishing hammer and a matching dolly held behind it, which flattens the hammer marks and smooths the surface without significantly changing the shape.
6. **Annealing.** Every operation work hardens the metal. Aluminum is annealed periodically — annealing temperatures for the common alloys fall roughly in the 650–775°F (345–415°C) range depending on grade — and shop practice uses soot or soap markers to judge temperature by hand. Steel is more forgiving but still stiffens with work.
7. **Fitting and finishing.** Panels are trimmed, edges wired or flanged for stiffness, joined by welding or riveting, and finished by filing and sanding.

### Jig chasing

Where several identical panels are needed, a shaped jig or form block is made and the metal is hammered against it. This trades setup time for repeatability and is the bridge between one-off hand work and tooled forming.

## Design guidelines

### Design in continuous curvature

Smooth, continuously curved surfaces are what the process makes well. Flat panels are actually harder — a large unsupported flat area oil-cans and will not stay flat — which is why hand-formed panels usually carry a slight crown even where the design reads as flat.

### Avoid sharp reverse curves

A tight reverse curve requires substantial local shrinking, which is the slowest and most skill-dependent operation in the shop. Every sharp transition between a convex and a concave region adds hours. Blending them into a longer transition costs nothing at design time.

### Break the panel up

A large panel with two difficult regions is often faster to make as two panels joined at a weld or a swage line than as one piece. Designing a joint line into a low-visibility area is a standard technique, not a compromise.

### Add a stiffening edge

A hand-formed panel with a raw cut edge is floppy and will not hold shape. Design in a flange, a wired edge, a return, or a swage — these are cheap to form by hand and transform the panel's rigidity.

### Material and gauge

Use annealed, formable grades: 1100 or 3003-O aluminum at roughly 0.050–0.080 in (1.2–2 mm), or 20 to 18 gauge mild steel. Hard tempers and high-strength alloys crack rather than stretch. Aluminum works faster and needs more annealing; steel is more forgiving of hammer errors and welds more easily. Gauge selection from the [sheet metal gauge chart](/charts/sheet-metal-gauge-chart), and flange development for any edges formed on a brake from the [bend radius and K-factor chart](/charts/sheet-metal-bend-radius-k-factor).

### Tolerance is a fitting exercise

A hand-formed panel is not a dimensioned part. It is made to a buck and fitted to its neighbors, and gaps are adjusted at assembly. Specify the surface with a buck or a template and the gap at assembly; do not put a linear tolerance on a compound curve.

| Consideration | Recommended | Why |
| --- | --- | --- |
| Surface | Continuous curvature, slight crown everywhere | Large flats oil-can and will not stay flat |
| Reverse curves | Blend into long transitions | Tight reverses need extensive shrinking |
| Panel breakdown | Split difficult regions into separate panels | Two easy panels beat one hard one |
| Edges | Flange, wire, or swage every free edge | Raw edges leave the panel floppy |
| Material | 1100 or 3003-O aluminum, or 20–18 ga mild steel | Hard tempers crack rather than stretch |
| Annealing | Plan interstage anneals for aluminum | Work hardening accumulates with every blow |
| Specification | Buck or template plus assembly gap | Compound curves are not linearly toleranced |

## Variants

- Dishing
- Jig Chasing
- Wheel Forming
- Planishing

## Cost drivers

Labor is essentially the entire cost. Material is a single sheet, the tools are hammers and a wheel, and the only significant capital is the English wheel and the buck. What varies is hours, and hours scale with how much shrinking the shape demands, how large the panel is, and how good the surface has to be before paint.

Volume breakpoints: panel beating is the right answer at one piece and stays right through perhaps a few dozen. Beyond that, a jig or form block for jig chasing repays itself in repeatability alone. Above a hundred or so, superforming or a single-sided hydroform tool becomes cheaper despite the tooling, and in the thousands, matched-die stamping wins outright.

1. **Design out the shrinking.** Convex, continuously curved shapes stretch into place; reverse curves and tight concave regions have to be shrunk, and shrinking is where the hours go.
2. **Build a buck early.** Reworking a panel because the target shape was ambiguous costs more than the buck did.
3. **Split large panels.** Two manageable panels and a weld seam are usually faster than one heroic piece.
4. **Make a form block for repeats.** Even three identical panels justify jig chasing over freehand work.
5. **Specify realistically.** Show-quality surface finish before paint can double the finishing hours over a functional panel that will be filled and sanded.

## FAQ

### Why can't compound curves be made by bending?

A flat sheet has no way to become curved in two directions at once without some fibers getting longer or shorter. Bending only changes curvature in one direction. Compound curvature requires stretching the metal in some regions and shrinking it in others, which is exactly what hammer, dolly, and wheel work does.

### What is an English wheel used for?

Rolling crown into a panel. The sheet passes between an upper flat wheel and a crowned lower anvil wheel, and each pass stretches the metal slightly along a track. It builds curvature slowly and controllably while leaving a smooth surface, which is why it does most of the shaping on large panels.

### What materials are used for panel beating?

Annealed formable grades: 1100 or 3003-O aluminum at roughly 0.050–0.080 in (1.2–2 mm), and 20 to 18 gauge (0.036–0.048 in) mild steel. Hard tempers and high-strength alloys crack instead of stretching. Aluminum shapes faster but needs frequent annealing; steel is more forgiving and welds more easily.

### Why does aluminum need annealing during panel beating?

Every hammer blow and wheel pass work hardens the metal, and hardened aluminum cracks rather than stretching. Annealing temperatures for the common panel alloys fall roughly in the 650–775°F (345–415°C) range depending on grade, and shop practice uses soot or soap markers to judge temperature without instruments.

### How many parts can be made by panel beating?

Realistically one to a few dozen. Above a handful of repeats, a form block for jig chasing pays for itself in consistency. Above a hundred or so, superforming or a single-sided hydroform tool becomes cheaper, and in the thousands, matched-die stamping wins decisively.

### How should a hand-formed panel be specified?

With a buck or profile templates for the surface, plus a gap requirement at assembly. A compound-curved hand-formed panel is not a dimensioned part — it is made to a physical reference and fitted to its neighbors, so putting linear tolerances on the curved surface is not meaningful.

## Alternative processes

- [Superforming](https://manufacturingprocesses.org/processes/forming/superforming.md): Superforming heats a superplastic aluminum alloy sheet and forms it with gas pressure over a single-sided tool, producing deep, complex panels.
- [Deep Drawing](https://manufacturingprocesses.org/processes/forming/deep-drawing.md): Deep drawing pushes a sheet metal blank through a die with a punch so the metal flows into a seamless cup or box deeper than its own diameter.
- [Metal Stamping](https://manufacturingprocesses.org/processes/forming/metal-stamping.md): Metal stamping shears and forms sheet metal between matched dies in a press, producing flat and shallow-formed parts at very high rates.
- [Hydroforming](https://manufacturingprocesses.org/processes/forming/hydroforming.md): Hydroforming presses sheet or tube into a single-sided die with high-pressure fluid, producing smooth seamless shapes that would need several stampings.

## Related processes

- [Metal Spinning](https://manufacturingprocesses.org/processes/forming/metal-spinning.md): Metal spinning presses a rotating metal disc against a mandrel with a roller or hand tool, forming it into an axially symmetric hollow shape.
- [Press Braking](https://manufacturingprocesses.org/processes/forming/press-braking.md): Press braking bends sheet metal by driving a punch into a V-die, forming one straight-line bend at a time.
- [Superforming](https://manufacturingprocesses.org/processes/forming/superforming.md): Superforming heats a superplastic aluminum alloy sheet and forms it with gas pressure over a single-sided tool, producing deep, complex panels.
- [Sheet Metal Fabrication](https://manufacturingprocesses.org/processes/forming/sheet-metal-fabrication.md): Sheet metal fabrication cuts, bends and joins flat stock into a finished assembly, typically laser cutting followed by press braking, hardware and welding.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/forming/panel-beating)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Paper Pulp Molding"
category: "Forming"
subcategory: "Wood"
materials: ["Wood"]
tolerances: "±0.020–0.060 in (±0.5–1.5 mm) for thick-wall and transfer molded parts; ±0.010–0.020 in (±0.25–0.5 mm) for thermoformed precision pulp."
volumes: "10,000–10,000,000+ parts per year"
lead_time: "4–10 weeks for tooling, substantially faster than plastic molding tools. Production cycles run from seconds to a few minutes per part depending on grade and thickness."
url: https://manufacturingprocesses.org/processes/forming/paper-pulp-molding
---

# Paper Pulp Molding

Paper pulp molding draws a slurry of cellulose fiber onto a perforated screen tool by vacuum, then dries the fiber mat into a rigid formed part.

- **Category**: [Forming](https://manufacturingprocesses.org/processes/forming.md)
- **Family**: Wood
- **Materials**: Wood
- **Typical tolerances**: ±0.020–0.060 in (±0.5–1.5 mm) for thick-wall and transfer molded parts; ±0.010–0.020 in (±0.25–0.5 mm) for thermoformed precision pulp.
- **Surface finish**: The screen side carries a visible mesh texture; the counter-tool side is smoother. Thermoformed precision pulp gives a smooth, consistent finish on both faces and accepts printing directly.
- **Typical volumes**: 10,000–10,000,000+ parts per year
- **Lead time**: 4–10 weeks for tooling, substantially faster than plastic molding tools. Production cycles run from seconds to a few minutes per part depending on grade and thickness.

## Overview

Paper pulp molding forms a part by drawing a dilute slurry of cellulose fiber — typically 0.5–2% fiber in water — onto a perforated screen tool under vacuum, so a fiber mat builds up on the tool face, then drying that mat into a rigid shape. The feedstock is recycled newsprint, kraft, or corrugated stock, and with no plastic added the finished part is curbside recyclable and usually compostable.

Four grades cover the market. Thick-wall (slush) molding produces heavy 0.12–0.32 in (3–8 mm) dunnage and plant pots. Transfer molding gives 0.06–0.12 in (1.5–3 mm) parts with one smooth face — egg cartons, cup carriers, electronics trays. Thermoformed (precision) pulp presses the wet mat between heated tools at roughly 350–430 °F (175–220 °C) for a smooth finish on both faces, thinner walls, and tighter tolerance. Processed pulp adds coatings or barrier treatments.

Tooling is a perforated aluminum or stainless screen, far cheaper and faster to make than an injection mold, which is why pulp molding has taken so much protective packaging away from expanded polystyrene.

## How it works

1. **Prepare the stock.** Recovered paper is pulped in water, screened for contaminants, and diluted to a working consistency of roughly 0.5–2% fiber by weight. Sizing agents for water resistance and any coloring are added here.
2. **Form.** The perforated screen tool is dipped into the slurry and vacuum is drawn through it, typically in the region of 0.4–0.8 bar. Water passes through the screen and fiber deposits on the face; mat thickness builds with immersion time, which is how wall thickness is controlled.
3. **Dewater and transfer.** Vacuum continues to pull free water out of the mat. In transfer molding, a matching counter-tool presses against the wet mat to consolidate it and pick it off the forming screen, which is what produces the second, smoother face.
4. **Dry.** Thick-wall parts go through a tunnel or through-air oven, typically for tens of minutes. Thermoformed pulp is pressed between heated matched tools at roughly 350–430 °F (175–220 °C) for seconds to a minute, driving off water and setting the shape in one operation — this is what gives precision pulp its thin, smooth, dimensionally consistent walls.
5. **Trim.** Edges are die cut or trimmed to final outline. Some tooling produces a net edge and needs no trimming.
6. **Post-treat if required.** Barrier coatings, moisture-resistant sizing, printing, or antistatic treatment are applied where the application demands them.

## Design guidelines

### Draft
3° is the minimum and 5–6° is comfortable. The wet mat has almost no green strength and must release cleanly from the screen, and drafted walls are also what allow finished parts to nest for shipping — a major part of the economic case against foam.

### Wall thickness
0.04–0.12 in (1–3 mm) for transfer and thermoformed parts, 0.12–0.32 in (3–8 mm) for thick-wall. Thickness is set by immersion time and is reasonably uniform, but sharp geometry causes the vacuum flow to concentrate and locally thicken or thin the mat.

### Radii
Generous radii everywhere: 0.12 in (3 mm) inside minimum, and more on any deep feature. Fiber does not conform to a sharp corner, and a corner that starves of fiber becomes a thin, weak spot.

### Depth of draw
Keep the draw depth within roughly the width of the feature. Deep narrow pockets shadow the vacuum and form thin, poorly consolidated walls.

### Nesting and stacking
Design nesting into the geometry from the start. Stack height per part is what determines shipping and warehousing cost of the packaging itself, and it is often the deciding factor in whether pulp beats foam or thermoformed plastic.

### Ribs and structure
Stiffness comes from geometry, not from material. Corrugations, ribs, and curved surfaces do the structural work; a flat pulp panel is floppy at any realistic wall thickness.

### Tolerances
Thick-wall and transfer molded parts hold roughly ±0.020–0.060 in (±0.5–1.5 mm); thermoformed precision pulp holds about ±0.010–0.020 in (±0.25–0.5 mm). Design the product interface with compliance in mind — pulp packaging works by cushioning, not by precise location.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Draft | 5–6° | 3° | Wet mat must release; parts must nest |
| Wall thickness | 0.06 in (1.5 mm) | 0.04–0.32 in (1–8 mm) | Set by immersion time |
| Inside radius | 0.20 in (5 mm) | 0.12 in (3 mm) | Fiber cannot conform to sharp corners |
| Draw depth | ≤ feature width | 1.5× width | Vacuum shadowing thins the wall |
| Tolerance | ±0.040 in (1 mm) | ±0.010 in (0.25 mm) thermoformed | Fiber mat and drying shrinkage |

## Cost drivers

Tooling is the reason pulp molding competes. A perforated aluminum or stainless screen tool costs a fraction of an injection mold and is produced in weeks rather than months, so the barrier to a new package design is low. Precision thermoformed pulp requires matched heated tools and costs more, but still less than comparable plastic tooling.

Running cost is water and energy. Fiber itself is cheap — recovered paper is among the least expensive feedstocks in manufacturing — but evaporating the water out of a formed mat is energy-intensive, and drying is the throughput bottleneck in every pulp molding plant. Wall thickness therefore maps directly onto cost, because thicker means longer drying.

Cycle time ranges from seconds for thin thermoformed parts to minutes for thick-wall dunnage, with multi-cavity rotary machines running continuously.

Volume breakpoints: pulp molding becomes economic from roughly 10,000 parts and scales to tens of millions. Below that, die-cut corrugated board is usually cheaper.

1. Reduce wall thickness — drying energy is the dominant running cost.
2. Design deep nesting; shipping air is what makes packaging expensive.
3. Get stiffness from corrugation and curvature rather than from thickness.
4. Skip barrier coatings unless the application requires them; they compromise recyclability as well as adding cost.
5. Use a single tool geometry to hold multiple product variants where cushioning tolerance allows.

## FAQ

### What draft angle does molded pulp need?

3° minimum, with 5–6° preferred. The wet fiber mat has very little strength and must release cleanly from the perforated screen tool. Generous draft also lets finished parts nest deeply, which is often the deciding economic factor for packaging.

### How thick can molded pulp walls be?

0.04–0.12 in (1–3 mm) for transfer and thermoformed grades and 0.12–0.32 in (3–8 mm) for thick-wall dunnage. Thickness is controlled by how long the screen tool stays immersed under vacuum, and thicker walls cost proportionally more drying energy and cycle time.

### What tolerance can molded pulp hold?

Roughly ±0.020–0.060 in (±0.5–1.5 mm) for thick-wall and transfer molded parts, and ±0.010–0.020 in (±0.25–0.5 mm) for thermoformed precision pulp. Pulp packaging works by cushioning rather than precise location, so product interfaces should be designed with compliance.

### Is molded pulp recyclable?

Yes, when made without barrier coatings. The feedstock is recovered paper and the part contains no plastic, so it goes into standard paper recycling streams and is generally compostable. Adding a plastic or wax barrier layer for moisture or grease resistance compromises both.

### What is the difference between thick-wall, transfer molded, and thermoformed pulp?

Thick-wall (slush) molding gives heavy 3–8 mm parts with rough surfaces for dunnage. Transfer molding presses the wet mat with a counter-tool for one smooth face at 1.5–3 mm. Thermoformed precision pulp dries the mat between heated matched tools at 350–430 °F (175–220 °C), giving smooth faces on both sides, thinner walls, and tighter tolerance.

### How does molded pulp tooling compare with injection mold tooling?

A perforated screen tool costs a small fraction of an injection mold and is delivered in weeks rather than months. That low barrier to entry, combined with the recyclability of the finished part, is why molded pulp has displaced expanded polystyrene across much of protective packaging.

## Alternative processes

- [Thermoforming](https://manufacturingprocesses.org/processes/forming/thermoforming.md): Thermoforming heats a plastic sheet until pliable and pulls or presses it over a single-sided mold, producing large thin-walled parts with low tooling cost.
- [Die Cutting](https://manufacturingprocesses.org/processes/cutting/die-cutting.md): Die cutting presses a shaped steel rule or rotary die through sheet stock, cutting flat outlines in paper, board, foam, film and textile.
- [Injection Molding](https://manufacturingprocesses.org/processes/forming/injection-molding.md): Injection molding forces molten thermoplastic into a steel mold under high pressure, producing high-volume identical plastic parts.
- [Rotation Molding](https://manufacturingprocesses.org/processes/forming/rotation-molding.md): Rotational molding tumbles powdered polymer inside a heated mold on two axes so it fuses into a seamless, stress-free hollow part with a uniform wall.
- [Compression Molding](https://manufacturingprocesses.org/processes/forming/compression-molding.md): Compression molding presses a measured charge of rubber or thermoset into an open heated mold that closes to cure it into its final shape.

## Related processes

- [Thermoforming](https://manufacturingprocesses.org/processes/forming/thermoforming.md): Thermoforming heats a plastic sheet until pliable and pulls or presses it over a single-sided mold, producing large thin-walled parts with low tooling cost.
- [Compression Molding](https://manufacturingprocesses.org/processes/forming/compression-molding.md): Compression molding presses a measured charge of rubber or thermoset into an open heated mold that closes to cure it into its final shape.
- [Die Cutting](https://manufacturingprocesses.org/processes/cutting/die-cutting.md): Die cutting presses a shaped steel rule or rotary die through sheet stock, cutting flat outlines in paper, board, foam, film and textile.
- [Rotation Molding](https://manufacturingprocesses.org/processes/forming/rotation-molding.md): Rotational molding tumbles powdered polymer inside a heated mold on two axes so it fuses into a seamless, stress-free hollow part with a uniform wall.
- [Composite Laminating](https://manufacturingprocesses.org/processes/forming/composite-laminating.md): Composite laminating lays reinforcing fiber into a mold and consolidates it with resin, curing into a stiff, light monocoque part.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/forming/paper-pulp-molding)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Plastic Extrusion"
category: "Forming"
subcategory: "Plastics and Rubber"
materials: ["Plastic"]
tolerances: "±0.005 in (±0.13 mm) on small dimensions held against a calibrator; ±0.020–0.060 in (±0.5–1.5 mm) on free-extruded surfaces of large profiles; angles ±1–2°. Bow and twist are specified separately per unit length."
volumes: "Quoted by weight — minimum runs of several hundred to a few thousand pounds; annual volumes reach millions of feet"
lead_time: "3–6 weeks for a custom profile die plus calibration tooling, including one or more die correction iterations; days per production run thereafter"
url: https://manufacturingprocesses.org/processes/forming/plastic-extrusion
---

# Plastic Extrusion

Plastic extrusion pushes molten polymer through a shaped die and cools it, producing continuous profile, sheet, pipe or film of constant cross-section.

- **Category**: [Forming](https://manufacturingprocesses.org/processes/forming.md)
- **Family**: Plastics and Rubber
- **Materials**: Plastic
- **Typical tolerances**: ±0.005 in (±0.13 mm) on small dimensions held against a calibrator; ±0.020–0.060 in (±0.5–1.5 mm) on free-extruded surfaces of large profiles; angles ±1–2°. Bow and twist are specified separately per unit length.
- **Surface finish**: Smooth gloss as extruded; matte or embossed using a textured calibrator or embossing roll; sheet gloss is set by the polishing stack
- **Typical volumes**: Quoted by weight — minimum runs of several hundred to a few thousand pounds; annual volumes reach millions of feet
- **Lead time**: 3–6 weeks for a custom profile die plus calibration tooling, including one or more die correction iterations; days per production run thereafter

## Overview

Plastic extrusion pushes molten polymer continuously through a shaped die and cools it into a constant cross-section: pipe, tube, window profile, wire insulation, sheet, and film. It is the highest-volume plastics process by weight and the only economical way to make a part whose length is arbitrary and whose cross-section never changes.

Almost every thermoplastic extrudes — rigid and flexible PVC, HDPE, LDPE, PP, ABS, polycarbonate, nylon, TPE — and co-extrusion combines two or three of them in a single pass. That is why a window profile can carry a rigid PVC body, a soft TPE gasket, and a UV-stable capstock all formed at once.

Tooling is a die plate plus calibration and cooling hardware, roughly an order of magnitude cheaper than an injection mold and typically 3–6 weeks out. Minimum orders are quoted in pounds of resin rather than in pieces, because startup scrap while the line stabilizes is the real setup cost.

## How it works

1. **Feed and melt.** Pellets drop into a single-screw extruder, typically 24:1 to 32:1 length-to-diameter with a compression ratio around 2:1–4:1. The screw conveys, compresses, and melts the resin through a combination of barrel heat and shear work. Hygroscopic resins — nylon, PC, PET, ABS — are dried first, exactly as for injection molding.

2. **Melt through the die.** Melt temperatures follow the resin: rigid PVC 330–390 °F (165–200 °C), HDPE 350–420 °F (175–215 °C), PP 400–480 °F (200–250 °C), ABS 410–480 °F (210–250 °C). The die converts round barrel flow into the profile cross-section, and its channels are balanced so every part of the section leaves at the same velocity. Unbalanced flow is what makes a profile twist.

3. **Calibrate and cool.** The hot extrudate is drawn immediately into a vacuum calibration sizer that pulls it against a fixed cooled surface, then through a water tank. Calibration is what makes tight tolerances possible — free-extruded profiles that touch nothing but air hold far less.

4. **Haul off.** A caterpillar puller draws the profile at constant speed. The ratio of haul-off speed to die output, the draw-down ratio, is the operator's main dimensional control, and it interacts with die swell — the elastic recovery that makes the extrudate expand as it leaves the die. Die swell can be substantial for polyolefins, which is why a die opening never matches the finished profile shape.

5. **Cut or coil.** Rigid profiles are flying-saw cut to length; flexible tube and wire insulation is coiled.

### How are sheet and film different?

Sheet uses a wide flat (coat-hanger) die feeding a three-roll polishing stack, which sets both thickness and surface gloss. Blown film extrudes an upward annular bubble that is inflated and cooled in air, then collapsed and wound. Both share the same first three steps and differ only in downstream equipment.

## Design guidelines

### Uniform wall

This is the whole discipline. Thick sections cool more slowly, shrink later, and drag the surrounding profile out of shape — the extruded equivalent of a sink mark, except it warps the entire cross-section rather than marking one face. Hold wall thickness within about ±10% across the profile, and where you cannot, core out the thick region rather than leaving it solid.

### Minimum wall

Rigid PVC and ABS profiles are practical from about 0.030 in (0.75 mm), with 0.050 in (1.3 mm) a safer design target. Below that the hot profile cannot support itself between die and calibrator and collapses under the calibration vacuum. Flexible materials tolerate thinner sections.

### Radii

No sharp corners, inside or out. Use at least 0.015–0.030 in (0.4–0.75 mm) on internal corners; larger is better. Sharp internal corners are stress risers in the part and hot spots in the die where melt hangs up and degrades.

### Symmetry and unsupported legs

Make the section symmetric about at least one axis where you can — asymmetric sections need more die-balancing iterations and are more prone to twist. Avoid long, thin, unsupported legs; a leg longer than about 10× its own thickness will flutter and wander. Tie it back with a web or thicken its tip slightly.

### Hollows

Hollow and multi-hollow profiles need a mandrel supported by a spider or bridge, which leaves weld lines where the melt recombines. They cost more to tool but hold better tolerances, because the internal surface is supported by an internal calibrator. Keep internal webs the same thickness as the outer wall.

### Tolerances

Specify tolerance only where it matters, and understand where it comes from. Dimensions held against a calibrator can reach ±0.005 in (±0.13 mm) on small sections. Free-extruded outside dimensions on large profiles are more like ±0.020–0.060 in (±0.5–1.5 mm), and angles run ±1–2°. Bow and twist are separate callouts specified per unit length. For matching polymer stiffness and strength to the requirement, see [/charts/material-properties](/charts/material-properties).

### Secondary operations

Anything that is not a constant cross-section — holes, notches, end machining, bends — is a separate operation on cut lengths. Design those features to locate off a surface the extrusion actually holds, such as a calibrated internal bore, rather than a free-cooled outside face.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Wall uniformity | Within ±10% across section | 2:1 thick-to-thin | Thick sections shrink last and warp the profile |
| Minimum wall (rigid) | 0.050 in (1.3 mm) | 0.030 in (0.75 mm) | Thinner sections collapse under calibration vacuum |
| Internal radius | 0.030 in (0.75 mm) or more | 0.015 in (0.4 mm) | Sharp corners are die hot spots and part stress risers |
| Unsupported leg | ≤ 5 × its thickness | about 10 × thickness | Long legs flutter and lose position |
| Hollow web thickness | Equal to the outer wall | — | Mismatched webs cool at different rates and pull the section |
| Tolerance, calibrated | ±0.005 in (±0.13 mm) | — | Only surfaces touching the calibrator hold this |
| Tolerance, free surface | ±0.020–0.060 in (±0.5–1.5 mm) | — | Nothing constrains a free-cooled surface |

## Cost drivers

Extrusion inverts the injection molding cost structure. The die is comparatively cheap, so the recurring costs — resin, line speed, and startup scrap — dominate. Resin is usually the majority of the delivered cost of a commodity profile, which makes cross-sectional area (pounds per foot) the single most important design variable.

Volume breakpoints are expressed in weight, not pieces. A custom profile normally carries a minimum run of several hundred to a few thousand pounds, because a line takes time and material to stabilize regardless of how much product you want.

Cost reduction:

1. **Take area out of the cross-section.** Every square inch of section is pounds per foot forever. Core out solid regions and replace them with ribbed or hollow sections that carry the same load.
2. **Widen tolerances on non-critical surfaces.** Every dimension held against a calibrator adds tooling and slows the line. Free-extruded surfaces are essentially free.
3. **Co-extrude instead of assembling.** A soft TPE seal co-extruded onto a rigid profile removes a separate gasket, an adhesive, and an assembly step.
4. **Consolidate into fewer profiles.** Two similar profiles mean two dies, two setups, and two rounds of startup scrap. One profile serving both applications, trimmed differently, usually wins.
5. **Design so the die can be balanced in one iteration.** Symmetric, uniform-wall sections with generous radii reach a sellable profile faster; complex asymmetric ones can take several die corrections, each of which is chargeable.

## FAQ

### What tolerance can plastic extrusion hold?

Dimensions held against a vacuum calibrator on a small section can reach ±0.005 in (±0.13 mm). Free-extruded surfaces on large profiles are more like ±0.020–0.060 in (±0.5–1.5 mm), and angles run ±1–2°. Bow and twist are separate callouts specified per unit length.

### What is the minimum wall thickness for an extruded plastic profile?

About 0.030 in (0.75 mm) for rigid PVC and ABS, with 0.050 in (1.3 mm) a safer design target. Below that the hot profile cannot support itself between the die and the calibrator and collapses under the calibration vacuum. Flexible materials tolerate thinner sections.

### Why does an extrusion die not match the profile shape?

Two effects intervene. Die swell — elastic recovery as the melt leaves the constraint of the die — expands the extrudate, and the haul-off then draws it back down. The die is cut to compensate for both, which is why custom profiles usually need one or more die correction iterations before first article approval.

### How much does an extrusion die cost compared with an injection mold?

Roughly an order of magnitude less. An extrusion die is a plate with a machined orifice plus calibration and cooling hardware, not a two-plate pressure-containing mold with an ejection system. The trade is that per-part cost is dominated by resin weight, so cross-sectional area matters far more than it does in molding.

### Can extruded profiles combine hard and soft materials?

Yes. Co-extrusion feeds two or three extruders into one die, so a rigid PVC window profile can carry a flexible TPE gasket and a UV-stable capstock in a single pass. It removes a separate gasket, its adhesive, and an assembly step, and it is standard practice in construction and automotive sealing profiles.

### How are holes and cutouts added to an extruded part?

As secondary operations on cut lengths, since extrusion can only produce a constant cross-section. Punching, drilling, notching, and end machining are done downstream, either in line with the puller or offline. Locate those features off a calibrated surface such as an internal bore, which the extrusion holds far better than a free-cooled outside face.

### Why do extruded profiles twist, and how do I prevent it?

Twist comes from unbalanced flow through the die and from non-uniform cooling, both of which are worsened by uneven wall thickness. Hold wall within about ±10% across the section, make the profile symmetric about at least one axis, and avoid heavy solid regions next to thin ones.

## Alternative processes

- [Pultrusion](https://manufacturingprocesses.org/processes/forming/pultrusion.md): Pultrusion pulls continuous fiber through a resin bath and a heated die, curing it into a constant-section composite profile of unlimited length.
- [Metal Extrusion](https://manufacturingprocesses.org/processes/forming/metal-extrusion.md): Metal extrusion pushes heated billet through a shaped die, producing continuous aluminum or copper profile of constant cross-section.
- [Roll Forming](https://manufacturingprocesses.org/processes/forming/roll-forming.md): Roll forming passes coil stock through a sequence of contoured roller stands that bend it progressively into a constant cross-section profile.
- [CNC Machining](https://manufacturingprocesses.org/processes/cutting/cnc-machining.md): CNC machining removes material from solid stock with a programmed cutting tool, producing accurate parts directly from a CAD model with no tooling.

## Related processes

- [Pultrusion](https://manufacturingprocesses.org/processes/forming/pultrusion.md): Pultrusion pulls continuous fiber through a resin bath and a heated die, curing it into a constant-section composite profile of unlimited length.
- [Blow Molding](https://manufacturingprocesses.org/processes/forming/blow-molding.md): Blow molding inflates a heated plastic tube or preform against the walls of a mold with compressed air, producing hollow one-piece containers and tanks.
- [Injection Molding](https://manufacturingprocesses.org/processes/forming/injection-molding.md): Injection molding forces molten thermoplastic into a steel mold under high pressure, producing high-volume identical plastic parts.
- [Metal Extrusion](https://manufacturingprocesses.org/processes/forming/metal-extrusion.md): Metal extrusion pushes heated billet through a shaped die, producing continuous aluminum or copper profile of constant cross-section.
- [Thermoforming](https://manufacturingprocesses.org/processes/forming/thermoforming.md): Thermoforming heats a plastic sheet until pliable and pulls or presses it over a single-sided mold, producing large thin-walled parts with low tooling cost.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/forming/plastic-extrusion)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Powder Metallurgy (Press and Sinter)"
category: "Forming"
subcategory: "Metal"
materials: ["Metal"]
tolerances: "Roughly ±0.001–0.002 in per inch on dimensions perpendicular to the pressing direction, controlled by die and core rod steel; dimensions along the pressing axis hold considerably looser unless the part is sized after sintering"
volumes: "25,000–5,000,000+ parts per year; below about 5,000–10,000 machining is usually cheaper"
lead_time: "12–20 weeks for die set design, build, and qualification; 3–6 weeks per production run thereafter"
url: https://manufacturingprocesses.org/processes/forming/powder-metallurgy-press-and-sinter
---

# Powder Metallurgy (Press and Sinter)

Press-and-sinter powder metallurgy compacts metal powder in a rigid die and sinters it below its melting point into a finished net-shape part.

- **Category**: [Forming](https://manufacturingprocesses.org/processes/forming.md)
- **Family**: Metal
- **Materials**: Metal
- **Typical tolerances**: Roughly ±0.001–0.002 in per inch on dimensions perpendicular to the pressing direction, controlled by die and core rod steel; dimensions along the pressing axis hold considerably looser unless the part is sized after sintering
- **Typical volumes**: 25,000–5,000,000+ parts per year; below about 5,000–10,000 machining is usually cheaper
- **Lead time**: 12–20 weeks for die set design, build, and qualification; 3–6 weeks per production run thereafter

## Overview

Press-and-sinter powder metallurgy compacts metal powder in a rigid die at 20–50 tons per square inch, then heats the resulting green compact below its melting point until the particles bond into a solid part. Iron and steel parts sinter at roughly 2,050–2,100°F (1,120–1,150°C) in a controlled atmosphere.

The parts come out net shape with essentially no scrap: material utilization approaches 100%, against 50% or worse when the same part is machined from bar. That, plus cycle times of seconds, is why PM makes automotive gears, sprockets, cams, bearing caps, synchronizer hubs, and hundreds of millions of self-lubricating bronze bushings a year.

The trade is porosity. A conventional pressed-and-sintered ferrous part typically finishes at 6.6–7.2 g/cm³ against 7.87 g/cm³ for solid iron, so 10–15% of the volume is pores, and strength and fatigue performance are lower than wrought material accordingly. On bearings that porosity is the product — the pores are filled with oil.

## How it works

1. **Powder blending.** Elemental or pre-alloyed powder is blended with graphite, alloying additions, and a solid lubricant that lets the compact eject from the die without galling.
2. **Compaction.** The blend is fed into a rigid die and pressed by upper and lower punches at roughly 20–50 tsi (275–690 MPa). Multi-level parts use multiple independently controlled punches so each level reaches similar density. The green compact holds together by mechanical interlocking alone and is fragile.
3. **Sintering.** The compact travels through a continuous belt or pusher furnace at roughly 2,050–2,100°F (1,120–1,150°C) for ferrous materials, held at temperature for something on the order of 20–45 minutes, in an endothermic or nitrogen-hydrogen atmosphere. The lubricant burns off in a delube zone first. Particles bond by diffusion; the part shrinks or grows slightly depending on the alloy system, and that dimensional change is a controlled process variable, not an accident. See [metal melting points](/charts/metal-melting-points) for how far below melting this sits.
4. **Sizing (optional).** A repressing operation in a sizing die brings critical dimensions into a tighter band and improves surface finish.
5. **Secondary operations.** Oil impregnation for self-lubricating bearings (typically filling 20–25% of the volume), resin impregnation to seal porosity for plating or pressure tightness, copper infiltration for higher strength, steam treatment for corrosion and wear resistance, heat treatment, and machining of features the press cannot form.

Density is the master variable. Everything about a PM part's strength, fatigue life, machinability, and corrosion behavior tracks its density, and density is set by compaction pressure, powder characteristics, and how uniformly the punches can pack a given geometry. Tall thin sections and multi-level shapes are hard to press to uniform density, which is why PM design rules focus on shapes that pack evenly.

## Design guidelines

### Think in one pressing direction

The die is rigid and the punches move along a single axis. That means no undercuts, no cross holes, no threads, no reverse tapers, and no features on the side wall — anything perpendicular to the pressing direction has to be machined afterward. Design the part as a profile extruded along the press axis.

### No draft required

Unlike casting or molding, PM walls parallel to the pressing direction need no draft at all — straight walls eject cleanly from a rigid die. Draft actually complicates tooling.

### Length-to-diameter ratio

Keep the pressed height under roughly 2.5–3× the minimum section width. Powder does not transmit pressure like a fluid; friction against the die wall means the middle of a tall part is pressed less than its ends, giving a density gradient and weak zones.

### Minimum wall and hole

Hold walls above about 0.060 in (1.5 mm) and holes above about 0.060–0.080 in (1.5–2 mm) diameter. Thin die sections and slender core rods break under repeated 50 tsi compaction.

### Radius, chamfer, and flat land

Sharp corners on the punch face wear and chip. Use radii wherever possible. Where a chamfer meets an outside diameter, leave a small flat land — typically about 0.010 in (0.25 mm) — so the punch does not end in a feather edge.

### Tolerances differ by axis

Dimensions perpendicular to the pressing direction — diameters, hole sizes — are controlled by rigid die and core rod steel and hold well, on the order of ±0.001–0.002 in per inch. Dimensions along the pressing axis depend on fill weight and press stroke and hold considerably looser. Never put the tightest tolerance on the press-axis dimension. For bearing and housing fits, see [ISO 286 fits and tolerances](/charts/iso-286-fits-tolerances); for property comparison against wrought material, the [material properties chart](/charts/material-properties).

### Design around the porosity

A conventional PM part is 10–15% porous. It cannot be leak-tight without resin impregnation, it does not electroplate without sealing, and its fatigue strength sits well below a wrought equivalent. If those matter, either specify a higher-density route or choose another process.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Undercuts and cross holes | None | Machine them afterward | Rigid die moves along one axis only |
| Draft | 0° | — | Straight walls eject cleanly from steel dies |
| Height ÷ minimum width | ≤ 2.5 | 3 | Die-wall friction creates a density gradient |
| Minimum wall | 0.080 in (2 mm) | 0.060 in (1.5 mm) | Thin die sections fail under compaction load |
| Minimum hole diameter | 0.080 in (2 mm) | 0.060 in (1.5 mm) | Slender core rods break |
| Flat land at chamfer | 0.010 in (0.25 mm) | — | Punches cannot end in a feather edge |
| Tolerance across press axis | ±0.001–0.002 in/in | — | Controlled by rigid die and core rod steel |
| Tolerance along press axis | Looser | — | Depends on fill weight and press stroke |

## Cost drivers

PM's economics rest on two things: near-total material utilization and very short cycle times. A compaction press makes several parts per minute with essentially no scrap, so per-part cost at volume is dominated by powder price and furnace throughput rather than labor. Powder costs more per pound than bar stock, but the part uses almost all of it, which flips the comparison against machining as soon as the machined version would generate significant chips.

Tooling is the barrier. A multi-level die set with independently controlled punches and core rods is a precision assembly and takes months. Every secondary operation — sizing, machining a cross hole, impregnation, heat treatment — adds a handling step that erodes the process's core advantage.

Volume breakpoints: below roughly 5,000–10,000 parts a year the die set will not amortize and machining wins. From 25,000 to several million a year is the classic PM window.

1. **Eliminate the secondary machining.** A cross hole or a thread turns a net-shape process into a machined part with extra steps. Redesign it out if you can.
2. **Reduce the number of levels.** Each pressed level needs its own punch and its own control axis. Two-level parts are dramatically cheaper to tool than four-level ones.
3. **Keep the press-axis dimension loose.** Tight axial tolerance forces a sizing operation on every part.
4. **Only specify high density where it is needed.** Higher density means higher compaction pressure, bigger presses, and sometimes double pressing and double sintering.
5. **Use the porosity.** Self-lubricating bearings, oil-retaining bushings, and filters exist because PM porosity is a feature, not a defect.

## FAQ

### How dense are pressed-and-sintered PM parts?

Typically 6.6–7.2 g/cm³ for ferrous parts, against 7.87 g/cm³ for solid iron — so 10–15% of the volume is porosity. Strength and fatigue performance scale with density and sit below wrought equivalents. Double pressing, copper infiltration, or a switch to metal injection molding gets closer to full density.

### Why can't PM parts have undercuts or cross holes?

Because the die is rigid steel and the punches only move along one axis. Any feature perpendicular to the pressing direction cannot be formed and must be machined after sintering, which adds a handling operation and gives back much of the net-shape advantage.

### Do powder metal parts need draft?

No. Walls parallel to the pressing direction eject cleanly from a rigid steel die, so zero draft is standard and draft actually complicates the tooling. This is a significant difference from casting and molding, where draft is mandatory.

### What tolerance can press-and-sinter hold?

Roughly ±0.001–0.002 in per inch on diameters and hole sizes, which are set by rigid die and core rod steel. Dimensions along the pressing axis depend on powder fill weight and press stroke and hold looser. A sizing operation after sintering tightens the axial dimensions at additional cost.

### Why are PM bearings self-lubricating?

The 20–25% interconnected porosity in a bronze bushing is deliberately vacuum-impregnated with oil. As the shaft rotates and the bearing warms, oil is drawn out to the running surface and reabsorbed when it cools. The porosity that limits structural PM parts is the entire point of the bearing.

### Powder metallurgy or metal injection molding?

Press-and-sinter for simple shapes that can be pressed along one axis, at high volume, with 10–15% porosity acceptable — gears, bushings, bearing caps. MIM for small complex geometry with undercuts and cross features, at 96–99% density, where the part could not be pressed in a rigid die at all.

## Alternative processes

- [Metal Injection Molding](https://manufacturingprocesses.org/processes/forming/metal-injection-molding.md): Metal injection molding molds a powder-and-binder feedstock exactly like a plastic, then debinds and sinters it into a dense small metal part.
- [CNC Machining](https://manufacturingprocesses.org/processes/cutting/cnc-machining.md): CNC machining removes material from solid stock with a programmed cutting tool, producing accurate parts directly from a CAD model with no tooling.
- [Investment Casting](https://manufacturingprocesses.org/processes/forming/investment-casting.md): Investment casting builds a ceramic shell around a wax pattern, melts the wax out and pours metal into the cavity, giving fine detail in almost any alloy.
- [Die Casting](https://manufacturingprocesses.org/processes/forming/die-casting.md): Die casting injects molten aluminum, zinc or magnesium into a hardened steel die under high pressure, producing thin-walled net-shape metal parts.

## Related processes

- [Metal Injection Molding](https://manufacturingprocesses.org/processes/forming/metal-injection-molding.md): Metal injection molding molds a powder-and-binder feedstock exactly like a plastic, then debinds and sinters it into a dense small metal part.
- [Forging](https://manufacturingprocesses.org/processes/forming/forging.md): Forging deforms solid metal under compressive load in a hammer, press or dies, so the grain flow follows the part contour and fatigue strength rises.
- [Heat Treatment](https://manufacturingprocesses.org/processes/forming/heat-treatment.md): Heat treatment cycles metal through controlled heating and cooling to change its hardness, strength, ductility or internal stress state.
- [Binder Jetting](https://manufacturingprocesses.org/processes/forming/binder-jetting.md): Binder jetting prints liquid binder into a powder bed to hold each layer, producing green parts that are later sintered or infiltrated.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/forming/powder-metallurgy-press-and-sinter)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Press Braking"
category: "Forming"
subcategory: "Metal"
materials: ["Metal"]
tolerances: "±0.010 in (±0.25 mm) on a single bend dimension, accumulating roughly ±0.010 in per additional bend; bend angle ±1°, or ±0.5° with in-process angle measurement"
volumes: "1–25,000 parts; above about 25,000 identical parts, hard tooling usually wins"
lead_time: "1–10 business days in a job shop; same-day for simple parts with material in stock. No tooling lead time"
url: https://manufacturingprocesses.org/processes/forming/press-braking
---

# Press Braking

Press braking bends sheet metal by driving a punch into a V-die, forming one straight-line bend at a time.

- **Category**: [Forming](https://manufacturingprocesses.org/processes/forming.md)
- **Family**: Metal
- **Materials**: Metal
- **Typical tolerances**: ±0.010 in (±0.25 mm) on a single bend dimension, accumulating roughly ±0.010 in per additional bend; bend angle ±1°, or ±0.5° with in-process angle measurement
- **Surface finish**: Inherits the incoming sheet finish; die shoulder marks on the outside of the bend are normal unless urethane or film-protected tooling is specified
- **Typical volumes**: 1–25,000 parts; above about 25,000 identical parts, hard tooling usually wins
- **Lead time**: 1–10 business days in a job shop; same-day for simple parts with material in stock. No tooling lead time

## Overview

Press braking bends sheet metal by driving a punch into a V-shaped die, forming one straight-line bend per stroke. It is the universal sheet metal forming process: no part-specific tooling, a few standard punches and dies covering thousands of part numbers, and a CNC backgauge that positions the blank for each bend in sequence.

Typical work is enclosures, chassis, brackets, panels, and channels in 0.024–0.250 in (0.6–6 mm) steel, stainless, and aluminum, at anything from one piece to tens of thousands. Because the tooling is generic, cost per part is machine time and setup, which makes press braking the natural partner to laser cutting in job-shop fabrication.

The number to design around is the inside bend radius, which in air bending is not set by the punch but by the die opening — roughly 16% of the V-width for mild steel. Get that wrong and the flat pattern is wrong.

## How it works

### Air bending, bottoming, or coining?

**Air bending** stops the punch short of the die bottom, so the sheet touches at just three points and the bend angle is a function of ram depth. One die opening covers a range of angles and thicknesses, and tonnage is lowest. This is the default.

**Bottoming** presses the sheet fully into the V, so the die angle sets the part angle. It needs roughly three to five times the air-bend force, but springback is smaller and repeatability better.

**Coining** drives the punch nose into the material, plastically deforming through the full thickness. It eliminates springback almost entirely and needs on the order of five to ten times air-bend force, so it is reserved for small, precise bends.

### The cycle

1. **Tool up.** Punch and die are loaded and the die opening chosen — the standard starting point is a V-width of about 8× material thickness.
2. **Program.** The CNC computes ram depth per bend from thickness, tooling, and target angle, and sets the backgauge positions for the bend sequence.
3. **Position.** The operator or robot slides the blank against the backgauge fingers.
4. **Bend.** The ram descends. Force for air bending mild steel follows the standard relation of roughly 575 × t² ÷ V tons per foot, with thickness t and die opening V in inches; stainless runs about 50% higher and aluminum roughly half.
5. **Springback.** The part relaxes elastically when the ram lifts — about 1–3° for mild steel, more for stainless, aluminum, and high-strength grades. The CNC overbends to compensate, and in-process angle measurement systems close the loop automatically.
6. **Sequence.** Bends are made in an order that never traps the part inside its own geometry. This is the real skill of press braking, and the reason bend sequence should be checked at design time.

Bend sequence and tool access explain most of the parts a fabricator says cannot be made: the geometry is fine, but the formed flange collides with the ram before the last bend can be reached.

## Design guidelines

### Inside bend radius

Use an inside radius of at least 1× material thickness for mild steel and annealed aluminum. Harder tempers need more: 6061-T6 commonly needs 2–4× thickness across the grain and more parallel to it. In air bending, the achieved radius follows the die opening — approximately 16% of the V-width — not the punch tip, so a sharp punch in a wide die still produces a large radius.

### One radius throughout

Keep every bend on the part at the same inside radius if you can. Each different radius is another tool change and another setup, and setup is most of the cost on short runs.

### Minimum flange length

A flange must be long enough to sit on both shoulders of the V-die: figure roughly 4× material thickness as a working minimum, or half the die opening plus the thickness. Shorter flanges slip into the die and form inconsistently, requiring special tooling.

### Hole and feature distance from a bend

Hold holes at least 2.5× material thickness plus the bend radius from the bend line. Closer holes distort. If the hole must be closer, either pierce it after bending or extend it into a slot that crosses the bend line entirely.

### Bend relief

Where a bend terminates at an edge, provide a relief notch at least as wide as the material thickness and deeper than the bend radius plus thickness. Without it the material tears at the end of the bend.

### Flat pattern and K-factor

Bend allowance depends on the K-factor — the position of the neutral axis as a fraction of thickness — which typically runs 0.33 for tight radii and approaches 0.5 for generous ones, with 0.42–0.45 a common working value for air-bent mild steel. Use the shop's actual bend deduction table rather than a nominal value; see the [bend radius and K-factor chart](/charts/sheet-metal-bend-radius-k-factor) and the [sheet metal gauge chart](/charts/sheet-metal-gauge-chart).

### Grain direction

Bend across the rolling direction wherever possible. Bends parallel to the grain crack at the outer fiber, most notably in 5000- and 6000-series aluminum and in high-strength steels.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Inside bend radius, mild steel | 1× thickness | 0.5× thickness with coining | Outer fiber cracks at tighter radii |
| Inside bend radius, 6061-T6 | 3–4× thickness | 2× thickness across grain | Hard tempers have little outer-fiber ductility |
| V-die opening | 8× thickness | 6–12× thickness | Sets achieved radius and required tonnage |
| Minimum flange length | 4× thickness | Half the V-width plus thickness | Flange must span both die shoulders |
| Hole to bend line | ≥ 2.5× thickness + radius | — | Stretching around the bend distorts the hole |
| Bend relief | Width ≥ thickness, depth > radius + thickness | — | Prevents tearing at the bend end |
| Same radius on all bends | Yes | — | Every distinct radius is another setup |
| Bend angle tolerance | ±1° | ±0.5° with angle measurement | Springback varies with coil lot and thickness |

## Cost drivers

Press braking has effectively no part-specific tooling cost, so the entire bill is setup plus cycle time plus handling. Setup dominates short runs: loading and aligning tooling, programming the sequence, and bending a first article can take longer than running fifty parts. Cycle time scales with the number of bends and with part size — a large panel needs two operators or a robot, which doubles the labor rate.

Volume breakpoints: 1–100 parts is where press braking is unbeatable, since a stamping die would never amortize. From 1,000 to 10,000 it is still usually competitive with hard tooling. Above roughly 25,000 identical parts a year, a progressive die or a roll-formed profile beats it on cycle time.

1. **Reduce the number of bends.** Each bend is a handling cycle. Two bends removed from a bracket is a real percentage of its cost.
2. **Use one bend radius and one material thickness across the whole assembly.** This lets the shop run multiple part numbers on a single setup.
3. **Keep flanges long enough to gauge on.** Flanges shorter than about 4× thickness force special tooling or a hand-held first article.
4. **Avoid bends that require the part to be reoriented repeatedly.** A sequence that can be run with one backgauge setup runs far faster.
5. **Design so no bend traps the part.** Check that the last bend can still be reached without the formed geometry hitting the ram — this is the most common cause of a redesign request from the shop floor.

## FAQ

### What is the minimum bend radius for press braking?

About 1× material thickness for mild steel and annealed aluminum. Hard tempers need more — 6061-T6 typically wants 2–4× thickness, and more if the bend runs parallel to the rolling direction. Coining can go tighter but requires roughly five to ten times the air-bend force.

### What die opening should I use on a press brake?

Start at about 8× material thickness. The die opening, not the punch tip, controls the achieved inside radius in air bending — roughly 16% of the V-width for mild steel — and it also sets the tonnage, which falls as the opening widens.

### What is the minimum flange length on a press brake?

Roughly 4× material thickness, or half the die opening plus one thickness, whichever is larger. The flange has to rest on both shoulders of the V-die. Shorter flanges slide into the die and form inconsistently, and require special tooling to make repeatably.

### How much springback should I expect?

About 1–3° for mild steel, more for stainless, aluminum, and high-strength grades. The CNC compensates by overbending, and angle-measurement systems correct in process, but springback varies between coil lots, which is why bend angles are typically toleranced at ±1°.

### What K-factor should I use for a flat pattern?

K-factor runs from about 0.33 at tight radii toward 0.5 at generous ones, with 0.42–0.45 common for air-bent mild steel. Use the fabricator's measured bend deduction table rather than a nominal figure, since it is specific to their tooling and material.

### Press braking or roll forming?

Press braking for varied parts, short runs, and anything under roughly 25,000 pieces, since it needs no dedicated tooling. Roll forming for long constant cross-section profiles at high volume, where its continuous line produces hundreds of feet per minute against one bend per stroke on a brake.

## Alternative processes

- [Roll Forming](https://manufacturingprocesses.org/processes/forming/roll-forming.md): Roll forming passes coil stock through a sequence of contoured roller stands that bend it progressively into a constant cross-section profile.
- [Metal Stamping](https://manufacturingprocesses.org/processes/forming/metal-stamping.md): Metal stamping shears and forms sheet metal between matched dies in a press, producing flat and shallow-formed parts at very high rates.
- [Tube and Section Bending](https://manufacturingprocesses.org/processes/forming/tube-and-section-bending.md): Tube and section bending draws or presses metal tube and profile around a former, with an internal mandrel wherever the wall would otherwise collapse.
- [Metal Extrusion](https://manufacturingprocesses.org/processes/forming/metal-extrusion.md): Metal extrusion pushes heated billet through a shaped die, producing continuous aluminum or copper profile of constant cross-section.

## Related processes

- [Sheet Metal Fabrication](https://manufacturingprocesses.org/processes/forming/sheet-metal-fabrication.md): Sheet metal fabrication cuts, bends and joins flat stock into a finished assembly, typically laser cutting followed by press braking, hardware and welding.
- [Laser Cutting](https://manufacturingprocesses.org/processes/cutting/laser-cutting.md): Laser cutting melts, burns or vaporises a narrow kerf through flat stock with a focused beam, producing profiles from a 2D file with no tooling.
- [Metal Stamping](https://manufacturingprocesses.org/processes/forming/metal-stamping.md): Metal stamping shears and forms sheet metal between matched dies in a press, producing flat and shallow-formed parts at very high rates.
- [Roll Forming](https://manufacturingprocesses.org/processes/forming/roll-forming.md): Roll forming passes coil stock through a sequence of contoured roller stands that bend it progressively into a constant cross-section profile.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/forming/press-braking)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Press Molding Ceramics"
category: "Forming"
subcategory: "Glass and Ceramics"
materials: ["Ceramic"]
tolerances: "Roughly ±0.5–1% of dimension for dry-pressed parts and ±1–2% for plastic-formed jiggered and ram-pressed ware. Ceramic tile dimensional classes are defined in ISO 13006 and ANSI A137.1."
volumes: "1,000–10,000,000+ pieces per year depending on variant"
lead_time: "6–16 weeks for steel dry-pressing tooling; 2–6 weeks for ram press dies or jigger molds. Production cycles run about one second (dry pressing) to a minute (ram pressing)."
url: https://manufacturingprocesses.org/processes/forming/press-molding-ceramics
---

# Press Molding Ceramics

Press molding ceramics forms clay between a die and a rotating profile tool or matched mold, producing flatware and repeat shapes quickly.

- **Category**: [Forming](https://manufacturingprocesses.org/processes/forming.md)
- **Family**: Glass and Ceramics
- **Materials**: Ceramic
- **Typical tolerances**: Roughly ±0.5–1% of dimension for dry-pressed parts and ±1–2% for plastic-formed jiggered and ram-pressed ware. Ceramic tile dimensional classes are defined in ISO 13006 and ANSI A137.1.
- **Typical volumes**: 1,000–10,000,000+ pieces per year depending on variant
- **Lead time**: 6–16 weeks for steel dry-pressing tooling; 2–6 weeks for ram press dies or jigger molds. Production cycles run about one second (dry pressing) to a minute (ram pressing).

## Overview

Press molding covers the ceramic forming processes that shape a body between tools under pressure rather than by casting or hand work. Three variants account for most production. Jiggering and jolleying press plastic clay against a rotating plaster mold with a profile tool, producing flatware and simple hollow ware in seconds per piece. Ram pressing squeezes plastic clay between two porous dies with an air-release cycle of roughly 20–60 seconds, making non-round shapes such as handles, platters, and cookware. Dry pressing compacts granulated powder at 4–8% moisture in a steel die at high pressure — the process behind essentially all ceramic tile and most technical ceramic blanks.

The reason to press rather than cast is cycle time. A slip casting takes hours; a jiggered plate takes seconds and a dry-pressed tile about a second. The trade is shape freedom: pressing needs an open die geometry, so undercuts and enclosed hollows are out.

Fired shrinkage runs roughly 6–10% for dry-pressed bodies and higher for plastic-formed ware, since plastic bodies carry far more water.

## How it works

### Jiggering and jolleying
1. A measured bat of plastic clay is dropped onto a rotating plaster mold that forms one face of the piece.
2. A metal profile tool — or, on a roller-head machine, a heated rotating roller — descends and spreads the clay against the mold, forming the second face. Jiggering forms the outside of flatware against a convex mold; jolleying forms the inside of hollow ware against a concave one.
3. The piece stays on the mold while the plaster draws water out, releases at leather-hard, and goes on to drying, bisque, glaze, and glost firing.

### Ram pressing
1. A slug of de-aired plastic clay is placed on the lower die.
2. The dies close under hydraulic pressure. Both dies are porous, and vacuum drawn through them pulls water from the clay while the pressure forms the shape.
3. Compressed air is blown back through the porous die to release the part — the reason the dies must be permeable.
4. Cycle time is roughly 20–60 seconds, and dies are typically a filled resin or specialized plaster with embedded tubing.

### Dry pressing
1. Spray-dried granulate at 4–8% moisture is fed by volume into a hardened steel die.
2. A uniaxial press compacts it, with tile presses commonly working in the region of 3,600–6,500 psi (25–45 MPa) and cycles around one second.
3. The green compact is strong enough to handle immediately and goes straight to drying and a single firing in a roller-hearth kiln.
4. For technical ceramics, isostatic pressing applies pressure hydraulically from all directions at far higher levels — 15,000–45,000 psi (100–300 MPa) — producing uniform green density in shapes such as spark plug insulators.

## Design guidelines

### Open die geometry only
Every pressing variant needs the part to release from a two-piece tool along one axis. No undercuts, no enclosed hollows, no re-entrant features. If the shape has them, it belongs in [slip casting](/processes/forming/ceramic-slip-casting) or [ceramic injection molding](/processes/forming/ceramic-injection-molding-cim).

### Uniform section
Density in a dry-pressed compact varies with the local compaction ratio. A section that is deeper in the pressing direction compacts differently from a shallow one, and the two then shrink differently in the kiln. Keep pressing depth as uniform as the design allows.

### Draft
1–3° minimum on dry-pressed parts, more on tall features. Ceramic granulate is abrasive and low-draft walls both stick and wear the die.

### Radii and edges
Fillet all internal corners at 0.02 in (0.5 mm) or more, and break external edges. Green compacts chip readily at sharp edges during handling, and fired ceramic chips even more readily.

### Aspect ratio in dry pressing
Uniaxial pressing produces a density gradient through the thickness because die-wall friction absorbs part of the applied load. Keep the pressed height modest relative to the width; tall parts need isostatic pressing to achieve uniform green density.

### Shrinkage allowance
Dry-pressed bodies shrink roughly 6–10% linearly on firing because they contain little water; plastic-formed jiggered and ram-pressed ware shrinks more. Tooling is sized to the measured figure for the specific body.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Geometry | Open, single draw direction | No undercuts | Part must release from the die |
| Draft (dry pressing) | 2–3° | 1° | Abrasive granulate galls the die |
| Internal radius | 0.04 in (1 mm) | 0.02 in (0.5 mm) | Green compacts chip at corners |
| Pressed height:width | Low | Isostatic pressing above | Die-wall friction causes density gradient |
| Firing shrinkage | 6–10% dry pressed | Body-specific | Tooling sized to measured shrinkage |

## Variants

- Jiggering
- Ram Pressing

## Cost drivers

Pressing is a tooling-and-throughput business. Dry pressing has the highest tooling cost — hardened steel dies that must resist an abrasive granulate — and by far the lowest cycle cost, at roughly a second per pressing on a multi-cavity tile press. Ram press dies are cheaper but wear faster and are replaced as consumables. Jigger molds are plaster and cheapest of all, but wear in tens to low hundreds of cycles and must be held in large numbers.

Firing dominates running cost in every variant. Tile is fired once in a continuous roller-hearth kiln; tableware is fired twice. Kiln throughput, not press throughput, is usually what caps a plant's output.

Volume breakpoints: jiggering suits a few thousand to millions of pieces of flatware a year. Ram pressing suits hundreds to tens of thousands of non-round pieces. Dry pressing needs high volume to justify steel tooling and is the standard route above roughly 100,000 units.

1. Design for a single draw direction; that decision determines whether pressing is available at all.
2. Keep pressing depth uniform to keep green density and therefore fired shrinkage uniform.
3. Match the variant to volume — plaster jigger molds for tableware runs, steel dies only at high volume.
4. Break every edge; chipping in handling is a major scrap source.
5. Design ware to stack densely in the kiln, since kiln capacity is usually the bottleneck.

## FAQ

### What is the difference between jiggering and jolleying?

Both press plastic clay between a rotating plaster mold and a profile tool. Jiggering forms flatware, with the mold defining the top face of the plate and the tool shaping the underside and foot. Jolleying forms hollow ware, with the mold defining the outside and the tool forming the inside.

### What pressure does ceramic tile pressing use?

Tile presses typically work in the region of 3,600–6,500 psi (25–45 MPa) on spray-dried granulate at 4–8% moisture, with cycles of roughly a second. Technical ceramics needing uniform green density through a tall section use isostatic pressing at 15,000–45,000 psi (100–300 MPa) instead.

### Why must ram press dies be porous?

Vacuum is drawn through the dies to pull water out of the plastic clay while it is being formed, and compressed air is then blown back through the same porosity to release the part. Without permeable dies, neither the dewatering nor the release would work.

### Can press molding produce undercuts?

No. Every pressing variant needs the part to release from a two-piece tool along a single axis. Undercuts, enclosed hollows, and re-entrant features require slip casting or ceramic injection molding instead.

### How much does a dry-pressed ceramic shrink?

Roughly 6–10% linearly on firing, considerably less than plastic-formed or slip-cast ware because the granulate contains only 4–8% moisture rather than 20% or more. Tooling is sized to the measured shrinkage of the specific body.

### Why does dry pressing produce a density gradient?

In uniaxial pressing, friction against the die wall absorbs part of the applied load, so regions further from the punch see less pressure and compact less. The result is a green density gradient that becomes a shrinkage gradient in the kiln. Keeping pressed height low relative to width limits it; isostatic pressing eliminates it.

## Alternative processes

- [Ceramic Slip Casting](https://manufacturingprocesses.org/processes/forming/ceramic-slip-casting.md): Ceramic slip casting pours liquid clay slip into a porous plaster mold that draws off water, leaving a solid layer that becomes the part.
- [Ceramic Injection Molding (CIM)](https://manufacturingprocesses.org/processes/forming/ceramic-injection-molding-cim.md): Ceramic injection molding molds a ceramic powder and binder feedstock, then debinds and sinters it into a dense technical ceramic part.
- [Clay Throwing](https://manufacturingprocesses.org/processes/forming/clay-throwing.md): Clay throwing forms a rotationally symmetric vessel by hand from a spinning ball of plastic clay on a potter's wheel.
- [Powder Metallurgy (Press and Sinter)](https://manufacturingprocesses.org/processes/forming/powder-metallurgy-press-and-sinter.md): Press-and-sinter powder metallurgy compacts metal powder in a rigid die and sinters it below its melting point into a finished net-shape part.
- [Binder Jetting](https://manufacturingprocesses.org/processes/forming/binder-jetting.md): Binder jetting prints liquid binder into a powder bed to hold each layer, producing green parts that are later sintered or infiltrated.

## Related processes

- [Ceramic Slip Casting](https://manufacturingprocesses.org/processes/forming/ceramic-slip-casting.md): Ceramic slip casting pours liquid clay slip into a porous plaster mold that draws off water, leaving a solid layer that becomes the part.
- [Ceramic Injection Molding (CIM)](https://manufacturingprocesses.org/processes/forming/ceramic-injection-molding-cim.md): Ceramic injection molding molds a ceramic powder and binder feedstock, then debinds and sinters it into a dense technical ceramic part.
- [Powder Metallurgy (Press and Sinter)](https://manufacturingprocesses.org/processes/forming/powder-metallurgy-press-and-sinter.md): Press-and-sinter powder metallurgy compacts metal powder in a rigid die and sinters it below its melting point into a finished net-shape part.
- [Clay Throwing](https://manufacturingprocesses.org/processes/forming/clay-throwing.md): Clay throwing forms a rotationally symmetric vessel by hand from a spinning ball of plastic clay on a potter's wheel.
- [Compression Molding](https://manufacturingprocesses.org/processes/forming/compression-molding.md): Compression molding presses a measured charge of rubber or thermoset into an open heated mold that closes to cure it into its final shape.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/forming/press-molding-ceramics)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Pultrusion"
category: "Forming"
subcategory: "Composites"
materials: ["Composite"]
tolerances: "Governed by ASTM D3917 for glass-reinforced pultruded shapes (EN 13706 in Europe). Wall thickness is held closest; overall width, straightness, and twist carry progressively looser bands. Do not assume machining-class tolerances on any pultruded dimension."
volumes: "Thousands to millions of linear feet; a custom die requires a long run to amortize"
lead_time: "10–20 weeks for a new die; standard catalog profiles ship from stock. Production runs are measured in feet per minute rather than parts per hour."
url: https://manufacturingprocesses.org/processes/forming/pultrusion
---

# Pultrusion

Pultrusion pulls continuous fiber through a resin bath and a heated die, curing it into a constant-section composite profile of unlimited length.

- **Category**: [Forming](https://manufacturingprocesses.org/processes/forming.md)
- **Family**: Composites
- **Materials**: Composite
- **Typical tolerances**: Governed by ASTM D3917 for glass-reinforced pultruded shapes (EN 13706 in Europe). Wall thickness is held closest; overall width, straightness, and twist carry progressively looser bands. Do not assume machining-class tolerances on any pultruded dimension.
- **Surface finish**: Die-formed surfaces are smooth and glossy where a surfacing veil is used, and show fiber texture without one. Cut ends expose raw fiber and are normally sealed for chemical or outdoor service.
- **Typical volumes**: Thousands to millions of linear feet; a custom die requires a long run to amortize
- **Lead time**: 10–20 weeks for a new die; standard catalog profiles ship from stock. Production runs are measured in feet per minute rather than parts per hour.

## Overview

Pultrusion pulls continuous reinforcement through a resin bath and then through a heated steel die that shapes and cures it, producing a constant-section composite profile in continuous length. Line speeds run roughly 12–60 in/min (0.3–1.5 m/min) with die temperatures of 250–390 °F (120–200 °C), and the product is cut to length on the fly, so length is effectively unlimited.

It is the highest fiber volume process in commercial composites — 40–70% depending on the reinforcement package — and the most anisotropic. Longitudinal roving carries the axial load; continuous strand mat or stitched fabric is added specifically to give the profile any transverse strength at all. Standard E-glass/polyester structural shapes are commonly published at around 30 ksi (207 MPa) longitudinal tensile strength and 2.5–3.0 Msi (17–21 GPa) flexural modulus, with transverse properties a small fraction of that.

Applications are structural and continuous: ladder rails, walkway grating, cable tray, concrete rebar, window lineals, utility crossarms, and I-beams and channels for corrosive environments where steel would not survive.

## How it works

1. **Creel and guide.** Glass or carbon roving is drawn from a creel of hundreds of packages, along with continuous strand mat, stitched fabric, or veil, and threaded through guide plates that place each reinforcement at its correct position in the cross-section.
2. **Impregnate.** The reinforcement passes through an open resin bath or, increasingly, a closed injection chamber immediately ahead of the die. Resins are unsaturated polyester for general structural work, vinyl ester for chemical resistance, epoxy for higher mechanical performance, and phenolic where fire, smoke, and toxicity performance governs.
3. **Preform.** A series of shaping plates progressively squeezes the wet package toward the final cross-section and drives out excess resin and air before the die.
4. **Cure in the die.** The chrome-plated tool steel die is typically 24–60 in (0.6–1.5 m) long and heated in zones to 250–390 °F (120–200 °C). Resin gels partway along and cures completely before the exit, so the profile leaves the die as a rigid solid. Exotherm inside the die is the process's controlling variable — too fast and the profile cracks internally, too slow and it exits uncured.
5. **Pull.** Reciprocating or caterpillar pullers grip the cured profile and provide the driving force for the whole line. Pull force, monitored continuously, is the primary process health indicator.
6. **Cut.** A flying cut-off saw with a diamond or carbide blade cuts to length without stopping the line.

## Design guidelines

### Constant cross-section, no exceptions
Pultrusion produces one profile per die and cannot vary the section along the length. Holes, notches, and cutouts are secondary operations, and every one of them cuts continuous fibers.

### Design for the fiber direction
Longitudinal properties are excellent and transverse properties are poor — often less than a quarter of the axial value. Never load a pultruded profile in a way that puts principal tension across the fibers, and never rely on the transverse strength of a flange to carry a bolt bearing load without added mat or fabric in that direction.

### Wall thickness
0.060–0.500 in (1.5–13 mm) is the normal range. Hold wall thickness as uniform as the section allows: a thick junction cures more slowly and exotherms more strongly than the thin walls around it, which is the usual source of internal cracking.

### Corner radii
Use inside radii of at least 0.060 in (1.5 mm) and preferably 0.125 in (3 mm). A sharp internal corner in the die is both a fiber-bridging site and a wear point in the tool.

### Connections
Bolted connections need generous edge distance — at least 3× hole diameter — and are limited by bearing strength, which is far below the axial tensile strength. Drill with carbide or diamond tooling and support the back face to avoid breakout delamination. Bonded and mechanically clamped connections often outperform bolts.

### UV and weathering
Polyester and vinyl ester surfaces chalk and fiber-bloom outdoors. Specify a surfacing veil, and a UV-stabilized resin or a painted or coextruded coating for exposed service.

### Tolerances
Dimensional tolerances for pultruded shapes are covered by ASTM D3917 (and EN 13706 in Europe); wall thickness is typically held closest, with overall width, straightness, and twist looser. Consult the standard rather than assuming machining-class tolerances.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Cross-section | Constant along length | No variation possible | Fixed die geometry |
| Wall thickness | 0.125–0.250 in (3–6 mm) | 0.060–0.500 in (1.5–13 mm) | Cure exotherm and pull force |
| Section uniformity | Within 2:1 | 3:1 | Thick junctions exotherm and crack |
| Inside radius | 0.125 in (3 mm) | 0.060 in (1.5 mm) | Fiber bridging and die wear |
| Bolt edge distance | 4× hole diameter | 3× | Low bearing and shear-out strength |
| Transverse load | Add mat or fabric | Roving alone | Roving carries no transverse load |

## Cost drivers

The die is the fixed cost and it is significant: a chrome-plated tool steel die machined to the exact cured profile, one per cross-section, plus the preform plates and guides that feed it. That cost only makes sense across long production runs, which is why pultrusion is a catalog business — most users buy standard shapes rather than commissioning a die.

Running cost is dominated by material, because the process converts feedstock into product with very little waste and very little labor. Glass roving and polyester resin are inexpensive; carbon and epoxy raise material cost several-fold and typically also require slower line speeds.

Line speed is the throughput lever, and it is set by cure kinetics rather than by machinery. A thicker section cures more slowly and runs slower, so cost per linear foot rises faster than cross-sectional area alone would suggest.

Volume breakpoints: a custom die needs thousands of linear feet of production to amortize. Below that, use a standard catalog profile, or fabricate from [composite laminating](/processes/forming/composite-laminating) if the quantity is genuinely low.

1. Use a standard catalog profile wherever possible — the die is the whole capital argument.
2. Keep wall thickness uniform to run the line as fast as cure allows.
3. Specify glass unless stiffness or weight genuinely requires carbon.
4. Design connections as clamped or bonded rather than bolted; bearing strength, not tensile strength, governs bolted joints.
5. Add a surfacing veil for any outdoor application — it is inexpensive insurance against fiber bloom.

## FAQ

### Can pultrusion make a variable cross-section?

No. The die is a fixed geometry and the profile is pulled through it continuously, so the cross-section is constant along the entire length. Any variation — holes, notches, tapers — is a secondary machining operation that cuts continuous fibers and locally weakens the profile.

### How strong are pultruded profiles?

Very strong along the fibers and weak across them. Standard E-glass/polyester structural shapes are commonly published around 30 ksi (207 MPa) longitudinal tensile with 2.5–3.0 Msi (17–21 GPa) flexural modulus, while transverse properties are a small fraction of that. Design loads must follow the axial direction.

### How fast does a pultrusion line run?

Roughly 12–60 in/min (0.3–1.5 m/min) for typical glass/polyester profiles. Speed is set by cure kinetics inside the heated die, not by machine capability, so thicker sections and slower-curing resins such as epoxy run more slowly.

### Why does pultrusion need continuous strand mat as well as roving?

Roving runs only along the pull direction and carries no transverse load at all. Continuous strand mat or stitched multi-axial fabric is added specifically to give the profile transverse and shear strength, which is what allows a bolted connection or a flange to work.

### What tolerances apply to pultruded shapes?

ASTM D3917 covers dimensional tolerances for glass-reinforced pultruded shapes in the US, and EN 13706 in Europe. Wall thickness is held closest, with width, straightness, and twist looser. Pultrusion is a structural process, not a precision one — machine any feature that needs a fit.

### When is pultrusion cheaper than aluminum extrusion?

Where corrosion, electrical insulation, or thermal break performance matters. Pultruded profiles do not corrode, do not conduct, and have a fraction of aluminum's thermal conductivity, which is why they dominate in cooling towers, chemical plants, and window thermal breaks despite a comparable or higher material cost.

## Alternative processes

- [Metal Extrusion](https://manufacturingprocesses.org/processes/forming/metal-extrusion.md): Metal extrusion pushes heated billet through a shaped die, producing continuous aluminum or copper profile of constant cross-section.
- [Plastic Extrusion](https://manufacturingprocesses.org/processes/forming/plastic-extrusion.md): Plastic extrusion pushes molten polymer through a shaped die and cools it, producing continuous profile, sheet, pipe or film of constant cross-section.
- [Composite Laminating](https://manufacturingprocesses.org/processes/forming/composite-laminating.md): Composite laminating lays reinforcing fiber into a mold and consolidates it with resin, curing into a stiff, light monocoque part.
- [Roll Forming](https://manufacturingprocesses.org/processes/forming/roll-forming.md): Roll forming passes coil stock through a sequence of contoured roller stands that bend it progressively into a constant cross-section profile.
- [Filament Winding](https://manufacturingprocesses.org/processes/forming/filament-winding.md): Filament winding wraps resin-impregnated continuous fiber onto a rotating mandrel in a controlled pattern, producing tubes and pressure vessels.

## Related processes

- [Filament Winding](https://manufacturingprocesses.org/processes/forming/filament-winding.md): Filament winding wraps resin-impregnated continuous fiber onto a rotating mandrel in a controlled pattern, producing tubes and pressure vessels.
- [Composite Laminating](https://manufacturingprocesses.org/processes/forming/composite-laminating.md): Composite laminating lays reinforcing fiber into a mold and consolidates it with resin, curing into a stiff, light monocoque part.
- [Plastic Extrusion](https://manufacturingprocesses.org/processes/forming/plastic-extrusion.md): Plastic extrusion pushes molten polymer through a shaped die and cools it, producing continuous profile, sheet, pipe or film of constant cross-section.
- [Metal Extrusion](https://manufacturingprocesses.org/processes/forming/metal-extrusion.md): Metal extrusion pushes heated billet through a shaped die, producing continuous aluminum or copper profile of constant cross-section.
- [DMC and SMC Molding](https://manufacturingprocesses.org/processes/forming/dmc-and-smc-molding.md): DMC and SMC molding compression-molds a pre-mixed dough or sheet of chopped fiber and thermoset resin into a stiff, dimensionally stable part.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/forming/pultrusion)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Rapid Prototyping"
category: "Forming"
subcategory: "Additive Manufacturing"
materials: ["Plastic", "Metal", "Composite"]
tolerances: "Process-dependent: ±0.004 in (±0.1 mm) for SLA, MSLA, PolyJet, and metal powder bed fusion; ±0.008–0.012 in (±0.2–0.3 mm) for MJF and SLS; ±0.020 in (±0.5 mm) for desktop FDM. Machined features are tighter than any of these."
volumes: "1–1,000 parts; unit cost is essentially flat with quantity"
lead_time: "1–5 business days for plastic parts; 5–15 business days for metal, where heat treatment and machining dominate. No tooling lead time in any case."
url: https://manufacturingprocesses.org/processes/forming/rapid-prototyping
---

# Rapid Prototyping

Rapid prototyping builds a part layer by layer directly from a CAD model, with no tooling and no minimum order quantity.

- **Category**: [Forming](https://manufacturingprocesses.org/processes/forming.md)
- **Family**: Additive Manufacturing
- **Materials**: Plastic, Metal, Composite
- **Typical tolerances**: Process-dependent: ±0.004 in (±0.1 mm) for SLA, MSLA, PolyJet, and metal powder bed fusion; ±0.008–0.012 in (±0.2–0.3 mm) for MJF and SLS; ±0.020 in (±0.5 mm) for desktop FDM. Machined features are tighter than any of these.
- **Surface finish**: Ra 20 µin (0.5 µm) on PolyJet and SLA up-facing surfaces through Ra 1,000 µin (25 µm) on coarse FDM and Ra 1,400 µin (35 µm) on EBM. See the [surface finish chart](/charts/surface-finish-chart) for comparison against machined and cast finishes.
- **Typical volumes**: 1–1,000 parts; unit cost is essentially flat with quantity
- **Lead time**: 1–5 business days for plastic parts; 5–15 business days for metal, where heat treatment and machining dominate. No tooling lead time in any case.

## Overview

Rapid prototyping is the practice of producing a physical part directly from a CAD model with no tooling and no minimum order quantity — in practice, additive manufacturing plus the low-volume processes that share its economics, such as [CNC machining](/processes/cutting/cnc-machining) and [vacuum casting](/processes/forming/vacuum-casting). The defining characteristic is that unit cost is nearly flat with quantity: the first part costs what the hundredth does, because there is no mold to amortize.

Lead time is the reason it exists. A part that would take 4–8 weeks to tool can be in hand in 1–5 business days, which changes how many design iterations a program can afford.

The trade-offs are real. Accuracy ranges from ±0.020 in (±0.5 mm) on desktop [FDM](/processes/forming/fused-deposition-modeling-fdm) to ±0.004 in (±0.1 mm) on [SLA](/processes/forming/stereolithography-sla) and metal powder bed fusion; material properties are process-specific rather than chosen freely; and above roughly 1,000–10,000 parts, molded and cast processes win decisively on unit cost.

## How it works

Every additive process follows the same five steps; what differs is how each layer is created.

1. **Prepare the model.** Export a watertight mesh (STL, 3MF, or STEP for machining) and check for inverted normals and non-manifold edges. Wall thicknesses that were legal in CAD may be below the chosen process's minimum.
2. **Orient and support.** Orientation sets accuracy, surface finish, strength direction, and cost simultaneously — it is the single most consequential decision in the workflow. Powder-bed processes ([SLS](/processes/forming/selective-laser-sintering-sls), [MJF](/processes/forming/multi-jet-fusion-mjf), [binder jetting](/processes/forming/binder-jetting)) need no supports; vat, extrusion, and metal powder-bed processes do.
3. **Slice.** Layer heights range from 0.0006 in (14 µm) on [material jetting](/processes/forming/material-jetting-polyjet) to 0.016 in (0.4 mm) on coarse FDM. Build time scales roughly inversely with layer height.
4. **Build.** Point-scanning processes (SLA, SLS, DMLS) take time proportional to cross-sectional area; whole-layer processes (MSLA, MJF, binder jetting) take time proportional to height only, which is why they scale better with part count.
5. **Post-process.** Support removal, washing, post-curing, depowdering, stress relief, or sintering, depending on the process. This step is routinely underestimated and is often the majority of the lead time for metal parts.

### Which process for which job?

| Process | Layer height | Typical tolerance | Materials | Best for |
| --- | --- | --- | --- | --- |
| SLA | 0.001–0.004 in (25–100 µm) | ±0.006 in (±0.15 mm) | Photopolymer | Smoothest surface, fine detail, master patterns |
| MSLA / DLP | 0.001–0.004 in (25–100 µm) | ±0.004 in (±0.1 mm) | Photopolymer | Many small parts per plate; dental, jewelry |
| FDM | 0.002–0.016 in (0.05–0.4 mm) | ±0.020 in (±0.5 mm) | Thermoplastic | Cheapest, largest, real engineering resins |
| SLS | 0.004 in (100 µm) | ±0.012 in (±0.3 mm) | Nylon, TPU | Functional plastic parts, no supports |
| MJF | 0.0031 in (80 µm) | ±0.008 in (±0.2 mm) | Nylon | Functional nylon at 10–10,000 pieces |
| PolyJet | 0.0006–0.0013 in (14–32 µm) | ±0.004 in (±0.1 mm) | Photopolymer | Multi-material, full color, overmold models |
| DMLS / SLM | 0.0008–0.0024 in (20–60 µm) | ±0.004 in (±0.1 mm) | Metal alloys | Metal geometry that cannot be machined |
| CNC machining | n/a | ±0.005 in (±0.13 mm) | Metal, plastic, wood | True material properties, tight tolerance |

## Design guidelines

### Choose the process before you finish the CAD
Minimum wall, minimum hole, and support requirements differ by a factor of three or more across processes. Designing a part to "3D printing" in the abstract produces geometry that is expensive in every one of them.

### Wall thickness by process
0.020 in (0.5 mm) for SLA and MJF, 0.030 in (0.8 mm) for SLS, 0.032 in (0.8 mm) for FDM, 0.040 in (1 mm) for metal powder bed fusion. Below these, parts either fail during the build or distort during cooling and cleaning.

### Orientation drives everything
Tolerance, surface finish, strength direction, support volume, and price all follow from how the part sits in the machine. State the critical surface on the drawing rather than leaving orientation to the operator.

### Do not print threads under M6
Print a pilot hole and tap it, or design for a heat-set or press-in insert. Printed threads in any additive process are dimensionally poor and strip early. The [tap drill chart](/charts/tap-drill-chart) gives pilot sizes and the [thread size chart](/charts/thread-size-chart) the corresponding major diameters.

### Add stock where tolerance matters
No additive process reliably holds a bearing fit. Print the feature 0.020–0.040 in (0.5–1.0 mm) undersize or oversize as appropriate and machine or ream it. See the [ISO 286 fits and tolerances chart](/charts/iso-286-fits-tolerances) for the fit classes and clearances involved.

### Design escape routes for powder and resin
Every enclosed volume in a powder or vat process must be openable. Powder needs 0.20 in (5 mm) ports; resin needs 0.12–0.16 in (3–4 mm) drains, two per cavity.

### Know when to leave
Above roughly 1,000 parts for small plastic components, [injection molding](/processes/forming/injection-molding) is cheaper per part despite the tooling. For 50–500 cosmetic parts, an SLA master plus [vacuum casting](/processes/forming/vacuum-casting) usually beats printing them all.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Wall thickness | 0.040 in (1.0 mm) | 0.020 in (0.5 mm) process-dependent | Build and cleaning survivability |
| Hole diameter | 0.12 in (3 mm) | 0.020–0.060 in (0.5–1.5 mm) | Small holes close in every process |
| Threaded feature | Insert or tapped | No printed thread under M6 | Layer structure strips under load |
| Machining stock | 0.040 in (1.0 mm) | 0.020 in (0.5 mm) | Additive cannot hold a bearing fit |
| Escape port | 0.20 in (5 mm) | 0.12 in (3 mm) | Powder and resin must exit |
| Overhang, supported processes | 45° from vertical | Support required | Each layer needs the one below |

## Variants

- Stereolithography (SLA)
- Selective Laser Sintering (SLS)
- Direct Metal Laser Sintering (DMLS)

## Cost drivers

Rapid prototyping has no tooling cost, so the entire bill is machine time, material, and post-processing labor. Machine time is dominated by build height in every process, and additionally by cross-sectional area in the point-scanning ones. Material cost tracks part volume, including support and, in powder processes, the un-fused powder that cannot be recycled.

Post-processing labor is the item that surprises people. Support removal, sanding, dyeing, depowdering, stress relief, and machining are hand operations that scale linearly with quantity and often exceed the machine cost. For metal parts, post-processing routinely costs more than the print.

Volume breakpoints worth memorizing: below 10 parts, almost any additive process is cheaper than tooling. From 50 to 500 cosmetic plastic parts, an SLA or MSLA master plus vacuum casting typically wins. From 500 to 10,000 functional nylon parts, MJF or SLS is competitive. Above 1,000–10,000 depending on part size, injection molding wins on unit cost and does not look back.

1. Reduce height before anything else; it is the strongest cost lever in every process.
2. Hollow, rib, or lattice solid volume rather than shrinking the envelope.
3. Batch parts into one build; machine setup and cooldown cycles are shared.
4. Loosen tolerances you do not need and call out only the features you do.
5. Match the process to the quantity honestly — printing 2,000 parts because the first 20 printed well is the most common and most expensive mistake in the category.

## FAQ

### Which rapid prototyping process should I use?

For the smoothest cosmetic model, SLA. For functional plastic parts, SLS or MJF. For the cheapest large part or a real engineering thermoplastic, FDM. For multi-material or full-color models, PolyJet. For metal, DMLS or SLM. For tight tolerance and true material properties, CNC machining.

### How fast is rapid prototyping?

1–5 business days for most plastic parts and 5–15 for metal, where stress relief and machining dominate. The comparison that matters is against tooling: an injection mold takes 4–8 weeks, so prototyping buys iteration cycles rather than just parts.

### At what quantity should I stop prototyping and buy tooling?

For small plastic parts, injection molding generally wins somewhere between 1,000 and 10,000 pieces depending on part size and complexity. Between about 50 and 500 cosmetic parts, an SLA master plus urethane vacuum casting usually beats both printing and tooling.

### How accurate are 3D printed parts?

It varies by an order of magnitude across processes: ±0.004 in (±0.1 mm) for SLA, MSLA, PolyJet, and metal powder bed fusion, ±0.008–0.012 in (±0.2–0.3 mm) for MJF and SLS, and ±0.020 in (±0.5 mm) for desktop FDM. No additive process holds a bearing fit — leave stock and machine the feature.

### Can prototype parts be used as end-use parts?

Some can. SLS and MJF nylon and metal powder bed fusion parts are used in production regularly. Photopolymer parts from SLA, MSLA, and PolyJet generally cannot, because they embrittle under UV and creep under sustained load.

### Why does part orientation matter so much?

Orientation simultaneously determines build height (cost), which surfaces carry supports (finish), the direction of the weak axis (strength), and the accuracy of specific features. It is the single most consequential choice in the workflow, which is why the critical surface should be called out on the drawing.

## Alternative processes

- [CNC Machining](https://manufacturingprocesses.org/processes/cutting/cnc-machining.md): CNC machining removes material from solid stock with a programmed cutting tool, producing accurate parts directly from a CAD model with no tooling.
- [Vacuum Casting (Urethane Casting)](https://manufacturingprocesses.org/processes/forming/vacuum-casting.md): Vacuum casting, known in the US as urethane casting, pours polyurethane resin into a silicone mold under vacuum to copy a master pattern for short runs.
- [Injection Molding](https://manufacturingprocesses.org/processes/forming/injection-molding.md): Injection molding forces molten thermoplastic into a steel mold under high pressure, producing high-volume identical plastic parts.
- [Thermoforming](https://manufacturingprocesses.org/processes/forming/thermoforming.md): Thermoforming heats a plastic sheet until pliable and pulls or presses it over a single-sided mold, producing large thin-walled parts with low tooling cost.
- [Investment Casting](https://manufacturingprocesses.org/processes/forming/investment-casting.md): Investment casting builds a ceramic shell around a wax pattern, melts the wax out and pours metal into the cavity, giving fine detail in almost any alloy.

## Related processes

- [Stereolithography (SLA)](https://manufacturingprocesses.org/processes/forming/stereolithography-sla.md): Stereolithography cures liquid photopolymer layer by layer with a scanning UV laser, giving the smoothest surfaces in additive manufacturing.
- [Selective Laser Sintering (SLS)](https://manufacturingprocesses.org/processes/forming/selective-laser-sintering-sls.md): Selective laser sintering fuses nylon powder layer by layer with a laser, so the surrounding loose powder supports the part and no supports are needed.
- [Fused Deposition Modeling (FDM)](https://manufacturingprocesses.org/processes/forming/fused-deposition-modeling-fdm.md): Fused deposition modeling extrudes molten thermoplastic filament along programmed paths so each layer welds to the one beneath it.
- [Multi Jet Fusion (MJF)](https://manufacturingprocesses.org/processes/forming/multi-jet-fusion-mjf.md): Multi Jet Fusion prints a fusing agent onto nylon powder and drives it with infrared energy, fusing whole layers at once for high throughput.
- [Direct Metal Laser Sintering (DMLS)](https://manufacturingprocesses.org/processes/forming/direct-metal-laser-sintering-dmls.md): Direct metal laser sintering fuses metal powder layer by layer with a laser, building dense metal parts anchored to a supported build plate.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/forming/rapid-prototyping)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Reaction Injection Molding"
category: "Forming"
subcategory: "Plastics and Rubber"
materials: ["Plastic"]
tolerances: "About ±0.020 in (±0.5 mm) on features under 12 in; on large panels expect roughly ±0.1–0.2% of the dimension. Thermal expansion of polyurethane is high, so specify the inspection temperature."
volumes: "250–10,000 parts per year; below that use urethane casting, above roughly 25,000 consider injection molding"
lead_time: "4–10 weeks for cast aluminum or nickel shell tooling; 1–5 minute cycles thereafter, plus trim and paint"
url: https://manufacturingprocesses.org/processes/forming/reaction-injection-molding
---

# Reaction Injection Molding

Reaction injection molding mixes two liquid reactants that polymerize inside a low-pressure mold, producing large tough polyurethane parts on light tooling.

- **Category**: [Forming](https://manufacturingprocesses.org/processes/forming.md)
- **Family**: Plastics and Rubber
- **Materials**: Plastic
- **Typical tolerances**: About ±0.020 in (±0.5 mm) on features under 12 in; on large panels expect roughly ±0.1–0.2% of the dimension. Thermal expansion of polyurethane is high, so specify the inspection temperature.
- **Surface finish**: Integral skin is dense and paintable as molded; textures are reproduced from the tool, and in-mold coating gives a Class A painted surface
- **Typical volumes**: 250–10,000 parts per year; below that use urethane casting, above roughly 25,000 consider injection molding
- **Lead time**: 4–10 weeks for cast aluminum or nickel shell tooling; 1–5 minute cycles thereafter, plus trim and paint

## Overview

Reaction injection molding (RIM) meters two liquid reactants — normally a polyol and an isocyanate — through an impingement mixhead and into a closed mold, where they polymerize into polyurethane in the cavity itself. Because the mold is filled with low-viscosity liquid rather than polymer melt, cavity pressure is a fraction of injection molding's and the tool can be cast aluminum or nickel shell rather than hardened steel.

That single fact defines the process. RIM makes large parts — vehicle fascias and body panels, medical equipment housings, agricultural and construction enclosures — at tooling costs and volumes that sit squarely between urethane casting and injection molding. The practical window is roughly 250–10,000 parts a year.

Chemistry also does something molding cannot: RIM tolerates thick sections and thick-to-thin transitions that would sink or void in an injection molded part.

## How it works

1. **Condition the components.** Polyol (the A side, carrying catalyst, chain extenders, blowing agent, and any filler) and isocyanate (the B side) are held in temperature-controlled day tanks, typically around 100–120 °F (38–50 °C), and continuously recirculated so they stay homogeneous and gas-free.

2. **Meter and impingement mix.** High-pressure pumps drive both streams at roughly 1,500–3,000 psi (100–200 bar) into a self-cleaning mixhead where they collide head-on in a small chamber. There is no mechanical mixer — the jets themselves do the mixing in milliseconds. Ratio accuracy is critical; an off-ratio shot cures to the wrong hardness no matter what the mold does.

3. **Fill at low pressure.** The mixed liquid leaves the mixhead and enters the cavity at low pressure through a runner designed to fill without turbulence. Cavity pressure typically stays under about 100 psi (7 bar), which is why clamp forces and tooling are so much lighter than injection molding.

4. **React in the mold.** The mold is held at roughly 130–160 °F (55–70 °C). Cream time is a few seconds, gel follows, and the exothermic reaction carries the part to demold strength. This is a chemical clock, not a cooling clock — cycle time depends on the formulation, not on wall thickness squared.

5. **Demold and post-cure.** Parts release in roughly 1–5 minutes for typical formulations. An internal mold release is normally compounded into the A side; external release agents supplement it. Structural and high-modulus parts are often post-cured in an oven to complete the reaction and stabilize dimensions before paint.

### What about structural RIM and reinforced RIM?

Blowing agents produce an integral-skin foam with a dense outer surface and a cellular core, giving high stiffness per pound at densities well below solid polyurethane. Reinforced RIM adds milled glass or flake to the A side for higher modulus and lower thermal expansion, at the cost of abrasive wear on the pumps and mixhead.

## Design guidelines

### Wall thickness

The range that makes RIM interesting is 0.100–0.500 in (2.5–12 mm), and unlike injection molding, thick sections are not automatically a defect. Cure is chemically driven, so a 10 mm section next to a 3 mm section is manageable where the same transition in a molded thermoplastic would sink or void. Even so, keep transitions gradual and avoid isolated heavy masses, which run hotter during the exotherm and can scorch.

### Draft

1–3° per side. Polyurethane grips tooling — parts are removed by hand or by air poppets on cast aluminum tools, not by an ejector plate — so err generous, and add draft on textured surfaces as you would for any molded part.

### Radii and ribs

Internal radii of at least 0.125 in (3 mm) on structural corners. Ribs are practical and useful; because the fill is low-viscosity liquid, a rib fills easily. Keep the rib base near 0.6–0.75 × the nominal wall to limit sink in the skin — RIM tolerates a thicker rib than the 0.5–0.6 × wall tabulated for thermoplastics on [/charts/injection-molding-design-guidelines](/charts/injection-molding-design-guidelines), because the skin is compliant and the cure is chemical rather than conduction-limited. Use ribs freely to stiffen large panels rather than adding wall.

### Part size and gating

RIM is one of the few plastics processes that scales comfortably to parts several square meters in area, because the low fill pressure means clamp force scales gently with projected area. Gate at the lowest point and let the cavity fill upward so air is pushed ahead of the front and out through vents at the high points. Trapped air becomes a surface void that is visible after paint.

### Surface and paint

Integral-skin RIM produces a dense, paintable outer skin. Class A automotive surfaces are achievable with in-mold coating, where the paint layer is sprayed into the open cavity before the shot. Expect to specify a primer and a flexible topcoat for exterior body panels.

### Inserts

Metal inserts, threaded bosses, and reinforcing frames can be placed in the cavity and encapsulated. Because fill pressure is low, inserts need far less support against washout than in injection molding — a genuine advantage for large bonded assemblies.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Wall thickness | 0.125–0.250 in (3–6 mm) | 0.100–0.500 in (2.5–12 mm) | Chemistry-driven cure tolerates thick sections |
| Wall transition | Gradual taper | Abrupt 3:1 step | Isolated heavy masses run hot during the exotherm |
| Draft | 2–3° per side | 1° | Polyurethane grips the tool; parts are hand-pulled |
| Internal radius | 0.125 in (3 mm) or more | — | Structural corners and flow-front smoothness |
| Rib base | 0.6–0.75 × wall | — | Low-viscosity fill makes ribs easy; skin still sinks |
| Cavity pressure | Under about 100 psi (7 bar) | — | Sets the low clamp force and cheap tooling |
| Part area | Up to several square meters | — | Clamp force scales gently with projected area |

## Cost drivers

RIM exists because of a tooling arbitrage. Fill pressure under about 100 psi means a cast aluminum or electroformed nickel tool is sufficient, at a small fraction of a hardened steel injection mold for the same large part. Against that, the per-part cost is higher: two-component liquid chemistry costs more per pound than commodity pellets, cycles run minutes rather than seconds, and most parts need trimming, flash removal, and paint.

Volume breakpoints:

- Under 100 parts: urethane casting in silicone molds is cheaper and faster.
- 250–10,000 parts a year: the RIM sweet spot, especially for parts too large for economical injection molding.
- Above roughly 25,000 parts a year for a mid-size part, injection molding's short cycle usually wins despite tooling.

Cost reduction:

1. **Use structural foam where stiffness allows.** Blowing the shot to a lower density cuts chemical cost per part directly and speeds demold.
2. **Rib instead of thickening.** Ribs fill easily in a low-viscosity system, so stiffness is cheap in geometry and expensive in wall.
3. **Consolidate parts.** RIM's size capability means one molding can replace a fabricated assembly of several thermoformed or sheet metal panels plus fasteners.
4. **Specify in-mold coating only where it is seen.** It buys a Class A surface but adds a spray step to every cycle.
5. **Design flash and gate locations for easy trim.** Trimming is manual on most RIM parts and is a real per-part labor line.

## FAQ

### How is reaction injection molding different from injection molding?

RIM fills the mold with two low-viscosity liquids that polymerize in the cavity, rather than with polymer melt. Cavity pressure stays under about 100 psi (7 bar) instead of thousands, so the tool can be cast aluminum rather than hardened steel and parts can be several square meters in area. The trade is a 1–5 minute cycle instead of seconds.

### What volumes make RIM the right choice?

Roughly 250–10,000 parts a year. Below about 100 parts, urethane casting in silicone molds is cheaper and faster; above roughly 25,000 parts a year for a mid-size part, injection molding's short cycle overcomes its tooling cost. RIM occupies the gap, especially for parts too large to injection mold economically.

### How thick can a RIM part be?

Practical walls run 0.100–0.500 in (2.5–12 mm), and thick-to-thin transitions are far better tolerated than in injection molding because cure is driven by chemistry rather than by heat conduction out through the wall. Still keep transitions gradual — isolated heavy masses run hot during the exotherm and can scorch.

### What draft angle does RIM need?

1–3° per side, and lean toward the high end. Polyurethane grips tooling and parts are typically released by hand or with air poppets rather than by an ejector plate, so generous draft matters more than it does on a steel injection mold with a full ejection system.

### Can RIM parts be painted to a Class A finish?

Yes. Integral-skin RIM produces a dense, paintable outer skin, and in-mold coating — spraying the paint layer into the open cavity before the shot — achieves automotive Class A surfaces. Exterior body panels normally still require a primer and a flexible topcoat to survive substrate movement.

### What is structural RIM?

A version using a blowing agent to produce an integral-skin foam: a dense outer skin over a cellular core. It delivers high stiffness per pound at densities well below solid polyurethane, cuts chemical cost per part, and demolds faster. Reinforced RIM instead adds milled glass for higher modulus and lower thermal expansion.

## Alternative processes

- [Thermoforming](https://manufacturingprocesses.org/processes/forming/thermoforming.md): Thermoforming heats a plastic sheet until pliable and pulls or presses it over a single-sided mold, producing large thin-walled parts with low tooling cost.
- [Vacuum Casting (Urethane Casting)](https://manufacturingprocesses.org/processes/forming/vacuum-casting.md): Vacuum casting, known in the US as urethane casting, pours polyurethane resin into a silicone mold under vacuum to copy a master pattern for short runs.
- [Injection Molding](https://manufacturingprocesses.org/processes/forming/injection-molding.md): Injection molding forces molten thermoplastic into a steel mold under high pressure, producing high-volume identical plastic parts.
- [Rotation Molding](https://manufacturingprocesses.org/processes/forming/rotation-molding.md): Rotational molding tumbles powdered polymer inside a heated mold on two axes so it fuses into a seamless, stress-free hollow part with a uniform wall.

## Related processes

- [Vacuum Casting (Urethane Casting)](https://manufacturingprocesses.org/processes/forming/vacuum-casting.md): Vacuum casting, known in the US as urethane casting, pours polyurethane resin into a silicone mold under vacuum to copy a master pattern for short runs.
- [Injection Molding](https://manufacturingprocesses.org/processes/forming/injection-molding.md): Injection molding forces molten thermoplastic into a steel mold under high pressure, producing high-volume identical plastic parts.
- [Rotation Molding](https://manufacturingprocesses.org/processes/forming/rotation-molding.md): Rotational molding tumbles powdered polymer inside a heated mold on two axes so it fuses into a seamless, stress-free hollow part with a uniform wall.
- [Compression Molding](https://manufacturingprocesses.org/processes/forming/compression-molding.md): Compression molding presses a measured charge of rubber or thermoset into an open heated mold that closes to cure it into its final shape.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/forming/reaction-injection-molding)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Roll Forming"
category: "Forming"
subcategory: "Metal"
materials: ["Metal"]
tolerances: "±0.010–0.030 in (±0.25–0.75 mm) on cross-section dimensions and about ±1° on formed angles; cut-length tolerance depends on whether cutoff is flying or stationary"
volumes: "20,000+ linear feet per profile; below that, press braking is usually cheaper"
lead_time: "12–20 weeks to design, cut, and debug a roll set; days to weeks per production run thereafter"
url: https://manufacturingprocesses.org/processes/forming/roll-forming
---

# Roll Forming

Roll forming passes coil stock through a sequence of contoured roller stands that bend it progressively into a constant cross-section profile.

- **Category**: [Forming](https://manufacturingprocesses.org/processes/forming.md)
- **Family**: Metal
- **Materials**: Metal
- **Typical tolerances**: ±0.010–0.030 in (±0.25–0.75 mm) on cross-section dimensions and about ±1° on formed angles; cut-length tolerance depends on whether cutoff is flying or stationary
- **Surface finish**: Inherits the coil finish, including pre-painted and galvanized coatings, which survive because forming contact is rolling rather than sliding
- **Typical volumes**: 20,000+ linear feet per profile; below that, press braking is usually cheaper
- **Lead time**: 12–20 weeks to design, cut, and debug a roll set; days to weeks per production run thereafter

## Overview

Roll forming feeds coil stock continuously through a line of contoured roller stands, each one adding a small increment of bend, until the strip emerges as a finished constant cross-section profile. Nothing is cut or stretched — the metal is progressively bent, and total deformation is spread over 10–30 stands so no single pass exceeds the material's forming limit.

It produces the long metal profiles of the built environment: metal roofing and siding panels, steel studs, door and window frames, garage door sections, shelf rails, guardrail, and automotive trim and structural rails. Stock thickness typically runs 0.010–0.250 in (0.25–6 mm) in steel, stainless, and aluminum, including pre-painted and galvanized coil, whose coating survives because the process bends rather than scrapes.

Line speeds of 50–300 ft/min (15–90 m/min) make it the cheapest way to make a profile by the mile. The catch is the tooling: a full roll set is one of the more expensive tool packages in sheet metal, and it makes exactly one cross-section.

## How it works

1. **Uncoil and level.** Coil is decoiled and passed through a leveler that removes coil set. Incoming flatness and consistent gauge matter more here than in most processes, because errors compound down the line.
2. **Pre-punch (optional).** Holes, slots, and notches are punched in the flat strip before forming. Punching flat is far cheaper than punching a formed profile, but any hole close to a future bend will distort as the metal wraps.
3. **Progressive forming.** The strip passes through pairs of driven rolls on successive stands. Each stand advances the bend a few degrees toward final shape. The sequence is designed as a "flower pattern" — the nested cross-sections at every stand, overlaid — and this pattern is the core engineering deliverable of a roll form tool.
4. **Strain management.** The metal at the outer edge of the strip has to travel a longer path than the metal at the center. Distributing that extra length over enough stands is what prevents edge wave, oil canning, and longitudinal bow. Adding stands is the standard fix for a profile that will not run.
5. **Sizing and straightening.** Final stands and a straightening head correct twist, camber, and bow. Even a well-designed line needs this: residual stress from the coil never fully disappears.
6. **Cutoff.** Parts are cut to length either by a flying shear or saw that travels with the moving strip, or by a stationary press with the line briefly stopped. Pre-cut blanks are also fed as an alternative for short parts.
7. **Post-forming.** Optional in-line operations include seam welding for tube, embossing, curving, and end forming.

Because forming is progressive and the tooling contacts the surface in rolling rather than sliding contact, pre-finished material comes through with its paint or coating intact — a significant advantage over stamping, where drawing operations scuff and stretch coatings.

## Design guidelines

### Constant cross-section, always

A roll-formed profile is the same at every point along its length. Any variation along the length — a taper, a change in width, a formed end — is a separate operation. Design as much as possible into the constant section.

### Bend radius

Inside radius should be at least 1× material thickness for mild steel and more for high-strength grades and hard aluminum tempers. Tighter radii can be rolled than press braked on the same material, because the bend is applied gradually, but the mechanics of outer-fiber elongation still apply. Pull thickness values from the [sheet metal gauge chart](/charts/sheet-metal-gauge-chart) and radius and K-factor values for the flat width development from the [bend radius and K-factor chart](/charts/sheet-metal-bend-radius-k-factor).

### Keep the profile symmetric

An asymmetric section pulls the strip sideways as it forms and comes off the line with twist and camber that the straightener must fight. Symmetry about the vertical centerline is the single biggest determinant of whether a profile runs easily.

### Watch the deep, narrow channel

Deep legs with narrow openings are hard to reach with rolls and are prone to springback the straightener cannot correct. Where a deep return is required, expect additional stands and a longer development.

### Pre-punched holes near bends

Holes punched flat, then rolled through a bend, distort. Keep pre-punched features at least 2× material thickness away from any bend line, or punch after forming — which costs more but preserves the feature.

### End condition and cut length

Cutoff leaves a burr and, on a flying shear, some end flare — a slight opening of the section at the cut. If the ends mate into a fitting, specify it, because end flare is corrected with an additional end-forming or restrike operation.

### Tolerances

Cross-section dimensions typically hold to ±0.010–0.030 in (±0.25–0.75 mm) and angles to about ±1°. Length tolerance depends on the cutoff method; a flying cutoff on a fast line is looser than a stop-and-cut press.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Cross-section | Constant along full length | Any variation is a second operation | Rolls produce one profile continuously |
| Inside bend radius | ≥ 1× thickness | Larger for high-strength grades | Outer fiber elongation limits are unchanged |
| Profile symmetry | Symmetric about the centerline | Asymmetric needs extra stands | Uneven forming loads twist the strip |
| Pre-punched hole to bend | ≥ 2× thickness | Punch after forming | Holes distort as the metal wraps the bend |
| Number of stands | 10–20 for typical profiles | 30+ for high-strength or complex sections | Strain must be spread to avoid edge wave |
| Cross-section tolerance | ±0.020 in (±0.5 mm) | ±0.010 in (±0.25 mm) | Springback varies with coil lot |
| Material thickness | 0.030–0.125 in (0.8–3 mm) | 0.010–0.250 in (0.25–6 mm) | Thin strip buckles, thick strip needs heavy stands |

## Cost drivers

Roll forming inverts the usual sheet metal economics. Tooling is high — a dedicated roll set with 10 to 30 stands of machined and hardened rolls is a major investment and takes months to design, cut, and debug — but running cost is among the lowest in metalworking. A line producing 200 ft/min with one operator has a per-foot cost that press braking cannot approach, and material utilization is near total because there is no skeleton scrap, only the cutoff kerf.

Volume breakpoints: below roughly 20,000–30,000 linear feet of a given profile, press braking or extrusion is cheaper because the roll tooling never amortizes. From 50,000 ft upward, roll forming pulls decisively ahead, and at millions of feet the line is essentially printing profile.

1. **Design a family of parts around one roll set.** Different lengths, hole patterns, and end forms from the same cross-section reuse the whole tool investment.
2. **Keep the profile symmetric.** Symmetric sections need fewer stands and less straightening, which cuts both tooling cost and scrap at startup.
3. **Punch in line and flat.** In-line pre-punching removes a whole secondary operation and its handling.
4. **Use pre-finished coil.** Pre-painted and galvanized stock comes off the line finished; roll forming's rolling contact does not damage coating the way drawing does.
5. **Widen the cross-section tolerance where you can.** Holding ±0.010 in on every dimension of a profile requires more stands and more straightening than ±0.030 in on the non-critical ones.

## FAQ

### How many roll stands does a profile need?

Typically 10–20 for common profiles, and 30 or more for complex sections or high-strength steel. The number is set by how much bend can be added per pass without exceeding the material's forming limit at the strip edge. Adding stands is the standard cure for edge wave and bow.

### What volume justifies roll forming?

Roughly 20,000–30,000 linear feet of a single profile as a lower bound, with the process pulling clearly ahead above 50,000 ft. Below that, press braking's zero tooling cost wins, since a full roll set is a months-long tooling investment that makes exactly one cross-section.

### Can holes be punched before roll forming?

Yes, and it is much cheaper than punching a formed profile — most lines pre-punch in line. Keep pre-punched features at least 2× material thickness from any bend line, because a hole that passes through a forming station distorts as the metal wraps around the radius.

### Why do roll-formed parts come out twisted or bowed?

Because metal at the strip edge travels a longer path than metal at the center, and residual coil stress relaxes unevenly. Asymmetric profiles are the worst offenders. The fixes are spreading strain over more stands, improving profile symmetry, and a final straightening head.

### Can pre-painted metal be roll formed?

Yes — it is one of the process's real advantages. Forming contact is rolling rather than sliding, so pre-painted and galvanized coil comes off the line with its coating intact. That eliminates a whole downstream finishing operation on products like roofing panels and door frames.

### Roll forming or aluminum extrusion?

Extrusion makes solid and hollow sections with varying wall thickness in one operation, but is limited to aluminum and other extrudable alloys and needs a press. Roll forming makes constant-thickness sections from any coil material, including high-strength and pre-painted steel, at very high line speed.

## Alternative processes

- [Press Braking](https://manufacturingprocesses.org/processes/forming/press-braking.md): Press braking bends sheet metal by driving a punch into a V-die, forming one straight-line bend at a time.
- [Metal Extrusion](https://manufacturingprocesses.org/processes/forming/metal-extrusion.md): Metal extrusion pushes heated billet through a shaped die, producing continuous aluminum or copper profile of constant cross-section.
- [Pultrusion](https://manufacturingprocesses.org/processes/forming/pultrusion.md): Pultrusion pulls continuous fiber through a resin bath and a heated die, curing it into a constant-section composite profile of unlimited length.
- [Metal Stamping](https://manufacturingprocesses.org/processes/forming/metal-stamping.md): Metal stamping shears and forms sheet metal between matched dies in a press, producing flat and shallow-formed parts at very high rates.

## Related processes

- [Press Braking](https://manufacturingprocesses.org/processes/forming/press-braking.md): Press braking bends sheet metal by driving a punch into a V-die, forming one straight-line bend at a time.
- [Metal Stamping](https://manufacturingprocesses.org/processes/forming/metal-stamping.md): Metal stamping shears and forms sheet metal between matched dies in a press, producing flat and shallow-formed parts at very high rates.
- [Metal Extrusion](https://manufacturingprocesses.org/processes/forming/metal-extrusion.md): Metal extrusion pushes heated billet through a shaped die, producing continuous aluminum or copper profile of constant cross-section.
- [Sheet Metal Fabrication](https://manufacturingprocesses.org/processes/forming/sheet-metal-fabrication.md): Sheet metal fabrication cuts, bends and joins flat stock into a finished assembly, typically laser cutting followed by press braking, hardware and welding.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/forming/roll-forming)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Rotation Molding"
category: "Forming"
subcategory: "Plastics and Rubber"
materials: ["Plastic"]
tolerances: "About ±1–2% of the dimension on mold-formed features; wall thickness commonly varies ±10–20% around nominal. Polyethylene shrinkage of 3.0–3.5% must be built into the tool."
volumes: "10–10,000 parts per year; the sweet spot is large parts at low to medium volume"
lead_time: "4–10 weeks for tooling depending on size and whether it is fabricated steel or cast aluminum; a 10–60 minute cycle per part thereafter"
url: https://manufacturingprocesses.org/processes/forming/rotation-molding
---

# Rotation Molding

Rotational molding tumbles powdered polymer inside a heated mold on two axes so it fuses into a seamless, stress-free hollow part with a uniform wall.

- **Category**: [Forming](https://manufacturingprocesses.org/processes/forming.md)
- **Family**: Plastics and Rubber
- **Materials**: Plastic
- **Typical tolerances**: About ±1–2% of the dimension on mold-formed features; wall thickness commonly varies ±10–20% around nominal. Polyethylene shrinkage of 3.0–3.5% must be built into the tool.
- **Surface finish**: Reproduces the mold surface on the outside; the inside surface is free-formed and slightly textured. Cast aluminum molds can carry a fine texture; fabricated steel molds are smoother but show weld lines
- **Typical volumes**: 10–10,000 parts per year; the sweet spot is large parts at low to medium volume
- **Lead time**: 4–10 weeks for tooling depending on size and whether it is fabricated steel or cast aluminum; a 10–60 minute cycle per part thereafter

## Overview

Rotational molding tumbles a measured charge of powdered polymer inside a heated, closed mold turning slowly on two perpendicular axes. The powder melts and fuses against the cavity wall layer by layer, producing a seamless, stress-free hollow part with a naturally uniform wall. Kayaks, water and chemical tanks, road barriers, agricultural bins, playground equipment, and vehicle fuel tanks are all rotomolded.

Roughly 80–90% of the volume is polyethylene — LLDPE, HDPE, and crosslinked PE — with PP, nylon, and PVC plastisol making up most of the balance. The process runs at essentially atmospheric pressure, so molds are fabricated sheet steel or cast aluminum shells rather than pressure-bearing tools. That makes tooling far cheaper than blow molding or injection molding for a large part.

The trade is speed: a full heat-and-cool cycle runs 10–60 minutes, against seconds for injection molding.

## How it works

1. **Charge.** A weighed shot of polymer powder, typically ground to around 35 mesh (roughly 500 µm), is loaded into one half of the open mold. Shot weight, not process pressure, sets wall thickness — double the charge and you double the wall.

2. **Heat and rotate.** The clamped mold enters an oven at 500–700 °F (260–370 °C) while rotating on two axes, usually 4–12 rpm on the major axis with a speed ratio around 4:1 between axes. The rotation is slow: powder tumbles and deposits, it does not fling outward. Centrifugal force plays no meaningful role.

3. **Fuse and densify.** Powder sticks to the hot cavity wall and coalesces. The control variable is peak internal air temperature (PIAT), measured with a wireless probe through the mold vent — around 400–430 °F (204–221 °C) for polyethylene. Under-cook and the part is porous with poor impact strength; over-cook and the polymer oxidizes, going yellow and brittle.

4. **Cool while still rotating.** The mold moves to a cooling station with forced air and often a water mist, still turning so the melt cannot pool. Cooling too fast warps the part and locks in stress; too slow costs cycle time and increases crystallinity in PE, which increases shrinkage.

5. **Demold.** The mold opens and the part is pulled. Polyethylene shrinks about 3.0–3.5%, which is high, and the part shrinks away from external surfaces but onto any male feature projecting into the cavity.

### Why is the wall so uniform?

Because deposition is driven by contact with a hot surface rather than by flow. Every square inch of cavity spends the same time in contact with tumbling powder, so material distributes itself without the flow-length and stretch problems that dominate injection and blow molding. This is the process's defining advantage: corners come out at full wall thickness rather than thinned.

## Design guidelines

### Wall thickness

Practical range is 0.060–0.500 in (1.5–12 mm), with 0.125–0.250 in (3–6 mm) covering most tanks and containers. Wall is set by shot weight and applies to the whole part, so you cannot locally thicken one region — if one area needs more material, the entire part gets it. Expect thickness variation on the order of ±10–20% around the nominal.

### Radii — the opposite rule from injection molding

Outside corners want a radius of at least 0.25 in (6 mm), and 0.5 in (13 mm) or more is far better. Inside corners must be even more generous, because powder bridges across a tight internal angle and leaves that corner thin and weak — exactly inverted from injection molding, where inside corners run thick. As a rule, make inside radii roughly twice the outside radius on the same corner.

### Draft

1–2° per side on external surfaces is enough, since the part shrinks away from the cavity. Any feature that projects into the cavity — a core forming a recess, a molded-in handle pocket — needs 3–5°, because the part shrinks onto it as it cools and grips hard.

### Stiffening

You cannot rib a rotomolded part the way you rib an injection molded one; there is no pressure to force material into a narrow slot. Stiffen with geometry instead: broad corrugations, domed panels, and kiss-offs, where two opposing walls are brought together in the tool so they fuse into an internal web. Kiss-offs are the standard way to add a shear web to a tank or a door and cost nothing but tool geometry.

### Threads, inserts, and flat surfaces

Molded-in threaded inserts, bosses, and metal fittings are held on posts in the mold and are captured as the polymer fuses around them. Avoid large flat panels — polyethylene shrinks 3% or more and a big unsupported flat will warp or oil-can. Crown it slightly or break it up with corrugations. For selecting between PE grades and alternatives on stiffness and service temperature, see [/charts/material-properties](/charts/material-properties).

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Wall thickness | 0.125–0.250 in (3–6 mm) | 0.060–0.500 in (1.5–12 mm) | Set globally by shot weight, not locally |
| Outside radius | 0.5 in (13 mm) or more | 0.25 in (6 mm) | Tight corners thin and lose impact strength |
| Inside radius | About 2 × the outside radius | — | Powder bridges tight internal angles |
| Draft, external | 1–2° per side | 1° | Part shrinks away from the cavity |
| Draft, over a core | 3–5° per side | 3° | Part shrinks onto anything projecting inward |
| Flat panel | Crown or corrugate it | — | 3%+ shrinkage warps large flats |
| Stiffening | Kiss-offs and corrugations | No narrow ribs | No pressure available to fill a slot |

## Cost drivers

Tooling is the cheapest of any hollow-part process at size, because the mold is an unpressurized shell — fabricated steel for large simple shapes, cast aluminum where detail matters. A rotomold for a 200-gallon tank costs a fraction of a blow mold for the same part and a small fraction of an injection mold. That is why rotational molding owns the large-part, low-to-medium-volume territory.

The recurring cost is time. A 10–60 minute cycle means each mold produces a handful of parts per shift, so throughput comes from running many molds on a carousel or rock-and-roll machine simultaneously. Powder also costs more per pound than pellets because of the grinding step.

Volume breakpoints:

- Under about 100 parts a year: fabricated sheet steel tooling, single mold, manual machine.
- 100–5,000: cast aluminum tooling with better detail and faster heat transfer, multiple molds on a carousel arm.
- Above roughly 10,000 parts a year for a small part, blow molding or injection molding usually takes over on unit cost.

Cost reduction:

1. **Take wall out globally.** Shot weight is the direct material cost. If only one region needs 6 mm, consider whether a kiss-off or corrugation can deliver the same stiffness at 4 mm everywhere.
2. **Fit more molds on the arm.** Cycle time is dominated by oven and cooling residence, not by part count, so a second and third mold on the same arm are nearly free in machine time.
3. **Use cast aluminum where the part warrants it.** It conducts heat much better than fabricated steel, which shortens the cycle and pays back over a long program.
4. **Design out secondary operations.** Molded-in inserts, threads, and fittings placed during the charge step avoid drilling, tapping, and welding after demold.
5. **Avoid deep, narrow recesses.** They trap heat, slow the cycle, and need extra draft to release.

## FAQ

### How is wall thickness controlled in rotational molding?

By shot weight alone. The powder charge you load into the mold determines the wall, and it applies uniformly to the whole part — you cannot thicken one region without thickening everything. Practical walls run 0.060–0.500 in (1.5–12 mm), with 0.125–0.250 in (3–6 mm) covering most tanks and containers.

### What corner radius does a rotomolded part need?

At least 0.25 in (6 mm) on outside corners, and 0.5 in (13 mm) or more is much better. Inside corners need roughly twice the outside radius, because powder bridges across a tight internal angle and leaves that corner thin — the reverse of injection molding, where inside corners come out thick.

### Why does rotational molding give a uniform wall?

Deposition is driven by contact with a hot mold surface rather than by pressurized flow. Every part of the cavity spends the same time in contact with the tumbling powder, so material distributes itself evenly and corners arrive at full thickness instead of thinned, which is the opposite of blow molding and thermoforming.

### Can you put ribs on a rotationally molded part?

Not narrow ones — the process runs at atmospheric pressure and nothing forces material into a slot. Stiffen with broad corrugations, domed panels, and kiss-offs, where two opposing walls are brought together in the tool and fuse into an internal web. Kiss-offs add a structural shear web at no material cost.

### How long is a rotational molding cycle?

Typically 10–60 minutes for the full heat-and-cool sequence, versus seconds for injection molding. Throughput therefore comes from running several molds simultaneously on a carousel or rock-and-roll machine rather than from shortening the cycle.

### When is rotational molding better than blow molding?

For large, thick-walled, low-to-medium-volume hollow parts. Rotomold tooling is an unpressurized shell that costs a fraction of a blow mold at the same size, the wall comes out uniform including in the corners, and there is no pinch-off seam. Blow molding wins above roughly 10,000 parts a year, where the much shorter cycle dominates.

## Alternative processes

- [Blow Molding](https://manufacturingprocesses.org/processes/forming/blow-molding.md): Blow molding inflates a heated plastic tube or preform against the walls of a mold with compressed air, producing hollow one-piece containers and tanks.
- [Thermoforming](https://manufacturingprocesses.org/processes/forming/thermoforming.md): Thermoforming heats a plastic sheet until pliable and pulls or presses it over a single-sided mold, producing large thin-walled parts with low tooling cost.
- [Reaction Injection Molding](https://manufacturingprocesses.org/processes/forming/reaction-injection-molding.md): Reaction injection molding mixes two liquid reactants that polymerize inside a low-pressure mold, producing large tough polyurethane parts on light tooling.
- [Injection Molding](https://manufacturingprocesses.org/processes/forming/injection-molding.md): Injection molding forces molten thermoplastic into a steel mold under high pressure, producing high-volume identical plastic parts.

## Related processes

- [Blow Molding](https://manufacturingprocesses.org/processes/forming/blow-molding.md): Blow molding inflates a heated plastic tube or preform against the walls of a mold with compressed air, producing hollow one-piece containers and tanks.
- [Thermoforming](https://manufacturingprocesses.org/processes/forming/thermoforming.md): Thermoforming heats a plastic sheet until pliable and pulls or presses it over a single-sided mold, producing large thin-walled parts with low tooling cost.
- [Reaction Injection Molding](https://manufacturingprocesses.org/processes/forming/reaction-injection-molding.md): Reaction injection molding mixes two liquid reactants that polymerize inside a low-pressure mold, producing large tough polyurethane parts on light tooling.
- [Dip Molding](https://manufacturingprocesses.org/processes/forming/dip-molding.md): Dip molding withdraws a heated former from liquid plastisol or latex, leaving a coating that cures into a flexible open-ended part such as a grip or cap.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/forming/rotation-molding)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Sand Casting"
category: "Forming"
subcategory: "Metal"
materials: ["Metal"]
tolerances: "About ±0.030 in (±0.8 mm) on the first inch plus roughly ±0.003 in per additional inch in green sand; no-bake and shell molds hold tighter. Allow up to 0.030 in (0.8 mm) mismatch across the parting line"
volumes: "1–100,000 parts per year; economical from a single piece"
lead_time: "1–4 weeks for a pattern, then 1–3 weeks per casting run; prototype castings from printed patterns in 1–2 weeks"
url: https://manufacturingprocesses.org/processes/forming/sand-casting
---

# Sand Casting

Sand casting pours molten metal into a cavity formed in bonded sand around a pattern, then breaks the mold away to release the part.

- **Category**: [Forming](https://manufacturingprocesses.org/processes/forming.md)
- **Family**: Metal
- **Materials**: Metal
- **Typical tolerances**: About ±0.030 in (±0.8 mm) on the first inch plus roughly ±0.003 in per additional inch in green sand; no-bake and shell molds hold tighter. Allow up to 0.030 in (0.8 mm) mismatch across the parting line
- **Surface finish**: Roughly 250–900 µin Ra (6–23 µm) as cast; no-bake and shell molds reach the low end of that range
- **Typical volumes**: 1–100,000 parts per year; economical from a single piece
- **Lead time**: 1–4 weeks for a pattern, then 1–3 weeks per casting run; prototype castings from printed patterns in 1–2 weeks

## Overview

Sand casting forms a cavity by packing bonded sand around a pattern, removing the pattern, and pouring molten metal into the void; the mold is destroyed to release the part. It remains the default process for large, heavy, or low-volume metal parts — engine blocks, pump housings, machine bases, manifolds, valve bodies.

Nothing else casts a multi-ton part economically, and nothing else reaches first article as cheaply: a wooden or 3D-printed pattern costs a fraction of a die and can be ready in days. Almost any pourable alloy works, including gray and ductile iron, aluminum, bronze, and carbon and stainless steels.

The price is precision and finish. Standard linear tolerance is about ±0.030 in (±0.8 mm) on the first inch, as-cast surfaces run 250–900 µin Ra, and every functional face needs machining stock. Economical from a single part up to roughly 100,000 per year.

## How it works

1. **Pattern.** A pattern of the part is made in wood, plastic, aluminum, or printed polymer, built deliberately oversize to compensate for solidification shrinkage — the shrink allowance is roughly 1% for aluminum and gray iron and closer to 2% for steel. Draft and machining stock are added to the pattern, not to the part model.
2. **Molding.** The pattern is placed in a flask and sand is packed around it. Green sand — silica sand with roughly 6–10% bentonite clay and 2–4% water — is the highest-volume medium. Chemically bonded no-bake sand (furan or phenolic urethane) gives better dimensional stability and finish for lower volumes and larger parts.
3. **Cope and drag.** The mold is made in two halves so the pattern can be withdrawn. The parting line between them is a permanent feature of the casting, visible as a fin.
4. **Cores.** Internal passages are formed by separate sand cores, usually shell or cold-box bonded, set into the mold on core prints. Cores are what make a water jacket or a hollow valve body possible.
5. **Gating and risers.** Sprue, runners, and gates are cut to fill the mold without turbulence; risers are reservoirs of liquid metal that feed the casting as it shrinks. Directional solidification — the casting freezing toward the riser — is what keeps shrinkage cavities out of the part.
6. **Pour.** Metal is poured with superheat above its melting point: gray iron at roughly 2,500–2,600°F (1,370–1,425°C), aluminum at 1,300–1,400°F (700–760°C). See [metal melting points](/charts/metal-melting-points).
7. **Cool and shake out.** Cooling runs from minutes for a small aluminum part to a day or more for a heavy steel casting. The sand mold is then broken up and most of the sand reclaimed.
8. **Cleaning.** Gates and risers are sawn or torched off, the parting fin is ground, and the casting is shot blasted. Heat treatment and machining follow.

Cooling rate governs properties: a heavy section cools slowly, giving coarser grain and lower strength than a thin section in the same casting. Chills — iron inserts placed in the mold — locally speed up sections that would otherwise be the weak spot.

## Design guidelines

### Uniform sections, and feed the heavy ones

Aim for a uniform wall. Minimum practical section is about 0.125 in (3 mm) in aluminum, 0.25 in (6 mm) in gray iron, and 0.25–0.375 in (6–10 mm) in steel; below that, metal freezes before the section fills. Where a heavy boss or junction is unavoidable it must be reachable by a riser, or it will contain a shrinkage cavity.

### Draft on every vertical face

Green sand molding needs 1.5–3° of draft to withdraw the pattern without tearing the mold; deep pockets need more, and rigid no-bake molds can approach 1°. Machined surfaces get draft too — it comes off later with the machining stock.

### Radius everything

Internal fillets should be at least 0.25 in (6 mm), or roughly half the adjoining wall thickness, whichever is larger. Sharp inside corners create hot spots and sand erosion; sharp outside corners break off in the mold and end up as inclusions in the metal.

### Machining allowance

Add 0.060–0.125 in (1.5–3 mm) of stock per surface on small and medium castings, and more on large ones or faces far from the parting line. Under-allowing is the most common cause of scrap: a casting that moves 0.050 in within tolerance leaves an uncleaned face.

### Design for a flat parting line

A parting line that follows a single plane is cheap. A stepped or contoured parting line costs pattern work and adds mismatch — expect up to about 0.030 in (0.8 mm) of offset across it, and never let a critical dimension depend on cope-to-drag alignment.

### Cores need prints and vents

Every core must be supported by core prints and must vent, or trapped gas will blow into the metal. Cores under about 0.5 in (12 mm) diameter, or slender cores longer than roughly 6× diameter, will shift or float; drill those features instead.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Minimum wall, aluminum | 0.19–0.25 in (5–6 mm) | 0.125 in (3 mm) | Thin sections freeze before the cavity fills |
| Minimum wall, gray iron | 0.25–0.375 in (6–10 mm) | 0.25 in (6 mm) | Iron loses fluidity fast in thin sections |
| Draft | 1.5–3° | 1° in no-bake sand | Pattern must withdraw without tearing sand |
| Internal fillet | 0.5× adjoining wall | 0.25 in (6 mm) | Sharp corners create hot spots and shrinkage |
| Machining stock | 0.060–0.125 in (1.5–3 mm) | More on large castings | Casting movement must not leave a black face |
| Cored hole diameter | ≥ 0.5 in (12 mm) | Drill anything smaller | Slender cores shift, float, and break |
| Parting-line mismatch | Design to tolerate 0.030 in (0.8 mm) | — | Cope and drag never align perfectly |

## Cost drivers

Pattern cost is low and per-part cost is high — the reverse of die casting. A wooden or printed pattern is inexpensive and can be modified with hand tools, which is why sand casting owns prototype and low-volume metal work. But every part consumes a new mold, so molding labor, sand handling, and cleaning dominate at volume. Cleaning-room labor — cutting gates, grinding parting lines, blasting — is routinely a third of the cost of a small casting, and it is driven directly by how many gates and risers the part needs.

Volume breakpoints: 1–100 parts on a hand-molded no-bake pattern; 100–5,000 on a matchplate pattern in an automated green sand line; above roughly 25,000–50,000 per year, permanent mold or die casting starts to win on aluminum, and shell molding or investment casting on steel.

1. **Simplify the cores.** Cores carry their own tooling, labor, and scrap rate. A part redesigned to need two cores instead of five drops sharply in cost.
2. **Keep the parting line flat.** It cuts pattern cost, reduces mismatch, and makes automated matchplate molding possible.
3. **Consolidate machined faces.** Group faces that must be machined onto the same setup axis; the machining bill often exceeds the casting bill.
4. **Specify the loosest tolerance that works.** Asking for ±0.010 in as cast forces a different process; ±0.030 in on the first inch is free.
5. **Choose an alloy for castability, not just strength.** Gray iron and aluminum-silicon alloys fill thin sections and feed well; high-strength steels and pure copper are much harder to cast sound.

## FAQ

### What tolerance can sand casting hold?

Plan on about ±0.030 in (±0.8 mm) on the first inch of any dimension, plus roughly ±0.003 in per additional inch, in green sand. No-bake and shell molds do better. Any dimension crossing the parting line must also absorb up to about 0.030 in of cope-to-drag mismatch.

### What is the minimum wall thickness for a sand casting?

Roughly 0.125 in (3 mm) in aluminum, 0.25 in (6 mm) in gray iron, and 0.25–0.375 in (6–10 mm) in steel. These are fluidity limits: below them the metal solidifies before the section fills, producing misruns and cold shuts.

### How much machining stock should I add to a sand casting?

0.060–0.125 in (1.5–3 mm) per surface on small and medium parts, and more on large castings or faces far from the parting line. Too little stock is the most common cause of scrap, since a casting that shifts within tolerance can leave an unmachined low spot.

### Why do sand castings need draft?

The pattern has to come out of the packed sand without tearing the mold wall. Green sand molding needs 1.5–3°; rigid chemically bonded no-bake sand can approach 1°. Draft is added to the pattern on top of the finished part geometry, then machined off where it matters.

### Sand casting or die casting?

Sand casting for large parts, low volumes, iron and steel, and anywhere the pattern still needs to change. Die casting for aluminum, zinc, and magnesium above roughly 10,000 parts a year, where its ±0.002 in/in tolerance and 32–63 µin Ra finish eliminate most machining.

### What metals can be sand cast?

Effectively any pourable metal: gray and ductile iron, aluminum-silicon alloys such as A356, bronzes and brasses, carbon and stainless steels, and nickel alloys. That alloy range is sand casting's main advantage over die casting, which is limited to low-melting-point metals.

## Alternative processes

- [Die Casting](https://manufacturingprocesses.org/processes/forming/die-casting.md): Die casting injects molten aluminum, zinc or magnesium into a hardened steel die under high pressure, producing thin-walled net-shape metal parts.
- [Investment Casting](https://manufacturingprocesses.org/processes/forming/investment-casting.md): Investment casting builds a ceramic shell around a wax pattern, melts the wax out and pours metal into the cavity, giving fine detail in almost any alloy.
- [CNC Machining](https://manufacturingprocesses.org/processes/cutting/cnc-machining.md): CNC machining removes material from solid stock with a programmed cutting tool, producing accurate parts directly from a CAD model with no tooling.
- [Forging](https://manufacturingprocesses.org/processes/forming/forging.md): Forging deforms solid metal under compressive load in a hammer, press or dies, so the grain flow follows the part contour and fatigue strength rises.

## Related processes

- [Investment Casting](https://manufacturingprocesses.org/processes/forming/investment-casting.md): Investment casting builds a ceramic shell around a wax pattern, melts the wax out and pours metal into the cavity, giving fine detail in almost any alloy.
- [Die Casting](https://manufacturingprocesses.org/processes/forming/die-casting.md): Die casting injects molten aluminum, zinc or magnesium into a hardened steel die under high pressure, producing thin-walled net-shape metal parts.
- [Centrifugal Casting](https://manufacturingprocesses.org/processes/forming/centrifugal-casting.md): Centrifugal casting pours molten metal into a spinning mold so centrifugal force packs it against the wall and drives inclusions toward the bore.
- [Heat Treatment](https://manufacturingprocesses.org/processes/forming/heat-treatment.md): Heat treatment cycles metal through controlled heating and cooling to change its hardness, strength, ductility or internal stress state.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/forming/sand-casting)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Selective Laser Melting (SLM)"
category: "Forming"
subcategory: "Additive Manufacturing"
materials: ["Metal"]
tolerances: "About ±0.004 in (±0.1 mm) up to 1 in (25 mm), then roughly ±0.2% of nominal. Machined features hold normal machining tolerances."
volumes: "1–500 parts"
lead_time: "5–15 business days including stress relief, cut-off, support removal, and any machining or hot isostatic pressing."
url: https://manufacturingprocesses.org/processes/forming/selective-laser-melting-slm
---

# Selective Laser Melting (SLM)

Selective laser melting fully melts metal powder with a laser under inert gas, producing parts at close to wrought density.

- **Category**: [Forming](https://manufacturingprocesses.org/processes/forming.md)
- **Family**: Additive Manufacturing
- **Materials**: Metal
- **Typical tolerances**: About ±0.004 in (±0.1 mm) up to 1 in (25 mm), then roughly ±0.2% of nominal. Machined features hold normal machining tolerances.
- **Surface finish**: Ra 200–500 µin (5–13 µm) as built on vertical walls, rising to Ra 800–1,200 µin (20–30 µm) on supported down-facing surfaces. Machining or polishing is required wherever fatigue performance matters, since as-built roughness is the dominant crack initiator.
- **Typical volumes**: 1–500 parts
- **Lead time**: 5–15 business days including stress relief, cut-off, support removal, and any machining or hot isostatic pressing.

## Overview

Selective laser melting (SLM) fully melts single-alloy metal powder with a high-power fiber laser under an inert atmosphere, producing parts at 99.5% density or better — close to wrought properties once heat treated. Layers are 0.0008–0.0024 in (20–60 µm) of 15–45 µm powder, and lasers range from 200 W on single-laser machines to multi-laser configurations of 1 kW and above for productivity.

Functionally, SLM and [DMLS](/processes/forming/direct-metal-laser-sintering-dmls) are the same process; the two names came from competing machine builders. ISO/ASTM classifies both as laser powder bed fusion (PBF-LB/M). The historic distinction — that sintering only partially fused a blended powder while melting fully liquefied a single alloy — no longer describes what commercial machines do.

The alloy set covers aluminum (AlSi10Mg, and increasingly high-strength grades), titanium, austenitic and precipitation-hardening stainless, nickel superalloys, cobalt chrome, tool steels, and pure copper on green- or blue-laser systems. Volumes of 1 to a few hundred parts, at accuracies near ±0.004 in (±0.1 mm).

## How it works

1. **Inert and preheat.** The chamber is purged to below roughly 0.1% oxygen with argon for reactive alloys or nitrogen for steels. Many machines preheat the substrate plate to 80–200 °C (175–390 °F) to reduce the thermal gradient and with it the residual stress.
2. **Recoat.** A 0.0008–0.0024 in (20–60 µm) layer of powder is spread; 0.0012 in (30 µm) is a common production setting, with 0.0024 in (60 µm) used where speed matters more than resolution.
3. **Melt.** The laser fully liquefies the powder and re-melts part of the layer beneath, so consecutive layers are metallurgically continuous rather than sintered together. Energy density — power divided by scan speed, hatch spacing, and layer thickness — is the master parameter: too little leaves lack-of-fusion porosity, too much causes keyholing and entrapped gas pores.
4. **Manage stress.** Scan strategies break each layer into stripes or islands and rotate the hatch angle (commonly 67°) between layers, which keeps any single direction from accumulating stress.
5. **Build to height.** Multi-laser machines divide the build area between two, four, or more lasers to raise the deposition rate, with overlap zones needing careful parameter matching.
6. **Heat treat on the plate,** then cut free by wire EDM, remove supports, and machine or HIP as the application requires. The sequence is identical to DMLS.

## Design guidelines

### Self-supporting geometry
45° from vertical is the working limit for unsupported downskins, and the single most valuable design move in SLM is to keep every surface above it. Chamfer instead of fillet on downward faces, and replace round horizontal holes with teardrops or diamonds.

### Wall thickness and aspect ratio
Minimum wall is about 0.016 in (0.4 mm); use 0.040 in (1 mm) or more for structural sections. Keep tall thin walls braced — an unbraced wall taller than roughly 50 times its thickness is likely to be knocked over by the recoater during the build.

### Uniform cross-section
Residual stress scales with the area of each melted layer and with abrupt section changes. Blending a bracket's transitions with generous fillets does more for build success than any parameter change.

### Powder removal
Every internal void needs an evacuation route. Provide ports of 0.08 in (2 mm) or larger, and remember that lattice structures and conformal channels hold powder tenaciously — specify how they will be cleaned before you design them in.

### Machining and inspection
Add 0.020–0.040 in (0.5–1.0 mm) of stock to fitted, sealing, and datum surfaces. Because internal porosity is not visible, safety-critical parts are typically CT scanned; design in wall thicknesses that CT can actually resolve.

### Copper and reflective alloys
Pure copper and high-conductivity alloys reflect most of a 1,070 nm fiber laser's energy. They need green (515 nm) or blue laser systems, and both machine availability and parameter maturity are more limited than for steel or titanium.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Overhang from vertical | 40° | 45° | Downskin quality collapses beyond this |
| Wall thickness | 0.040 in (1.0 mm) | 0.016 in (0.4 mm) | Recoater loads and distortion |
| Wall aspect ratio | 30:1 | 50:1 | Recoater strikes unbraced walls |
| Horizontal hole | Teardrop or diamond | 0.3 in (8 mm) round | Crown sags without support |
| Powder escape port | 0.16 in (4 mm) | 0.08 in (2 mm) | Powder must have an exit |
| Machining stock | 0.040 in (1.0 mm) | 0.020 in (0.5 mm) | Post-cut-off distortion |

## Cost drivers

The cost structure is identical to DMLS: build height sets machine hours, melted volume sets powder consumption, and post-processing — stress relief, plate cut-off, support removal, machining, and HIP where required — often exceeds the print cost. Multi-laser machines improve the deposition rate but do not change the fact that Z height governs.

Powder cost varies by alloy over roughly an order of magnitude between stainless and titanium or nickel superalloy, and requalification of reused powder (sieving, oxygen and morphology checks) is a real overhead in regulated work.

Volume breakpoints: 1–50 parts for complex geometry is where SLM is unambiguously the right answer. Simple shapes at any quantity belong in [CNC machining](/processes/cutting/cnc-machining); a few hundred or more of a complex shape usually justifies [investment casting](/processes/forming/investment-casting) tooling.

1. Orient for minimum height, then for minimum support area.
2. Design self-supporting angles and hole profiles — every avoided support is avoided labor.
3. Hollow, rib, or lattice bulk volume.
4. Fill the plate; heat treatment and cut-off costs are shared across everything on it.
5. Call out machined features individually rather than applying a tight blanket tolerance to the whole part.

## FAQ

### What is the difference between SLM and DMLS?

In current practice there is none worth designing around. Both are laser powder bed fusion of metal, both fully melt the powder, and ISO/ASTM covers them under the single term PBF-LB/M. The names originated with competing machine builders in the 1990s.

### What density do SLM parts reach?

Above 99.5% of theoretical with developed parameters, and above 99.9% is routine on qualified alloys. Remaining porosity comes from lack of fusion at low energy density or keyhole collapse at high energy density; hot isostatic pressing closes both.

### Why do SLM machines rotate the scan direction between layers?

Each melt track contracts as it solidifies, pulling along its own axis. Rotating the hatch pattern — 67° per layer is a common choice — prevents that contraction from accumulating in one direction and reduces both residual stress and anisotropy.

### Can SLM print copper?

Yes, but not on a standard machine. Pure copper reflects most of the 1,070 nm infrared light a conventional fiber laser produces. Green (515 nm) or blue diode systems couple far better and are what commercial copper SLM uses; availability and qualified parameter sets are still limited compared with steel or titanium.

### How thin can an SLM wall be?

About 0.016 in (0.4 mm) is achievable, but 0.040 in (1 mm) is the sensible structural minimum. Also watch aspect ratio: an unbraced wall taller than roughly 30–50 times its thickness is liable to be struck and deflected by the recoater blade.

### Do SLM parts need heat treatment?

Yes. Stress relief while the part is still on the build plate is mandatory for anything but the simplest geometry, and most alloys then need their conventional solution and aging cycle to reach datasheet properties. Fatigue-critical parts add hot isostatic pressing.

## Alternative processes

- [Direct Metal Laser Sintering (DMLS)](https://manufacturingprocesses.org/processes/forming/direct-metal-laser-sintering-dmls.md): Direct metal laser sintering fuses metal powder layer by layer with a laser, building dense metal parts anchored to a supported build plate.
- [Investment Casting](https://manufacturingprocesses.org/processes/forming/investment-casting.md): Investment casting builds a ceramic shell around a wax pattern, melts the wax out and pours metal into the cavity, giving fine detail in almost any alloy.
- [CNC Machining](https://manufacturingprocesses.org/processes/cutting/cnc-machining.md): CNC machining removes material from solid stock with a programmed cutting tool, producing accurate parts directly from a CAD model with no tooling.
- [Metal Injection Molding](https://manufacturingprocesses.org/processes/forming/metal-injection-molding.md): Metal injection molding molds a powder-and-binder feedstock exactly like a plastic, then debinds and sinters it into a dense small metal part.
- [Electron Beam Melting (EBM)](https://manufacturingprocesses.org/processes/forming/electron-beam-melting-ebm.md): Electron beam melting fuses metal powder with an electron beam in vacuum at high preheat temperature, which cuts residual stress in titanium parts.

## Related processes

- [Direct Metal Laser Sintering (DMLS)](https://manufacturingprocesses.org/processes/forming/direct-metal-laser-sintering-dmls.md): Direct metal laser sintering fuses metal powder layer by layer with a laser, building dense metal parts anchored to a supported build plate.
- [Electron Beam Melting (EBM)](https://manufacturingprocesses.org/processes/forming/electron-beam-melting-ebm.md): Electron beam melting fuses metal powder with an electron beam in vacuum at high preheat temperature, which cuts residual stress in titanium parts.
- [Binder Jetting](https://manufacturingprocesses.org/processes/forming/binder-jetting.md): Binder jetting prints liquid binder into a powder bed to hold each layer, producing green parts that are later sintered or infiltrated.
- [Directed Energy Deposition (DED)](https://manufacturingprocesses.org/processes/forming/directed-energy-deposition-ded.md): Directed energy deposition feeds metal wire or powder into a moving melt pool, building up large features or repairing existing parts.
- [Rapid Prototyping](https://manufacturingprocesses.org/processes/forming/rapid-prototyping.md): Rapid prototyping builds a part layer by layer directly from a CAD model, with no tooling and no minimum order quantity.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/forming/selective-laser-melting-slm)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Selective Laser Sintering (SLS)"
category: "Forming"
subcategory: "Additive Manufacturing"
materials: ["Plastic"]
tolerances: "±0.012 in (±0.3 mm) up to about 4 in (100 mm), then ±0.3% of nominal. Critical bores are normally reamed after the build."
volumes: "1–5,000 parts; cost per part is nearly flat with quantity"
lead_time: "3–7 business days. A full-height build is 20–40 hours of print time plus 8–24 hours of controlled cooling before parts can be removed."
url: https://manufacturingprocesses.org/processes/forming/selective-laser-sintering-sls
---

# Selective Laser Sintering (SLS)

Selective laser sintering fuses nylon powder layer by layer with a laser, so the surrounding loose powder supports the part and no supports are needed.

- **Category**: [Forming](https://manufacturingprocesses.org/processes/forming.md)
- **Family**: Additive Manufacturing
- **Materials**: Plastic
- **Typical tolerances**: ±0.012 in (±0.3 mm) up to about 4 in (100 mm), then ±0.3% of nominal. Critical bores are normally reamed after the build.
- **Surface finish**: Ra 200–500 µin (5–13 µm) as built, a uniform matte with slight grain. Chemical vapor smoothing takes it to roughly Ra 40–120 µin (1–3 µm) and seals surface porosity.
- **Typical volumes**: 1–5,000 parts; cost per part is nearly flat with quantity
- **Lead time**: 3–7 business days. A full-height build is 20–40 hours of print time plus 8–24 hours of controlled cooling before parts can be removed.

## Overview

Selective laser sintering (SLS) fuses nylon powder layer by layer with a CO2 laser inside a chamber held just below the polymer's melting point — around 340 °F (170 °C) for PA12. Because the surrounding un-sintered powder cake supports every layer, SLS needs no support structures, which frees the designer to build undercuts, internal channels, captive assemblies, and lattices that no molded or machined part could carry.

The workhorse material is PA12 (nylon 12), with PA11, glass- and mineral-filled nylons, carbon-filled grades, and TPU also in routine production. Parts reach roughly 95–100% of bulk density, and PA12 comes out near 7,000 psi (48 MPa) tensile with 10–20% elongation — functional properties, not merely representative ones.

Typical accuracy is ±0.012 in (±0.3 mm) or ±0.3% of nominal, whichever is greater, at a 0.004 in (100 µm) layer. Surfaces are uniformly matte and slightly grainy. SLS is economical from one part into the low thousands, especially where small parts nest densely in three dimensions.

## How it works

1. **Powder dosing and preheat.** Fresh and reclaimed PA12 are blended — commonly 30–50% virgin — and the chamber is brought within a few degrees of the polymer's melting range. PA12 melts at roughly 350–365 °F (178–185 °C) and the bed sits near 340 °F (170 °C). That narrow window is the whole trick: the laser only supplies the last few degrees.
2. **Recoat.** A roller or blade spreads a 0.003–0.006 in (75–150 µm) powder layer, typically 0.004 in (100 µm), across the build area.
3. **Scan.** A 30–100 W CO2 laser traces the cross-section, sintering particles to each other and to the layer below. Contour and fill passes run at different powers to control edge accuracy.
4. **Index and repeat.** The piston drops one layer and the cycle repeats. Because the cake supports everything, parts are nested in all three axes rather than sitting on a platform, and packing density becomes the main lever on cost.
5. **Controlled cooldown.** The finished cake cools in the machine or a take-out frame for 8–24 hours. This is not idle time — cooling too fast produces differential shrinkage, curl, and warp, worst on large flat parts.
6. **Breakout and cleaning.** The cake is broken open, parts extracted, and residual powder removed by brushing and bead blasting. Internal channels are cleared through escape holes with compressed air.
7. **Finishing (optional).** Dyeing (usually black), vibratory tumbling, or chemical vapor smoothing follow. Vapor smoothing seals surface porosity, which matters for cleanability in medical and food-contact applications.

## Design guidelines

### Wall thickness
0.030 in (0.8 mm) is the practical minimum; use 0.040 in (1.0 mm) for anything load-bearing or larger than a few inches. Thin unsupported walls survive the build but distort during cooldown.

### Escape holes
Any enclosed volume traps un-sintered powder. Provide at least two escape holes of 0.20 in (5 mm) diameter on opposite faces so powder can be blown through rather than shaken out. Long internal channels need intermediate ports.

### Bulk sections
Avoid solid sections thicker than about 0.8 in (20 mm). Heat accumulates in the cake, the region keeps sintering after the laser has moved on, and the part grows and warps. Core out thick regions or replace them with a ribbed shell or lattice.

### Clearances for moving and captive parts
SLS prints assemblies in place. Allow 0.020 in (0.5 mm) clearance on small features and 0.040 in (1.0 mm) on larger ones — the heat-affected zone around each scan grows the surface slightly, and a tight gap sinters solid.

### Holes and detail
Keep holes at 0.060 in (1.5 mm) or larger; smaller ones close up with partially sintered powder. Embossed and engraved detail needs 0.030 in (0.8 mm) of width and depth, and text wants 0.12 in (3 mm) cap height to survive bead blasting.

### Living hinges and snap fits
PA12 takes a snap fit well but makes a poor as-built living hinge; the sintered structure carries more void content than molded nylon and fatigues faster.

### Threaded features
Below M6, print a clearance hole and use a heat-set or press-in insert rather than a printed thread. If you intend to tap the nylon directly, the [tap drill chart](/charts/tap-drill-chart) gives the pilot sizes.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Wall thickness | 0.040 in (1.0 mm) | 0.030 in (0.8 mm) | Thin walls distort during cooldown |
| Escape hole | 0.30 in (8 mm) | 0.20 in (5 mm) | Powder must be blown out, not shaken |
| Hole diameter | 0.060 in (1.5 mm) | 0.040 in (1.0 mm) | Smaller holes sinter partly closed |
| Clearance, moving | 0.040 in (1.0 mm) | 0.020 in (0.5 mm) | Heat-affected zone grows the surface |
| Embossed detail | 0.040 in (1.0 mm) | 0.030 in (0.8 mm) | Bead blasting erodes finer features |
| Solid section | 0.6 in (15 mm) | 0.8 in (20 mm) | Heat soak causes growth and warp |

## Cost drivers

SLS is priced on the volume of build envelope a part consumes, not on setup — there is no tooling and no meaningful per-order fixed cost. Two things follow. First, packing density matters enormously: a bureau nesting parts at 8–12% of chamber volume amortizes the same powder and machine hours across many more parts than one running a single component. Second, unit cost is nearly flat with quantity, the opposite of a molded process.

Powder is the other real cost. Only part of the un-sintered cake can be reused, because powder near sintered regions degrades thermally; refresh ratios of 30–50% virgin are standard, so you effectively pay for powder that never became a part.

Volume breakpoints: SLS is competitive from 1 to roughly 5,000 parts a year. Beyond that — particularly where walls are thin and uniform — [injection molding](/processes/forming/injection-molding) wins on unit cost once tooling amortizes.

1. Reduce part volume rather than bounding box: hollow, rib, or lattice bulk sections.
2. Batch orders so parts nest into a single build.
3. Consolidate assemblies into one printed part; SLS removes fasteners and assembly labor at no geometric cost.
4. Skip dyeing and vapor smoothing on hidden components.
5. Keep the tallest dimension modest — build height drives machine hours more directly than part count does.

## FAQ

### Does SLS need support structures?

No. The un-sintered powder cake surrounds and supports every layer, so overhangs, undercuts, and internal geometry print without supports. That is why SLS parts can be nested in all three axes and why complex assemblies can be printed as a single piece.

### What tolerance can SLS hold?

±0.012 in (±0.3 mm) up to about 4 in (100 mm), then ±0.3% of the dimension. Accuracy depends on where in the build a part sits and how it was oriented, so critical bores are usually reamed afterward.

### How big do SLS escape holes need to be?

At least 0.20 in (5 mm) diameter, and you want two on opposite faces of any enclosed cavity so compressed air can drive the powder through. A single hole traps powder no matter how long the part is shaken.

### How strong are SLS nylon parts?

PA12 comes out near 7,000 psi (48 MPa) tensile with 10–20% elongation at 95–100% of bulk density. Z-axis strength is lower than XY, but the gap is typically 10–25% rather than the much larger anisotropy seen in FDM.

### Why are SLS parts usually gray or dyed black?

Raw sintered PA12 is off-white to light gray with a porous matte surface that picks up handling marks. Dyeing black is inexpensive, penetrates a few tenths of a millimeter, and hides both the porosity and the powder texture.

### Can SLS parts be made watertight?

Not reliably as built, because the sintered structure retains interconnected porosity. Chemical vapor smoothing or an epoxy or urethane sealer closes the surface for low-pressure use; for genuine pressure containment, machine or mold the part instead.

### How does SLS compare with MJF?

Both fuse PA12 powder without supports. MJF applies fusing agent by inkjet and fuses whole layers with infrared lamps, giving higher throughput, slightly tighter tolerance (±0.008 in / ±0.2 mm), and denser, more isotropic parts. SLS offers the wider material menu, including PA11, TPU, and filled grades.

## Alternative processes

- [Multi Jet Fusion (MJF)](https://manufacturingprocesses.org/processes/forming/multi-jet-fusion-mjf.md): Multi Jet Fusion prints a fusing agent onto nylon powder and drives it with infrared energy, fusing whole layers at once for high throughput.
- [Injection Molding](https://manufacturingprocesses.org/processes/forming/injection-molding.md): Injection molding forces molten thermoplastic into a steel mold under high pressure, producing high-volume identical plastic parts.
- [Fused Deposition Modeling (FDM)](https://manufacturingprocesses.org/processes/forming/fused-deposition-modeling-fdm.md): Fused deposition modeling extrudes molten thermoplastic filament along programmed paths so each layer welds to the one beneath it.
- [CNC Machining](https://manufacturingprocesses.org/processes/cutting/cnc-machining.md): CNC machining removes material from solid stock with a programmed cutting tool, producing accurate parts directly from a CAD model with no tooling.
- [Vacuum Casting (Urethane Casting)](https://manufacturingprocesses.org/processes/forming/vacuum-casting.md): Vacuum casting, known in the US as urethane casting, pours polyurethane resin into a silicone mold under vacuum to copy a master pattern for short runs.

## Related processes

- [Multi Jet Fusion (MJF)](https://manufacturingprocesses.org/processes/forming/multi-jet-fusion-mjf.md): Multi Jet Fusion prints a fusing agent onto nylon powder and drives it with infrared energy, fusing whole layers at once for high throughput.
- [Rapid Prototyping](https://manufacturingprocesses.org/processes/forming/rapid-prototyping.md): Rapid prototyping builds a part layer by layer directly from a CAD model, with no tooling and no minimum order quantity.
- [Stereolithography (SLA)](https://manufacturingprocesses.org/processes/forming/stereolithography-sla.md): Stereolithography cures liquid photopolymer layer by layer with a scanning UV laser, giving the smoothest surfaces in additive manufacturing.
- [Direct Metal Laser Sintering (DMLS)](https://manufacturingprocesses.org/processes/forming/direct-metal-laser-sintering-dmls.md): Direct metal laser sintering fuses metal powder layer by layer with a laser, building dense metal parts anchored to a supported build plate.
- [Fused Deposition Modeling (FDM)](https://manufacturingprocesses.org/processes/forming/fused-deposition-modeling-fdm.md): Fused deposition modeling extrudes molten thermoplastic filament along programmed paths so each layer welds to the one beneath it.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/forming/selective-laser-sintering-sls)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Sheet Metal Fabrication"
category: "Forming"
subcategory: "Metal"
materials: ["Metal"]
tolerances: "±0.005 in (±0.13 mm) on laser-cut features; about ±0.010 in (±0.25 mm) per bend, accumulating across bends; welded assemblies are substantially looser unless fixtured or machined after welding"
volumes: "1–10,000 parts per year; above about 10,000 stamping or roll forming usually wins"
lead_time: "3–10 business days typical, 1–3 days for quick-turn simple parts; no tooling lead time"
url: https://manufacturingprocesses.org/processes/forming/sheet-metal-fabrication
---

# Sheet Metal Fabrication

Sheet metal fabrication cuts, bends and joins flat stock into a finished assembly, typically laser cutting followed by press braking, hardware and welding.

- **Category**: [Forming](https://manufacturingprocesses.org/processes/forming.md)
- **Family**: Metal
- **Materials**: Metal
- **Typical tolerances**: ±0.005 in (±0.13 mm) on laser-cut features; about ±0.010 in (±0.25 mm) per bend, accumulating across bends; welded assemblies are substantially looser unless fixtured or machined after welding
- **Surface finish**: Inherits the incoming sheet finish; commonly grained, then powder coated, anodized, or plated
- **Typical volumes**: 1–10,000 parts per year; above about 10,000 stamping or roll forming usually wins
- **Lead time**: 3–10 business days typical, 1–3 days for quick-turn simple parts; no tooling lead time

## Overview

Sheet metal fabrication is the job-shop workflow that turns flat stock into a finished assembly: cut the flat pattern, bend it, install hardware, weld what needs joining, and finish it. No part-specific tooling is involved at any step, so a design can go from a CAD model to parts in hand in days, and change again the following week.

It is the default for enclosures, chassis, brackets, panels, racks, and machine guarding, in 0.024–0.250 in (0.6–6 mm) cold-rolled steel, galvanized, 304 and 316 stainless, and 5052 and 6061 aluminum. Volumes run from one prototype to about 10,000 a year, above which stamping's hard tooling starts to win.

The cost structure is entirely setup and labor: cut time, bend count, weld inches, and hardware count. Every one of those is a design variable, which is why sheet metal parts respond so strongly to design-for-manufacture work.

## How it works

1. **Flat pattern.** The 3D model is unfolded into a flat blank. This is not just geometry — bend allowance depends on material, thickness, inside radius, and the shop's tooling, expressed as a K-factor or a bend deduction table. Send the 3D model, not someone else's flat pattern; the fabricator's numbers differ from yours.
2. **Cutting.** Fiber laser cutting dominates for its speed and edge quality; CNC turret punching is competitive for parts with many repeated holes and can add forms such as louvers, extrusions, and countersinks in the same operation. Waterjet handles thick plate and heat-sensitive materials.
3. **Deburring.** Sheared and punched edges carry a burr, laser edges carry dross. Deburring is a separate operation, whether by timesaver machine, tumbling, or hand.
4. **Forming.** Press braking makes one bend per stroke against a CNC backgauge. Bend sequence matters: a part must never be trapped by its own geometry before the last bend is reached.
5. **Hardware.** Self-clinching nuts, studs, and standoffs are pressed into the flat or partially formed part, cold-flowing the sheet into a groove on the fastener. Each type has a minimum sheet thickness and a minimum edge distance.
6. **Joining.** MIG, TIG, and spot welding, plus riveting and adhesive bonding. Welding sheet metal distorts it — heat input has to be controlled and sequenced, and weld shrinkage moves the assembly out of tolerance if it is not planned for.
7. **Finishing.** Grinding welds, graining, then powder coating, anodizing, plating, or passivation.

Tolerance stacks up along that chain. Cut features hold to about ±0.005 in, a single bend to roughly ±0.010 in, and a multi-bend welded assembly to something much looser — which is why critical dimensions belong on cut features rather than across welded joints wherever possible.

## Design guidelines

### One thickness, one bend radius

Design the whole assembly in one material and gauge, and use the same inside bend radius everywhere. Every change of thickness is a different setup and a different flat pattern rule; every different radius is another tool change on the brake. Pull working thicknesses from the [sheet metal gauge chart](/charts/sheet-metal-gauge-chart).

### Bend radius

Inside radius of at least 1× material thickness for mild steel and annealed aluminum; 6061-T6 typically needs 2–4×, and more when bending parallel to the grain. Flat pattern development from the [bend radius and K-factor chart](/charts/sheet-metal-bend-radius-k-factor).

### Hole size, spacing, and distance from bends

Punched holes should be at least 1× material thickness in diameter (1.5–2× in stainless), at least 2× thickness from an edge, and at least 2.5× thickness plus the bend radius from a bend line. Laser cutting relaxes the diameter rule but not the distortion rule near bends.

### Minimum flange length

About 4× material thickness. Shorter flanges cannot sit on both shoulders of the V-die and need special tooling.

### Bend relief and corner relief

Provide a relief notch at the end of any bend that terminates at an edge — at least one material thickness wide and deeper than the bend radius plus thickness. At the intersection of two bends, add a corner relief so the material does not pile up and tear.

### Hardware

Self-clinching fasteners need a minimum sheet thickness (check the specific part number), a minimum distance to the nearest edge, and clearance on the back side for the installation anvil. Installing hardware after forming is often impossible because the press cannot reach — plan installation before the bends that would block it.

### Weld design

Specify weld length, not continuous beads, wherever strength allows: stitch welds put in a fraction of the heat and cause a fraction of the distortion. Keep welds away from tightly toleranced features, and give the fabricator a datum scheme that does not run across a welded joint.

### Tolerances

Do not stack tolerances across bends. Dimension from a single datum edge, and put the tight dimensions on laser-cut features in the flat.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Inside bend radius | 1× thickness | 2–4× for 6061-T6 | Outer fiber cracks at tighter radii in hard tempers |
| Minimum flange | 4× thickness | Half the V-die opening plus thickness | Flange must span both die shoulders |
| Hole diameter | ≥ 1× thickness | 1.5–2× in stainless | Punch strength; relaxed for laser cutting |
| Hole to edge | ≥ 2× thickness | 1.5× thickness | Thin webs bulge and tear |
| Hole to bend line | ≥ 2.5× thickness + radius | — | Metal stretching around the bend ovals the hole |
| Bend relief | Width ≥ thickness, depth > radius + thickness | — | Prevents tearing at the end of the bend |
| Cut feature tolerance | ±0.005 in (±0.13 mm) | — | Laser and turret accuracy |
| Bend dimension tolerance | ±0.010 in (±0.25 mm) per bend | Accumulates across bends | Springback and gauging variation stack up |

## Cost drivers

With no hard tooling, cost is machine time plus labor plus material. Cutting time scales with cut length, not part area, so a part covered in small holes costs more than a solid one of the same size. Bend count is the largest single driver on formed parts, because each bend is a separate handling cycle. Welding is more expensive still — weld inches, plus grinding, plus the distortion correction that follows. Hardware is priced per piece installed. Material utilization matters at volume: parts that nest tightly on a sheet cost meaningfully less per piece.

Volume breakpoints: 1–10 pieces is prototype territory where setup dominates completely and the per-part price barely falls with quantity. From 100 to 1,000 the setup amortizes and the price drops sharply. Above roughly 10,000 a year, stamping or roll forming begins to win on cycle time despite the tooling investment.

1. **Cut the bend count.** Every bend removed is handling time on every part, forever.
2. **Replace welds with tabs and slots or hardware.** Self-locating tabs also improve assembly accuracy by removing fixture dependence.
3. **Nest-friendly outlines.** Rectangular-ish parts nest far better than parts with sprawling arms; nesting efficiency shows up directly in the material line.
4. **Standardize gauge and hardware across the assembly.** One thickness and one fastener type across five parts collapses setups and purchasing.
5. **Tolerance realistically.** ±0.005 in on cut features is free; ±0.005 in across a welded joint requires fixturing, straightening, or machining after welding.

## FAQ

### What tolerance can sheet metal fabrication hold?

About ±0.005 in (±0.13 mm) on laser-cut features and roughly ±0.010 in (±0.25 mm) per bend, with bend tolerances accumulating across a multi-bend part. Welded assemblies are looser still. Put critical dimensions on cut features in the flat rather than across bends or welds.

### What is the minimum bend radius for sheet metal?

At least 1× material thickness for cold-rolled steel and annealed aluminum. 6061-T6 typically needs 2–4× thickness, and more when the bend runs parallel to the rolling direction. Using one radius throughout the assembly avoids tool changes on the press brake.

### How close can a hole be to a bend?

At least 2.5× material thickness plus the inside bend radius, measured from the hole edge to the bend line. Closer holes distort into ovals as the metal stretches around the bend. If a hole must be closer, either machine it after forming or extend it into a slot that crosses the bend entirely.

### Why do I need to send a 3D model instead of a flat pattern?

Bend allowance depends on the fabricator's specific tooling, material, and measured K-factor. A flat pattern developed with different assumptions will produce a part whose flange lengths are wrong by a few hundredths of an inch per bend, and the error accumulates.

### When should I switch from fabrication to stamping?

Around 10,000 parts a year for a simple part, sooner if it has many bends or holes. Stamping trades a multi-month tooling investment for a cycle time measured in fractions of a second; fabrication trades zero tooling for minutes of labor per part.

### How do I avoid distortion when welding sheet metal?

Use stitch welds instead of continuous beads wherever the strength allows, alternate weld locations to balance heat input, keep welds away from tightly toleranced features, and avoid running the datum scheme across a welded joint. Weld shrinkage is predictable in direction but hard to hold to tight numbers.

## Alternative processes

- [Metal Stamping](https://manufacturingprocesses.org/processes/forming/metal-stamping.md): Metal stamping shears and forms sheet metal between matched dies in a press, producing flat and shallow-formed parts at very high rates.
- [CNC Machining](https://manufacturingprocesses.org/processes/cutting/cnc-machining.md): CNC machining removes material from solid stock with a programmed cutting tool, producing accurate parts directly from a CAD model with no tooling.
- [Metal Extrusion](https://manufacturingprocesses.org/processes/forming/metal-extrusion.md): Metal extrusion pushes heated billet through a shaped die, producing continuous aluminum or copper profile of constant cross-section.
- [Roll Forming](https://manufacturingprocesses.org/processes/forming/roll-forming.md): Roll forming passes coil stock through a sequence of contoured roller stands that bend it progressively into a constant cross-section profile.

## Related processes

- [Press Braking](https://manufacturingprocesses.org/processes/forming/press-braking.md): Press braking bends sheet metal by driving a punch into a V-die, forming one straight-line bend at a time.
- [Laser Cutting](https://manufacturingprocesses.org/processes/cutting/laser-cutting.md): Laser cutting melts, burns or vaporises a narrow kerf through flat stock with a focused beam, producing profiles from a 2D file with no tooling.
- [Punching and Blanking](https://manufacturingprocesses.org/processes/cutting/punching-and-blanking.md): Punching and blanking shear sheet metal between a punch and a die, producing holes and flat outlines at very high speed.
- [Arc Welding](https://manufacturingprocesses.org/processes/joining/arc-welding.md): Arc welding melts and fuses metal with an electric arc between an electrode and the work, shielded from the atmosphere by gas or flux.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/forming/sheet-metal-fabrication)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Steam Bending"
category: "Forming"
subcategory: "Wood"
materials: ["Wood"]
volumes: "1–5,000 pieces per year"
lead_time: "Days per batch: roughly one hour of steaming per inch (25 mm) of thickness, a minute to bend, then 1–7 days held on the former while the piece dries and sets."
url: https://manufacturingprocesses.org/processes/forming/steam-bending
---

# Steam Bending

Steam bending plasticises solid timber with saturated steam so it can be bent around a former and held until it dries into shape.

- **Category**: [Forming](https://manufacturingprocesses.org/processes/forming.md)
- **Family**: Wood
- **Materials**: Wood
- **Typical volumes**: 1–5,000 pieces per year
- **Lead time**: Days per batch: roughly one hour of steaming per inch (25 mm) of thickness, a minute to bend, then 1–7 days held on the former while the piece dries and sets.

## Overview

Steam bending plasticizes solid timber with saturated steam at atmospheric pressure — around 212 °F (100 °C) — so that it can be bent around a former and held until it dries into the new shape. The working rule is roughly one hour of steaming per inch (25 mm) of thickness, and the wood must be bent within about a minute of leaving the box, before it cools out of its plastic range.

The physics that makes it work is asymmetric: wet, hot wood will compress by a large fraction of its length but will only stretch about 1–2% before the fibers tear. Practical bending therefore uses a steel compression strap with fixed end stops, which forces the whole bend into compression and keeps the outer fibers from ever going into tension. With a strap, good species reach radius-to-thickness ratios around 2.5:1 to 3:1; without one, 10:1 to 15:1 is the realistic limit.

Species matters more than technique. White oak, ash, beech, elm, and hickory bend well; most softwoods and many tropical hardwoods do not bend usefully at any ratio.

## How it works

1. **Select and prepare the stock.** Straight grain running the full length is essential — grain runout across the bend is a guaranteed failure point, as is any knot. Air-dried or green stock at roughly 20–30% moisture content bends far better than kiln-dried timber below 12%, and material dried below about 12% often will not bend at all without re-conditioning.
2. **Steam.** The wood is held in a steam box at atmospheric pressure and near 212 °F (100 °C) for approximately one hour per inch (25 mm) of thickness. Steaming does not add much moisture; its job is to raise the wood's temperature so the lignin and hemicellulose soften.
3. **Set up the former.** Everything must be ready before the wood comes out — former, clamps or strap, and end stops. There is no time to improvise.
4. **Bend immediately.** Working time is on the order of 30–90 seconds. The bend is made in one continuous movement against the compression strap, with the strap's end stops bearing hard against the ends of the workpiece so the wood is driven into compression rather than allowed to stretch.
5. **Clamp and hold.** The piece stays on the former under restraint while it cools and dries.
6. **Dry to set the shape.** The bend does not become permanent until the wood dries back to around 12% moisture content or below — typically 1–7 days at ambient, faster in a drying kiln. Removing the piece early gives most of the bend back.
7. **Release and trim.** Some springback always remains, typically a few percent of the bend angle, so formers are made tighter than the target radius and pieces are trimmed to fit after release.

## Design guidelines

### Choose the species first
Bending quality is a species property. White oak, ash, beech, elm, hickory, and to a lesser extent walnut and birch bend well. Spruce, pine, fir, and most tropical hardwoods including mahogany bend poorly regardless of preparation. No amount of steam makes a bad bending species behave like a good one.

### Radius-to-thickness ratio
With a compression strap and end stops, good species reach approximately 2.5:1 to 3:1 (radius to thickness). Without a strap — bending free — expect 10:1 to 15:1 before the outer fibers tear. If the design needs a tighter bend than the species and setup can deliver, either reduce the thickness or move to [wood laminating](/processes/forming/wood-laminating).

### Always use a compression strap for tight bends
Hot wet wood compresses readily but tears in tension after roughly 1–2% strain. The strap plus end stops moves the neutral axis to the outer face so that essentially the whole section works in compression. This single fixture is the difference between a 3:1 and a 12:1 bend.

### Grain and defects
Grain must run continuously along the length of the piece. Runout, knots, and included bark are failure initiators, and a piece that would be structurally fine as a straight member can fail at a bend.

### Allow for springback
Overbend the former by a few percent of the angle. Springback varies with species, thickness, radius, and drying regime, so the first piece off a new former is a calibration piece.

### Section shape
Rectangular sections bend predictably. Complex profiles and moldings are usually bent as rectangular blanks and shaped afterward, because a profiled section distorts unevenly and its thin regions fail first.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Species | White oak, ash, beech, elm | Softwoods bend poorly | Bending is a species property |
| Radius:thickness, with strap | 4:1 | 2.5:1 | Compression limit of hot wet wood |
| Radius:thickness, unsupported | 15:1 | 10:1 | Outer fibers tear beyond 1–2% strain |
| Moisture content | 20–30% | Below 12% will not bend | Lignin must soften with heat and water |
| Steaming time | 1 h per inch (25 mm) | — | Heat must reach the core |
| Working time after steam | Under 60 s | 90 s | Wood stiffens as it cools |
| Hold on former | Until 12% MC | Days | Shape is not set until dry |

## Variants

- Circle Bending
- Open Bending

## Cost drivers

Steam bending has very low tooling cost and high labor and loss cost, which is the opposite profile from most forming processes. A former is a shaped piece of plywood or hardwood plus a steel strap; it can be made in a day and costs almost nothing to modify.

Cycle time is dominated by two waits. Steaming takes an hour per inch of thickness, and drying on the former takes days. That means formers, not people or machines, are the throughput constraint — a shop making bent chair backs owns many identical formers so that pieces can sit on them while new ones are bent.

Loss rate is the real cost driver. Even in an experienced shop, a proportion of pieces fail in tension, wrinkle in compression, or spring back out of tolerance, and each failure has already absorbed selected stock and steaming time. Selecting straight-grained stock aggressively is cheaper than the failures it prevents.

Volume breakpoints: steam bending suits 1 to a few thousand pieces a year. Where the geometry allows and springback must be tightly controlled, [wood laminating](/processes/forming/wood-laminating) is more repeatable at moderate volumes, and molded plywood or [composite laminating](/processes/forming/composite-laminating) takes over at higher ones.

1. Select straight-grained, defect-free stock; it costs less than the scrap it avoids.
2. Build multiple identical formers — drying time, not bending time, sets throughput.
3. Use green or air-dried stock rather than kiln-dried wherever the supply allows.
4. Bend rectangular blanks and machine the profile afterward.
5. Design bends to the loosest radius the product tolerates; the failure rate rises sharply as the ratio tightens.

## FAQ

### How long should wood be steamed before bending?

Roughly one hour per inch (25 mm) of thickness in a steam box at atmospheric pressure and about 212 °F (100 °C). The purpose is to heat the wood through, not to wet it, so longer steaming does not help and can degrade the timber.

### Which woods bend best?

White oak, ash, beech, elm, and hickory are the classic bending species, with walnut and birch acceptable. Softwoods such as pine, spruce, and fir bend poorly, as do most tropical hardwoods including mahogany. Species choice determines the achievable radius more than any aspect of technique.

### How tight a radius can steam bending achieve?

With a steel compression strap and end stops, good species reach roughly 2.5:1 to 3:1 radius-to-thickness. Bending free without a strap, expect 10:1 to 15:1. Anything tighter than the species allows should be laminated from thin plies instead.

### Why is a compression strap necessary?

Hot wet wood compresses readily but tears after only about 1–2% tensile strain. A steel strap on the outside of the bend, with end stops bearing against the ends of the workpiece, prevents the outer fibers from stretching and forces the entire section into compression, where wood is far more tolerant.

### Can kiln-dried wood be steam bent?

Poorly, and often not at all below about 12% moisture content. Kiln drying sets the lignin in a way that does not fully reverse with steaming. Green or air-dried stock at roughly 20–30% moisture content is the correct feedstock for bending work.

### How much springback should I allow?

A few percent of the bend angle, varying with species, thickness, radius, and how thoroughly the piece dries on the former. There is no universal figure, so the first piece off a new former is treated as a calibration piece and the former is adjusted accordingly.

### How long must a bend stay on the former?

Until the wood dries back to roughly 12% moisture content — typically 1–7 days at ambient conditions, faster in a drying kiln. The bend is not permanent until the wood is dry, and releasing early returns most of the curvature.

## Alternative processes

- [Wood Laminating](https://manufacturingprocesses.org/processes/forming/wood-laminating.md): Wood laminating glues thin veneers or kerfed strips over a form so the assembly holds a curve permanently once the adhesive has cured.
- [CNC Machining](https://manufacturingprocesses.org/processes/cutting/cnc-machining.md): CNC machining removes material from solid stock with a programmed cutting tool, producing accurate parts directly from a CAD model with no tooling.
- [Composite Laminating](https://manufacturingprocesses.org/processes/forming/composite-laminating.md): Composite laminating lays reinforcing fiber into a mold and consolidates it with resin, curing into a stiff, light monocoque part.
- [Joinery](https://manufacturingprocesses.org/processes/joining/joinery.md): Joinery cuts interlocking geometry into timber so the parts locate and hold each other, with or without adhesive or fasteners.
- [Tube and Section Bending](https://manufacturingprocesses.org/processes/forming/tube-and-section-bending.md): Tube and section bending draws or presses metal tube and profile around a former, with an internal mandrel wherever the wall would otherwise collapse.

## Related processes

- [Wood Laminating](https://manufacturingprocesses.org/processes/forming/wood-laminating.md): Wood laminating glues thin veneers or kerfed strips over a form so the assembly holds a curve permanently once the adhesive has cured.
- [Joinery](https://manufacturingprocesses.org/processes/joining/joinery.md): Joinery cuts interlocking geometry into timber so the parts locate and hold each other, with or without adhesive or fasteners.
- [Tube and Section Bending](https://manufacturingprocesses.org/processes/forming/tube-and-section-bending.md): Tube and section bending draws or presses metal tube and profile around a former, with an internal mandrel wherever the wall would otherwise collapse.
- [Composite Laminating](https://manufacturingprocesses.org/processes/forming/composite-laminating.md): Composite laminating lays reinforcing fiber into a mold and consolidates it with resin, curing into a stiff, light monocoque part.
- [Timber Frame Structures](https://manufacturingprocesses.org/processes/joining/timber-frame-structures.md): Timber frame structures assemble large solid or engineered timber members into a load-bearing frame using cut joints and steel connectors.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/forming/steam-bending)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Stereolithography (SLA)"
category: "Forming"
subcategory: "Additive Manufacturing"
materials: ["Plastic"]
tolerances: "±0.006 in (±0.15 mm) over the first inch, then about ±0.001 in/in (±0.1% of nominal); ±0.004 in (±0.1 mm) on small, well-supported features. Vendors quote between ±0.1% and ±0.2% of dimension."
volumes: "1–100 parts; a few hundred when the SLA part serves as a master for urethane casting"
lead_time: "1–3 business days for typical prototypes, same day on desktop machines. A 6 in (150 mm) tall build runs 8–20 hours plus 1–2 hours of washing and post-cure."
url: https://manufacturingprocesses.org/processes/forming/stereolithography-sla
---

# Stereolithography (SLA)

Stereolithography cures liquid photopolymer layer by layer with a scanning UV laser, giving the smoothest surfaces in additive manufacturing.

- **Category**: [Forming](https://manufacturingprocesses.org/processes/forming.md)
- **Family**: Additive Manufacturing
- **Materials**: Plastic
- **Typical tolerances**: ±0.006 in (±0.15 mm) over the first inch, then about ±0.001 in/in (±0.1% of nominal); ±0.004 in (±0.1 mm) on small, well-supported features. Vendors quote between ±0.1% and ±0.2% of dimension.
- **Surface finish**: Ra 20–100 µin (0.5–2.5 µm) as built on up-facing and vertical surfaces, the smoothest of the additive processes. Sanding and polishing reach below Ra 10 µin (0.25 µm). Down-facing surfaces are rougher where supports contact.
- **Typical volumes**: 1–100 parts; a few hundred when the SLA part serves as a master for urethane casting
- **Lead time**: 1–3 business days for typical prototypes, same day on desktop machines. A 6 in (150 mm) tall build runs 8–20 hours plus 1–2 hours of washing and post-cure.

## Overview

Stereolithography (SLA) builds parts by curing liquid photopolymer with a scanning UV laser, one 0.001–0.004 in (25–100 µm) layer at a time. It gives the smoothest as-built surface of any additive process — commonly Ra 20–100 µin (0.5–2.5 µm) — and holds roughly ±0.006 in (±0.15 mm) over the first inch, which is why it remains the default for appearance models, master patterns for urethane casting, fluidic prototypes, and any part where a visible layer line counts as a defect.

Feedstocks are acrylate and epoxy photopolymers, not true thermoplastics: standard rigid, tough (ABS-like), elastomeric, castable, biocompatible dental grades, and high-temperature resins with heat deflection near 460 °F (238 °C). Build envelopes run from about 5.7 × 5.7 × 7.3 in (145 × 145 × 185 mm) on desktop machines to roughly 30 × 30 × 22 in (750 × 750 × 550 mm) on large industrial frames.

SLA is economical from one part to a few hundred. Photopolymers yellow and embrittle under UV and creep under sustained load, so long-life structural parts usually belong in [SLS](/processes/forming/selective-laser-sintering-sls) or molded thermoplastic.

## How it works

1. **Slice and support.** The model is oriented, sliced at 0.001–0.004 in (25–100 µm) — 0.002 in (50 µm) is the common default — and a support scaffold is generated. Contact points are typically 0.4–0.6 mm across so they snap off cleanly.
2. **Recoat.** The platform steps by one layer thickness and a recoater blade (top-down machines) or the resin film itself (bottom-up machines) re-establishes a flat layer over the last cured cross-section.
3. **Expose.** Galvanometer mirrors sweep the laser across the cross-section. Industrial systems use a 355 nm solid-state laser with a 0.003–0.010 in (80–250 µm) spot; desktop machines use 405 nm diodes. Cure depth follows the Beer–Lambert relationship between exposure energy and the resin's penetration depth, so each pass deliberately over-cures into the layer below to weld the two together.
4. **Peel or dip.** Bottom-up machines separate the part from the vat film every layer, and this peel force is what limits large flat cross-sections. Typical vertical build rate is 0.2–0.8 in/h (5–20 mm/h).
5. **Drain and wash.** The build drains, then washes in isopropyl alcohol or TPM for 5–20 minutes to strip uncured resin. Over-washing swells and softens thin features.
6. **Remove supports.** Cut or snap supports while the part is still green and comparatively soft; it is far harder after post-cure.
7. **UV post-cure.** 15–60 minutes of UV at 60–80 °C (140–175 °F) completes crosslinking. This step is not optional — a green part can be 30–50% weaker and noticeably more compliant, and every published resin datasheet is quoted post-cured.

## Design guidelines

### Wall thickness
Supported walls print reliably at 0.020 in (0.5 mm). Give unsupported or tall walls 0.040 in (1 mm); thinner sections flex under recoating and peel forces and finish wavy.

### Hollowing and drainage
Hollow any section thicker than about 0.4 in (10 mm). Trapped resin adds weight, cures unevenly, and can crack the shell weeks later. Use a 0.080 in (2 mm) shell with at least two drain holes of 0.14 in (3.5 mm) or larger at opposite ends, so air can enter as resin leaves.

### Minimum features and text
Embossed and engraved detail needs 0.016 in (0.4 mm) of width and 0.016 in (0.4 mm) of height or depth. Text below about 0.08 in (2 mm) cap height fills in during washing.

### Holes and channels
Keep through-holes at or above 0.020 in (0.5 mm) — smaller ones cure shut. Internal channels below 0.040 in (1 mm) are difficult to clear of resin at any length, so plan a flush port.

### Orientation
Tilt the part 10–20° off the platform. That shrinks each layer's cross-section (lower peel force), moves supports off cosmetic faces, and avoids the stair-stepping that shows on shallow slopes. Up-facing and vertical walls come out smoothest; down-facing surfaces carry support witness marks.

### Fits and clearances
Allow 0.004–0.008 in (0.1–0.2 mm) for a slip fit and 0.012 in (0.3 mm) for a moving fit. Cured photopolymer has little yield plateau, so interference fits crack rather than deform — the [ISO 286 fits and tolerances chart](/charts/iso-286-fits-tolerances) shows how these compare with machined clearances.

### Threads
Model threads only at M4/#8 and above. Below that, print a pilot hole and tap it, or design for a heat-set insert; the [tap drill chart](/charts/tap-drill-chart) gives pilot sizes. Printed threads in brittle resin strip after a few cycles.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Supported wall | 0.040 in (1.0 mm) | 0.020 in (0.5 mm) | Thin walls flex during recoat and peel |
| Unsupported wall | 0.060 in (1.5 mm) | 0.040 in (1.0 mm) | No backing against separation forces |
| Hole diameter | 0.060 in (1.5 mm) | 0.020 in (0.5 mm) | Smaller holes cure shut during the wash |
| Embossed detail | 0.024 in (0.6 mm) | 0.016 in (0.4 mm) | Laser spot size and cure bleed set the floor |
| Drain hole | 0.20 in (5 mm) | 0.14 in (3.5 mm) | Resin viscosity limits drainage |
| Clearance, moving | 0.016 in (0.4 mm) | 0.008 in (0.2 mm) | Cure bleed closes tight gaps |
| Layer height | 0.002 in (50 µm) | 0.001 in (25 µm) | Build time doubles at half the layer |

## Cost drivers

SLA pricing is driven by build height first and resin volume second. Laser SLA time scales with number of layers times scanned area, so the same part standing 6 in (150 mm) tall costs far more than laid flat. Resin is the next lever: engineering, castable, and dental grades run several times the price of standard resin, and support structures are consumed at full resin cost.

Labor is the third. Support removal, sanding of witness marks, and painting or clear-coating are hand operations that often exceed machine cost on cosmetic parts.

Volume breakpoints: SLA is usually cheapest from 1 to roughly 50 parts. From about 50 to 500, printing one SLA master and casting the rest in urethane ([vacuum casting](/processes/forming/vacuum-casting)) is normally less expensive. Above 1,000–2,000 parts, tooling for [injection molding](/processes/forming/injection-molding) starts to pay back.

1. Lay the part down or tilt it; never build tall unless a cosmetic face demands it.
2. Hollow bulk sections and add drain holes — you are billed for resin volume.
3. Nest several parts into one build rather than ordering them separately.
4. Accept 0.004 in (100 µm) layers on non-cosmetic geometry; it roughly halves build time versus 0.002 in (50 µm).
5. Specify standard resin unless a datasheet property is genuinely required.

## FAQ

### What tolerance can SLA hold?

Plan on ±0.006 in (±0.15 mm) over the first inch and roughly ±0.001 in per additional inch. Small, well-supported features can come in at ±0.004 in (±0.1 mm). Accuracy degrades on tall, thin geometry because post-cure shrinkage accumulates over the Z height.

### What layer height should I use for SLA?

0.002 in (50 µm) is the standard compromise. Drop to 0.001 in (25 µm) only for fine detail or curved cosmetic surfaces, since it roughly doubles build time. 0.004 in (100 µm) is fine for bulky non-cosmetic geometry.

### How thin can an SLA wall be?

0.020 in (0.5 mm) if the wall is supported along both edges, and 0.040 in (1 mm) if it stands alone. Below that the wall deflects under recoating and peel forces during the build and comes out wavy.

### Why do SLA parts need post-curing?

Laser exposure only partially crosslinks the resin. A 15–60 minute UV post-cure at 60–80 °C (140–175 °F) completes the reaction. Without it a part can be 30–50% weaker and measurably more compliant than the datasheet values, which are always quoted post-cured.

### Are SLA parts suitable for outdoor or long-term use?

Generally no. Photopolymers yellow and embrittle under UV and creep under sustained load. For functional parts that must last, use a sintered nylon process such as SLS or MJF, or move to molded thermoplastic.

### Is SLA stronger than FDM?

SLA is isotropic — there is no weak Z axis — and stiffer, so it wins on small, thin, detailed geometry. But standard SLA resins are brittle, often under 10% elongation at break, while FDM in ABS, PC, or nylon absorbs far more impact. Choose SLA for accuracy and finish, FDM for toughness.

### Why does my hollow SLA part need a hole in it?

Hollow parts trap liquid resin, which adds weight, cures unevenly, and can crack the shell. Design in at least two drain holes of 0.14 in (3.5 mm) or larger at opposite ends of the cavity so air can enter as resin drains.

## Alternative processes

- [Selective Laser Sintering (SLS)](https://manufacturingprocesses.org/processes/forming/selective-laser-sintering-sls.md): Selective laser sintering fuses nylon powder layer by layer with a laser, so the surrounding loose powder supports the part and no supports are needed.
- [Fused Deposition Modeling (FDM)](https://manufacturingprocesses.org/processes/forming/fused-deposition-modeling-fdm.md): Fused deposition modeling extrudes molten thermoplastic filament along programmed paths so each layer welds to the one beneath it.
- [Multi Jet Fusion (MJF)](https://manufacturingprocesses.org/processes/forming/multi-jet-fusion-mjf.md): Multi Jet Fusion prints a fusing agent onto nylon powder and drives it with infrared energy, fusing whole layers at once for high throughput.
- [CNC Machining](https://manufacturingprocesses.org/processes/cutting/cnc-machining.md): CNC machining removes material from solid stock with a programmed cutting tool, producing accurate parts directly from a CAD model with no tooling.
- [Vacuum Casting (Urethane Casting)](https://manufacturingprocesses.org/processes/forming/vacuum-casting.md): Vacuum casting, known in the US as urethane casting, pours polyurethane resin into a silicone mold under vacuum to copy a master pattern for short runs.

## Related processes

- [Masked Stereolithography (MSLA / DLP)](https://manufacturingprocesses.org/processes/forming/masked-stereolithography-msla-dlp.md): Masked stereolithography cures a whole resin layer at once through an LCD or DLP image, so build time depends on height rather than part count.
- [Material Jetting (PolyJet)](https://manufacturingprocesses.org/processes/forming/material-jetting-polyjet.md): Material jetting sprays droplets of photopolymer and cures them layer by layer, allowing several materials and colors in one build.
- [Rapid Prototyping](https://manufacturingprocesses.org/processes/forming/rapid-prototyping.md): Rapid prototyping builds a part layer by layer directly from a CAD model, with no tooling and no minimum order quantity.
- [Selective Laser Sintering (SLS)](https://manufacturingprocesses.org/processes/forming/selective-laser-sintering-sls.md): Selective laser sintering fuses nylon powder layer by layer with a laser, so the surrounding loose powder supports the part and no supports are needed.
- [Vacuum Casting (Urethane Casting)](https://manufacturingprocesses.org/processes/forming/vacuum-casting.md): Vacuum casting, known in the US as urethane casting, pours polyurethane resin into a silicone mold under vacuum to copy a master pattern for short runs.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/forming/stereolithography-sla)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Superforming"
category: "Forming"
subcategory: "Metal"
materials: ["Metal"]
volumes: "10–10,000 parts per year; above roughly 10,000 conventional stamping wins on cycle time"
lead_time: "8–16 weeks for tooling and process development; 4–10 weeks per production run, constrained by the long cycle time"
url: https://manufacturingprocesses.org/processes/forming/superforming
---

# Superforming

Superforming heats a superplastic aluminum alloy sheet and forms it with gas pressure over a single-sided tool, producing deep, complex panels.

- **Category**: [Forming](https://manufacturingprocesses.org/processes/forming.md)
- **Family**: Metal
- **Materials**: Metal
- **Surface finish**: The tool-contact side reproduces the tool finish; the gas side is free-formed and slightly less controlled
- **Typical volumes**: 10–10,000 parts per year; above roughly 10,000 conventional stamping wins on cycle time
- **Lead time**: 8–16 weeks for tooling and process development; 4–10 weeks per production run, constrained by the long cycle time

## Overview

Superforming exploits superplasticity: certain fine-grained alloys, held in a narrow temperature window and strained slowly enough, stretch to several times their original length without necking. Superplastic aluminum grades such as AA5083 SPF reach elongations of several hundred percent at roughly 840–970°F (450–520°C), against 20–25% at room temperature.

The process heats a clamped sheet to that window and forms it against a single-sided tool with gas pressure over a period of minutes rather than a fraction of a second. Because only one tool half exists and forming pressures are low — on the order of 1–3 MPa (145–435 psi) — tooling is a small fraction of a matched press die.

That makes it the answer for deep, complex, low-volume panels: aircraft interior and nacelle components, rail vehicle cladding, specialty and low-volume automotive body panels, architectural cladding, and medical equipment housings. Cycle times of 20 minutes to a couple of hours cap it near 10,000 parts a year.

## How it works

1. **Blank and clamp.** A sheet of superplastic-grade material is loaded into the press and clamped around its full perimeter between heated platens, forming a gas-tight seal. Nothing feeds in from the flange — every bit of the formed shape comes from stretching the sheet inside the clamp line.
2. **Heat soak.** The blank is brought to the forming temperature — roughly 840–970°F (450–520°C) for superplastic aluminum, and around 1,650°F (900°C) for Ti-6Al-4V — and held until uniform.
3. **Gas forming.** Inert gas, usually argon, is admitted on one side and the sheet balloons into the tool. Pressure is ramped on a programmed schedule, not simply applied: superplastic flow only occurs within a narrow strain-rate band, typically in the region of 10⁻⁴ to 10⁻³ per second, and the control system adjusts pressure continuously to hold strain rate inside it. Push too fast and the material necks and tears like ordinary sheet.
4. **Hold and cool.** The formed sheet is held against the tool briefly, then the part is removed and cooled.
5. **Trim.** The clamped perimeter is scrap and is trimmed off, usually by five-axis routing or waterjet.

### The four tool arrangements

**Cavity forming** blows the sheet directly into a female tool. It is the simplest arrangement and thins most severely at the deepest point.

**Bubble forming** first blows a free bubble away from a male tool, then reverses the pressure to wrap that bubble down over the tool. Pre-stretching the sheet before it touches anything distributes the thinning far more evenly, which is how deep male shapes are made.

**Backpressure forming** applies gas to both sides, with a net differential driving the forming. The superimposed hydrostatic pressure suppresses internal cavitation — microvoid formation at grain boundaries — which some alloys need in order to reach full elongation with sound material.

**Diaphragm forming** uses a superplastic sheet as a driving membrane to press a non-superplastic material, such as a composite laminate or a conventional alloy, against a tool.

## Design guidelines

### All the material comes from thinning

There is no draw-in. The sheet is clamped at the perimeter, so the surface area of the finished part must come entirely from stretching the blank. A part whose surface area is twice the blank's plan area will finish, on average, at half the blank thickness — and the thinning is not uniform. Start every design by estimating the surface-area ratio, and choose blank thickness from that rather than from the desired final wall.

### Thinning concentrates at depth and at corners

The deepest region of a cavity-formed part and the tightest corner radii thin the most, because they stretch last and stretch most. If a specific minimum wall is required somewhere, say so on the drawing; the supplier will pick the process variant and blank thickness to hit it, and may switch from cavity to bubble forming for exactly this reason.

### Radii can be tight

Because forming pressure acts everywhere normal to the surface and the material is extremely ductile at temperature, corner radii much tighter than press-brake or draw practice are achievable. The limit is thinning at the radius, not cracking.

### One good surface

The sheet contacts the tool on one side only. That side takes the tool's finish and detail; the gas side is free-formed and slightly less controlled. Put cosmetic and datum surfaces on the tool side.

### Draft is helpful but not mandatory

The part shrinks slightly on cooling and can be lifted from a single-sided tool with modest draft. A small amount — a degree or two — makes extraction reliable.

### Post-form properties

Time at forming temperature affects temper. Superplastic aluminum grades are typically used in a non-heat-treatable condition, so the formed part's strength comes from the alloy rather than a subsequent age. Do not assume a T6 temper survives the cycle. Blank thickness selection from the [sheet metal gauge chart](/charts/sheet-metal-gauge-chart), and secondary bend and flange development from the [bend radius and K-factor chart](/charts/sheet-metal-bend-radius-k-factor).

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Blank thickness | Sized from the surface-area ratio | — | All formed area comes from thinning the blank |
| Thinning | Design for 40–60% at depth | Locally more at tight corners | Deep regions stretch last and stretch most |
| Forming variant | Bubble forming for deep male shapes | Cavity forming thins worst at depth | Pre-stretching distributes strain evenly |
| Corner radii | Tighter than cold forming allows | Limited by local thinning | Ductility at temperature is very high |
| Cosmetic surface | On the tool side | Gas side is free-formed | Only one side contacts the tool |
| Draft | 1–2° | 0° possible with shrinkage | Aids extraction from a single-sided tool |
| Perimeter | Add clamped flange as trim scrap | — | The seal area cannot be part of the part |

## Variants

- Cavity Forming
- Bubble Forming
- Backpressure Forming
- Diaphragm Forming

## Cost drivers

Superforming trades tooling cost for cycle time. The tool is single-sided and sees only low pressure, so it can be cast iron, steel, or ceramic and costs a fraction of a matched press die — that is the entire economic argument. Against it, cycle times of 20 minutes to a couple of hours mean a press makes a handful of parts per shift, so machine time per part is high, and the material itself is a specialty superplastic grade at a premium over commodity sheet. Trimming the clamped perimeter is a five-axis operation on every part and is a real cost line.

Volume breakpoints: from tens of parts, superforming beats matched-die stamping because there is no die to pay for. It stays competitive to a few thousand a year. Above roughly 10,000, conventional stamping's cycle time wins decisively despite the tooling cost, and above that only a geometry that stamping physically cannot make justifies staying with the process.

1. **Consolidate panels.** The saving comes from replacing a multi-piece stamped-and-welded assembly with one formed panel and deleting the joints, sealant, and fixtures.
2. **Choose the right variant.** Moving from cavity to bubble forming can cut required blank thickness substantially, which reduces both material cost and cycle time.
3. **Keep the blank small.** Blank size sets both material cost and the amount of trim scrap.
4. **Design in the details.** Beads, flanges, and mounting features formed in the tool cost nothing per part and remove downstream operations.

## FAQ

### What makes an alloy superplastic?

A very fine, stable grain structure combined with a narrow temperature window and a slow strain rate. Under those conditions deformation occurs largely by grain boundary sliding rather than dislocation motion, so the sheet stretches uniformly instead of necking. AA5083 SPF, AA2004, AA7475, and Ti-6Al-4V are the common superplastic grades.

### What temperature is superforming done at?

Roughly 840–970°F (450–520°C) for superplastic aluminum grades and around 1,650°F (900°C) for Ti-6Al-4V. The window is narrow in both directions: too cold and superplastic flow does not occur, too hot and the fine grain structure that enables it coarsens away.

### Why is superforming so slow?

Because superplastic flow only happens within a narrow strain-rate band, on the order of 10⁻⁴ to 10⁻³ per second. Gas pressure is ramped on a programmed schedule to hold the strain rate inside that band. Forming faster causes the material to neck and tear like ordinary sheet, so cycle times run from 20 minutes to a couple of hours.

### How much does the material thin during superforming?

It depends entirely on the surface-area ratio, because the sheet is clamped at the perimeter and nothing feeds in. A part with twice the blank's plan area averages half the blank thickness, and the deepest regions and tightest corners thin more than that. Blank thickness is chosen from this calculation, not from the target wall.

### What is bubble forming and why use it?

The sheet is first blown into a free bubble away from a male tool, then the pressure is reversed to wrap the bubble down over it. Pre-stretching the sheet before it touches anything distributes thinning far more evenly than blowing directly into a cavity, which is what makes deep male shapes achievable with a reasonable blank thickness.

### What volume suits superforming?

From tens of parts up to a few thousand a year. The single-sided low-pressure tool costs a fraction of a matched press die, which is decisive at low volume. Above roughly 10,000 parts a year, conventional stamping's fraction-of-a-second cycle time overtakes it unless the geometry is one stamping physically cannot produce.

## Alternative processes

- [Hydroforming](https://manufacturingprocesses.org/processes/forming/hydroforming.md): Hydroforming presses sheet or tube into a single-sided die with high-pressure fluid, producing smooth seamless shapes that would need several stampings.
- [Panel Beating](https://manufacturingprocesses.org/processes/forming/panel-beating.md): Panel beating shapes sheet metal by hand with hammers, dollies and an English wheel, producing one-off compound-curved panels without any dies.
- [Metal Stamping](https://manufacturingprocesses.org/processes/forming/metal-stamping.md): Metal stamping shears and forms sheet metal between matched dies in a press, producing flat and shallow-formed parts at very high rates.
- [Composite Laminating](https://manufacturingprocesses.org/processes/forming/composite-laminating.md): Composite laminating lays reinforcing fiber into a mold and consolidates it with resin, curing into a stiff, light monocoque part.

## Related processes

- [Panel Beating](https://manufacturingprocesses.org/processes/forming/panel-beating.md): Panel beating shapes sheet metal by hand with hammers, dollies and an English wheel, producing one-off compound-curved panels without any dies.
- [Deep Drawing](https://manufacturingprocesses.org/processes/forming/deep-drawing.md): Deep drawing pushes a sheet metal blank through a die with a punch so the metal flows into a seamless cup or box deeper than its own diameter.
- [Hydroforming](https://manufacturingprocesses.org/processes/forming/hydroforming.md): Hydroforming presses sheet or tube into a single-sided die with high-pressure fluid, producing smooth seamless shapes that would need several stampings.
- [Metal Stamping](https://manufacturingprocesses.org/processes/forming/metal-stamping.md): Metal stamping shears and forms sheet metal between matched dies in a press, producing flat and shallow-formed parts at very high rates.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/forming/superforming)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Swaging"
category: "Forming"
subcategory: "Metal"
materials: ["Metal"]
tolerances: "Roughly ±0.002–0.005 in (±0.05–0.13 mm) on swaged outside diameter; inside diameter holds a comparable tolerance only when a mandrel is used"
volumes: "100–1,000,000+ parts; practical in dozens on a manual machine"
lead_time: "2–6 weeks including a new die set; days for repeat work on existing tooling"
url: https://manufacturingprocesses.org/processes/forming/swaging
---

# Swaging

Swaging hammers or presses the end of a tube or bar inward against a die to reduce or taper its diameter without removing any material.

- **Category**: [Forming](https://manufacturingprocesses.org/processes/forming.md)
- **Family**: Metal
- **Materials**: Metal
- **Typical tolerances**: Roughly ±0.002–0.005 in (±0.05–0.13 mm) on swaged outside diameter; inside diameter holds a comparable tolerance only when a mandrel is used
- **Surface finish**: Burnished by the repeated die contact, typically smoother than the incoming bar or tube surface
- **Typical volumes**: 100–1,000,000+ parts; practical in dozens on a manual machine
- **Lead time**: 2–6 weeks including a new die set; days for repeat work on existing tooling

## Overview

Swaging reduces or tapers the end of a tube or bar by hammering it inward against shaped dies. In rotary swaging — the production form — a spindle spins a set of dies past a ring of rollers, and each roller passage drives the dies closed and lets them spring open again, delivering thousands of light blows per minute while the workpiece is fed in.

Nothing is removed. The material is displaced, so the metal cold works, gets stronger, and takes on a burnished surface, and a swaged tube end has continuous grain rather than a cut edge. Diameters can be held to a few thousandths of an inch.

It is how cable and wire rope fittings are attached, how tube ends are reduced for assembly, how tapered tubes such as golf shafts and antenna sections are made, and how screwdriver blades and pointed bar ends are formed. Tooling is simple and inexpensive, so swaging is economic from small quantities up to millions.

## How it works

1. **Die set selection.** Two or four die segments are ground to the finished profile — a straight reduction, a taper, or a combination. The dies define the finished outside diameter directly.
2. **Spindle rotation.** The dies sit in slots in a rotating spindle, with backers behind them. A stationary outer ring holds a set of rollers. As the spindle turns, each backer passes a roller and is driven inward, closing the dies; between rollers, centrifugal force opens them.
3. **Blow rate.** The result is a rapid succession of light hammer blows — commonly on the order of 1,000–5,000 per minute depending on spindle speed and roller count. Each individual blow does very little work, which is precisely why the metal flows smoothly rather than cracking.
4. **Feed.** The workpiece is pushed into the closing dies. Because material is displaced rather than cut, it flows both radially inward and axially — a swaged tube gets longer as it gets smaller.
5. **Mandrel (optional).** For tube work where the inside diameter matters, a mandrel is inserted so the wall is compressed between the dies and the mandrel. Without one, the ID follows whatever the wall does, and thin walls can buckle inward.
6. **Multiple passes.** Area reduction per pass is limited, commonly in the range of 10–30% depending on material. Deeper reductions are staged through successive die sets, with the metal work hardening at each stage.

### Cold work is the point

Because the operation is cold, the swaged region ends up stronger than the parent material and with a smooth, burnished surface. On ductile alloys this is free strength. On material that is already hard or has low ductility, the same cold work causes cracking — swaging is a ductile-metal process, and heavily cold-worked material may need an intermediate anneal.

## Design guidelines

### Reduce gradually

Keep area reduction modest per pass — commonly 10–30% depending on material — and stage larger reductions through multiple die sets. Trying to take too much in one pass causes the metal to pile up ahead of the dies, producing folds and laps that show up as cracks in service.

### Tapers should be gentle

A long shallow taper feeds smoothly through the dies. A steep taper makes material accumulate at the shoulder rather than flow, which needs staged dies and more passes. Where a design allows a longer transition, take it.

### Use a mandrel when the bore matters

Swaging a tube without internal support controls only the outside diameter; the wall thickens and the bore follows unpredictably, and thin walls can buckle. A mandrel makes the inside diameter a controlled dimension and keeps the wall uniform. Say on the drawing whether the OD, the ID, or the wall is the controlled feature — all three cannot be.

### Expect the part to grow longer

Displaced material has to go somewhere, and a proportion of it goes axially. Overall length after swaging is longer than the blank, and the growth has to be accounted for either as a cut-to-length allowance or a trim operation.

### Ductile materials only

Low-carbon steel, austenitic stainless, copper, brass, aluminum, and titanium all swage. Hardened steels, cast alloys, and heavily cold-worked stock crack. Where a part needs a hardened final condition, swage first and heat treat afterward.

### Sharp geometry does not swage

The process makes bodies of revolution with gradual transitions. Sharp shoulders, steps with square corners, and non-round sections are not swageable — the metal simply will not fill them under light repeated blows.

For fits between a swaged end and its mating part, see [ISO 286 fits and tolerances](/charts/iso-286-fits-tolerances); for base material comparison, the [material properties chart](/charts/material-properties).

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Area reduction per pass | 10–20% | ≈30% in ductile alloys | Excess reduction folds and laps the metal |
| Taper transition | Long and shallow | Steep tapers need staged dies | Material must flow, not pile up |
| Bore control | Use a mandrel | ID uncontrolled without one | Wall thickens inward under the dies |
| Controlled dimension | Choose OD, ID, or wall | Cannot control all three | Displaced material has to go somewhere |
| Length allowance | Add for axial growth | — | Volume is conserved; the part lengthens |
| Material | Ductile, annealed or soft temper | Hardened stock cracks | Cold work accumulates with every blow |
| Geometry | Bodies of revolution, gradual transitions | No square shoulders | Light blows will not fill sharp corners |

## Variants

- Rotary Swaging
- Hydraulic Swaging

## Cost drivers

Swaging has the cheapest tooling of any of the bulk forming processes: a set of two or four ground die segments, plus a mandrel if the bore is controlled. Machines are small and setups are quick, so the entry cost is low enough that swaging is economic on hundreds of parts. At volume, cycle times of seconds and near-zero material loss make it very cheap per piece — there are no chips, unlike turning the same taper.

Volume breakpoints: dozens of parts are practical on a manual machine. From 1,000 to 100,000, a dedicated die set and semi-automatic feed pay back quickly. In the millions, fully automated swagers with in-line feed and part handling take over.

1. **Design one reduction rather than several.** Every additional stage is another die set and another pass.
2. **Lengthen the taper.** Gentle transitions swage in fewer passes with less die wear.
3. **Only control the bore when you need it.** A mandrel adds tooling, setup, and lubrication requirements.
4. **Swage instead of turning a taper.** The material savings are total — nothing becomes a chip — and the swaged surface is burnished, often eliminating a finishing operation.
5. **Specify an annealed starting condition.** Cracking from over-hardened stock is the most common avoidable scrap in swaging.

## FAQ

### How much can be reduced in one swaging pass?

Commonly 10–30% area reduction per pass, depending on material ductility. Larger reductions are staged through successive die sets. Attempting too much in one pass makes the metal pile up ahead of the dies and fold over on itself, creating laps that behave like cracks.

### Does swaging control the inside diameter of a tube?

Only if a mandrel is used. Without internal support the dies control the outside diameter, the wall thickens inward, and the bore follows unpredictably — thin walls can buckle. Specify which of OD, ID, or wall thickness is the controlled feature, because all three cannot be held at once.

### Does swaging make the part stronger?

Yes. The operation is cold, so the swaged region work hardens and gains strength, and the repeated die contact leaves a burnished surface. The same mechanism is the process limit: material that is already hard or has low ductility cracks instead of flowing, so parts needing a hardened final state are swaged first and heat treated afterward.

### Why does a swaged part get longer?

Because no material is removed. Volume is conserved, so metal displaced radially inward by the dies also flows axially. The blank has to be cut short of the finished length, or the part trimmed afterward, to account for the growth.

### What materials can be swaged?

Ductile metals: low-carbon steel, austenitic stainless, copper, brass, aluminum, and titanium. Hardened steel, cast alloys, and heavily cold-worked stock crack under the repeated blows. An intermediate anneal restores ductility when several stages of reduction are needed.

### Swaging or turning a taper on a lathe?

Turning is more flexible and holds tighter tolerances, but it cuts the material away as chips and severs the grain flow. Swaging keeps all the material, work hardens the formed region, leaves a burnished surface, and runs in seconds per part with inexpensive tooling — but it only makes gradual, round transitions.

## Alternative processes

- [CNC Turning](https://manufacturingprocesses.org/processes/cutting/cnc-turning.md): CNC turning rotates the workpiece against a stationary tool, producing round parts with excellent concentricity and short cycle times.
- [Forging](https://manufacturingprocesses.org/processes/forming/forging.md): Forging deforms solid metal under compressive load in a hammer, press or dies, so the grain flow follows the part contour and fatigue strength rises.
- [Hydroforming](https://manufacturingprocesses.org/processes/forming/hydroforming.md): Hydroforming presses sheet or tube into a single-sided die with high-pressure fluid, producing smooth seamless shapes that would need several stampings.
- [Metal Extrusion](https://manufacturingprocesses.org/processes/forming/metal-extrusion.md): Metal extrusion pushes heated billet through a shaped die, producing continuous aluminum or copper profile of constant cross-section.

## Related processes

- [Tube and Section Bending](https://manufacturingprocesses.org/processes/forming/tube-and-section-bending.md): Tube and section bending draws or presses metal tube and profile around a former, with an internal mandrel wherever the wall would otherwise collapse.
- [Forging](https://manufacturingprocesses.org/processes/forming/forging.md): Forging deforms solid metal under compressive load in a hammer, press or dies, so the grain flow follows the part contour and fatigue strength rises.
- [Metal Extrusion](https://manufacturingprocesses.org/processes/forming/metal-extrusion.md): Metal extrusion pushes heated billet through a shaped die, producing continuous aluminum or copper profile of constant cross-section.
- [Roll Forming](https://manufacturingprocesses.org/processes/forming/roll-forming.md): Roll forming passes coil stock through a sequence of contoured roller stands that bend it progressively into a constant cross-section profile.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/forming/swaging)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Thermoforming"
category: "Forming"
subcategory: "Plastics and Rubber"
materials: ["Plastic"]
tolerances: "±0.030 in (±0.75 mm) on vacuum-formed tool-side features; ±0.015 in (±0.4 mm) pressure formed and on CNC-trimmed edges. Non-tool-side dimensions carry all sheet thickness variation."
volumes: "50–50,000 parts per year for heavy gauge; thin-gauge roll-fed packaging runs into the millions"
lead_time: "1–2 weeks for prototype wood or epoxy tooling; 2–4 weeks for a production aluminum tool; days per run thereafter"
url: https://manufacturingprocesses.org/processes/forming/thermoforming
---

# Thermoforming

Thermoforming heats a plastic sheet until pliable and pulls or presses it over a single-sided mold, producing large thin-walled parts with low tooling cost.

- **Category**: [Forming](https://manufacturingprocesses.org/processes/forming.md)
- **Family**: Plastics and Rubber
- **Materials**: Plastic
- **Typical tolerances**: ±0.030 in (±0.75 mm) on vacuum-formed tool-side features; ±0.015 in (±0.4 mm) pressure formed and on CNC-trimmed edges. Non-tool-side dimensions carry all sheet thickness variation.
- **Surface finish**: The sheet's own surface (gloss, matte, or extruded grain) is reproduced on the non-tool side; the tool side takes the mold finish, and pressure forming can reproduce molded textures
- **Typical volumes**: 50–50,000 parts per year for heavy gauge; thin-gauge roll-fed packaging runs into the millions
- **Lead time**: 1–2 weeks for prototype wood or epoxy tooling; 2–4 weeks for a production aluminum tool; days per run thereafter

## Overview

Thermoforming heats a thermoplastic sheet until it is rubbery, then pulls or presses it over a single-sided mold and cools it in shape. Because the tool forms only one face and never sees melt pressure, tooling is a small fraction of an injection mold's cost and can be cut from aluminum in 2–4 weeks — which makes thermoforming the standard answer for large, thin-walled parts at hundreds to low tens of thousands per year.

It spans a wide range: 0.010 in blister packs and clamshells at one end; 0.250 in machine enclosures, vehicle interior panels, refrigerator liners, and shower trays at the other. Common materials are HIPS, ABS, HDPE, PP, PETG, PVC, polycarbonate, and acrylic.

The defining constraint is that the sheet thins as it stretches. A part drawn near its practical limit typically retains only 20–50% of the starting sheet gauge at the deepest corner, so you buy sheet by the thinnest section you need rather than the average.

## How it works

1. **Clamp and heat.** The sheet is clamped in a frame and passed under radiant heaters until it reaches its forming window: PETG 275–320 °F (135–160 °C), HIPS and ABS 300–360 °F (150–180 °C), PP 290–330 °F (145–165 °C), polycarbonate 375–400 °F (190–205 °C). Semi-crystalline materials such as PP have a narrow window and sag quickly; amorphous materials are far more forgiving.

2. **Pre-stretch (optional).** For deep draws, the sheet is bubble-blown or a mechanical plug pushes it partway into the cavity before vacuum is applied. Plug assist moves material into the deep regions before it touches cold tool surfaces and freezes.

3. **Form.** In vacuum forming, the tool evacuates the space beneath the sheet and atmospheric pressure — a hard ceiling of 14.7 psi (1 bar) — pushes the sheet onto the mold. Pressure forming adds compressed air on the back side at 40–100 psi (3–7 bar), giving four to seven times the forming force and with it sharper radii, crisper detail, and molded-in texture.

4. **Cool.** The sheet gives up heat to the aluminum tool, which is water-cooled on production tooling. Cooling dominates the cycle: roughly 20 seconds for thin-gauge packaging up to several minutes for a 0.250 in enclosure panel.

5. **Trim.** The formed sheet is removed and the part cut free. Thin-gauge packaging is trimmed with steel-rule dies in line; heavy-gauge parts go to a 3- or 5-axis CNC router, which is also where holes and cutouts are made.

### What does twin-sheet thermoforming add?

Two sheets are heated simultaneously, formed in opposing tool halves, and pressed together while still hot so they fuse at designed contact areas. The result is a hollow, double-walled part — pallets, fuel tanks, structural doors — with far higher stiffness per pound than a single-sheet part, at roughly double the tooling and cycle.

## Design guidelines

### Depth of draw

The ratio of draw depth to the smallest opening width is the governing number. Straight vacuum forming is comfortable up to about 1:1. Plug assist or pressure forming pushes it to 2:1 and beyond, but every increment costs wall thickness in the corners. Female (cavity) tools distribute material better on deep parts than male (plug) tools.

### Starting gauge and wall thinning

Order sheet by the thinnest section you need, not the average. In a hard draw the deepest corner retains 20–50% of the starting gauge — a 0.187 in (4.75 mm) sheet drawn aggressively can finish under 0.060 in (1.5 mm) at the corners. Uniform draw depth across the part is the single best defense.

### Draft

Male tools need 3° per side minimum, and 5° is safer, because the part shrinks onto the tool as it cools and has to be stripped off. Female cavities need only 1–2°, since the part shrinks away from the wall. Textured surfaces need more: budget roughly 1° extra per 0.001 in (0.025 mm) of texture depth, the same relationship tabulated for molded textures on [/charts/injection-molding-design-guidelines](/charts/injection-molding-design-guidelines).

### Radii

Minimum inside radius equal to the starting sheet thickness; 2–3× thickness is much better. Small radii are where the sheet stretches fastest and thins most, and on a vacuum-formed part they simply round over, because 1 bar cannot force the sheet into a tight corner. Pressure forming holds noticeably sharper corners for exactly this reason.

### Ribs, undercuts, and detail

Ribs must be wide and shallow — a formed sheet cannot fill a narrow slot. Undercuts require a moving core, a stripper ring, or enough material flexibility to snap the part off; each adds tool cost. Molded-in text and logos are practical in pressure forming and marginal in vacuum forming.

### Tolerances and trimming

Tool-side features hold roughly ±0.030 in (±0.75 mm) in vacuum forming and ±0.015 in (±0.4 mm) in pressure forming. The non-tool side carries all the thickness variation, so never dimension across the sheet. Trimmed edges and hole positions come from the CNC trim fixture and hold about ±0.015 in (±0.4 mm) — put locating features there rather than on formed geometry.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Depth-to-width ratio | ≤ 1:1 vacuum forming | 2:1+ with plug assist | Deeper draws thin the corners past usable wall |
| Draft, male tool | 5° per side | 3° per side | Part shrinks onto the tool and must strip off |
| Draft, female tool | 2° per side | 1° per side | Part shrinks away from the cavity |
| Inside radius | 2–3 × sheet thickness | 1 × sheet thickness | Corners stretch most; 1 bar cannot fill sharp detail |
| Formed wall retention | 50% of starting gauge | 20% at the deepest corner | Sets the sheet gauge you have to buy |
| Forming pressure | 14.7 psi (1 bar) vacuum | 40–100 psi (3–7 bar) pressure forming | More pressure buys detail and sharper radii |
| Tolerance, tool side | ±0.030 in (±0.75 mm) vacuum | ±0.015 in (±0.4 mm) pressure | The non-tool side absorbs all thickness variation |

## Variants

- Vacuum Forming
- Pressure Forming
- Plug-assisted Forming
- Twin Sheet Thermoforming

## Cost drivers

The trade against injection molding is straightforward. Thermoform tooling is a single-sided aluminum form — wood or filled epoxy for prototypes — that costs a small fraction of a two-plate injection mold and arrives in 2–4 weeks. Per-part cost is higher, because you buy extruded sheet, throw away the trim skeleton, and run a longer cycle.

Volume breakpoints:

- Under 50 parts: wood or epoxy prototype tooling, hand trim.
- 50–5,000: machined aluminum single-cavity tool with CNC trim — the classic heavy-gauge sweet spot.
- 5,000–50,000: temperature-controlled cast aluminum tool, pressure forming, dedicated trim fixture.
- Above roughly 50,000 parts a year for a small part, injection molding usually wins on unit cost despite the tooling.

Cost reduction:

1. **Even out the draw depth.** One deep pocket forces you to buy heavy sheet for the entire panel. Redistributing depth lets you drop a gauge, which is a direct material saving on every part.
2. **Nest parts on the sheet.** The trim skeleton is pure scrap on virgin material; multi-up tools and tight nesting cut it substantially, and most thermoformers regrind and blend it back in house.
3. **Choose vacuum over pressure forming when cosmetics allow.** Pressure-forming tools are pressure vessels with a matched top half and cost significantly more.
4. **Put tight tolerances on trimmed features.** CNC-trimmed holes and edges hold about ±0.015 in (±0.4 mm); formed geometry does not.
5. **Use twin-sheet only where stiffness demands it.** It roughly doubles both tooling and cycle, but it replaces an assembly of two formed halves plus fasteners.

## FAQ

### How deep can you thermoform a part?

Straight vacuum forming handles a depth-to-minimum-width ratio of about 1:1 comfortably. Plug assist and pressure forming push that to 2:1 and beyond, but every increment thins the corners further. Female cavity tools distribute material better than male plug tools on deep parts.

### How much does a thermoformed part thin out?

In a hard draw the deepest corner typically retains only 20–50% of the starting sheet gauge. Specify sheet by the thinnest section your part needs, not the average — a 0.187 in (4.75 mm) sheet drawn aggressively can finish under 0.060 in (1.5 mm) in the corners.

### What is the difference between vacuum forming and pressure forming?

Vacuum forming relies on atmospheric pressure alone, a hard ceiling of 14.7 psi (1 bar). Pressure forming adds 40–100 psi (3–7 bar) of compressed air behind the sheet, giving four to seven times the forming force. That buys sharper radii, molded-in texture and lettering, and roughly double the dimensional precision, at higher tooling cost.

### What draft angle does thermoforming need?

Male (plug) tools need at least 3° per side and preferably 5°, because the cooling part shrinks onto the tool and must be stripped off. Female (cavity) tools need only 1–2°, since the part pulls away from the wall. Add roughly 1° per 0.001 in (0.025 mm) of texture depth on textured surfaces.

### When is thermoforming cheaper than injection molding?

Below roughly 5,000–50,000 parts a year, and especially for large, thin-walled parts. Thermoform tooling is single-sided aluminum at a small fraction of an injection mold's cost and arrives in 2–4 weeks. Above that volume the higher per-part cost of sheet and trim scrap usually tips the balance back to injection molding.

### What tolerance can thermoforming hold?

About ±0.030 in (±0.75 mm) on vacuum-formed tool-side features and ±0.015 in (±0.4 mm) with pressure forming. CNC-trimmed edges and holes also hold about ±0.015 in (±0.4 mm), so locating features should be trimmed rather than formed.

### What is twin-sheet thermoforming used for?

Hollow, double-walled parts such as pallets, fuel tanks, and structural doors. Two sheets are formed in opposing tool halves and pressed together while hot so they fuse at designed contact areas, giving much higher stiffness per pound than a single sheet at roughly double the tooling and cycle time.

## Alternative processes

- [Injection Molding](https://manufacturingprocesses.org/processes/forming/injection-molding.md): Injection molding forces molten thermoplastic into a steel mold under high pressure, producing high-volume identical plastic parts.
- [Rotation Molding](https://manufacturingprocesses.org/processes/forming/rotation-molding.md): Rotational molding tumbles powdered polymer inside a heated mold on two axes so it fuses into a seamless, stress-free hollow part with a uniform wall.
- [Reaction Injection Molding](https://manufacturingprocesses.org/processes/forming/reaction-injection-molding.md): Reaction injection molding mixes two liquid reactants that polymerize inside a low-pressure mold, producing large tough polyurethane parts on light tooling.
- [Blow Molding](https://manufacturingprocesses.org/processes/forming/blow-molding.md): Blow molding inflates a heated plastic tube or preform against the walls of a mold with compressed air, producing hollow one-piece containers and tanks.

## Related processes

- [Vacuum Casting (Urethane Casting)](https://manufacturingprocesses.org/processes/forming/vacuum-casting.md): Vacuum casting, known in the US as urethane casting, pours polyurethane resin into a silicone mold under vacuum to copy a master pattern for short runs.
- [Blow Molding](https://manufacturingprocesses.org/processes/forming/blow-molding.md): Blow molding inflates a heated plastic tube or preform against the walls of a mold with compressed air, producing hollow one-piece containers and tanks.
- [Plastic Extrusion](https://manufacturingprocesses.org/processes/forming/plastic-extrusion.md): Plastic extrusion pushes molten polymer through a shaped die and cools it, producing continuous profile, sheet, pipe or film of constant cross-section.
- [Rotation Molding](https://manufacturingprocesses.org/processes/forming/rotation-molding.md): Rotational molding tumbles powdered polymer inside a heated mold on two axes so it fuses into a seamless, stress-free hollow part with a uniform wall.
- [Die Cutting](https://manufacturingprocesses.org/processes/cutting/die-cutting.md): Die cutting presses a shaped steel rule or rotary die through sheet stock, cutting flat outlines in paper, board, foam, film and textile.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/forming/thermoforming)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Tube and Section Bending"
category: "Forming"
subcategory: "Metal"
materials: ["Metal"]
tolerances: "Roughly ±1° on bend angle and plane of bend, and about ±0.030 in (±0.75 mm) on leg lengths; tolerances accumulate along a multi-bend part, so dimension from a single datum end"
volumes: "1–500,000 parts; economical from a single piece if tooling already exists"
lead_time: "Days to 2 weeks against existing tooling; 4–8 weeks if a new radius tool set is required"
url: https://manufacturingprocesses.org/processes/forming/tube-and-section-bending
---

# Tube and Section Bending

Tube and section bending draws or presses metal tube and profile around a former, with an internal mandrel wherever the wall would otherwise collapse.

- **Category**: [Forming](https://manufacturingprocesses.org/processes/forming.md)
- **Family**: Metal
- **Materials**: Metal
- **Typical tolerances**: Roughly ±1° on bend angle and plane of bend, and about ±0.030 in (±0.75 mm) on leg lengths; tolerances accumulate along a multi-bend part, so dimension from a single datum end
- **Surface finish**: Retains the incoming tube finish; die marks along the tangent and slight wiper scoring on the inside of the bend are normal
- **Typical volumes**: 1–500,000 parts; economical from a single piece if tooling already exists
- **Lead time**: Days to 2 weeks against existing tooling; 4–8 weeks if a new radius tool set is required

## Overview

Tube and section bending forms round tube, square tube, pipe, and structural profile around a die without cutting or welding. The dominant production method is rotary draw bending: the tube is clamped to a rotating bend die and drawn around it, usually with a flexible mandrel inside to stop the wall from collapsing.

It is how handrails, roll cages, exhaust systems, hydraulic and fuel lines, furniture frames, bicycle frames, and heat exchanger coils are made. Bending replaces welded elbows with a continuous member, which is stronger, cleaner, and — in fluid systems — lower in pressure drop.

Two numbers govern feasibility. The bend ratio, centerline radius divided by tube outside diameter, is comfortable at 2D or more and needs increasing tooling support below that. The wall factor, outside diameter divided by wall thickness, decides whether a mandrel is required: thin-wall tube with a high wall factor collapses without one.

## How it works

### Rotary draw bending

1. **Clamp.** The tube is clamped against the bend die by a clamp die. The bend die's groove radius is the centerline radius of the finished bend, so each radius needs its own tool set.
2. **Support.** A pressure die presses against the tube along the tangent and follows it into the bend. Inside the tube, a mandrel — usually a plug with linked flexible balls — supports the wall through the arc. A wiper die sits immediately behind the tangent point on the inside of the bend, filling the gap where wrinkles would otherwise start.
3. **Draw.** The bend die rotates and draws the tube around itself. The outside of the bend stretches and thins; the inside compresses and thickens.
4. **Retract and index.** Tooling releases, the tube is advanced and rotated to the next bend's plane, and the cycle repeats. A CNC bender handles feed length, rotation, and bend angle automatically.

### Other methods

**Compression bending** wraps the tube around a fixed form with a following roller. It is simpler and cheaper but produces more distortion, so it suits large radii and non-critical work.

**Roll bending** passes the tube through three rolls to make large-radius sweeps, coils, and rings — no mandrel, no tight radii.

**Ram (press) bending** pushes a former into the tube between two supports. It is the fastest and crudest method, normal for conduit and heavy-wall pipe where ovality does not matter.

**Ring rolling** is a related hot process: a pierced forged preform is rolled between a main roll and a mandrel to produce a seamless ring for bearing races, flanges, and gear blanks, with grain flow following the circumference.

Springback is present in every method — roughly 1–3° for mild steel, more for stainless, aluminum, and high-strength grades — and the machine overbends to compensate. Because springback varies with material lot and wall thickness, the first article of every new batch is normally checked and the program trimmed.

## Design guidelines

### Bend radius

Specify centerline radius, not inside or outside radius, because that is how tooling is described. A centerline radius of 2× tube outside diameter (2D) is the comfortable standard and often runs with simple tooling. 1.5D is routine with a mandrel and wiper die. 1D is achievable in the right material but demands full tooling, careful setup, and tolerance for more thinning and ovality.

### Standardize on one radius

Each distinct centerline radius requires its own bend die, clamp die, pressure die, and mandrel. A part using a single radius throughout is dramatically cheaper to tool than one using three, and reusing a radius the fabricator already owns can eliminate tooling cost entirely.

### Wall factor decides the mandrel

Outside diameter divided by wall thickness is the wall factor. Thick-wall tube at a generous radius bends unsupported; as the wall factor rises past roughly 25–30, or the bend ratio drops below about 2D, a mandrel becomes necessary to keep the tube from collapsing on the inside of the bend. Tube wall is frequently specified by gauge rather than decimal thickness — see the [sheet metal gauge chart](/charts/sheet-metal-gauge-chart) — and alloy formability differences are summarized in the [material properties chart](/charts/material-properties).

### Straight tangent between bends

Leave enough straight length between bends for the clamp die to grip — a working minimum is about 1× tube diameter of clamping length beyond the tangent point, and 2–3× diameter is far more comfortable. Bends placed back to back with no straight between them require special tooling or cannot be made at all.

### Expect thinning and ovality

The outside of the bend thins 10–25% at typical radii, and the section goes oval, commonly by a few percent. Tighter radii make both worse. If a wall thickness or a roundness limit is functional, state it on the drawing — a 15% maximum thinning callout and a 5% ovality limit are common, and they directly constrain how tight the radius can be.

### Orient the weld seam

On welded tube, position the seam near the neutral axis of the bend — roughly 90° from the plane of the bend — so it is neither stretched on the outside nor compressed on the inside. Seams placed on the outside of a tight bend split.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Centerline radius | ≥ 2× tube OD | 1× OD with full tooling | Thinning and ovality rise steeply below 2D |
| Number of distinct radii | One | Each radius is a full tool set | Bend, clamp, pressure die and mandrel per radius |
| Wall factor (OD ÷ wall) | < 25 bends unsupported | Above that, mandrel required | Thin walls collapse on the inside of the bend |
| Straight between bends | 2–3× tube diameter | ≈1× diameter | Clamp die needs grip length |
| Wall thinning | Design for 15% | 25% at tight radii | Outer fiber stretches around the bend |
| Ovality | < 5% | 8% | Section flattens as it is drawn around the die |
| Weld seam position | Near the neutral axis | Never on the outside of the bend | Stretched seams split |
| Bend angle tolerance | ±1° | ±0.5° with first-article correction | Springback varies with material lot |

## Variants

- Mandrel Bending
- Ring Rolling

## Cost drivers

Tooling cost is per radius, not per part, and that single fact drives most bending economics. A CNC bender with the right tooling in stock produces parts with only a setup charge; a part specifying an unusual radius forces a new tool set that must be paid for before the first bend. Cycle time is a few seconds per bend, so per-part cost on a simple two-bend part is dominated by handling and setup rather than machine time.

Volume breakpoints: 1–100 parts is entirely setup-driven, and reusing existing tooling matters more than anything else in the design. From 1,000 to 100,000, a dedicated tool set amortizes easily and CNC cycle time dominates. Very high volumes may justify a dedicated multi-stack bender that makes several radii without a tool change.

1. **Use one radius throughout.** It is the single largest cost lever in the process.
2. **Ask the fabricator what tooling they own.** Designing to an existing bend die can remove the entire tooling line from the quote.
3. **Open the radius.** Going from 1D to 2D reduces thinning, ovality, scrap, and tooling complexity simultaneously.
4. **Leave generous straights.** Short tangents force special clamp tooling or additional operations.

## FAQ

### What is the minimum bend radius for tube bending?

A centerline radius of 2× the tube outside diameter (2D) is the comfortable standard. 1.5D is routine with a mandrel and wiper die, and 1D is achievable in ductile material with full tooling, at the cost of more wall thinning and ovality. Always specify centerline radius, since that is how tooling is described.

### When does tube bending need a mandrel?

When the wall factor — outside diameter divided by wall thickness — rises past roughly 25–30, or when the bend ratio drops below about 2D. Without internal support in those conditions the tube collapses and wrinkles on the inside of the bend. Thick-wall tube at generous radii bends unsupported.

### How much does a tube wall thin during bending?

Typically 10–25% on the outside of the bend at normal radii, with the inside thickening. Ovality of a few percent is also normal. Both get worse as the radius tightens. If wall thickness or roundness is functional, state a limit — 15% maximum thinning and 5% ovality are common callouts.

### Why does using one bend radius matter so much?

Every distinct centerline radius needs its own bend die, clamp die, pressure die, and mandrel. A part with three radii needs three complete tool sets. Designing to a single radius, and preferably to one the fabricator already owns, can remove the entire tooling line from a quote.

### How much straight length is needed between bends?

A working minimum of about one tube diameter of clamping length beyond the tangent point, with two to three diameters much more comfortable. The clamp die has to grip somewhere. Bends placed back to back with no straight between them require special tooling or are not possible.

### Where should the weld seam go on a bent tube?

Near the neutral axis, roughly 90° from the plane of the bend, so the seam is neither stretched on the outside nor compressed on the inside. A seam positioned on the outside of a tight bend is a common cause of splitting.

## Alternative processes

- [Hydroforming](https://manufacturingprocesses.org/processes/forming/hydroforming.md): Hydroforming presses sheet or tube into a single-sided die with high-pressure fluid, producing smooth seamless shapes that would need several stampings.
- [Arc Welding](https://manufacturingprocesses.org/processes/joining/arc-welding.md): Arc welding melts and fuses metal with an electric arc between an electrode and the work, shielded from the atmosphere by gas or flux.
- [Metal Extrusion](https://manufacturingprocesses.org/processes/forming/metal-extrusion.md): Metal extrusion pushes heated billet through a shaped die, producing continuous aluminum or copper profile of constant cross-section.
- [CNC Wire Bending](https://manufacturingprocesses.org/processes/forming/cnc-wire-bending.md): CNC wire bending feeds round or flat wire through a programmable head that bends it in sequence into a finished 2D or 3D form.

## Related processes

- [Press Braking](https://manufacturingprocesses.org/processes/forming/press-braking.md): Press braking bends sheet metal by driving a punch into a V-die, forming one straight-line bend at a time.
- [Roll Forming](https://manufacturingprocesses.org/processes/forming/roll-forming.md): Roll forming passes coil stock through a sequence of contoured roller stands that bend it progressively into a constant cross-section profile.
- [Swaging](https://manufacturingprocesses.org/processes/forming/swaging.md): Swaging hammers or presses the end of a tube or bar inward against a die to reduce or taper its diameter without removing any material.
- [Hydroforming](https://manufacturingprocesses.org/processes/forming/hydroforming.md): Hydroforming presses sheet or tube into a single-sided die with high-pressure fluid, producing smooth seamless shapes that would need several stampings.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/forming/tube-and-section-bending)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Vacuum Casting (Urethane Casting)"
category: "Forming"
subcategory: "Plastics and Rubber"
materials: ["Plastic"]
tolerances: "±0.3% of the nominal dimension, with a minimum of about ±0.010 in (±0.25 mm). Accuracy depends on the master pattern and drifts slightly as the silicone mold ages."
volumes: "10–100 parts; 15–25 parts per silicone mold before it must be replaced"
lead_time: "5–15 working days from master pattern to first parts, including silicone cure; a few parts per mold per day thereafter"
url: https://manufacturingprocesses.org/processes/forming/vacuum-casting
---

# Vacuum Casting (Urethane Casting)

Vacuum casting, known in the US as urethane casting, pours polyurethane resin into a silicone mold under vacuum to copy a master pattern for short runs.

- **Category**: [Forming](https://manufacturingprocesses.org/processes/forming.md)
- **Family**: Plastics and Rubber
- **Materials**: Plastic
- **Typical tolerances**: ±0.3% of the nominal dimension, with a minimum of about ±0.010 in (±0.25 mm). Accuracy depends on the master pattern and drifts slightly as the silicone mold ages.
- **Surface finish**: Reproduces the master exactly — polished masters give gloss, bead-blasted masters give an even matte, and light textures transfer directly
- **Typical volumes**: 10–100 parts; 15–25 parts per silicone mold before it must be replaced
- **Lead time**: 5–15 working days from master pattern to first parts, including silicone cure; a few parts per mold per day thereafter

## Overview

Vacuum casting — called urethane casting in the US — pours a two-part polyurethane resin into a silicone rubber mold under vacuum, so the resin fills every detail without entrapped air. The silicone mold is itself cast around a master pattern, almost always a stereolithography or CNC-machined part, which means the process reproduces a single prototype into a small batch of production-like copies.

It occupies a narrow but important window: 10–100 parts, delivered in one to two weeks, with the appearance, feel, and approximate mechanical behavior of injection molded parts. Resins are formulated to mimic ABS, PP, PC, and filled compounds, plus rubber-like grades from roughly 30 to 90 Shore A and optically clear grades.

The constraint is mold life. A silicone tool yields roughly 15–25 parts before it degrades, so larger batches mean more molds, not more shots.

## How it works

1. **Build the master.** A stereolithography, PolyJet, or machined pattern is finished to the exact surface you want on the final parts — the silicone reproduces everything, including layer lines and sanding scratches. Any texture, polish, or lettering has to exist on the master.

2. **Gate and box the master.** The pattern is suspended in a casting frame with sprue and vent risers attached, and a parting line is marked on it with tape to plan where the mold will be cut open.

3. **Cast the silicone.** Two-part RTV silicone is degassed and poured around the master, then cured — typically several hours in an oven at around 105–160 °F (40–70 °C). Cure time is the long pole in the first-article schedule.

4. **Cut and extract.** The cured block is cut open along the planned parting line, usually by hand with a scalpel in a deliberately jagged path so the two halves key back together precisely. The master is removed, leaving a cavity that is an exact negative.

5. **Cast the part.** The mold is closed, taped, and placed in a vacuum chamber. Polyurethane A and B components are metered, mixed, and poured under vacuum so no air is trapped in the resin or the cavity. Pot life is short — often only a few minutes — so mixing, degassing, and pouring are one continuous operation.

6. **Cure and demold.** The filled mold goes into an oven, commonly 160–175 °F (70–80 °C) for 30–60 minutes depending on the resin, then the flexible silicone is peeled back off the part.

### Why can silicone molds handle undercuts?

Because the mold itself stretches. Silicone elongates enough to release moderate undercuts, snap features, and even light textures that would require a side action in a steel tool. This is the main geometric advantage over injection molding, and it is why vacuum casting can produce a finished-looking prototype with living hinges and integral clips in one piece.

## Design guidelines

### Wall thickness

Minimum around 0.030 in (0.75 mm) for a small part; 0.060–0.120 in (1.5–3 mm) is the comfortable range. Thin sections cool and cure unevenly and are prone to short fills at the extremities of a long flow path. Unlike injection molding, thick sections are tolerated — but heavy masses exotherm, which can discolor the resin and increase shrinkage locally.

### Draft

1–3° is preferred and makes demolding cleaner, but zero draft is workable because the silicone flexes. This is a genuine design freedom: parts intended for eventual injection molding should still carry production draft so that the prototype validates the real geometry.

### Undercuts and detail

Moderate undercuts, snap hooks, and light textures release directly from the silicone with no tooling action. Deep, sharp undercuts still risk tearing the mold and shortening its already limited life.

### Tolerances

The standard figure is ±0.3% of the nominal dimension with a minimum of about ±0.010 in (±0.25 mm) — so a 100 mm dimension carries about ±0.3 mm. Accuracy depends on the master, on silicone shrinkage, and on how many shots the mold has taken, since dimensions drift slightly as the tool ages. Where a part must be dimensionally critical, take it from the first parts out of a fresh mold.

### Text, texture, and finish

Whatever the master carries, the parts carry. A polished master gives gloss parts; a bead-blasted master gives an even matte. Molded-in text down to small point sizes reproduces well. Parts can also be painted, primed, and pad printed exactly like injection molded ABS, which is why vacuum casting dominates pre-production appearance models.

### Material selection

Resins are specified by target behavior rather than by polymer name — an "ABS-like" grade matching approximate stiffness and impact, a "PP-like" grade for living hinges, rubber-like grades at 30–90 Shore A, optically clear grades for lenses. They approximate but do not equal the real thermoplastic, particularly in long-term heat and UV performance. For the properties of the production polymers being simulated, see [/charts/material-properties](/charts/material-properties).

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Wall thickness | 0.060–0.120 in (1.5–3 mm) | 0.030 in (0.75 mm) | Thin walls short-fill at the end of a long flow path |
| Draft | 1–3° per side | 0° | Silicone flexes, but production draft validates the real design |
| Undercuts | Moderate, released by mold flex | Deep sharp undercuts | Tearing shortens an already short mold life |
| Tolerance | ±0.3% of dimension | Minimum ±0.010 in (±0.25 mm) | Master accuracy plus silicone shrinkage |
| Parts per mold | 15–25 | — | Silicone degrades from heat and resin chemistry |
| Surface | Whatever the master carries | — | The silicone reproduces every scratch and layer line |

## Cost drivers

The cost structure is dominated by labor and by mold life. Each silicone mold is hand-cut, hand-taped, hand-poured, and hand-demolded, so per-part cost stays roughly flat with quantity instead of falling — the opposite of injection molding. The master pattern and the first silicone tool are the fixed costs, and they are small compared with any steel tooling.

Volume breakpoints:

- 1–5 parts: 3D print them directly. A silicone tool is not worth building.
- 10–100 parts: the vacuum casting window, using one to five molds.
- 100–1,000 parts: possible, but you are now paying for five to fifty silicone tools and the labor of running them. Compare hard against aluminum bridge tooling for injection molding.
- Over 1,000 parts: injection molding, essentially always.

Cost reduction:

1. **Design the parting line into the master.** A clean, planar split cuts silicone consumption and makes each demold faster and less damaging to the mold.
2. **Keep the part small.** Silicone is billed by volume and a large casting frame consumes a great deal of it. Splitting a large part into bonded sections is often cheaper than one big mold.
3. **Batch color into the resin.** Pigmenting the shot avoids painting every part, which is usually the largest secondary-operation line item.
4. **Reuse one mold for the whole batch where quality allows.** Parts 20–25 out of a mold are dimensionally and cosmetically the weakest; if the application tolerates it, you avoid building a second tool.
5. **Get the master right first.** Every defect on the pattern is reproduced on every part in the batch, and fixing it means a new mold, not a process adjustment.

## FAQ

### How many parts can you get from one silicone mold?

Roughly 15–25, depending on part geometry, resin chemistry, and cure temperature. The silicone degrades from repeated heat exposure and from the polyurethane's reaction, so larger batches require building additional molds rather than taking more shots from one.

### What tolerance can vacuum casting hold?

The standard figure is ±0.3% of the nominal dimension with a minimum of about ±0.010 in (±0.25 mm), so a 100 mm dimension carries roughly ±0.3 mm. Accuracy inherits from the master pattern and drifts slightly as the mold ages, so take dimensionally critical parts from the first shots out of a fresh mold.

### Is vacuum casting the same as urethane casting?

Yes — they are the same process under two names. European and Asian suppliers generally say vacuum casting, describing the vacuum chamber used to fill the mold air-free; US suppliers generally say urethane casting, describing the polyurethane resin. Silicone tooling, RTV casting, and polyurethane casting all refer to the same method.

### Can vacuum cast parts have undercuts?

Yes, and this is one of the process's main advantages. The silicone mold flexes enough to release moderate undercuts, snap hooks, living hinges, and light textures that would require a side action in a steel injection mold. Deep, sharp undercuts still risk tearing the mold and shortening its already limited life.

### When should I choose injection molding over vacuum casting?

Above roughly 100 parts. Vacuum casting cost stays nearly flat per part because every mold is hand-built and hand-run, so a batch of 500 means building and running ten or more silicone tools. At that point an aluminum bridge tool for injection molding is usually cheaper and gives real production material properties.

### Do vacuum cast parts have the same properties as the plastic they imitate?

They approximate them. Polyurethane resins are formulated to match the stiffness, impact behavior, and feel of ABS, PP, or PC closely enough for fit, function, and appearance testing. They differ meaningfully in long-term heat resistance, UV stability, and chemical resistance, so do not use them to qualify a part for those requirements.

## Alternative processes

- [Injection Molding](https://manufacturingprocesses.org/processes/forming/injection-molding.md): Injection molding forces molten thermoplastic into a steel mold under high pressure, producing high-volume identical plastic parts.
- [Stereolithography (SLA)](https://manufacturingprocesses.org/processes/forming/stereolithography-sla.md): Stereolithography cures liquid photopolymer layer by layer with a scanning UV laser, giving the smoothest surfaces in additive manufacturing.
- [Multi Jet Fusion (MJF)](https://manufacturingprocesses.org/processes/forming/multi-jet-fusion-mjf.md): Multi Jet Fusion prints a fusing agent onto nylon powder and drives it with infrared energy, fusing whole layers at once for high throughput.
- [CNC Machining](https://manufacturingprocesses.org/processes/cutting/cnc-machining.md): CNC machining removes material from solid stock with a programmed cutting tool, producing accurate parts directly from a CAD model with no tooling.

## Related processes

- [Stereolithography (SLA)](https://manufacturingprocesses.org/processes/forming/stereolithography-sla.md): Stereolithography cures liquid photopolymer layer by layer with a scanning UV laser, giving the smoothest surfaces in additive manufacturing.
- [Rapid Prototyping](https://manufacturingprocesses.org/processes/forming/rapid-prototyping.md): Rapid prototyping builds a part layer by layer directly from a CAD model, with no tooling and no minimum order quantity.
- [Reaction Injection Molding](https://manufacturingprocesses.org/processes/forming/reaction-injection-molding.md): Reaction injection molding mixes two liquid reactants that polymerize inside a low-pressure mold, producing large tough polyurethane parts on light tooling.
- [Injection Molding](https://manufacturingprocesses.org/processes/forming/injection-molding.md): Injection molding forces molten thermoplastic into a steel mold under high pressure, producing high-volume identical plastic parts.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/forming/vacuum-casting)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Wood Laminating"
category: "Forming"
subcategory: "Wood"
materials: ["Wood"]
tolerances: "Springback is typically 1–3% of the bend curvature with a rigid adhesive, against 5–10% for a steamed bend. Forms are made slightly overbent and calibrated on the first assembly."
volumes: "1–5,000 pieces per year"
lead_time: "1–3 weeks including form making. In production, cure time per glue-up is several hours at room temperature or minutes with hot pressing or radio-frequency curing."
url: https://manufacturingprocesses.org/processes/forming/wood-laminating
---

# Wood Laminating

Wood laminating glues thin veneers or kerfed strips over a form so the assembly holds a curve permanently once the adhesive has cured.

- **Category**: [Forming](https://manufacturingprocesses.org/processes/forming.md)
- **Family**: Wood
- **Materials**: Wood
- **Typical tolerances**: Springback is typically 1–3% of the bend curvature with a rigid adhesive, against 5–10% for a steamed bend. Forms are made slightly overbent and calibrated on the first assembly.
- **Typical volumes**: 1–5,000 pieces per year
- **Lead time**: 1–3 weeks including form making. In production, cure time per glue-up is several hours at room temperature or minutes with hot pressing or radio-frequency curing.

## Overview

Wood laminating glues a stack of thin laminae — sawn strips or veneers — over a form so the assembly holds the curve permanently once the adhesive cures. Because each lamina is thin, the bending strain it sees is small: a lamina of thickness t bent to radius R experiences a peak strain of about t/2R, so a 1/16 in (1.6 mm) strip on a 6 in (150 mm) radius sees roughly 0.5% strain, comfortably within what most hardwoods tolerate. Stack enough of them and a tight curve becomes achievable in a species that could never be steam bent.

The payoff over [steam bending](/processes/forming/steam-bending) is springback: a properly glued lamination with a rigid adhesive returns roughly 1–3% of its curvature, against 5–10% or more for a steamed bend, and it does not creep back over years. It also works in any species, including those that will not bend at all.

The variants are veneer lamination (0.02–0.12 in / 0.5–3 mm plies, the Eames-chair route), solid wood lamination from thicker sawn strips, and kerfing, where saw cuts across the back of a solid board let it bend at the cost of strength.

## How it works

1. **Size the laminae.** Choose lamina thickness from the target radius. As a working rule, keep thickness at or below about R/50 for species that bend easily and R/100 for stiff or brittle ones, then divide the finished thickness by that figure to get the number of plies.
2. **Prepare the stack.** Laminae are resawn and thicknessed, and — unlike plywood — the grain of every ply runs in the same direction, along the length of the bend, because the object is a structural bent member rather than a dimensionally stable panel. Keep all plies from the same board where appearance matters.
3. **Apply adhesive.** Spread is applied to one or both faces of every glue line. Adhesive choice governs springback: urea-formaldehyde and resorcinol cure rigid and creep very little, epoxy cures rigid and fills gaps, while PVA — even the cross-linking grades — remains slightly thermoplastic and allows the bend to creep open over time. For bent lamination, a rigid-curing adhesive is not a preference but a requirement.
4. **Clamp to the form.** Vacuum bagging applies up to roughly 12 psi (0.8 bar), which is sufficient for thin veneers over a well-fitting form. Thicker solid laminae need mechanical clamping or a press at substantially higher pressure, because the plies must be forced into intimate contact along every glue line.
5. **Cure.** Urea-formaldehyde cures in several hours at room temperature; a hot press or radio-frequency curing brings that down to minutes. Cure fully before removing the clamps — an under-cured glue line is exactly where creep and springback come from.
6. **Release and machine.** The assembly comes off the form close to shape and is then trimmed, profiled, and sanded like any solid wood component.

### Kerfing
Kerfing takes a different approach: parallel saw kerfs are cut across the back of a solid board, leaving a thin skin of perhaps 1/16 in (1.5 mm), and the board is bent so the kerfs close. It is fast and needs no form, but the result is weak and the kerfs must be filled and hidden, so it is used for architectural trim and non-structural curves rather than load-bearing parts.

## Design guidelines

### Lamina thickness sets the radius
Peak bending strain in a lamina is approximately its thickness divided by twice the bend radius. Keep that strain below roughly 1% and the ply bends without cracking. In practice: thickness at or below R/50 for easily bent species, R/100 for stiff or brittle ones.

### Use a rigid adhesive
Urea-formaldehyde is the traditional choice for bent lamination precisely because it cures glass-hard and does not creep. Resorcinol adds water resistance; epoxy adds gap filling. PVA of any grade will let the lamination slowly open, sometimes over months.

### Grain direction
All plies run the same direction in a structural bent lamination — this is not plywood. Cross-banding adds dimensional stability but destroys the bending stiffness and strength along the member.

### Glue line thickness and clamping pressure
Bent laminations fail at starved or gapped glue lines. Vacuum pressure at roughly 12 psi (0.8 bar) is adequate for thin veneers over a smooth form; thicker laminae need real clamping pressure and a form that matches the assembly closely along its whole length.

### Odd ply counts and symmetric stacks
An odd number of plies with a symmetric arrangement keeps internal stresses balanced and reduces twist, particularly on wide laminations.

### Design for springback of 1–3%
Small, but not zero. Make the form slightly tighter than the target radius and calibrate on the first assembly.

### Moisture content
Laminate at the moisture content the finished part will live at, typically 6–10% for interior use. A lamination glued wet and dried in service will distort.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Lamina thickness | R/100 | R/50 | Keeps bending strain near or below 1% |
| Adhesive | Urea-formaldehyde, resorcinol, epoxy | Avoid PVA | Thermoplastic adhesives creep |
| Grain direction | All plies parallel | No cross-banding | Cross-band destroys bending strength |
| Clamping pressure | Press or clamps for solid laminae | 12 psi (0.8 bar) vacuum for veneer | Glue lines must close completely |
| Ply count | Odd, symmetric | — | Balances internal stress, limits twist |
| Springback allowance | 1–3% | — | Form must be slightly overbent |

## Variants

- Kerfing
- Solid Wood Lamination
- Veneer Lamination

## Cost drivers

The form is the tooling, and it is inexpensive: a stack of shaped plywood or MDF ribs, often cut on a [CNC router](/processes/cutting/cnc-machining), plus a caul or vacuum bag. A two-part matched form costs more but clamps more evenly and is worth it in production.

The recurring cost is labor and clamp time. Every glue-up occupies a form for the full cure period, so form count sets throughput exactly as it does in steam bending — with the important difference that a hot press or radio-frequency curing can cut cure time from hours to minutes and turn a form into a production tool rather than a bottleneck.

Material yield is the other factor. Resawing thick stock into thin laminae loses a saw kerf per ply, so a lamination made from many thin plies can consume noticeably more rough lumber than its finished volume suggests. Where appearance matters and plies must be sequence-matched from one board, yield falls further.

Volume breakpoints: bent lamination is economical from one piece to several thousand a year. Above that, molded plywood pressed in matched heated dies is faster, and [composite laminating](/processes/forming/composite-laminating) or [DMC/SMC molding](/processes/forming/dmc-and-smc-molding) takes over where wood is not required.

1. Use the thickest lamina the radius allows — fewer plies means fewer glue lines and less resawing loss.
2. Build several forms; cure time, not glue-up time, limits output.
3. Consider hot pressing or RF curing to turn hours of clamp time into minutes.
4. Sequence-match plies from a single board only on visible edges.
5. Use kerfing for hidden non-structural curves where a full lamination is not justified.

## FAQ

### How thick should each lamina be?

Thin enough that bending strain stays near or below 1%. Peak strain is roughly the lamina thickness divided by twice the bend radius, so a useful rule is thickness at or below R/50 for easily bent species and R/100 for stiff or brittle ones. A 6 in (150 mm) radius therefore wants plies of about 1/16 in (1.6 mm) or less.

### Which adhesive should be used for bent lamination?

One that cures rigid: urea-formaldehyde is the traditional choice, with resorcinol for water resistance and epoxy where gap filling is needed. Avoid PVA of any grade — it stays slightly thermoplastic and lets the lamination creep open over months or years.

### How much springback does a wood lamination have?

Roughly 1–3% of the curvature with a rigid adhesive and a full cure, compared with 5–10% or more for steam bending. Forms are made slightly tighter than the target radius, and the first assembly off a new form is used to calibrate the correction.

### Is laminating better than steam bending?

For repeatability, yes. Laminating works in any species, holds its shape far better, and reaches tighter radii. Steam bending keeps continuous grain through the bend, has no visible glue lines, and needs almost no tooling — which is why it survives for chair parts and boat frames where appearance and grain continuity matter.

### Do the plies in a bent lamination cross like plywood?

No. All plies run parallel along the length of the member. Cross-banding is what gives plywood its dimensional stability as a panel, but it would destroy the bending stiffness and strength of a structural laminated member.

### What is kerfing and when should it be used?

Parallel saw kerfs are cut across the back of a solid board leaving a thin skin, so the board bends as the kerfs close. It is fast and needs no form, but it is weak and the kerfs must be filled and hidden. Use it for architectural trim and hidden curves, not for load-bearing parts.

## Alternative processes

- [Steam Bending](https://manufacturingprocesses.org/processes/forming/steam-bending.md): Steam bending plasticises solid timber with saturated steam so it can be bent around a former and held until it dries into shape.
- [Composite Laminating](https://manufacturingprocesses.org/processes/forming/composite-laminating.md): Composite laminating lays reinforcing fiber into a mold and consolidates it with resin, curing into a stiff, light monocoque part.
- [CNC Machining](https://manufacturingprocesses.org/processes/cutting/cnc-machining.md): CNC machining removes material from solid stock with a programmed cutting tool, producing accurate parts directly from a CAD model with no tooling.
- [Joinery](https://manufacturingprocesses.org/processes/joining/joinery.md): Joinery cuts interlocking geometry into timber so the parts locate and hold each other, with or without adhesive or fasteners.
- [Compression Molding](https://manufacturingprocesses.org/processes/forming/compression-molding.md): Compression molding presses a measured charge of rubber or thermoset into an open heated mold that closes to cure it into its final shape.

## Related processes

- [Steam Bending](https://manufacturingprocesses.org/processes/forming/steam-bending.md): Steam bending plasticises solid timber with saturated steam so it can be bent around a former and held until it dries into shape.
- [3D Thermal Laminating](https://manufacturingprocesses.org/processes/forming/3d-thermal-laminating.md): 3D thermal laminating bonds a decorative foil or veneer to a contoured substrate with heat and vacuum so it wraps around the profiled edges.
- [Joinery](https://manufacturingprocesses.org/processes/joining/joinery.md): Joinery cuts interlocking geometry into timber so the parts locate and hold each other, with or without adhesive or fasteners.
- [Composite Laminating](https://manufacturingprocesses.org/processes/forming/composite-laminating.md): Composite laminating lays reinforcing fiber into a mold and consolidates it with resin, curing into a stiff, light monocoque part.
- [Timber Frame Structures](https://manufacturingprocesses.org/processes/joining/timber-frame-structures.md): Timber frame structures assemble large solid or engineered timber members into a load-bearing frame using cut joints and steel connectors.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/forming/wood-laminating)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: category
name: "Cutting"
processes: 15
url: https://manufacturingprocesses.org/processes/cutting
---

# Cutting

15 manufacturing processes in the cutting family.

Cutting processes remove material to arrive at the finished geometry, by
mechanical shear, by melting and blowing, by erosion, or by chemistry.

Because most cutting processes need little or no dedicated tooling, they
dominate prototyping and low volumes, and they set the tolerance ceiling
for parts that other processes cannot hold.

## Processes

| Process | Tolerances | Typical volumes | Lead time |
| --- | --- | --- | --- |
| [5-Axis CNC Machining](https://manufacturingprocesses.org/processes/cutting/5-axis-cnc-machining.md) | ±0.005 in (±0.13 mm) standard; ±0.001 in (±0.025 mm) precision; single-setup work avoids the 0.002–0.005 in (0.05–0.13 mm) datum shift multi-setup machining carries | 1–500 parts typical; dedicated production cells run higher | 5–15 business days for prototypes; 3–6 weeks for production quantities |
| [CNC Machining](https://manufacturingprocesses.org/processes/cutting/cnc-machining.md) | ±0.005 in (±0.13 mm) standard shop practice, equivalent to ISO 2768-m; ±0.001 in (±0.025 mm) for precision work; ±0.0002 in (±0.005 mm) on ground and precision-turned features. | 1–10,000 parts; unit cost drops steeply from 1 to about 10 pieces, then flattens | 3–10 business days for typical prototype and low-volume work, with 1–3 day expedite widely available. Production runs are scheduled against machine capacity rather than tooling lead time. |
| [CNC Milling](https://manufacturingprocesses.org/processes/cutting/cnc-milling.md) | ±0.005 in (±0.13 mm) standard; ±0.001 in (±0.025 mm) precision; ±0.0005 in (±0.013 mm) on selected features | 1–1,000 parts typical; economical to roughly 10,000 depending on cycle time | 3–10 business days for prototypes; 2–4 weeks for production quantities |
| [CNC Turning](https://manufacturingprocesses.org/processes/cutting/cnc-turning.md) | ±0.005 in (±0.13 mm) standard; ±0.001 in (±0.025 mm) on diameters; ±0.0005 in (±0.013 mm) achievable; concentricity within 0.0005 in (0.013 mm) TIR in a single chucking | 1–10,000 parts; bar-fed production runs well beyond 100,000 | 3–10 business days for prototypes; 2–4 weeks for production quantities |
| [Die Cutting](https://manufacturingprocesses.org/processes/cutting/die-cutting.md) | ±0.010–0.030 in (±0.25–0.75 mm) for steel rule dies; ±0.005–0.015 in (±0.13–0.4 mm) for rotary and hardened punch-and-die tooling; print-to-cut registration typically ±1/32 in (±0.8 mm) | Digital knife 1–500; steel rule 500–500,000; rotary dies 100,000+ | Steel rule dies in 2–5 business days; rotary dies 2–4 weeks; digital knife cutting same day |
| [Electrical Discharge Machining](https://manufacturingprocesses.org/processes/cutting/electrical-discharge-machining.md) | Wire EDM ±0.0005 in (±0.013 mm) commercial, ±0.0001 in (±0.0025 mm) achievable with multiple skim passes; sinker EDM ±0.001 in (±0.025 mm) typical, ±0.0005 in (±0.013 mm) achievable | 1–1,000 parts; standard for tooling, dies and low-volume precision components | 1–3 weeks; sinker work adds electrode fabrication time |
| [Glass Scoring](https://manufacturingprocesses.org/processes/cutting/glass-scoring.md) | ±1/32 in (±0.8 mm) with careful hand or CNC-table work; ±1/16 in (±1.6 mm) is common for architectural glass | 1–100,000+; equally suited to one-off work and automated CNC cutting tables | Minutes per cut; glass shops typically cut to size same-day |
| [Laser Cutting](https://manufacturingprocesses.org/processes/cutting/laser-cutting.md) | ±0.005 in (±0.13 mm) on thin sheet; ±0.010 in (±0.25 mm) on plate and over long dimensions; edge taper roughly 0.5–1° | 1–10,000 parts; no tooling, so single pieces are practical and unit cost falls mainly through nesting | 1–5 business days; same-day is common for simple sheet parts |
| [Photochemical Machining](https://manufacturingprocesses.org/processes/cutting/photochemical-machining.md) | ±10% of material thickness as standard (±0.001 in / ±0.025 mm on 0.010 in / 0.25 mm stock); ±20% of thickness for conservative planning; ±0.0005 in (±0.013 mm) on thin foil | 1–1,000,000+; economical at both extremes because there is no hard tooling | Phototools in 1–3 days; prototypes in 1–2 weeks; production 2–4 weeks |
| [Plasma Cutting](https://manufacturingprocesses.org/processes/cutting/plasma-cutting.md) | Conventional plasma ±0.030–0.060 in (±0.75–1.5 mm); high-definition plasma ±0.010–0.020 in (±0.25–0.5 mm); 1–3° of bevel on the cut face | 1–5,000 parts; no tooling required | 1–5 business days; same-day for simple plate profiles |
| [Punching and Blanking](https://manufacturingprocesses.org/processes/cutting/punching-and-blanking.md) | ±0.003–0.010 in (±0.08–0.25 mm) on punched features; ±0.002 in (±0.05 mm) achievable in a well-maintained progressive die; features punched in one station relate to each other more tightly than features across stations | Turret punching 1–5,000; progressive dies 10,000–10,000,000+ | Turret punching 1–5 business days; hard tooling 4–12 weeks, then production in hours per thousand |
| [Sawing](https://manufacturingprocesses.org/processes/cutting/sawing.md) | Band saw ±0.030 in (±0.75 mm) typical, ±0.010 in (±0.25 mm) with careful setup; cold saw ±0.005–0.010 in (±0.13–0.25 mm); abrasive saw ±0.020 in (±0.5 mm) or looser | 1 to unlimited; bundle cutting makes high volumes efficient | Minutes per cut; same-day service at any metal supplier or fabrication shop |
| [Swiss Screw Machining](https://manufacturingprocesses.org/processes/cutting/swiss-screw-machining.md) | ±0.0005 in (±0.013 mm) routine; ±0.0002 in (±0.005 mm) achievable on diameters; ±0.001 in (±0.025 mm) on lengths | 500–1,000,000+ parts | 2–4 weeks for first articles including setup; production releases in days once the setup is proven |
| [Tube and Profile Laser Cutting](https://manufacturingprocesses.org/processes/cutting/tube-and-profile-laser-cutting.md) | ±0.005 in (±0.13 mm) on feature position within a setup; roughly ±0.020 in (±0.5 mm) cumulative over multi-meter lengths; overall accuracy is also bounded by the mill tolerance of the incoming tube | 10–50,000 parts; bundle loading makes production runs efficient | 3–10 business days; production runs quote by the bar |
| [Water Jet Cutting](https://manufacturingprocesses.org/processes/cutting/water-jet-cutting.md) | ±0.005 in (±0.13 mm) typical on thin material; ±0.003 in (±0.076 mm) achievable; ±0.010–0.020 in (±0.25–0.5 mm) on thick plate; 1–3° of natural taper unless a compensating head is used | 1–5,000 parts; no tooling, so single pieces are routine | 1–5 business days; often same-week for single parts |

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/cutting)*


<!-- ---------------- -->

---
type: process
name: "5-Axis CNC Machining"
category: "Cutting"
subcategory: "Mechanical"
materials: ["Metal", "Plastic", "Composite"]
tolerances: "±0.005 in (±0.13 mm) standard; ±0.001 in (±0.025 mm) precision; single-setup work avoids the 0.002–0.005 in (0.05–0.13 mm) datum shift multi-setup machining carries"
volumes: "1–500 parts typical; dedicated production cells run higher"
lead_time: "5–15 business days for prototypes; 3–6 weeks for production quantities"
url: https://manufacturingprocesses.org/processes/cutting/5-axis-cnc-machining
---

# 5-Axis CNC Machining

5-axis CNC machining tilts and rotates the part or the head so the tool reaches every face and stays short and rigid on complex surfaces.

- **Category**: [Cutting](https://manufacturingprocesses.org/processes/cutting.md)
- **Family**: Mechanical
- **Materials**: Metal, Plastic, Composite
- **Typical tolerances**: ±0.005 in (±0.13 mm) standard; ±0.001 in (±0.025 mm) precision; single-setup work avoids the 0.002–0.005 in (0.05–0.13 mm) datum shift multi-setup machining carries
- **Surface finish**: 125 µin Ra (3.2 µm) as-machined; 32–63 µin (0.8–1.6 µm) with finishing passes; 16 µin (0.4 µm) achievable on contours
- **Typical volumes**: 1–500 parts typical; dedicated production cells run higher
- **Lead time**: 5–15 business days for prototypes; 3–6 weeks for production quantities

## Overview

5-axis CNC machining adds two rotary axes to the three linear ones so the cutter can approach the work from almost any direction. It runs in two distinct modes: **3+2 positional**, where the rotaries index and lock and the machine then cuts exactly like a 3-axis mill, and **simultaneous 5-axis**, where all five axes move together to sweep a tool along a curved surface.

The engineering value is less about exotic geometry than about setup elimination — five faces machined in one fixturing with no datum shift between them — and tool shortening, because tilting the head lets a short, rigid cutter reach into a cavity that a 3-axis machine could only touch with a long, deflecting tool.

Tolerances match 3-axis work at **±0.005 in (±0.13 mm)** standard and **±0.001 in (±0.025 mm)** precision; what improves is the positional relationship between features on different faces. Typical parts: impellers, turbine blades, aerospace brackets, mold cores, orthopedic implants and anything with compound-angle holes.

## How it works

1. **Machine configuration.** A trunnion machine rotates the part on an A/C table and suits small to medium work; a swivel-head machine rotates the spindle on a B axis and suits large or heavy parts; hybrids split the rotation between head and table. The configuration decides the working envelope and how much the part weighs into the accuracy budget.
2. **CAM programming.** Toolpaths are generated with explicit tool-axis control, gouge checking against the full surface model, and collision checking of holder, spindle nose, fixture and machine castings. Full kinematic simulation is standard practice, not an optional step.
3. **Post-processing and kinematics.** The control's tool-center-point management keeps the tool tip on the programmed point while the rotaries move, so the programmer works in part coordinates instead of machine coordinates.
4. **Fixturing.** The part is held on a minimal footprint — a dovetail fixture, tombstone or sacrificial tab of **0.25–0.50 in (6–13 mm)** — so five sides stay open to the tool.
5. **Roughing in 3+2.** Bulk removal is done with the rotaries locked, which is more rigid and lets the machine run 3-axis feed rates.
6. **Simultaneous finishing.** Ball-nose or barrel cutters sweep the contour. Residual cusp height follows **h ≈ s²/(8R)** for stepover *s* and effective cutter radius *R* — so a barrel cutter with a large-radius flank produces the same scallop as a ball nose at several times the stepover.
7. **Probing and inspection.** In-process probing re-establishes the part origin after roughing and checks critical features before the part comes off the machine, since refixturing to inspect defeats the point of single-setup work.

## Design guidelines

### Prefer 3+2 to simultaneous
Any face that is planar or can be reached along a fixed tool axis should be programmed as 3+2. Simultaneous motion is for surfaces where the tool axis must change during the cut — blades, blends, impeller flanks. 3+2 is more rigid, faster and easier to verify.

### Design in the clamping stock
Add a **0.25–0.50 in (6–13 mm)** sacrificial boss or dovetail tab to the model, and note where it may be removed. Without it, the shop has to guess a holding scheme and often ends up adding a setup — the exact cost the process was chosen to avoid.

### Keep tool overhang short
The reason to tilt is rigidity. Keep tool length-to-diameter under **4:1** wherever the geometry allows; deflection scales with the cube of overhang, so an 8:1 tool deflects roughly eight times as much under the same load as a 4:1 tool.

### Internal radii still follow the 3-axis rule
Rotary axes do not make corners sharp. Internal vertical corner radius should be at least **1/3 of the pocket depth**, and the same radius should be reused so one cutter finishes the part.

### Dimension compound-angle features with GD&T
Holes and bosses on compound angles should be located with true position from a clear datum reference frame, not with chained angular dimensions — see [GD&T symbols](/charts/gdt-symbols). Use standard hole and thread sizes from the [drill size chart](/charts/drill-size-chart) and [tap drill chart](/charts/tap-drill-chart), and put a spotface normal to the hole axis wherever a fastener head must seat.

### Specify finish only where it is functional
As-machined contours run 125 µin Ra (3.2 µm), finishing passes give 32–63 µin (0.8–1.6 µm), and 16 µin (0.4 µm) is achievable — but finishing time is set by stepover, so a blanket fine-finish callout across a whole impeller is expensive ([surface finish chart](/charts/surface-finish-chart)).

| Material | Suitability | Watch for |
|---|---|---|
| 7075 / 6061 aluminum | Excellent | Most aerospace 5-axis work; high removal rates |
| Ti-6Al-4V | Common but slow | 100–200 SFM; heat management and tool wear dominate |
| Inconel 718 / 625 | Difficult | Very low speeds, ceramic or coated carbide tooling |
| 17-4 PH, 15-5 PH stainless | Good | Machine in condition A, then age |
| P20 / H13 tool steel | Good | Mold and die cores; often finished by EDM |
| PEEK, Ultem | Excellent | Implant and fixture work; stable and clean-cutting |
| Carbon fiber laminate | Abrasive | Diamond tooling, dust extraction, no coolant contamination |

| Feature | Recommended | Limit | Why |
|---|---|---|---|
| Tool length : diameter | ≤ 4:1 | ~6:1 tilted | Deflection grows with overhang cubed |
| Clamping stock | 0.25–0.50 in (6–13 mm) | None | Something has to hold the part |
| Internal corner radius | ≥ 1/3 pocket depth | Tool radius | Rotary axes do not sharpen corners |
| Finishing stepover | Set by cusp h ≈ s²/(8R) | — | Halving the scallop costs 40% more passes |
| Undercut access | Reachable with a tilted tool | Lollipop cutter needed | Tool body must clear the surface |

## Cost drivers

5-axis time bills at a higher rate than 3-axis, and the programming behind it takes longer because every move has to be collision-checked against the whole machine. The process pays for itself through setup count, not through hourly rate.

**Setups avoided.** A part needing five faces machined takes three or four setups on a 3-axis machine, each with its own fixture, offsets, inspection and datum shift. If one 5-axis setup replaces three 3-axis setups, it usually wins outright — and it removes the 0.002–0.005 in (0.05–0.13 mm) of stack-up those setups introduce.

**Finishing stepover.** On contoured surfaces, cycle time is dominated by the number of finishing passes. Because cusp height goes with the square of stepover, doubling the allowable scallop lets the stepover grow about 40% and removes roughly 30% of the passes.

**Programming and simulation.** A one-off impeller can carry more programming hours than machine hours.

**Fixture design.** Purpose-built dovetail and tombstone fixtures are a real line item on the first order and free on every reorder.

Four ways to take cost out:

1. Model the clamping tab yourself so the shop is not designing workholding on your behalf.
2. Keep surfaces 3+2-machinable wherever function permits; save simultaneous motion for the surfaces that need it.
3. Apply a fine surface finish callout only to sealing, bearing and aerodynamic surfaces.
4. If the part has fewer than three machined faces, quote it as [3-axis milling](/processes/cutting/cnc-milling) instead — 5-axis capability you do not use is still on the invoice.

## FAQ

### What is the difference between 3+2 and simultaneous 5-axis machining?

In 3+2 machining the two rotary axes index to an angle and lock, and the machine then cuts as a rigid 3-axis mill from that orientation. In simultaneous 5-axis all five axes move together so the tool axis changes continuously along the cut. 3+2 is faster, more rigid and easier to verify; simultaneous is required only for surfaces such as impeller flanks and turbine blades where the tool axis must change during the cut.

### Is 5-axis machining more accurate than 3-axis?

The achievable tolerance on any single feature is about the same — ±0.005 in (±0.13 mm) standard and ±0.001 in (±0.025 mm) precision. What improves is the relationship between features on different faces, because they are cut in one fixturing instead of being re-datumed, removing the 0.002–0.005 in (0.05–0.13 mm) of stack-up that each additional setup introduces.

### When is 5-axis machining worth the higher hourly rate?

When it replaces multiple setups or lets a much shorter tool reach a deep feature. A part needing four or five machined faces typically takes three or four 3-axis setups, each with fixturing, offsets and inspection. One 5-axis setup replacing three 3-axis setups usually costs less overall despite the higher machine rate.

### How do I control surface finish on a 5-axis contour?

Finish on a swept surface is set by the residual cusp between passes, approximated by h ≈ s²/(8R) for stepover s and effective cutter radius R. Because the relationship is quadratic, halving the cusp height requires about 40% more passes. Barrel or circle-segment cutters present a much larger effective radius than a ball nose and reach the same cusp height at several times the stepover.

### Do I need to design a fixture for a 5-axis part?

You should at least model the clamping stock. Adding a 0.25–0.50 in (6–13 mm) sacrificial boss or dovetail tab and noting where it can be removed lets the shop hold the part on a minimal footprint with five sides open. Without it, workholding gets improvised and often adds the second setup the process was meant to eliminate.

### Does 5-axis machining allow undercuts?

It allows features that a fixed vertical tool axis cannot reach, which covers many geometries described as undercuts, because the tool can be tilted to approach from the side. True internal undercuts still need a lollipop or T-slot cutter, and the whole tool body — not just the tip — has to clear the surrounding surfaces.

## Alternative processes

- [CNC Milling](https://manufacturingprocesses.org/processes/cutting/cnc-milling.md): CNC milling removes material with a rotating multi-flute cutter moved along programmed paths, producing prismatic parts from solid stock.
- [Electrical Discharge Machining](https://manufacturingprocesses.org/processes/cutting/electrical-discharge-machining.md): EDM erodes conductive metal with controlled electrical sparks across a dielectric gap, cutting hardened material and sharp internal corners.
- [Investment Casting](https://manufacturingprocesses.org/processes/forming/investment-casting.md): Investment casting builds a ceramic shell around a wax pattern, melts the wax out and pours metal into the cavity, giving fine detail in almost any alloy.
- [Direct Metal Laser Sintering (DMLS)](https://manufacturingprocesses.org/processes/forming/direct-metal-laser-sintering-dmls.md): Direct metal laser sintering fuses metal powder layer by layer with a laser, building dense metal parts anchored to a supported build plate.
- [Selective Laser Melting (SLM)](https://manufacturingprocesses.org/processes/forming/selective-laser-melting-slm.md): Selective laser melting fully melts metal powder with a laser under inert gas, producing parts at close to wrought density.

## Related processes

- [CNC Milling](https://manufacturingprocesses.org/processes/cutting/cnc-milling.md): CNC milling removes material with a rotating multi-flute cutter moved along programmed paths, producing prismatic parts from solid stock.
- [CNC Turning](https://manufacturingprocesses.org/processes/cutting/cnc-turning.md): CNC turning rotates the workpiece against a stationary tool, producing round parts with excellent concentricity and short cycle times.
- [CNC Machining](https://manufacturingprocesses.org/processes/cutting/cnc-machining.md): CNC machining removes material from solid stock with a programmed cutting tool, producing accurate parts directly from a CAD model with no tooling.
- [Swiss Screw Machining](https://manufacturingprocesses.org/processes/cutting/swiss-screw-machining.md): Swiss screw machining feeds bar stock through a guide bushing so cutting happens right next to the support, holding tight tolerances on slender parts.
- [Electrical Discharge Machining](https://manufacturingprocesses.org/processes/cutting/electrical-discharge-machining.md): EDM erodes conductive metal with controlled electrical sparks across a dielectric gap, cutting hardened material and sharp internal corners.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/cutting/5-axis-cnc-machining)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "CNC Machining"
category: "Cutting"
subcategory: "Mechanical"
materials: ["Wood", "Plastic", "Composite", "Metal"]
tolerances: "±0.005 in (±0.13 mm) standard shop practice, equivalent to ISO 2768-m; ±0.001 in (±0.025 mm) for precision work; ±0.0002 in (±0.005 mm) on ground and precision-turned features."
volumes: "1–10,000 parts; unit cost drops steeply from 1 to about 10 pieces, then flattens"
lead_time: "3–10 business days for typical prototype and low-volume work, with 1–3 day expedite widely available. Production runs are scheduled against machine capacity rather than tooling lead time."
url: https://manufacturingprocesses.org/processes/cutting/cnc-machining
---

# CNC Machining

CNC machining removes material from solid stock with a programmed cutting tool, producing accurate parts directly from a CAD model with no tooling.

- **Category**: [Cutting](https://manufacturingprocesses.org/processes/cutting.md)
- **Family**: Mechanical
- **Materials**: Wood, Plastic, Composite, Metal
- **Typical tolerances**: ±0.005 in (±0.13 mm) standard shop practice, equivalent to ISO 2768-m; ±0.001 in (±0.025 mm) for precision work; ±0.0002 in (±0.005 mm) on ground and precision-turned features.
- **Surface finish**: Ra 125 µin (3.2 µm) as machined by default; Ra 63 µin (1.6 µm) with a finishing pass and Ra 32 µin (0.8 µm) with light finish cuts. Below Ra 16 µin (0.4 µm) requires grinding, lapping, or polishing.
- **Typical volumes**: 1–10,000 parts; unit cost drops steeply from 1 to about 10 pieces, then flattens
- **Lead time**: 3–10 business days for typical prototype and low-volume work, with 1–3 day expedite widely available. Production runs are scheduled against machine capacity rather than tooling lead time.

## Overview

CNC machining removes material from a solid block or bar with programmed cutting tools, producing finished parts directly from a CAD model with no tooling of any kind. Standard shop practice holds ±0.005 in (±0.13 mm); precision work holds ±0.001 in (±0.025 mm), and ground or precision-turned features reach ±0.0002 in (±0.005 mm) — an order of magnitude tighter than any additive process.

It is the most material-agnostic process in manufacturing. Aluminum, stainless and tool steels, titanium, brass and copper, engineering plastics such as POM, PEEK, and PC, glass and carbon composites, and hardwood and sheet stock all machine on the same machines with different tooling and parameters. Crucially, the part has the full, certified properties of wrought stock — no layer boundaries, no porosity, no sintering shrinkage.

The economics are the mirror image of molding: no tooling cost, high per-part cost, and a unit price that falls with quantity only as programming and fixturing amortize. Typical range is 1 to 10,000 parts, delivered in 3–10 business days.

## How it works

1. **CAM programming.** The solid model is imported, stock and work offsets defined, and toolpaths generated for each operation. Programming is a fixed cost per part number, which is why the first piece is disproportionately expensive and the tenth is not.
2. **Workholding and setup.** The part is clamped in a vise, on a fixture plate, or in a chuck. Every orientation the tool must approach from is a separate *setup*, and each setup adds labor plus a positional stack-up of roughly 0.002–0.005 in (0.05–0.13 mm) between features machined in different setups. This is why 5-axis machines earn their cost: they reach five faces in one setup.
3. **Roughing.** Bulk material is removed at high feed and depth of cut, leaving 0.010–0.020 in (0.25–0.5 mm) of stock. Material removal rate here is what sets cycle time on any part with significant volume to clear.
4. **Semi-finishing and finishing.** Light radial depths at higher surface speed produce the final geometry and surface. Cutting speed varies enormously by material — carbide in 6061 aluminum runs several times the surface speed usable in 304 stainless and roughly five to ten times that usable in Ti-6Al-4V — which is why material choice moves cycle time and price so much.
5. **Hole making.** Drilling for clearance and pilot holes, then reaming or boring where a bore tolerance or finish is specified, then tapping or thread milling.
6. **Secondary setups.** The part is re-fixtured to machine remaining faces, sometimes on soft jaws cut to match the first-operation geometry.
7. **Deburr and inspect.** Edge breaks are a hand or tumbling operation. Dimensional inspection ranges from calipers on general tolerances to CMM reports on toleranced and geometrically controlled features.

## Design guidelines

### Internal corner radii
Every internal vertical corner takes the radius of the tool that cut it — there is no sharp internal corner in a milled pocket. Specify a radius of at least one third the pocket depth, and make it 15–20% larger than the nominal tool radius so the cutter sweeps an arc rather than engaging the full slot width, which chatters and burns tools.

### Pocket depth
Stay within 4× tool diameter for standard tooling; reduced-neck and long-reach tools go to 6–10× at lower feeds and with visible taper. Depth and corner radius are linked — a 0.5 in (12 mm) deep pocket with a 0.125 in (3 mm) corner radius forces a long, thin cutter and costs several times what a 0.25 in (6 mm) radius would.

### Wall thickness
0.031 in (0.8 mm) minimum in metal and 0.060 in (1.5 mm) in plastic. Watch aspect ratio too: a wall taller than about 10× its thickness deflects away from the cutter and finishes thin and wavy.

### Holes and threads
Use standard drill diameters — see the [drill size chart](/charts/drill-size-chart). Depth is limited to roughly 4× diameter with a standard drill and 10× with peck cycles; deeper needs gun drilling. A tapped depth of 1.5× thread diameter develops essentially full strength in similar materials, so deeper is wasted machine time. Use standard sizes from the [thread size chart](/charts/thread-size-chart) and pilots from the [tap drill chart](/charts/tap-drill-chart), and thread mill large, blind, or hard-material threads instead of tapping them.

### Undercuts
Undercuts need T-slot, dovetail, or lollipop cutters. Design them to standard widths — retaining-ring grooves and O-ring glands are standardized for exactly this reason. Non-standard undercuts mean custom tooling.

### Tolerances and datums
Put ISO 2768-m (or a blanket ±0.005 in) in the title block and call out only the features that genuinely need more; every tightened dimension adds inspection time and may add a grinding operation. Where a fit is involved, specify the fit class rather than a symmetric tolerance — the [ISO 286 fits and tolerances chart](/charts/iso-286-fits-tolerances) gives the values, and the [GD&T symbols chart](/charts/gdt-symbols) covers the geometric controls that communicate function better than a bilateral tolerance.

### Surface finish
Ra 125 µin (3.2 µm) is the default machined finish. Ra 63 µin (1.6 µm) costs a finishing pass, Ra 32 µin (0.8 µm) costs light cuts, and below Ra 16 µin (0.4 µm) means grinding, lapping, or polishing — see the [surface finish chart](/charts/surface-finish-chart). Specify a finer finish only where it does work: sealing faces, sliding surfaces, fatigue-critical fillets.

### Non-metals
Abrasive materials — carbon and glass composites, MDF, filled plastics — consume carbide quickly and want diamond-coated or PCD tooling. Unfilled thermoplastics need sharp positive-rake geometry and good chip evacuation, because rubbing generates heat the polymer cannot conduct away and the chip re-welds to the surface.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Internal corner radius | 1/3 of pocket depth | Tool radius + 15% | Full-radius engagement chatters |
| Pocket depth | 4× tool diameter | 10× with special tooling | Tool deflection and chatter |
| Wall thickness, metal | 0.060 in (1.5 mm) | 0.031 in (0.8 mm) | Deflection under cutting load |
| Wall thickness, plastic | 0.080 in (2 mm) | 0.060 in (1.5 mm) | Lower stiffness, heat build-up |
| Hole depth | 4× diameter | 10× with peck drilling | Chip evacuation |
| Tapped depth | 1.5× diameter | 3× diameter | No added strength beyond 1.5× |
| Text | Engraved 0.020 in (0.5 mm) wide | 0.012 in (0.3 mm) | Smallest practical engraving cutter |
| General tolerance | ±0.005 in (ISO 2768-m) | ±0.001 in precision | Every tightened dimension adds cost |

## Cost drivers

Machining is billed in spindle hours plus setup labor plus material, and the three respond to different design decisions.

**Cycle time** follows material removal volume and machinability. Roughing away 80% of a billet costs real money; so does a material that must be cut slowly. Aluminum 6061 is the cheapest common metal to machine, stainless costs meaningfully more, and titanium and nickel superalloys more again — both from slow cutting speeds and from tool consumption.

**Setups** are the step change. Each new orientation adds fixturing and handling labor and introduces a tolerance stack between setups. A part that can be finished in two setups on a 3-axis mill is dramatically cheaper than one needing five, which is the entire economic argument for 5-axis work.

**Tolerance and finish** multiply both. A blanket ±0.001 in can double the price of a part where only two features needed it.

Volume breakpoints: unit cost falls steeply from 1 to about 10 parts as programming and fixturing amortize, then flattens — going from 100 to 1,000 pieces changes the price far less than going from 1 to 10. Against [injection molding](/processes/forming/injection-molding), machining loses somewhere between 100 and 1,000 parts for small plastic components. Against [investment casting](/processes/forming/investment-casting) or [die casting](/processes/forming/die-casting), the crossover is typically a few hundred to a few thousand.

1. Loosen every tolerance that does not do work, and call out the few that do.
2. Design so the part can be finished in as few setups as possible.
3. Increase internal corner radii — it lets the shop use a larger, stiffer, faster tool.
4. Use standard drill, tap, and stock sizes; non-standard means special tooling or extra operations.
5. Choose 6061 over 7075 and either over stainless unless the property is genuinely required.

## FAQ

### What tolerance can CNC machining hold?

±0.005 in (±0.13 mm) is standard shop practice and equivalent to ISO 2768-m. Precision work reaches ±0.001 in (±0.025 mm), and ground or precision-turned features get to ±0.0002 in (±0.005 mm). Cost rises sharply below ±0.001 in, so tighten only the features that need it.

### Why do CNC pockets have rounded internal corners?

Because a rotating cutter cannot produce a sharp internal vertical corner — the corner takes the radius of the tool. Specify a radius of at least one third the pocket depth, and make it 15–20% larger than the nominal tool radius so the cutter sweeps an arc rather than engaging the full slot width.

### How deep can a CNC pocket be?

About 4× the cutting tool diameter with standard tooling. Reduced-neck and long-reach tools extend that to 6–10× at lower feeds and with more chatter and taper. Depth and corner radius are linked: a deep pocket with a small corner radius forces a long, thin, expensive tool.

### How deep should a tapped hole be?

1.5× the thread diameter develops essentially full thread strength when the fastener and part are similar materials. Deeper adds machine time and tap breakage risk without adding strength. In soft materials such as aluminum with a steel fastener, 2× diameter is a reasonable allowance.

### Why does the number of setups matter so much?

Each setup is a re-fixturing operation with its own labor, and features machined in different setups carry a positional stack-up of roughly 0.002–0.005 in (0.05–0.13 mm) between them. Reducing setups cuts both cost and tolerance stack, which is why 5-axis machining is worth its higher hourly rate on complex parts.

### When is CNC machining cheaper than injection molding?

Below roughly 100–1,000 parts for small plastic components, depending on complexity and mold cost. Machining has no tooling charge but a high per-part cost that flattens rather than falls with quantity, while molding has the opposite profile.

## Alternative processes

- [Injection Molding](https://manufacturingprocesses.org/processes/forming/injection-molding.md): Injection molding forces molten thermoplastic into a steel mold under high pressure, producing high-volume identical plastic parts.
- [Investment Casting](https://manufacturingprocesses.org/processes/forming/investment-casting.md): Investment casting builds a ceramic shell around a wax pattern, melts the wax out and pours metal into the cavity, giving fine detail in almost any alloy.
- [Sheet Metal Fabrication](https://manufacturingprocesses.org/processes/forming/sheet-metal-fabrication.md): Sheet metal fabrication cuts, bends and joins flat stock into a finished assembly, typically laser cutting followed by press braking, hardware and welding.
- [Direct Metal Laser Sintering (DMLS)](https://manufacturingprocesses.org/processes/forming/direct-metal-laser-sintering-dmls.md): Direct metal laser sintering fuses metal powder layer by layer with a laser, building dense metal parts anchored to a supported build plate.
- [Metal Injection Molding](https://manufacturingprocesses.org/processes/forming/metal-injection-molding.md): Metal injection molding molds a powder-and-binder feedstock exactly like a plastic, then debinds and sinters it into a dense small metal part.

## Related processes

- [CNC Milling](https://manufacturingprocesses.org/processes/cutting/cnc-milling.md): CNC milling removes material with a rotating multi-flute cutter moved along programmed paths, producing prismatic parts from solid stock.
- [CNC Turning](https://manufacturingprocesses.org/processes/cutting/cnc-turning.md): CNC turning rotates the workpiece against a stationary tool, producing round parts with excellent concentricity and short cycle times.
- [5-Axis CNC Machining](https://manufacturingprocesses.org/processes/cutting/5-axis-cnc-machining.md): 5-axis CNC machining tilts and rotates the part or the head so the tool reaches every face and stays short and rigid on complex surfaces.
- [Swiss Screw Machining](https://manufacturingprocesses.org/processes/cutting/swiss-screw-machining.md): Swiss screw machining feeds bar stock through a guide bushing so cutting happens right next to the support, holding tight tolerances on slender parts.
- [CNC Engraving](https://manufacturingprocesses.org/processes/finishing/cnc-engraving.md): CNC engraving cuts text and artwork into a surface with a small pointed or ball-nose tool following a programmed path.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/cutting/cnc-machining)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "CNC Milling"
category: "Cutting"
subcategory: "Mechanical"
materials: ["Metal", "Plastic", "Wood", "Composite"]
tolerances: "±0.005 in (±0.13 mm) standard; ±0.001 in (±0.025 mm) precision; ±0.0005 in (±0.013 mm) on selected features"
volumes: "1–1,000 parts typical; economical to roughly 10,000 depending on cycle time"
lead_time: "3–10 business days for prototypes; 2–4 weeks for production quantities"
url: https://manufacturingprocesses.org/processes/cutting/cnc-milling
---

# CNC Milling

CNC milling removes material with a rotating multi-flute cutter moved along programmed paths, producing prismatic parts from solid stock.

- **Category**: [Cutting](https://manufacturingprocesses.org/processes/cutting.md)
- **Family**: Mechanical
- **Materials**: Metal, Plastic, Wood, Composite
- **Typical tolerances**: ±0.005 in (±0.13 mm) standard; ±0.001 in (±0.025 mm) precision; ±0.0005 in (±0.013 mm) on selected features
- **Surface finish**: 125 µin Ra (3.2 µm) as-machined; 63 µin (1.6 µm) with a finish pass; 32 µin (0.8 µm) fine; 16 µin (0.4 µm) achievable
- **Typical volumes**: 1–1,000 parts typical; economical to roughly 10,000 depending on cycle time
- **Lead time**: 3–10 business days for prototypes; 2–4 weeks for production quantities

## Overview

CNC milling removes material from solid stock with a rotating multi-flute cutter driven along programmed toolpaths, producing prismatic parts with flat faces, pockets, slots, holes and 3D contours. It is the default process for functional metal and plastic parts in quantities of one to a few thousand, and the only one that goes from CAD model to finished part in aluminum, steel, stainless, titanium or engineering plastics with no tooling at all.

A 3-axis machining center holds **±0.005 in (±0.13 mm)** as a routine shop standard and **±0.001 in (±0.025 mm)** where the drawing calls for it. Surfaces come off the machine near 125 µin Ra (3.2 µm) and can be finished to 32 µin (0.8 µm).

The constraints are geometric rather than material: every feature must be reachable by a rotating tool of finite diameter and finite length, which is why internal vertical corners are always radiused and deep narrow pockets cost more than shallow wide ones.

## How it works

1. **CAM programming.** The solid model is imported, stock defined, and toolpaths generated against a tool library, work offsets and cutting parameters. Programming is real cost on a one-off part and near zero amortized across a run.
2. **Workholding and setup.** The blank is clamped in a vise, on soft jaws, or in a dedicated fixture. Each face that must be machined is a *setup*, and every setup after the first re-establishes datums — budget roughly 0.002–0.005 in (0.05–0.13 mm) of positional stack-up between features cut in different setups.
3. **Zeroing.** An edge finder or touch probe sets the part origin and each tool's length offset is measured. On probed machines this takes seconds; done manually, minutes per tool.
4. **Roughing.** Bulk material comes off with adaptive or trochoidal paths using a small radial engagement — roughly 5–15% of cutter diameter — at full axial depth, keeping chip load constant and tool temperature down. Roughing leaves 0.010–0.020 in (0.25–0.50 mm) of stock on finished surfaces.
5. **Finish passes.** Radial depths of 0.002–0.010 in (0.05–0.25 mm) remove the roughing stock and set final tolerance and finish. This is where the tolerance on the drawing turns into cycle time.
6. **Hole making.** Drilling to standard sizes, peck cycles beyond about 4× diameter of depth, reaming or boring where the ID tolerance is tighter than a drill can hold, and tapping or thread milling.
7. **Deburring and inspection.** Brush or manual deburring, then calipers and gage pins for open tolerances and a CMM for GD&T callouts.

Typical carbide surface speeds show where the hours go: aluminum runs at 500–1,000 SFM (150–300 m/min), carbon and alloy steel at 250–450 SFM (75–140 m/min), 304/316 stainless at 150–300 SFM (45–90 m/min), and Ti-6Al-4V at 100–200 SFM (30–60 m/min). Titanium is not difficult to cut so much as slow to cut, and the machine bills by the hour.

## Design guidelines

### Internal corner radii
A rotating tool cannot cut a sharp internal vertical corner. Specify a corner radius of at least **1/3 the pocket depth**, and make it slightly larger than a standard tool radius — 0.135 in rather than 0.125 in for a 1/4 in cutter — because a corner matching the tool exactly wraps the cutter in full engagement and chatters.

### Pocket depth and tool reach
Keep pocket depth under **4× the diameter of the tool the geometry forces you to use**. Reduced-neck and long-reach tooling extends that to roughly 6–10× at lower feed rates, since deflection grows with the cube of overhang. Allow **0.020 in (0.5 mm)** of floor radius unless a sharp floor is functionally required.

### Wall thickness
Minimum **0.030 in (0.76 mm)** in metal and **0.060 in (1.5 mm)** in plastic; 0.060 in and 0.125 in (3 mm) respectively are the thicknesses that machine cleanly. Keep unsupported wall height under roughly 10× wall thickness.

### Holes and threads
Design around stocked drill sizes rather than arbitrary diameters ([drill size chart](/charts/drill-size-chart)). Standard hole depth is 4× diameter; deeper needs peck cycles or through-coolant tooling. Threads need only **1× diameter of full thread in steel and 1.5–2× in aluminum and plastics** — 3× adds tapping time and tap-breakage risk without adding strength ([tap drill chart](/charts/tap-drill-chart)).

### Tolerances and datums
Default the drawing to ISO 2768-m and tighten only the dimensions carrying a fit or an interface. Put bearing bores, dowel holes and shaft fits on an [ISO 286 fit class](/charts/iso-286-fits-tolerances), and express position and orientation with [GD&T callouts](/charts/gdt-symbols) rather than chained ± dimensions.

### Surface finish
125 µin Ra (3.2 µm) as-machined, 63 µin (1.6 µm) with a finish pass, 32 µin (0.8 µm) fine, and 16 µin (0.4 µm) achievable with light passes; smoother than that belongs to grinding ([surface finish chart](/charts/surface-finish-chart)).

| Material | Machinability | Watch for |
|---|---|---|
| 6061 / 7075 aluminum | Excellent | Highest removal rates, lowest cost per part |
| Brass C360 | Excellent | Free-cutting, holds fine detail |
| 1018 / 4140 steel | Good | Heavier cutting forces, slower than aluminum |
| 303 / 304 / 316 stainless | Moderate | Work hardens; constant feed, no dwelling |
| Ti-6Al-4V | Difficult | Low speed, high heat, flood coolant |
| Inconel 718 | Difficult | Rapid tool wear drives cost per part |
| Acetal, PEEK | Excellent | Dimensionally stable, clean chips |
| ABS / polycarbonate | Moderate | Melts and smears; single-flute tools, air blast |
| Carbon fiber / G10 | Abrasive | Diamond tooling and dust extraction |

| Feature | Recommended | Limit | Why |
|---|---|---|---|
| Internal corner radius | ≥ 1/3 pocket depth | Tool radius | Round tools cut round corners |
| Pocket depth | ≤ 4× tool diameter | ~10× long-reach | Deflection and chatter |
| Wall thickness, metal | 0.060 in (1.5 mm) | 0.030 in (0.76 mm) | Walls deflect under cutting load |
| Wall thickness, plastic | 0.125 in (3 mm) | 0.060 in (1.5 mm) | Low stiffness, heat softening |
| Hole depth | ≤ 4× diameter | ~10× with pecking | Chip evacuation |
| Thread depth | 1–2× diameter | 3× diameter | Extra engagement adds no strength |
| Engraved detail | ≥ 0.020 in (0.5 mm) | Smallest stocked cutter | Small tools break and run slowly |

## Cost drivers

Cost is machine hours plus setup plus material, and machine hours are set by how much metal comes off and how well finished the part must be when cutting stops.

**Setups dominate at low volume.** Each reorientation adds fixturing, offsets, programming and inspection. Going from three setups to two usually saves more than loosening a tolerance.

**Removed volume drives cycle time.** A part machined from a 4 × 4 × 2 in block that finishes at 15% of the blank weight spends most of its cycle roughing; buy stock close to the finished envelope.

**Tolerance and finish are multipliers.** Moving from ±0.005 in to ±0.001 in means slower finish passes, in-process probing, controlled inspection and higher scrap risk. Across twenty dimensions it can double the price.

**Tooling and volume.** Cutters under 1/8 in (3 mm), long-reach tools and custom form tools run slowly and break. Programming and first-article inspection are one-time charges — noise at 50 parts, most of the invoice at one.

Five ways to take cost out:

1. Use one internal corner radius throughout so a single cutter finishes the part.
2. Keep every pocket shallower than 4× the diameter of the tool it forces.
3. Pick holes and threads from the [drill size chart](/charts/drill-size-chart) and [tap drill chart](/charts/tap-drill-chart).
4. Leave the drawing at ISO 2768-m and tighten only the interfaces that matter.
5. If the part truly needs four or more machined faces, quote it as [5-axis](/processes/cutting/5-axis-cnc-machining) — one setup often beats three cheaper ones.

## FAQ

### What tolerance can CNC milling hold?

±0.005 in (±0.13 mm) is the routine shop standard and costs nothing extra, ±0.001 in (±0.025 mm) is normal work on selected features, and ±0.0005 in (±0.013 mm) is achievable with finish passes, probing and controlled inspection conditions. Allow an extra 0.002–0.005 in (0.05–0.13 mm) of stack-up between features cut in different setups.

### What is the minimum wall thickness for CNC milling?

0.030 in (0.76 mm) in metal and 0.060 in (1.5 mm) in plastic are workable minimums, but 0.060 in (1.5 mm) and 0.125 in (3 mm) machine cleanly without chatter. Keep unsupported wall height under about 10× the wall thickness — deflection under cutting load, not strength, is what limits thin walls.

### Why do CNC milled parts always have rounded internal corners?

The cutter is round, so the corner it leaves is round, with a radius equal to the tool radius. A tool small enough for a tiny radius is also short and slow. Specify a corner radius of at least one third of the pocket depth and reuse the same radius throughout so one tool can finish the part.

### How deep can a CNC milled pocket be?

About 4× the cutter diameter with standard tooling, extending to roughly 6–10× with reduced-neck and long-reach tools at reduced feed rates, because deflection grows with the cube of tool overhang. Deeper features are better handled by machining from both sides or by sinker EDM.

### What surface finish does CNC milling produce?

About 125 µin Ra (3.2 µm) as-machined. A dedicated finish pass reaches 63 µin (1.6 µm), fine finishing 32 µin (0.8 µm), and 16 µin (0.4 µm) is achievable with light passes and sharp tooling. Below that, grinding or polishing is the appropriate process.

### What materials can be CNC milled?

Aluminum, brass, copper, carbon and alloy steel, stainless, titanium, Inconel, magnesium and most engineering plastics. Machinability rather than possibility decides cost: aluminum runs at 500–1,000 SFM (150–300 m/min) while Ti-6Al-4V runs at 100–200 SFM (30–60 m/min), so the titanium version of the same part occupies the machine far longer.

## Alternative processes

- [Die Casting](https://manufacturingprocesses.org/processes/forming/die-casting.md): Die casting injects molten aluminum, zinc or magnesium into a hardened steel die under high pressure, producing thin-walled net-shape metal parts.
- [Investment Casting](https://manufacturingprocesses.org/processes/forming/investment-casting.md): Investment casting builds a ceramic shell around a wax pattern, melts the wax out and pours metal into the cavity, giving fine detail in almost any alloy.
- [Metal Injection Molding](https://manufacturingprocesses.org/processes/forming/metal-injection-molding.md): Metal injection molding molds a powder-and-binder feedstock exactly like a plastic, then debinds and sinters it into a dense small metal part.
- [Direct Metal Laser Sintering (DMLS)](https://manufacturingprocesses.org/processes/forming/direct-metal-laser-sintering-dmls.md): Direct metal laser sintering fuses metal powder layer by layer with a laser, building dense metal parts anchored to a supported build plate.
- [Sheet Metal Fabrication](https://manufacturingprocesses.org/processes/forming/sheet-metal-fabrication.md): Sheet metal fabrication cuts, bends and joins flat stock into a finished assembly, typically laser cutting followed by press braking, hardware and welding.

## Related processes

- [CNC Turning](https://manufacturingprocesses.org/processes/cutting/cnc-turning.md): CNC turning rotates the workpiece against a stationary tool, producing round parts with excellent concentricity and short cycle times.
- [5-Axis CNC Machining](https://manufacturingprocesses.org/processes/cutting/5-axis-cnc-machining.md): 5-axis CNC machining tilts and rotates the part or the head so the tool reaches every face and stays short and rigid on complex surfaces.
- [CNC Machining](https://manufacturingprocesses.org/processes/cutting/cnc-machining.md): CNC machining removes material from solid stock with a programmed cutting tool, producing accurate parts directly from a CAD model with no tooling.
- [Swiss Screw Machining](https://manufacturingprocesses.org/processes/cutting/swiss-screw-machining.md): Swiss screw machining feeds bar stock through a guide bushing so cutting happens right next to the support, holding tight tolerances on slender parts.
- [Electrical Discharge Machining](https://manufacturingprocesses.org/processes/cutting/electrical-discharge-machining.md): EDM erodes conductive metal with controlled electrical sparks across a dielectric gap, cutting hardened material and sharp internal corners.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/cutting/cnc-milling)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "CNC Turning"
category: "Cutting"
subcategory: "Mechanical"
materials: ["Metal", "Plastic"]
tolerances: "±0.005 in (±0.13 mm) standard; ±0.001 in (±0.025 mm) on diameters; ±0.0005 in (±0.013 mm) achievable; concentricity within 0.0005 in (0.013 mm) TIR in a single chucking"
volumes: "1–10,000 parts; bar-fed production runs well beyond 100,000"
lead_time: "3–10 business days for prototypes; 2–4 weeks for production quantities"
url: https://manufacturingprocesses.org/processes/cutting/cnc-turning
---

# CNC Turning

CNC turning rotates the workpiece against a stationary tool, producing round parts with excellent concentricity and short cycle times.

- **Category**: [Cutting](https://manufacturingprocesses.org/processes/cutting.md)
- **Family**: Mechanical
- **Materials**: Metal, Plastic
- **Typical tolerances**: ±0.005 in (±0.13 mm) standard; ±0.001 in (±0.025 mm) on diameters; ±0.0005 in (±0.013 mm) achievable; concentricity within 0.0005 in (0.013 mm) TIR in a single chucking
- **Surface finish**: 125 µin Ra (3.2 µm) as-turned; 32 µin (0.8 µm) with a finish pass; 16 µin (0.4 µm) achievable
- **Typical volumes**: 1–10,000 parts; bar-fed production runs well beyond 100,000
- **Lead time**: 3–10 business days for prototypes; 2–4 weeks for production quantities

## Overview

CNC turning spins the workpiece in a spindle while a single-point tool traverses along and across it, generating everything round: shafts, pins, bushings, spacers, hydraulic fittings, connector bodies and threaded studs. Bar-fed lathes take stock roughly 0.25–3 in (6–75 mm) in diameter and cut a complete part in seconds to a few minutes; chuck-fed machines take castings, forgings and slugs a foot or more across.

Diameters hold **±0.001 in (±0.025 mm)** as normal work, and features cut in the same chucking are concentric within about **0.0005 in (0.013 mm) TIR** — the strongest single reason to turn a part rather than mill it round.

Adding live tooling, a Y axis and a sub-spindle makes the lathe a mill-turn center that finishes cross-holes, flats and back-end features without a second machine, eliminating both the setup cost and the tolerance stack a transfer would introduce.

## How it works

1. **Stock feeding.** A bar feeder pushes stock through the spindle for parts within bar capacity; larger or non-round blanks are held in a chuck or collet. Bar work is what makes turning cheap per part — the machine reloads itself.
2. **Facing and rough turning.** The end is faced and the OD reduced in passes of roughly 0.050–0.150 in (1.3–3.8 mm) depth of cut at feeds around 0.008–0.015 in/rev in steel. Constant surface speed control keeps SFM steady as the diameter shrinks.
3. **Finish turning.** Depth of cut drops to 0.005–0.020 in (0.13–0.50 mm) and feed to 0.002–0.006 in/rev. Finish is governed mostly by feed and tool nose radius — theoretical roughness approximates **Ra ≈ f²/(32r)** — so halving the feed cuts roughness roughly fourfold.
4. **Centerline hole making.** Center drill, then drill, then ream or bore. Boring is what holds an ID to a tight tolerance and true roundness; a drill alone will not.
5. **Grooving, threading and knurling.** Retaining-ring grooves and thread reliefs are plunged with form tools; threads are single-pointed over multiple passes or cut with a die head.
6. **Live-tool operations.** Cross-holes, flats and keyways are cut with driven tools while the main spindle indexes to an angular position, avoiding a separate milling setup.
7. **Part-off and back working.** A parting blade roughly 0.080–0.125 in (2–3 mm) wide separates the part. On a sub-spindle machine it is caught, its back end faced, chamfered and drilled, and the finished piece drops out complete.

## Design guidelines

### Length-to-diameter ratio
Held in the chuck alone, keep unsupported length under **3× diameter**; with a tailstock center, 8:1 is comfortable. Beyond that, deflection and chatter take over and the part needs a steady rest — or a [Swiss machine](/processes/cutting/swiss-screw-machining), which supports the bar right at the cut and handles 20:1 routinely.

### Wall thickness on turned tubes
**0.020 in (0.5 mm)** is the metal minimum, but below about 0.040 in (1 mm) the chuck jaws themselves distort the part while it is cut. Soft jaws, an expanding mandrel or reduced clamping pressure are what make thin-wall work possible, and roundness rather than diameter becomes the limiting characteristic.

### Thread reliefs and shoulder undercuts
Give every external thread a runout groove at least **1.5× the thread pitch** wide, and every shoulder that must seat flat an undercut. Without one, the incomplete thread and the tool-nose fillet hold the mating part off its face.

### Corner radii and chamfers
Every shoulder carries the insert's nose radius, typically **0.008–0.031 in (0.2–0.8 mm)** — draw it rather than implying a sharp corner. Put a 0.010–0.030 in × 45° chamfer on every entering edge; it costs nothing on the lathe and removes a deburring operation.

### Holes, threads and fits
Centerline holes should use stocked drill sizes ([drill size chart](/charts/drill-size-chart)) and tapped bores standard tap drills ([tap drill chart](/charts/tap-drill-chart)). Use standard thread sizes from the [thread size chart](/charts/thread-size-chart), since a non-standard pitch needs a custom insert. For press, slip and running fits, call out an [ISO 286 fit class](/charts/iso-286-fits-tolerances) rather than inventing a bilateral tolerance.

### Tolerance and finish
±0.005 in (±0.13 mm) is the default, ±0.001 in (±0.025 mm) on diameters is routine, and ±0.0005 in (±0.013 mm) needs a dedicated finish pass and stable temperature. Finish runs 125 µin Ra (3.2 µm) as-turned and 32 µin (0.8 µm) after a finish pass; below 8 µin (0.2 µm), grind or roller-burnish ([surface finish chart](/charts/surface-finish-chart)).

| Material | Turnability | Watch for |
|---|---|---|
| Brass C360 | Excellent | The free-cutting benchmark: best finish, longest tool life |
| 12L14 / 1215 steel | Excellent | Chips break cleanly; ideal for bar work |
| 6061 / 2011 aluminum | Excellent | 2011 chips better than 6061 in high-volume bar work |
| 303 stainless | Good | Use over 304 wherever corrosion requirements allow |
| 304 / 316 stainless | Moderate | Work hardens; long stringy chips |
| 4140 / 4340 alloy steel | Good | Turn before hardening; hard turning needs CBN |
| Ti-6Al-4V | Difficult | Low speeds, flood coolant, fire risk from fine chips |
| Acetal, PEEK, PTFE | Excellent | Sharp positive tooling; PTFE creeps after cutting |
| Nylon | Moderate | Absorbs moisture, so diameters drift after machining |

| Feature | Recommended | Limit | Why |
|---|---|---|---|
| L:D, chuck only | ≤ 3:1 | ~4:1 | Deflection and chatter |
| L:D, with tailstock | ≤ 8:1 | ~10:1 | Center supports the free end |
| Wall thickness | 0.040 in (1 mm) | 0.020 in (0.5 mm) | Clamping force distorts thin tubes |
| Shoulder radius | 0.015 in (0.4 mm) | 0.008 in (0.2 mm) | Sharper inserts chip and wear fast |
| Thread relief width | ≥ 1.5× pitch | 1× pitch | Tool must run out cleanly |
| Bore depth | ≤ 4× bore diameter | ~6× with carbide bar | Bar deflection scales with length cubed |
| Entering chamfer | 0.020 in × 45° | — | Removes burrs, eases assembly |

## Cost drivers

Turning is priced on cycle time, and cycle time on a bar-fed machine is short — which is why turned parts are usually the cheapest machined parts you can buy.

**Bar capacity is the first cost cliff.** A part that fits through the spindle runs from bar with automatic reload. One diameter larger and it becomes chuck work: manual loading, one part per cycle, several times the cost per piece.

**Tools and operations.** Each groove, thread, cross-hole and back-side feature is another turret position and more seconds per part. A design that fits inside a standard turret load runs unattended.

**Secondary operations.** A cross-hole that could have been cut with live tooling but instead goes to a mill adds a setup, a fixture and a tolerance stack.

**Tolerance and material.** Diameters at ±0.001 in (±0.025 mm) come free with a finish pass; below ±0.0005 in (±0.013 mm) you pay for thermal stability and in-process gaging. Free-machining bar costs more per pound and repays it in cycle time and tool life.

Four ways to take cost out:

1. Keep the largest diameter inside standard bar capacity so the job runs bar-fed.
2. Consolidate features onto one end so the part finishes without a second chucking.
3. Choose 303 over 304 and 12L14 over 1018 wherever corrosion and strength allow.
4. Call out [ISO 286 fits](/charts/iso-286-fits-tolerances) only on diameters that mate, and leave the rest at ISO 2768-m.

## FAQ

### What tolerance can CNC turning hold?

±0.005 in (±0.13 mm) as a default, ±0.001 in (±0.025 mm) on diameters as normal precision work, and ±0.0005 in (±0.013 mm) with a dedicated finish pass and stable temperature. Features cut in the same chucking hold concentricity within about 0.0005 in (0.013 mm) TIR.

### How long can a turned part be relative to its diameter?

About 3× diameter held in the chuck alone and 8× with a tailstock center. Past that the part deflects away from the tool and chatters. Swiss screw machining supports the bar at the cutting point with a guide bushing and routinely runs 20:1 and beyond in the same material.

### What surface finish does CNC turning produce?

125 µin Ra (3.2 µm) as-turned, 32 µin (0.8 µm) after a finish pass, and 16 µin (0.4 µm) achievable. Finish comes mostly from feed rate and nose radius, approximated by Ra ≈ f²/(32r), so a lighter feed or a larger nose radius improves it directly. Below 8 µin (0.2 µm), grinding or roller burnishing is the right process.

### What is the minimum wall thickness for a turned tube?

0.020 in (0.5 mm) is the practical floor in metal, but below about 0.040 in (1 mm) the chuck jaws distort the part while it is being cut, so roundness rather than diameter becomes the limiting characteristic. Soft jaws, an expanding mandrel or reduced clamping pressure are what make thin-wall turning work.

### Do I need a thread relief on a turned thread?

Yes for any thread that runs into a shoulder. A relief groove at least 1.5× the thread pitch wide lets the threading tool run out cleanly and lets the mating part seat against the shoulder. Without one, the incomplete thread and the tool-nose fillet hold the assembly off its face.

### Can CNC turning cut cross-holes and flats?

Yes, on a machine with live tooling. Driven tools cut cross-holes, flats and keyways while the main spindle indexes to an angular position, so the part comes off complete. Without live tooling those features become a second setup on a mill, adding cost and a positional tolerance stack.

## Alternative processes

- [Swaging](https://manufacturingprocesses.org/processes/forming/swaging.md): Swaging hammers or presses the end of a tube or bar inward against a die to reduce or taper its diameter without removing any material.
- [Metal Injection Molding](https://manufacturingprocesses.org/processes/forming/metal-injection-molding.md): Metal injection molding molds a powder-and-binder feedstock exactly like a plastic, then debinds and sinters it into a dense small metal part.
- [Powder Metallurgy (Press and Sinter)](https://manufacturingprocesses.org/processes/forming/powder-metallurgy-press-and-sinter.md): Press-and-sinter powder metallurgy compacts metal powder in a rigid die and sinters it below its melting point into a finished net-shape part.
- [Metal Extrusion](https://manufacturingprocesses.org/processes/forming/metal-extrusion.md): Metal extrusion pushes heated billet through a shaped die, producing continuous aluminum or copper profile of constant cross-section.
- [Investment Casting](https://manufacturingprocesses.org/processes/forming/investment-casting.md): Investment casting builds a ceramic shell around a wax pattern, melts the wax out and pours metal into the cavity, giving fine detail in almost any alloy.

## Related processes

- [CNC Milling](https://manufacturingprocesses.org/processes/cutting/cnc-milling.md): CNC milling removes material with a rotating multi-flute cutter moved along programmed paths, producing prismatic parts from solid stock.
- [Swiss Screw Machining](https://manufacturingprocesses.org/processes/cutting/swiss-screw-machining.md): Swiss screw machining feeds bar stock through a guide bushing so cutting happens right next to the support, holding tight tolerances on slender parts.
- [5-Axis CNC Machining](https://manufacturingprocesses.org/processes/cutting/5-axis-cnc-machining.md): 5-axis CNC machining tilts and rotates the part or the head so the tool reaches every face and stays short and rigid on complex surfaces.
- [CNC Machining](https://manufacturingprocesses.org/processes/cutting/cnc-machining.md): CNC machining removes material from solid stock with a programmed cutting tool, producing accurate parts directly from a CAD model with no tooling.
- [Sawing](https://manufacturingprocesses.org/processes/cutting/sawing.md): Sawing separates stock with a toothed blade — band, circular or cold saw — and is the default first operation on bar, tube and plate.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/cutting/cnc-turning)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Die Cutting"
category: "Cutting"
subcategory: "Mechanical"
materials: ["Plastic", "Composite", "Wood"]
tolerances: "±0.010–0.030 in (±0.25–0.75 mm) for steel rule dies; ±0.005–0.015 in (±0.13–0.4 mm) for rotary and hardened punch-and-die tooling; print-to-cut registration typically ±1/32 in (±0.8 mm)"
volumes: "Digital knife 1–500; steel rule 500–500,000; rotary dies 100,000+"
lead_time: "Steel rule dies in 2–5 business days; rotary dies 2–4 weeks; digital knife cutting same day"
url: https://manufacturingprocesses.org/processes/cutting/die-cutting
---

# Die Cutting

Die cutting presses a shaped steel rule or rotary die through sheet stock, cutting flat outlines in paper, board, foam, film and textile.

- **Category**: [Cutting](https://manufacturingprocesses.org/processes/cutting.md)
- **Family**: Mechanical
- **Materials**: Plastic, Composite, Wood
- **Typical tolerances**: ±0.010–0.030 in (±0.25–0.75 mm) for steel rule dies; ±0.005–0.015 in (±0.13–0.4 mm) for rotary and hardened punch-and-die tooling; print-to-cut registration typically ±1/32 in (±0.8 mm)
- **Typical volumes**: Digital knife 1–500; steel rule 500–500,000; rotary dies 100,000+
- **Lead time**: Steel rule dies in 2–5 business days; rotary dies 2–4 weeks; digital knife cutting same day

## Overview

Die cutting presses a shaped blade through sheet stock to cut a flat outline, and covers three tooling families that behave quite differently. **Steel rule dies** — hardened rule bent to shape and set into a plywood board — cut paper, board, corrugated, foam, rubber, gasket material, film and textile on a flatbed or clamshell press. **Rotary dies**, either flexible plates or solid engraved cylinders, do the same work continuously from a web at high speed. **Digital knife cutting** does it with a drag or oscillating blade and no tooling at all.

Standard 2-point steel rule is **0.028 in (0.7 mm)** thick, which sets the cut width, and US die height is conventionally 0.918 in. Tolerance runs **±0.010–0.030 in (±0.25–0.75 mm)** for steel rule and tighter for rotary and hardened tooling.

Tooling is cheap and arrives in days rather than weeks, which is why die cutting owns packaging, gaskets and converted materials.

## How it works

1. **Die construction.** A die board, typically 3/4 in (18 mm) plywood, is laser-cut with a narrow kerf following the part outline. Steel rule is bent to shape and driven into that kerf, and ejection rubber is bonded alongside every rule to push the material back off the blade after the cut.
2. **Cutting rule versus creasing rule.** Cutting rule has a sharpened edge and severs the material. Creasing rule is blunt and rounded, and works against a matrix channel on the opposite platen to compress a fold line into board without cutting it — the reason a folding carton bends where it is supposed to.
3. **Press selection.** Flatbed platen presses suit short runs, heavy board and thick gasket stock; rotary presses run continuously from a web at high speed for labels, cartons and adhesive-backed parts.
4. **Makeready.** Packing is built up under the die until every rule cuts to the same depth across the whole sheet. This is the operator skill in the process and the reason a first-off sheet is not the same as the hundredth.
5. **The stroke.** The rule penetrates against a hardened cutting plate, or against the anvil roll in a rotary press. Rule that cuts too deep destroys the cutting plate; too shallow and parts do not separate.
6. **Nicks and bridges.** Small gaps left in the rule leave the part attached to the surrounding sheet so it can travel through the press without falling out, then break free at stripping.
7. **Stripping and blanking.** Waste is pushed out of the sheet and parts are separated, either in-line on a production press or by hand on short runs.

## Design guidelines

### Plan nicks where they will not show
Every nick leaves a small witness on the part edge. Place them on non-visible edges, away from sealing surfaces and away from corners, and tell the die maker where they are acceptable rather than leaving it to chance.

### Minimum feature size
Steel rule dies cut holes and internal features down to about **0.125 in (3 mm)**, and rule-to-rule spacing should also stay at or above **0.125 in (3 mm)** so the die board retains strength between kerfs. Smaller holes need a punch inserted into the die rather than bent rule.

### Design creases with the grain
Paperboard folds cleanly parallel to the grain direction and cracks when folded across it. Lay out cartons so the primary folds run with the grain, and expect the crease geometry — matrix channel width and depth — to be selected from the board caliper and rule thickness by the die maker.

### Allow for registration on printed work
Print-to-cut registration is typically around **±1/32 in (±0.8 mm)**. Add bleed to any printed area that runs to a cut edge, and keep critical graphics away from the trim.

### Match the tooling family to the volume
Digital knife cutting has no tooling at all and suits samples and short runs. A steel rule die pays for itself over a few hundred to a few thousand pieces. A rotary die is a larger investment justified by continuous web production in the hundreds of thousands.

### Respect material thickness limits
Steel rule handles most converted materials up to roughly **0.125 in (3 mm)**; thicker foam and rubber need taller rule and more press tonnage, and compressible materials cut with a slight edge crush that has to be acceptable.

| Material | Die cutting suitability | Notes |
|---|---|---|
| Paperboard, folding carton | Excellent | Cut and crease in one stroke; mind grain direction |
| Corrugated board | Excellent | Rotary dies for volume, flatbed for heavy flute |
| Foam (PE, PU, EVA) | Excellent | Compressible edge; taller rule for thick stock |
| Rubber and gasket sheet | Excellent | Cork, neoprene, silicone; a core gasket process |
| Adhesive-backed film and label stock | Excellent | Kiss-cutting through the face only, leaving the liner intact |
| Thin plastic film and sheet | Good | Brittle films crack; warmer stock cuts better |
| Textiles, felt, leather | Good | Fraying is the design constraint, not the cut |
| Sheet metal | Not applicable | Use [punching and blanking](/processes/cutting/punching-and-blanking) |

| Feature | Recommended | Limit | Why |
|---|---|---|---|
| Material thickness | ≤ 0.125 in (3 mm) | Taller rule for thick foam | Rule height and press tonnage |
| Minimum hole | 0.25 in (6 mm) | 0.125 in (3 mm) | Bent rule cannot hold a smaller shape |
| Rule-to-rule spacing | ≥ 0.25 in (6 mm) | 0.125 in (3 mm) | Die board strength between kerfs |
| Cut width (kerf) | 0.028 in (0.7 mm) | Rule thickness | 2-point rule is standard |
| Print-to-cut registration | ±1/32 in (±0.8 mm) | — | Add bleed on printed edges |
| Corner radius | ≥ 0.06 in (1.5 mm) | Sharp | Rule cannot bend to a sharp point cleanly |

## Cost drivers

Die cutting is priced as cheap tooling plus fast press time, so the questions are how many tooling families you need and how well the part nests.

**Tooling family.** A steel rule die is inexpensive and delivered in days. A rotary die costs considerably more and takes weeks, but runs continuously. Digital knife cutting eliminates tooling entirely at a much lower throughput.

**Sheet yield.** Cost per part tracks how many parts fit on a sheet or across a web width. A part sized to fit the standard sheet or web with minimal trim can be dramatically cheaper than one that wastes a strip on every pass.

**Makeready.** Setting the die and building packing so every rule cuts evenly is a fixed cost per run, which is why short runs carry a high per-part price and long runs do not.

**Material.** On packaging and gasket work, the substrate is frequently the largest single line item, so board caliper and grade choice matters more than the cutting.

**Secondary operations.** Stripping waste, blanking, gluing and folding are separate steps priced separately.

Four ways to take cost out:

1. Size the part to the standard sheet or web width before finalizing dimensions.
2. Use digital knife cutting for prototypes and design changes; commit to a die once the geometry is frozen.
3. Combine multiple part numbers onto one die where they share material and caliper.
4. Keep internal features at 0.125 in (3 mm) and above so bent rule can make them without inserted punches.

## FAQ

### What is the difference between a steel rule die and a rotary die?

A steel rule die is hardened rule bent to shape and set into a plywood board, used flat in a platen press; it costs little and arrives in days. A rotary die is a cylinder — flexible plate or solid engraved steel — that cuts continuously from a moving web at much higher speed, costs considerably more and takes weeks to make. Volume decides which one is right.

### What tolerance can die cutting hold?

±0.010–0.030 in (±0.25–0.75 mm) for steel rule dies, tightening to ±0.005–0.015 in (±0.13–0.4 mm) for rotary and hardened punch-and-die tooling. Print-to-cut registration on printed work is typically around ±1/32 in (±0.8 mm), which is why printed areas running to a cut edge need bleed.

### What is the smallest hole a steel rule die can cut?

About 0.125 in (3 mm), and rule-to-rule spacing should stay at or above the same figure so the die board keeps its strength between kerfs. Smaller holes require a hardened punch inserted into the die rather than bent rule, which adds cost but is entirely routine.

### What is the difference between cutting rule and creasing rule?

Cutting rule has a sharpened edge and severs the material. Creasing rule is blunt and rounded and works against a matrix channel on the opposite platen, compressing a fold line into the board without cutting through it. A folding carton die carries both, which is how the blank is cut and its fold lines formed in a single stroke.

### Why does grain direction matter in die cutting board?

Paperboard folds cleanly parallel to the grain and cracks along the crease when folded across it. Carton layouts should therefore orient primary folds with the grain. It is a layout decision made before tooling, and it cannot be corrected in the press.

### What are nicks and why are they in my part?

Nicks, or bridges, are small gaps deliberately left in the cutting rule so the part stays attached to the surrounding sheet as it travels through the press, then breaks free at stripping. Each leaves a small witness on the edge, so specify where they are acceptable — on non-visible edges, away from sealing surfaces and away from corners.

## Alternative processes

- [Laser Cutting](https://manufacturingprocesses.org/processes/cutting/laser-cutting.md): Laser cutting melts, burns or vaporises a narrow kerf through flat stock with a focused beam, producing profiles from a 2D file with no tooling.
- [Water Jet Cutting](https://manufacturingprocesses.org/processes/cutting/water-jet-cutting.md): Water jet cutting drives a high-pressure jet of water and abrasive through the material, cutting almost anything with no heat-affected zone.
- [Punching and Blanking](https://manufacturingprocesses.org/processes/cutting/punching-and-blanking.md): Punching and blanking shear sheet metal between a punch and a die, producing holes and flat outlines at very high speed.
- [Photochemical Machining](https://manufacturingprocesses.org/processes/cutting/photochemical-machining.md): Photochemical machining masks sheet metal photographically and etches away the unmasked areas, cutting burr-free flat parts with no hard tooling.

## Related processes

- [Punching and Blanking](https://manufacturingprocesses.org/processes/cutting/punching-and-blanking.md): Punching and blanking shear sheet metal between a punch and a die, producing holes and flat outlines at very high speed.
- [Foil Blocking and Embossing](https://manufacturingprocesses.org/processes/finishing/foil-blocking-and-embossing.md): Foil blocking and embossing press a heated die into the surface to transfer metallic foil, raise a relief, or do both at once.
- [Screen Printing](https://manufacturingprocesses.org/processes/finishing/screen-printing.md): Screen printing pushes ink through a stencilled mesh onto the surface below, laying down one opaque color per pass.
- [Paper Pulp Molding](https://manufacturingprocesses.org/processes/forming/paper-pulp-molding.md): Paper pulp molding draws a slurry of cellulose fiber onto a perforated screen tool by vacuum, then dries the fiber mat into a rigid formed part.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/cutting/die-cutting)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Electrical Discharge Machining"
category: "Cutting"
subcategory: "Thermal"
materials: ["Metal"]
tolerances: "Wire EDM ±0.0005 in (±0.013 mm) commercial, ±0.0001 in (±0.0025 mm) achievable with multiple skim passes; sinker EDM ±0.001 in (±0.025 mm) typical, ±0.0005 in (±0.013 mm) achievable"
volumes: "1–1,000 parts; standard for tooling, dies and low-volume precision components"
lead_time: "1–3 weeks; sinker work adds electrode fabrication time"
url: https://manufacturingprocesses.org/processes/cutting/electrical-discharge-machining
---

# Electrical Discharge Machining

EDM erodes conductive metal with controlled electrical sparks across a dielectric gap, cutting hardened material and sharp internal corners.

- **Category**: [Cutting](https://manufacturingprocesses.org/processes/cutting.md)
- **Family**: Thermal
- **Materials**: Metal
- **Typical tolerances**: Wire EDM ±0.0005 in (±0.013 mm) commercial, ±0.0001 in (±0.0025 mm) achievable with multiple skim passes; sinker EDM ±0.001 in (±0.025 mm) typical, ±0.0005 in (±0.013 mm) achievable
- **Surface finish**: 250 µin Ra (6.3 µm) rough cut; 16 µin (0.4 µm) after skim passes; 4–8 µin (0.1–0.2 µm) with fine-finish circuits
- **Typical volumes**: 1–1,000 parts; standard for tooling, dies and low-volume precision components
- **Lead time**: 1–3 weeks; sinker work adds electrode fabrication time

## Overview

Electrical discharge machining removes metal with thousands of controlled electrical sparks per second across a gap flooded with dielectric fluid. There is no cutting force and no mechanical contact, so hardness is irrelevant — fully hardened D2 at 62 HRC erodes at the same rate as annealed stock, and tungsten carbide, Inconel and titanium cut as readily as tool steel.

Two forms dominate. **Wire EDM** pulls a brass or coated wire, typically **0.010 in (0.25 mm)** diameter, through the part like a bandsaw blade, cutting any 2D or tapered profile to **±0.0001 in (±0.0025 mm)** on well-controlled work. **Sinker (die-sink) EDM** plunges a shaped graphite or copper electrode to burn its mirror image into the workpiece, producing blind cavities and internal corners far sharper than any milling cutter.

The one hard requirement is electrical conductivity: EDM cuts metals and graphite, and cannot touch ceramics, glass or plastics.

## How it works

1. **Setup and dielectric.** The work is submerged in deionized water (wire) or hydrocarbon oil (sinker). The dielectric insulates the gap until breakdown voltage is reached, then flushes the eroded debris away — flushing quality, not spark energy, is what usually limits speed.
2. **Start hole (wire only).** An internal profile needs a hole to thread the wire through, typically **0.020–0.080 in (0.5–2 mm)**, drilled conventionally or by small-hole EDM. Profiles that break out to the edge do not need one.
3. **Pulse generation.** The generator applies pulses across a gap of a few thousandths of an inch. Each discharge melts and vaporizes a microscopic crater; pulse on-time and current set both removal rate and finish, trading one directly against the other.
4. **Rough cut.** The first pass uses high energy and removes most of the material, leaving a rough surface near 250 µin Ra (6.3 µm) and a recast layer of roughly **0.0002–0.001 in (5–25 µm)**.
5. **Skim passes.** Wire EDM then makes successive low-energy passes over the same profile, each removing a few ten-thousandths of an inch. Skims are what deliver both the final tolerance and a finish down to 16 µin Ra (0.4 µm) or better; each one is another full lap of the profile and another slice of the bill.
6. **Sinker electrode sequence.** Sinker work typically uses a roughing electrode and one or more finishing electrodes, each made undersized by the spark gap it will burn — **0.0005–0.002 in (0.013–0.05 mm) per side** on finishing settings.
7. **Post-processing.** For fatigue-critical parts, the recast and heat-affected layer is removed by additional skims, polishing or chemical etching, and the drawing should say so explicitly.

## Design guidelines

### Wire diameter sets the internal corner radius
A wire EDM internal corner cannot be sharper than the wire radius plus the spark gap. Standard **0.010 in (0.25 mm)** wire gives about a **0.006 in (0.15 mm)** corner radius; fine **0.004 in (0.10 mm)** wire gets to roughly **0.003 in (0.076 mm)** at a much slower cutting rate. Draw the radius; do not specify a sharp internal corner on a wire-cut profile.

### Sinker EDM is how you get a genuinely sharp corner
A sinker electrode reproduces its own shape, so the practical internal corner radius drops to about **0.002 in (0.05 mm)** — the reason mold and die cavities with square internal corners are burned rather than milled.

### Allow for overburn
The cut is always larger than the tool by the spark gap. Wire kerf is roughly wire diameter plus 0.002 in (0.05 mm) total; sinker cavities finish 0.0005–0.002 in (0.013–0.05 mm) per side larger than the electrode. Machine offsets handle this, but it drives electrode design and dictates the minimum slot width you can ask for.

### Specify surface integrity, not just roughness
Every EDM surface carries a recast layer. A rough-cut-only part keeps 0.0002–0.001 in (5–25 µm) of recast plus a heat-affected zone beneath it, which lowers fatigue strength. If the part is cyclically loaded, call out the number of skim passes or a post-process rather than leaving it to the shop.

### Thickness, taper and stock preparation
Wire EDM cuts material up to about **12–16 in (300–400 mm)** thick on standard machines, and most machines can cut tapers up to about 30°. Stress-relieve hardened stock before wire cutting — releasing residual stress mid-cut is the most common cause of a part that closes on the wire or walks out of tolerance.

### Finish expectations
Rough cut 250 µin Ra (6.3 µm); after skims 16 µin (0.4 µm); with fine-finish circuits 4–8 µin (0.1–0.2 µm). The texture is matte and non-directional, which is why EDM finishes are specified deliberately on mold surfaces ([surface finish chart](/charts/surface-finish-chart)).

| Material | EDM behavior | Notes |
|---|---|---|
| Hardened tool steel (A2, D2, H13, S7) | Excellent | Cut after heat treat; hardness does not slow it |
| Tungsten carbide | Good | Slower and harder on wire; standard for punches and dies |
| Inconel, Hastelloy | Good | Cuts at a similar rate to steel regardless of strength |
| Titanium alloys | Good | Common for aerospace and medical profiles |
| Copper, brass, aluminum | Fast | High conductivity gives high removal rates |
| Graphite | Cuts readily | Also the standard sinker electrode material |
| Ceramics, glass, plastics | Not possible | Non-conductive — use [water jet](/processes/cutting/water-jet-cutting) |

| Feature | Recommended | Limit | Why |
|---|---|---|---|
| Internal corner radius (wire) | 0.006 in (0.15 mm) | 0.003 in (0.076 mm) | Wire radius plus spark gap |
| Internal corner radius (sinker) | 0.005 in (0.13 mm) | 0.002 in (0.05 mm) | Electrode fabrication plus gap |
| Wire start hole | 0.040 in (1 mm) | 0.020 in (0.5 mm) | Wire must thread through |
| Material thickness | ≤ 8 in (200 mm) | 12–16 in (300–400 mm) | Flushing and wire guidance |
| Taper angle | ≤ 15° | ~30° | Wire guide geometry |
| Tolerance | ±0.0005 in (±0.013 mm) | ±0.0001 in (±0.0025 mm) | Each tighter band adds skim passes |

## Variants

- Die Sink EDM
- Wire EDM

## Cost drivers

EDM is priced almost entirely by machine hours, and it is slow — removal rates are far below milling. What it buys is geometry and hardness that milling cannot reach.

**Cut area, not part complexity.** Wire EDM cost tracks profile length multiplied by stock thickness. A complicated outline in thin plate is cheap; a simple outline in 4 in (100 mm) plate is not. Complexity itself is free.

**Skim passes.** Each pass is another complete lap of the profile. A part needing ±0.0001 in (±0.0025 mm) and a 16 µin (0.4 µm) finish may take three or four skims after the rough cut, several times the cost of a separation cut.

**Electrode fabrication (sinker).** Electrodes are milled from graphite or copper and are frequently the dominant cost. A cavity needing rough and finish electrodes carries two complete machining jobs before the burn starts.

**Hardness is not a cost driver.** This is the key economic fact: cutting hardened 62 HRC tool steel costs the same as cutting annealed stock, so parts can be heat treated first and finished afterward with no distortion allowance.

Four ways to take cost out:

1. Rough the part by [milling](/processes/cutting/cnc-milling) and use EDM only for the features that need it.
2. Specify the loosest tolerance and finish the function allows — every skim pass is a full lap.
3. Draw corner radii of 0.006 in (0.15 mm) or larger so standard 0.010 in wire can be used at full speed.
4. Stress-relieve hardened stock before wire cutting to avoid scrap from parts that move mid-cut.

## FAQ

### What tolerance can wire EDM hold?

±0.0005 in (±0.013 mm) is normal commercial work and ±0.0001 in (±0.0025 mm) is achievable with multiple skim passes and a temperature-controlled machine. Each tighter band costs another full lap of the profile, so the tolerance you specify translates directly into passes and therefore into price.

### Can EDM cut any material?

Any electrically conductive material, including fully hardened tool steel, tungsten carbide, Inconel, titanium, copper and graphite. Hardness does not slow the process, which is why parts are heat treated first and EDM'd afterward. Ceramics, glass and plastics are non-conductive and cannot be EDM'd — those go to water jet cutting.

### What is the smallest internal corner radius wire EDM can cut?

About 0.006 in (0.15 mm) with standard 0.010 in (0.25 mm) wire, since the corner radius equals the wire radius plus the spark gap. Fine 0.004 in (0.10 mm) wire reaches roughly 0.003 in (0.076 mm) but cuts much more slowly. For sharper internal corners, sinker EDM reaches about 0.002 in (0.05 mm).

### What is the recast layer in EDM and does it matter?

Each spark melts a microscopic crater and some of that melted metal resolidifies on the surface as a recast layer, roughly 0.0002–0.001 in (5–25 µm) thick after a rough cut, with a heat-affected zone beneath it. It reduces fatigue strength, so for cyclically loaded parts you should specify skim passes, polishing or chemical etching to remove it rather than leaving the decision to the shop.

### How thick a part can wire EDM cut?

Standard machines cut up to about 12–16 in (300–400 mm) of stock thickness, and most can cut tapers to roughly 30°. Cost scales with profile length times thickness, so a thick part with a simple outline can easily cost more than a thin part with an intricate one.

### Why does EDM stock need to be stress relieved first?

Hardened and heavily machined stock carries residual stress. As the wire cuts through it, that stress redistributes and the part moves — closing on the wire, bowing, or walking out of tolerance mid-cut. Stress relieving before wire cutting is the standard preventive step and the most common cause of scrap when skipped.

## Alternative processes

- [Water Jet Cutting](https://manufacturingprocesses.org/processes/cutting/water-jet-cutting.md): Water jet cutting drives a high-pressure jet of water and abrasive through the material, cutting almost anything with no heat-affected zone.
- [CNC Milling](https://manufacturingprocesses.org/processes/cutting/cnc-milling.md): CNC milling removes material with a rotating multi-flute cutter moved along programmed paths, producing prismatic parts from solid stock.
- [Laser Cutting](https://manufacturingprocesses.org/processes/cutting/laser-cutting.md): Laser cutting melts, burns or vaporises a narrow kerf through flat stock with a focused beam, producing profiles from a 2D file with no tooling.
- [Photochemical Machining](https://manufacturingprocesses.org/processes/cutting/photochemical-machining.md): Photochemical machining masks sheet metal photographically and etches away the unmasked areas, cutting burr-free flat parts with no hard tooling.

## Related processes

- [CNC Milling](https://manufacturingprocesses.org/processes/cutting/cnc-milling.md): CNC milling removes material with a rotating multi-flute cutter moved along programmed paths, producing prismatic parts from solid stock.
- [5-Axis CNC Machining](https://manufacturingprocesses.org/processes/cutting/5-axis-cnc-machining.md): 5-axis CNC machining tilts and rotates the part or the head so the tool reaches every face and stays short and rigid on complex surfaces.
- [Water Jet Cutting](https://manufacturingprocesses.org/processes/cutting/water-jet-cutting.md): Water jet cutting drives a high-pressure jet of water and abrasive through the material, cutting almost anything with no heat-affected zone.
- [Grinding, Sanding and Polishing](https://manufacturingprocesses.org/processes/finishing/grinding-sanding-and-polishing.md): Grinding, sanding and polishing remove a thin layer of material with bonded or coated abrasive to correct form or improve surface finish.
- [Heat Treatment](https://manufacturingprocesses.org/processes/forming/heat-treatment.md): Heat treatment cycles metal through controlled heating and cooling to change its hardness, strength, ductility or internal stress state.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/cutting/electrical-discharge-machining)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Glass Scoring"
category: "Cutting"
subcategory: "Mechanical"
materials: ["Glass"]
tolerances: "±1/32 in (±0.8 mm) with careful hand or CNC-table work; ±1/16 in (±1.6 mm) is common for architectural glass"
volumes: "1–100,000+; equally suited to one-off work and automated CNC cutting tables"
lead_time: "Minutes per cut; glass shops typically cut to size same-day"
url: https://manufacturingprocesses.org/processes/cutting/glass-scoring
---

# Glass Scoring

Glass scoring runs a hardened wheel across the surface to create a controlled fracture line, then breaks the sheet cleanly along it.

- **Category**: [Cutting](https://manufacturingprocesses.org/processes/cutting.md)
- **Family**: Mechanical
- **Materials**: Glass
- **Typical tolerances**: ±1/32 in (±0.8 mm) with careful hand or CNC-table work; ±1/16 in (±1.6 mm) is common for architectural glass
- **Surface finish**: The broken face is as-fractured and sharp, with a lightly crushed band along the score line; edges are then seamed, ground or polished as specified
- **Typical volumes**: 1–100,000+; equally suited to one-off work and automated CNC cutting tables
- **Lead time**: Minutes per cut; glass shops typically cut to size same-day

## Overview

Glass scoring cuts flat glass without removing any material. A hardened wheel run across the surface under load creates a shallow median crack, and bending the sheet propagates that crack cleanly through the full thickness. There is no kerf, no dust and no coolant beyond a drop of cutting oil.

It works by hand up to about **1/4 in (6 mm)** and to roughly **1/2 in (12 mm)** with running pliers or a bench breaker; thicker glass goes to a water jet or a diamond saw. Straight lines and gentle outside curves are practical, but **inside corners are impossible** — the crack runs where the stress field sends it, not where the drawing wants it.

Two rules govern everything: score once and never re-score the same line, and break promptly after scoring.

## How it works

1. **Wheel selection.** A tungsten carbide wheel is chosen by included angle, matched to thickness — shallower angles around 120° for thin glass, more obtuse angles up to about 160° for thick. The wrong angle either fails to open a useful crack or crushes the surface.
2. **Scoring.** The wheel is drawn across the surface in a single steady pass with a light film of cutting oil. A correct score looks like a faint line, not a deep scratch: excessive pressure produces flaking along the score and a ragged break.
3. **Never re-score.** Running the wheel down the same line a second time damages the wheel and cannot deepen the existing crack, because the wheel now rides in a groove instead of loading fresh surface.
4. **Breaking.** The sheet is bent to put the scored face in tension — over a fulcrum, by tapping from underneath, or with running pliers. The median crack propagates through the thickness in an instant.
5. **Break promptly.** A score does not stay sharp. The crack tip blunts over time, and a stale score breaks unpredictably and off-line, so scoring and breaking are treated as one operation.
6. **Grozing.** Shapes that cannot be broken out in one piece are nibbled back with grozing pliers, then ground to the line.
7. **Edge finishing.** A raw broken edge is sharp and full of microcracks, and is the weakest part of the finished piece. Edges are seamed (arrised), ground flat or polished depending on the application, and this step is not optional on anything that will be handled or tempered.

## Design guidelines

### Straight lines and outside curves only
The break follows the stress field, not the score, once it starts. Gentle convex curves break out reliably; concave curves and inside corners do not. Any shape with an internal corner must be nibbled and ground, or cut by [water jet](/processes/cutting/water-jet-cutting) instead.

### Minimum strip width
About **0.4 in (10 mm)** by hand, narrower with running pliers. Below that the strip breaks up rather than snapping along the score.

### Tempered glass cannot be scored at all
Toughened glass is in a balanced state of surface compression and core tension; any penetration of the surface causes the whole pane to disintegrate. **All cutting, drilling and edge work must be completed before tempering** — a change to a tempered pane means a new pane.

### Laminated glass is a two-sided operation
Score and break both glass plies, then part the interlayer with heat or a blade. The edge quality of laminated work is inherently rougher than monolithic glass.

### Specify the edge, not just the outline
A raw broken edge is a safety hazard and the main strength limit of the finished piece. State whether the edge should be seamed, ground or polished. Any glass destined for tempering must be edge-worked first, because a flaw at the edge becomes the failure origin.

### Score the right face on coated glass
Low-E and other soft coatings are scored from the uncoated side. Confirm coating orientation before cutting, since it also determines which way the pane faces when installed.

| Glass type | Scoring suitability | Notes |
|---|---|---|
| Annealed float glass | Excellent | The standard case; hand or CNC table |
| Low-iron float | Excellent | Same behavior as standard float |
| Coated / low-E glass | Good | Score from the uncoated face |
| Laminated glass | Workable | Score both faces, then part the interlayer |
| Patterned / rolled glass | Fair | Score the smooth face; breaks less predictably |
| Borosilicate | Fair | Harder and less forgiving than soda-lime |
| Wired glass | Poor | Score and break the glass, then cut the wire |
| Tempered / toughened glass | Impossible | Shatters; must be cut before tempering |
| Over 1/2 in (12 mm) thick | Impractical | Use water jet or a diamond saw |

| Feature | Recommended | Limit | Why |
|---|---|---|---|
| Thickness, hand scoring | ≤ 1/4 in (6 mm) | ~1/2 in (12 mm) with a breaker | Crack must run the full thickness |
| Minimum strip width | ≥ 0.75 in (19 mm) | ~0.4 in (10 mm) | Narrow strips break up |
| Curve type | Convex only | Gentle radii | The break follows the stress field |
| Inside corners | Not achievable | Groze and grind, or water jet | Crack cannot turn a corner |
| Time from score to break | Immediate | Minutes | The crack tip blunts with time |
| Edge condition | Seamed or ground | As-broken is sharp | Edge flaws govern strength |

## Cost drivers

Scoring itself is nearly free — a wheel, a straightedge and a few seconds. Cost lives in the material, in the edge work and in breakage.

**Sheet yield.** Glass is bought as full sheets and the offcuts are usually unusable. Sizing parts so they nest into standard stock sheets is the largest single lever on cost.

**Edge finishing.** Seaming, grinding and polishing are separate operations, and on a small part the polished edge can cost more than the glass. Specify only the edge class the application needs.

**Breakage.** Scrap from bad breaks, handling damage and chipped corners is a normal part of the process and is priced into the job. Thicker and larger panes carry higher breakage risk.

**Thickness.** Thicker glass is more expensive per unit area, harder to break cleanly, and past about 1/2 in (12 mm) moves to a different process entirely.

**Sequence with tempering.** Because tempered glass cannot be cut, every dimensional change means a new pane. Confirm sizes before the tempering step, not after.

Three ways to take cost out:

1. Design pane sizes that nest into standard sheet dimensions.
2. Specify seamed edges where the glass will be captured in a frame and reserve polished edges for exposed edges.
3. Keep all outlines to straight lines and convex curves so the shape can be broken out rather than ground.

## FAQ

### How thick a glass can be cut by scoring?

Up to about 1/4 in (6 mm) by hand and roughly 1/2 in (12 mm) with running pliers or a bench breaker. Above that the crack no longer propagates reliably through the full thickness, and water jet cutting or a diamond saw is the right process.

### Why can't tempered glass be cut?

Toughened glass holds its surface in compression balanced against tension in the core. Any penetration of that compressed surface releases the stored energy and the entire pane disintegrates into fragments. All cutting, drilling and edge work must be finished before tempering, so a dimensional change means a new pane rather than a rework.

### Why should you never score the same line twice?

The second pass cannot deepen the existing crack because the wheel rides in the groove it already made rather than loading fresh surface, and it damages the wheel. A single steady pass with correct pressure produces a faint line and a clean break; a re-scored line produces flaking and a ragged edge.

### How soon after scoring should glass be broken?

Immediately. The tip of the median crack blunts over time, and a score left standing breaks unpredictably and wanders off the line. Production practice treats scoring and breaking as a single operation rather than two separate steps.

### Can glass scoring cut inside corners?

No. Once a break starts it follows the stress field rather than the score, so concave curves and internal corners cannot be broken out. Those shapes are nibbled back with grozing pliers and ground to the line, or cut by water jet, which has no such restriction and also produces no heat affected zone.

### What tolerance can glass cutting hold?

About ±1/32 in (±0.8 mm) with careful hand work or on a CNC scoring table, and ±1/16 in (±1.6 mm) is commonly accepted for architectural glazing. Because scoring removes no material there is no kerf to compensate, but the break can wander slightly from the score line.

## Alternative processes

- [Water Jet Cutting](https://manufacturingprocesses.org/processes/cutting/water-jet-cutting.md): Water jet cutting drives a high-pressure jet of water and abrasive through the material, cutting almost anything with no heat-affected zone.
- [Sawing](https://manufacturingprocesses.org/processes/cutting/sawing.md): Sawing separates stock with a toothed blade — band, circular or cold saw — and is the default first operation on bar, tube and plate.
- [Grinding, Sanding and Polishing](https://manufacturingprocesses.org/processes/finishing/grinding-sanding-and-polishing.md): Grinding, sanding and polishing remove a thin layer of material with bonded or coated abrasive to correct form or improve surface finish.

## Related processes

- [Glassblowing](https://manufacturingprocesses.org/processes/forming/glassblowing.md): Glassblowing gathers molten glass on a blowpipe and inflates it, by hand or by machine, into hollow ware such as bottles, jars and vessels.
- [Lampworking](https://manufacturingprocesses.org/processes/forming/lampworking.md): Lampworking softens glass rod and tube in a bench torch and forms it directly, producing scientific glassware, neon tube and small artwork.
- [Water Jet Cutting](https://manufacturingprocesses.org/processes/cutting/water-jet-cutting.md): Water jet cutting drives a high-pressure jet of water and abrasive through the material, cutting almost anything with no heat-affected zone.
- [Grinding, Sanding and Polishing](https://manufacturingprocesses.org/processes/finishing/grinding-sanding-and-polishing.md): Grinding, sanding and polishing remove a thin layer of material with bonded or coated abrasive to correct form or improve surface finish.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/cutting/glass-scoring)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Laser Cutting"
category: "Cutting"
subcategory: "Thermal"
materials: ["Metal", "Plastic", "Wood", "Composite"]
tolerances: "±0.005 in (±0.13 mm) on thin sheet; ±0.010 in (±0.25 mm) on plate and over long dimensions; edge taper roughly 0.5–1°"
volumes: "1–10,000 parts; no tooling, so single pieces are practical and unit cost falls mainly through nesting"
lead_time: "1–5 business days; same-day is common for simple sheet parts"
url: https://manufacturingprocesses.org/processes/cutting/laser-cutting
---

# Laser Cutting

Laser cutting melts, burns or vaporises a narrow kerf through flat stock with a focused beam, producing profiles from a 2D file with no tooling.

- **Category**: [Cutting](https://manufacturingprocesses.org/processes/cutting.md)
- **Family**: Thermal
- **Materials**: Metal, Plastic, Wood, Composite
- **Typical tolerances**: ±0.005 in (±0.13 mm) on thin sheet; ±0.010 in (±0.25 mm) on plate and over long dimensions; edge taper roughly 0.5–1°
- **Surface finish**: Edge quality is graded to ISO 9013 (perpendicularity and mean height of profile) rather than Ra; nitrogen-cut stainless is bright and oxide-free, oxygen-cut mild steel leaves an oxide layer
- **Typical volumes**: 1–10,000 parts; no tooling, so single pieces are practical and unit cost falls mainly through nesting
- **Lead time**: 1–5 business days; same-day is common for simple sheet parts

## Overview

Laser cutting focuses a beam onto flat stock and melts, burns or vaporizes a narrow kerf through it while an assist gas blows the molten material out the bottom. It needs no tooling — the part comes straight from a DXF — which is why it dominates sheet metal prototyping and short-run fabrication.

Two machine types cover most work. **Fiber lasers** (1.06 µm wavelength) cut metals: mild steel to about 1 in (25 mm) on common 4–6 kW machines, stainless and aluminum somewhat less, with 10–20 kW machines extending that to roughly 1.5 in (38 mm). **CO2 lasers** (10.6 µm) cut metal more slowly but are the machine for acrylic, plywood, MDF, leather, paper and textiles.

Kerf runs **0.004–0.015 in (0.1–0.4 mm)** depending on thickness, edges come off within about a degree of square, and profile tolerance is **±0.005 in (±0.13 mm)** on thin sheet.

## How it works

1. **Nesting and programming.** Parts are nested on the sheet to maximize yield, with lead-ins placed in scrap, common-line cutting where two parts can share an edge, and micro-tabs where small parts would otherwise drop and tip.
2. **Beam delivery and focus.** The beam is focused to a spot on the order of 0.004–0.010 in (0.1–0.25 mm). Focus position is set relative to thickness and gas: at or near the surface for thin oxygen cutting, and deeper into the material for nitrogen cutting of thick stainless.
3. **Piercing.** Before each contour the beam dwells to pierce through. Pierce time grows sharply with thickness and is a meaningful share of cycle time on plate, which is why a nest with many small holes costs more than its cut length suggests.
4. **Assist gas selection.** Oxygen is used on mild steel, where the exothermic reaction with iron adds energy and roughly doubles achievable thickness — at the price of an oxide layer on the cut face. Nitrogen is used on stainless and aluminum to blow the melt out without oxidation, giving a bright, weld-ready and paint-ready edge at higher gas cost and lower speed. Compressed air is common on thin material.
5. **Contour cutting.** A capacitive height sensor holds nozzle standoff within a few thousandths of an inch while the head follows the profile. Sharp corners are slowed or looped to avoid dwelling and burning.
6. **Part removal.** Parts drop to a slat bed or are held by micro-tabs and broken out. Small parts, thin webs and long thin strips are where nesting problems appear.
7. **Secondary work.** Dross removal or edge deburring where required, then forming, welding or finishing. Oxygen-cut mild steel needs the oxide layer removed before painting or welding.

## Design guidelines

### Minimum hole diameter
Keep holes at least **1× material thickness** in mild steel and **1.5× thickness** in stainless and aluminum. Smaller holes taper badly and the pierce disturbs the edge, because the beam has to blow through the full thickness in a spot barely larger than itself.

### Minimum slot width and web
Slots should be at least **1× material thickness** wide, and the web left between two cutouts or between a cutout and the edge should also be at least **1× thickness**. Narrower webs overheat, warp and can burn through, especially in stainless.

### Corner radii
Internal corners pick up the kerf radius automatically, roughly 0.005 in (0.13 mm). On plate, add a radius of at least half the material thickness to sharp internal corners so the head does not dwell in the corner and burn it.

### Design for the bend, not just the cut
If the part will be formed, add bend reliefs at the ends of bend lines and keep holes at least 1.5× material thickness plus the bend radius away from the bend, or they will distort. See the [bend radius and K-factor chart](/charts/sheet-metal-bend-radius-k-factor) and specify stock from the [sheet metal gauge chart](/charts/sheet-metal-gauge-chart).

### Specify the edge condition you need
Nitrogen-cut stainless and aluminum come off bright, oxide-free and ready to weld or paint. Oxygen-cut mild steel carries an oxide layer that must be removed first. Thermal cut quality is classified by ISO 9013 (perpendicularity tolerance and mean height of profile), which is the right language for a drawing note.

### Know what must not go in the machine
PVC and other chlorinated plastics release hydrogen chloride when lasered — toxic to operators and corrosive to the machine — and no reputable shop will run them. Polycarbonate burns and discolors rather than cutting cleanly. Galvanized and coated steels cut but require zinc fume extraction.

| Material | Machine / gas | Practical limit and notes |
|---|---|---|
| Mild steel | Fiber, oxygen | ~1 in (25 mm) at 4–6 kW; oxide edge needs removal before paint or weld |
| Stainless steel | Fiber, nitrogen | ~1 in (25 mm) at high power; bright oxide-free edge |
| Aluminum | Fiber, nitrogen | ~0.5–0.75 in (12–19 mm); reflective, prone to dross |
| Brass, copper | Fiber | Thin sections; highly reflective, needs back-reflection protection |
| Titanium | Fiber, argon | Argon shielding prevents a brittle oxidized edge |
| Acrylic (PMMA) | CO2 | Flame-polished transparent edge; the showcase laser material |
| Plywood, MDF, leather | CO2 | Charred edge, extraction required |
| Polycarbonate | Poor | Burns and yellows; mechanical cutting is better |
| PVC, vinyl, PTFE | Never | Releases hydrogen chloride or fluorine compounds |

| Feature | Recommended | Limit | Why |
|---|---|---|---|
| Hole diameter | ≥ 1.5× thickness | 1× thickness (mild steel) | Small holes taper and blow out on pierce |
| Slot width | ≥ 1× thickness | 0.8× thickness | Kerf and heat need room |
| Web between features | ≥ 1× thickness | 0.8× thickness | Thin webs overheat and warp |
| Internal corner radius | ≥ 0.5× thickness | Kerf radius, ~0.005 in | Prevents corner burn on plate |
| Hole to bend line | ≥ 1.5× thickness + bend radius | — | Holes distort inside the bend zone |
| Tab / thin strip width | ≥ 2× thickness | 1× thickness | Thin strips warp from heat |

## Cost drivers

Laser cutting carries no tooling cost, so price is machine time plus material, and machine time is dominated by two things: how far the head travels and how many times it has to pierce.

**Cut length and pierce count.** A part with fifty small holes may take longer than a part twice its size with none. Every hole is a pierce plus a contour.

**Thickness.** Speed falls steeply with thickness while pierce time climbs. Doubling plate thickness costs far more than double the time.

**Material and assist gas.** Nitrogen cutting consumes large volumes of gas and runs slower than oxygen cutting; a bright oxide-free edge on stainless is a real premium over an oxide edge on mild steel.

**Nesting efficiency.** You pay for the sheet, not the part. A shape that nests tightly, or that can share cut lines with its neighbor, can shift material cost noticeably. Ask the shop to nest before finalizing the outline.

**Setup and programming.** Small, but it is why a single piece costs more per part than fifty of the same piece off one sheet.

Four ways to take cost out:

1. Consolidate or enlarge small holes — hole count drives pierce count, which drives cycle time.
2. Choose the thinnest gauge that meets the structural need ([sheet metal gauge chart](/charts/sheet-metal-gauge-chart)).
3. Accept an oxygen-cut oxide edge on mild steel wherever the part will be blasted or coated anyway.
4. Design outlines that tile: rectangles and shapes with parallel edges nest tighter than free-form profiles.

## FAQ

### How thick a material can a laser cut?

A 4–6 kW fiber laser cuts mild steel to about 1 in (25 mm) with oxygen assist, stainless to roughly the same with nitrogen at high power, and aluminum to about 0.5–0.75 in (12–19 mm). Machines at 10–20 kW extend mild steel to roughly 1.5 in (38 mm). Above that, plasma or water jet cutting is faster and cheaper.

### What tolerance does laser cutting hold?

±0.005 in (±0.13 mm) on thin sheet is normal, opening to ±0.010 in (±0.25 mm) on plate and across long dimensions. The cut face also carries about 0.5–1° of taper, which matters when the edge is a mating surface rather than a free profile.

### What is the minimum hole size for laser cutting?

About 1× material thickness in mild steel and 1.5× thickness in stainless and aluminum. Below that the pierce disturbs the edge and the hole comes out tapered, because the beam must blow through the full thickness in a spot barely wider than the kerf. Smaller holes should be drilled or punched afterward.

### What is the difference between a fiber laser and a CO2 laser?

A fiber laser emits at 1.06 µm, which metals absorb efficiently, so it cuts steel, stainless, aluminum, brass and copper faster and with lower running cost. A CO2 laser emits at 10.6 µm, which organics absorb well, making it the machine for acrylic, plywood, MDF, leather and paper. Fiber lasers cut clear acrylic poorly because the material is largely transparent at that wavelength.

### Which materials should never be laser cut?

PVC and other chlorinated plastics, including vinyl, release hydrogen chloride gas that is toxic to operators and corrodes the machine. PTFE releases fluorine compounds. Polycarbonate burns and yellows instead of cutting cleanly. Galvanized and coated steel can be cut but requires zinc fume extraction.

### Does laser cutting leave a heat affected zone?

Yes — it is a thermal process, so a narrow zone along the cut face is heated and rapidly cooled, typically a few thousandths of an inch on sheet and wider on plate. It is small enough to ignore for most fabrication, but for hardened, tempered or fatigue-critical parts, water jet cutting removes the question entirely because it produces no heat affected zone.

### Should I use oxygen or nitrogen assist gas?

Oxygen on mild steel, where the exothermic reaction with iron adds energy and roughly doubles the thickness the machine can handle, leaving an oxide layer that must be removed before painting or welding. Nitrogen on stainless and aluminum, which blows the melt out without oxidation and leaves a bright edge ready to weld or paint, at higher gas cost and lower speed.

## Alternative processes

- [Water Jet Cutting](https://manufacturingprocesses.org/processes/cutting/water-jet-cutting.md): Water jet cutting drives a high-pressure jet of water and abrasive through the material, cutting almost anything with no heat-affected zone.
- [Plasma Cutting](https://manufacturingprocesses.org/processes/cutting/plasma-cutting.md): Plasma cutting blows an ionised gas jet through conductive metal, melting and ejecting a kerf far faster than a laser on thick plate.
- [Punching and Blanking](https://manufacturingprocesses.org/processes/cutting/punching-and-blanking.md): Punching and blanking shear sheet metal between a punch and a die, producing holes and flat outlines at very high speed.
- [Photochemical Machining](https://manufacturingprocesses.org/processes/cutting/photochemical-machining.md): Photochemical machining masks sheet metal photographically and etches away the unmasked areas, cutting burr-free flat parts with no hard tooling.
- [Die Cutting](https://manufacturingprocesses.org/processes/cutting/die-cutting.md): Die cutting presses a shaped steel rule or rotary die through sheet stock, cutting flat outlines in paper, board, foam, film and textile.

## Related processes

- [Tube and Profile Laser Cutting](https://manufacturingprocesses.org/processes/cutting/tube-and-profile-laser-cutting.md): Tube laser cutting rotates and feeds tube or profile through a laser head, cutting holes, slots and joint geometry along its length.
- [Sheet Metal Fabrication](https://manufacturingprocesses.org/processes/forming/sheet-metal-fabrication.md): Sheet metal fabrication cuts, bends and joins flat stock into a finished assembly, typically laser cutting followed by press braking, hardware and welding.
- [Press Braking](https://manufacturingprocesses.org/processes/forming/press-braking.md): Press braking bends sheet metal by driving a punch into a V-die, forming one straight-line bend at a time.
- [Punching and Blanking](https://manufacturingprocesses.org/processes/cutting/punching-and-blanking.md): Punching and blanking shear sheet metal between a punch and a die, producing holes and flat outlines at very high speed.
- [Laser Marking and Engraving](https://manufacturingprocesses.org/processes/finishing/laser-marking-and-engraving.md): Laser marking alters the surface with a focused beam — annealing, etching or engraving it — to make permanent, contact-free marks.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/cutting/laser-cutting)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Photochemical Machining"
category: "Cutting"
subcategory: "Chemical"
materials: ["Metal"]
tolerances: "±10% of material thickness as standard (±0.001 in / ±0.025 mm on 0.010 in / 0.25 mm stock); ±20% of thickness for conservative planning; ±0.0005 in (±0.013 mm) on thin foil"
volumes: "1–1,000,000+; economical at both extremes because there is no hard tooling"
lead_time: "Phototools in 1–3 days; prototypes in 1–2 weeks; production 2–4 weeks"
url: https://manufacturingprocesses.org/processes/cutting/photochemical-machining
---

# Photochemical Machining

Photochemical machining masks sheet metal photographically and etches away the unmasked areas, cutting burr-free flat parts with no hard tooling.

- **Category**: [Cutting](https://manufacturingprocesses.org/processes/cutting.md)
- **Family**: Chemical
- **Materials**: Metal
- **Typical tolerances**: ±10% of material thickness as standard (±0.001 in / ±0.025 mm on 0.010 in / 0.25 mm stock); ±20% of thickness for conservative planning; ±0.0005 in (±0.013 mm) on thin foil
- **Surface finish**: Unetched faces retain the mill finish of the incoming sheet; etched faces are matte and lightly textured. Edges are entirely burr-free with a characteristic cusp at mid-thickness where the two etch fronts meet
- **Typical volumes**: 1–1,000,000+; economical at both extremes because there is no hard tooling
- **Lead time**: Phototools in 1–3 days; prototypes in 1–2 weeks; production 2–4 weeks

## Overview

Photochemical machining — also called chemical etching or photo etching — masks sheet metal photographically and dissolves away everything the mask does not protect. There is no punch, no cutting force and no heat, so parts come out completely **burr-free, stress-free and free of any heat affected zone or work hardening**. That combination is why it is the standard process for shims, lead frames, encoder discs, EMI shields, fine mesh screens, flexures and precision springs.

Stock runs **0.0005–0.060 in (0.013–1.5 mm)**, with most work between 0.001 and 0.030 in (0.025–0.75 mm). Tolerance scales with thickness at roughly **±10% of material thickness**, so 0.010 in (0.25 mm) stock holds about ±0.001 in (±0.025 mm).

Complexity is free. A part with a thousand holes etches in exactly the same time as a plain rectangle, and the only tooling is a piece of film.

## How it works

1. **Phototool generation.** The CAD outline is plotted at high resolution onto film or glass, in two halves that register front to back. Tooling cost is trivial compared with a die, and a revised design means a new plot rather than new steel.
2. **Cleaning and lamination.** The sheet is chemically cleaned and a photoresist — usually dry film — is laminated to both faces. Any contamination here becomes a defect later.
3. **Exposure and development.** UV light through the phototool cures the resist in the pattern; developing washes away the uncured resist and exposes bare metal exactly where material is to be removed.
4. **Etching.** The sheet passes through a spray etcher that floods both faces. Ferric chloride is the workhorse etchant for steels, stainless, copper alloys and nickel, run at controlled concentration (commonly around 42–45° Baumé) and temperature (roughly 120–130 °F / 49–54 °C). Aluminum and titanium use different chemistries.
5. **Isotropic attack.** Etching removes material in all directions at once, so it undercuts the resist as it goes down. Etching from both sides simultaneously halves the depth each front must travel and leaves the characteristic cusp where the two fronts meet at mid-thickness.
6. **Stripping and rinsing.** Resist is stripped, the sheet is rinsed and dried, and parts are broken out of the retaining tabs that hold them in the panel.
7. **Optional half-etching.** Etching one side only produces fold lines, counterbores, recesses and part marking. Depth is targeted at 30–60% of thickness and controlled to roughly ±10% of the stock thickness.

## Design guidelines

### Feature size is governed by thickness
The minimum hole diameter, slot width and web between features is about **1.0–1.2× material thickness**. This is a direct consequence of isotropic etching: the etch spreads sideways as fast as it goes down, so no feature can be much narrower than the sheet is thick.

### Tolerance scales with thickness, not with part size
Plan on **±10% of material thickness** as the standard tolerance, with ±20% as a conservative figure for large panels. On 0.005 in (0.13 mm) foil that is ±0.0005 in (±0.013 mm); on 0.040 in (1 mm) stock it is ±0.004 in (±0.1 mm). Thin stock is where the process is most precise. Specify stock as a decimal thickness rather than a gauge number, and cross-check the two against the [sheet metal gauge chart](/charts/sheet-metal-gauge-chart).

### Do not draw sharp internal corners
Etched corners come out radiused. Specify a radius of at least **0.5× material thickness** on internal corners rather than a sharp corner the process cannot make.

### The edge is not square
Etching from both faces leaves an hourglass profile with a cusp at mid-thickness. Where a square edge is functionally required — a bearing surface, a precise slot width for a mating tongue — dimension to the narrowest point and state it, or choose a different process.

### Use half-etching deliberately
Half-etched fold lines let a flat part be bent by hand into a formed shape without tooling. Half-etched pockets make counterbores and recesses. Depth control of about ±10% of stock thickness is the practical limit, so do not use a half-etch where the residual thickness is structurally critical.

### Plan tabs and nesting
Parts are held in the panel by small tabs and broken out, and each tab leaves a witness mark. Say where tabs are acceptable. Since price is per panel area, nesting drives cost — panels are typically in the 12 × 18 in to 24 × 36 in (300 × 450 mm to 600 × 900 mm) range depending on the shop.

| Material | Etchability | Notes |
|---|---|---|
| 301 / 302 / 304 / 316 stainless | Excellent | The most common PCM material; ferric chloride |
| Copper, brass, phosphor bronze | Excellent | Fast etching; standard for shields and lead frames |
| Beryllium copper | Excellent | Springs and contacts; etch then heat treat |
| Nickel, Inconel, Kovar, Invar | Good | Slower etch rates; common in electronics packaging |
| Spring steel, shim steel | Excellent | Burr-free shims are a signature application |
| Molybdenum, tungsten | Specialty | Not every shop runs the chemistry |
| Aluminum | Good | Different chemistry, coarser minimum features |
| Titanium | Specialty | Requires aggressive chemistry and specific handling |
| Plastics, glass, ceramics | Not possible | No etchant chemistry in this process |

| Feature | Recommended | Limit | Why |
|---|---|---|---|
| Material thickness | 0.001–0.030 in (0.025–0.75 mm) | 0.0005–0.060 in (0.013–1.5 mm) | Feature size and tolerance scale with thickness |
| Hole diameter | ≥ 1.2× thickness | 1.0× thickness | Etch spreads sideways as fast as down |
| Slot width | ≥ 1.2× thickness | 1.0× thickness | Same isotropic limit |
| Web between features | ≥ 1× thickness | 0.8× thickness | Undercut from both sides |
| Internal corner radius | ≥ 0.5× thickness | Sharp corners not achievable | Isotropic etch rounds corners |
| Half-etch depth | 30–60% of thickness | ±10% of thickness control | Depth control limit |
| Standard tolerance | ±10% of thickness | ±20% on large panels | Undercut variation across the panel |

## Cost drivers

Photochemical machining is priced by panel area and by the number of panels, which produces an unusual cost structure: complexity is free, and area is everything.

**Panel area and nesting.** You pay for sheet processed, not features produced. A part that nests at 90% yield costs a fraction of one that nests at 50%, and adding features to the part changes nothing.

**Material.** Thin precision-rolled stainless, beryllium copper and nickel alloys are a real share of the panel cost, particularly at production volumes.

**Thickness.** Thicker stock etches longer, holds looser tolerance and needs coarser minimum features, so thickness pushes cost up in two directions at once.

**Phototooling.** A one-time charge measured in hundreds rather than thousands, with a lead time of days. This is what makes the process viable for prototypes and for designs that are still changing.

**Secondary operations.** Forming half-etched fold lines, plating, and passivation are separate steps priced separately.

Four ways to take cost out:

1. Send the part outline to the shop for nesting before finalizing dimensions — panel yield dominates price.
2. Use the thinnest stock that meets the functional requirement; it etches faster and holds tighter tolerance.
3. Combine multiple part numbers onto one panel when they share material and thickness.
4. Add features freely — holes, text, patterns and half-etched fold lines cost nothing extra, so consolidate parts rather than adding assembly steps.

## FAQ

### What tolerance can photochemical machining hold?

Roughly ±10% of the material thickness as a standard, with ±20% as a conservative planning figure on large panels. On 0.010 in (0.25 mm) stock that is about ±0.001 in (±0.025 mm), and on 0.005 in (0.13 mm) foil about ±0.0005 in (±0.013 mm). The process is most precise on thin material, which is the opposite of most cutting processes.

### What is the minimum feature size for chemical etching?

About 1.0–1.2× the material thickness for hole diameter, slot width and the web between features. Etching is isotropic, so the etchant spreads sideways as fast as it works down, and no feature can be much narrower than the sheet is thick. That is why thin stock gives fine features and thick stock does not.

### Why are photochemically machined parts burr-free?

Nothing mechanical touches the part. Material is dissolved rather than sheared, cut or melted, so there is no plastic deformation to raise a burr, no work hardening, no heat affected zone and no residual stress. For thin springs, shims and flexures this matters as much as the dimensional accuracy.

### What thickness range can be photochemically machined?

About 0.0005–0.060 in (0.013–1.5 mm), with most production between 0.001 and 0.030 in (0.025–0.75 mm). Above that range, etch times get long, features get coarse relative to the part, and laser or water jet cutting becomes the better option.

### What is half-etching used for?

Etching from one side only, typically to 30–60% of the thickness, creates fold lines that let a flat part be bent by hand into a formed shape, as well as counterbores, recesses and permanent part marking. Depth control is roughly ±10% of stock thickness, so half-etching should not be used where the residual thickness carries a critical load.

### Does photochemical machining need tooling?

Only a phototool — a high-resolution plot on film or glass, costing a small fraction of a die and produced in 1–3 days. That is why the process suits prototypes and evolving designs as well as million-piece production runs, and why adding features to a part costs nothing.

## Alternative processes

- [Laser Cutting](https://manufacturingprocesses.org/processes/cutting/laser-cutting.md): Laser cutting melts, burns or vaporises a narrow kerf through flat stock with a focused beam, producing profiles from a 2D file with no tooling.
- [Punching and Blanking](https://manufacturingprocesses.org/processes/cutting/punching-and-blanking.md): Punching and blanking shear sheet metal between a punch and a die, producing holes and flat outlines at very high speed.
- [Water Jet Cutting](https://manufacturingprocesses.org/processes/cutting/water-jet-cutting.md): Water jet cutting drives a high-pressure jet of water and abrasive through the material, cutting almost anything with no heat-affected zone.
- [Electrical Discharge Machining](https://manufacturingprocesses.org/processes/cutting/electrical-discharge-machining.md): EDM erodes conductive metal with controlled electrical sparks across a dielectric gap, cutting hardened material and sharp internal corners.
- [Electroforming](https://manufacturingprocesses.org/processes/forming/electroforming.md): Electroforming grows a metal shell by electrodeposition onto a mandrel that is afterwards removed, producing thin parts with sub-micron detail.

## Related processes

- [Photo Etching](https://manufacturingprocesses.org/processes/finishing/photo-etching.md): Photo etching masks a surface photographically and etches shallow detail into it for decoration, marking or texture.
- [Punching and Blanking](https://manufacturingprocesses.org/processes/cutting/punching-and-blanking.md): Punching and blanking shear sheet metal between a punch and a die, producing holes and flat outlines at very high speed.
- [Electroforming](https://manufacturingprocesses.org/processes/forming/electroforming.md): Electroforming grows a metal shell by electrodeposition onto a mandrel that is afterwards removed, producing thin parts with sub-micron detail.
- [Laser Cutting](https://manufacturingprocesses.org/processes/cutting/laser-cutting.md): Laser cutting melts, burns or vaporises a narrow kerf through flat stock with a focused beam, producing profiles from a 2D file with no tooling.
- [Electropolishing](https://manufacturingprocesses.org/processes/finishing/electropolishing.md): Electropolishing reverses plating: current dissolves the surface fastest at the peaks, leaving a bright, deburred, passive finish.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/cutting/photochemical-machining)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Plasma Cutting"
category: "Cutting"
subcategory: "Thermal"
materials: ["Metal"]
tolerances: "Conventional plasma ±0.030–0.060 in (±0.75–1.5 mm); high-definition plasma ±0.010–0.020 in (±0.25–0.5 mm); 1–3° of bevel on the cut face"
volumes: "1–5,000 parts; no tooling required"
lead_time: "1–5 business days; same-day for simple plate profiles"
url: https://manufacturingprocesses.org/processes/cutting/plasma-cutting
---

# Plasma Cutting

Plasma cutting blows an ionised gas jet through conductive metal, melting and ejecting a kerf far faster than a laser on thick plate.

- **Category**: [Cutting](https://manufacturingprocesses.org/processes/cutting.md)
- **Family**: Thermal
- **Materials**: Metal
- **Typical tolerances**: Conventional plasma ±0.030–0.060 in (±0.75–1.5 mm); high-definition plasma ±0.010–0.020 in (±0.25–0.5 mm); 1–3° of bevel on the cut face
- **Surface finish**: Cut faces carry visible drag lines, 1–3° of bevel and some dross; edge quality is classified to ISO 9013 rather than by Ra
- **Typical volumes**: 1–5,000 parts; no tooling required
- **Lead time**: 1–5 business days; same-day for simple plate profiles

## Overview

Plasma cutting forces gas through a constricted nozzle and an electric arc, ionizing it into a plasma jet hot enough to melt the workpiece while the gas velocity blows the molten metal out of the kerf. The workpiece must be electrically conductive, which limits it to metals — mild steel, stainless and aluminum in practice.

Its territory is thick conductive plate. Quality cuts run from about **0.030 in to 2 in (0.8–50 mm)**, with severance cuts on high-current systems beyond that, and above roughly 0.75 in (19 mm) plasma is generally faster and cheaper per foot than a laser.

The tradeoffs are dimensional. Kerf is **0.06–0.15 in (1.5–4 mm)**, the cut face carries **1–3° of bevel**, and conventional plasma holds only **±0.030–0.060 in (±0.75–1.5 mm)**, tightening to **±0.010–0.020 in (±0.25–0.5 mm)** on high-definition systems.

## How it works

1. **Arc initiation.** A pilot arc is struck between electrode and nozzle, then transferred to the workpiece once contact is established. The work must be grounded, which is why the process is limited to conductive material.
2. **Gas selection.** Compressed air is the general-purpose choice and is standard for mild steel. Oxygen gives the squarest edge and least dross on mild steel; nitrogen, or nitrogen with hydrogen or argon mixtures, is used on stainless and aluminum to avoid oxidation and improve edge color.
3. **Constriction and swirl.** The nozzle orifice constricts the arc into a high-velocity column, and a swirl ring spins the gas so the jet stays coherent. That swirl direction is why one side of the kerf comes out squarer than the other.
4. **Piercing.** The torch pierces from above with the head raised, then drops to cutting height. Piercing is hard on consumables and pierce count is a real driver of nozzle and electrode life.
5. **Cutting.** The torch follows the profile at a height held by arc-voltage or capacitive control. Speed is high — far faster than water jet, and faster than laser on thick plate — but the kerf is wide and the tolerance loose by comparison.
6. **Dross and bevel.** Too fast leaves high-speed dross on the bottom edge; too slow leaves low-speed dross that is harder to remove. The cut face carries 1–3° of bevel and a heat affected zone of roughly 0.010–0.050 in (0.25–1.3 mm).
7. **Post-processing.** Dross is ground or chipped off. Air plasma cutting of mild steel leaves a nitrogen-enriched cut face that can cause weld porosity, so critical weld preps are ground back to clean metal.

## Design guidelines

### Do not use plasma for holes
Round holes are plasma's weakest feature. Keep any plasma-cut hole at least **1.5–2× material thickness** in diameter, and expect taper and a rough entry. Bolt holes, dowel holes and anything with a fit requirement should be drilled or punched afterward — call them out that way on the drawing.

### Budget the kerf
At 0.06–0.15 in (1.5–4 mm), plasma kerf is one to two orders of magnitude wider than an EDM kerf and several times a laser's. It matters for nesting, for narrow webs, and for any feature dimensioned to a cut edge.

### Allow for bevel and specify the good side
The swirling gas makes one side of the kerf squarer than the other. On a typical torch with clockwise swirl, the squarer face is on the right of the direction of travel, so outside profiles are cut clockwise and internal holes counterclockwise to keep the good side on the part. If one face of the part must be square, say which one.

### Leave grind stock on weld preps
Air plasma cutting of mild steel leaves a nitrogen-enriched surface on the cut face that can produce porosity in a subsequent weld. Specify that weld-prep faces be ground back to clean metal, or use an oxygen or nitrogen-based process gas.

### Choose high-definition plasma when tolerance matters
Conventional plasma holds ±0.030–0.060 in (±0.75–1.5 mm); high-definition systems with finer nozzles and better gas control reach ±0.010–0.020 in (±0.25–0.5 mm) with less bevel. If the drawing needs better than that, the part belongs on a [laser](/processes/cutting/laser-cutting) or [water jet](/processes/cutting/water-jet-cutting).

### Mind distortion on thin and long parts
Plasma puts more heat into the work than laser does. Long thin parts and thin sheet can bow. Below about 0.1 in (2.5 mm), laser cutting is generally both more accurate and less distorting. Cutting speed also tracks the material's melting point and thermal conductivity, which is why copper and brass cut slowly despite being perfectly conductive — see [metal melting points](/charts/metal-melting-points).

| Material | Plasma suitability | Notes |
|---|---|---|
| Mild steel / carbon steel plate | Excellent | The core application; oxygen gives the squarest edge |
| Stainless steel | Good | Nitrogen or nitrogen mixes; edge discolors, needs cleanup before polish |
| Aluminum | Good | Nitrogen or air; dross removal is harder than on steel |
| Copper, brass | Marginal | High conductivity and heat capacity; slow and rough |
| Thin sheet under 0.1 in (2.5 mm) | Marginal | Heat distortion; laser is better |
| Non-conductive materials | Not possible | The arc requires a conductive path to the work |

| Feature | Recommended | Limit | Why |
|---|---|---|---|
| Plate thickness | 0.25–1.5 in (6–38 mm) | ~2 in (50 mm) quality cut | The range where plasma beats laser on cost |
| Hole diameter | ≥ 2× thickness | 1.5× thickness | Small holes come out tapered and rough |
| Kerf allowance | 0.06–0.15 in (1.5–4 mm) | — | Wide kerf affects nesting and webs |
| Web between features | ≥ 2× thickness | 1× thickness | Heat input distorts narrow webs |
| Tolerance expectation | ±0.030 in (±0.75 mm) | ±0.010 in (±0.25 mm) high-definition | Arc wander and bevel |
| Weld prep faces | Grind after cutting | — | Nitrogen-enriched face causes weld porosity |

## Cost drivers

Plasma is the low cost per foot option on thick conductive plate, and its economics are driven by consumables and speed rather than by tooling.

**Thickness and current.** Higher amperage cuts thicker and faster but consumes more power and consumables. Within its range, plasma cutting speed on plate is high enough that machine time is rarely the dominant cost.

**Consumables.** Nozzles and electrodes are wear items with a finite number of pierces and arc-on hours. A nest full of small internal cutouts consumes consumables far faster than the same cut length as open profiles.

**Pierce count.** Piercing is the hardest thing a torch does. Fewer, larger internal features means longer consumable life.

**Secondary operations.** Dross grinding, weld-prep grinding and any drilling of holes that plasma cannot hold to tolerance all add labor after the cut.

**High-definition versus conventional.** A high-definition system costs more per hour but can eliminate a downstream machining operation by holding ±0.010–0.020 in (±0.25–0.5 mm).

Four ways to take cost out:

1. Use plasma for outlines and drill the holes that carry a fit; do not ask the torch for precision it does not have.
2. Consolidate small internal cutouts to reduce pierce count and consumable wear.
3. Reserve stainless and aluminum for laser or water jet where edge appearance matters; plasma edges on those materials need cleanup.
4. Below about 0.25 in (6 mm), price it against [laser cutting](/processes/cutting/laser-cutting) before committing — plasma's advantage appears in thick plate.

## FAQ

### How thick can plasma cut?

Quality cuts run from about 0.030 in to 2 in (0.8–50 mm) depending on system current, with severance cuts possible beyond that on high-amperage systems. Above roughly 0.75 in (19 mm), plasma is generally faster and cheaper per foot of cut than a fiber laser, which is where it earns its place in a fabrication shop.

### What tolerance can plasma cutting hold?

Conventional plasma holds ±0.030–0.060 in (±0.75–1.5 mm). High-definition systems with finer nozzles and tighter gas control reach ±0.010–0.020 in (±0.25–0.5 mm) with less bevel. Anything tighter than that should go to laser or water jet cutting rather than to a plasma table.

### Why are plasma-cut holes poor quality?

The kerf is wide at 0.06–0.15 in (1.5–4 mm), the arc has to pierce and then immediately turn a tight radius, and the cut face carries 1–3° of bevel. Keep plasma-cut holes at least 1.5–2× material thickness in diameter, and drill or ream any hole that carries a fit.

### Can plasma cut aluminum and stainless steel?

Yes — both are conductive, which is the only hard requirement. Nitrogen or nitrogen-based mixtures are used instead of oxygen to limit oxidation. Edges discolor and dross removal is harder than on mild steel, so where edge appearance matters, laser or water jet cutting is usually the better choice.

### Do I need to grind a plasma cut before welding?

For critical welds, yes. Air plasma cutting of mild steel leaves a nitrogen-enriched layer on the cut face that can cause porosity in a subsequent weld. Either specify that weld-prep faces be ground back to clean metal, or have the cut made with an oxygen or nitrogen process gas rather than shop air.

### Plasma or laser — which should I use?

Thickness and tolerance decide. Below about 0.25 in (6 mm), laser is more accurate, cleaner and usually cheaper. Above roughly 0.75 in (19 mm) in mild steel, plasma cuts faster and at lower cost per foot. In between, the choice comes down to whether the part needs the laser's ±0.005 in (±0.13 mm) and square edge.

## Alternative processes

- [Laser Cutting](https://manufacturingprocesses.org/processes/cutting/laser-cutting.md): Laser cutting melts, burns or vaporises a narrow kerf through flat stock with a focused beam, producing profiles from a 2D file with no tooling.
- [Water Jet Cutting](https://manufacturingprocesses.org/processes/cutting/water-jet-cutting.md): Water jet cutting drives a high-pressure jet of water and abrasive through the material, cutting almost anything with no heat-affected zone.
- [Sawing](https://manufacturingprocesses.org/processes/cutting/sawing.md): Sawing separates stock with a toothed blade — band, circular or cold saw — and is the default first operation on bar, tube and plate.
- [Punching and Blanking](https://manufacturingprocesses.org/processes/cutting/punching-and-blanking.md): Punching and blanking shear sheet metal between a punch and a die, producing holes and flat outlines at very high speed.

## Related processes

- [Laser Cutting](https://manufacturingprocesses.org/processes/cutting/laser-cutting.md): Laser cutting melts, burns or vaporises a narrow kerf through flat stock with a focused beam, producing profiles from a 2D file with no tooling.
- [Water Jet Cutting](https://manufacturingprocesses.org/processes/cutting/water-jet-cutting.md): Water jet cutting drives a high-pressure jet of water and abrasive through the material, cutting almost anything with no heat-affected zone.
- [Arc Welding](https://manufacturingprocesses.org/processes/joining/arc-welding.md): Arc welding melts and fuses metal with an electric arc between an electrode and the work, shielded from the atmosphere by gas or flux.
- [Sheet Metal Fabrication](https://manufacturingprocesses.org/processes/forming/sheet-metal-fabrication.md): Sheet metal fabrication cuts, bends and joins flat stock into a finished assembly, typically laser cutting followed by press braking, hardware and welding.
- [Sawing](https://manufacturingprocesses.org/processes/cutting/sawing.md): Sawing separates stock with a toothed blade — band, circular or cold saw — and is the default first operation on bar, tube and plate.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/cutting/plasma-cutting)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Punching and Blanking"
category: "Cutting"
subcategory: "Mechanical"
materials: ["Metal", "Plastic"]
tolerances: "±0.003–0.010 in (±0.08–0.25 mm) on punched features; ±0.002 in (±0.05 mm) achievable in a well-maintained progressive die; features punched in one station relate to each other more tightly than features across stations"
volumes: "Turret punching 1–5,000; progressive dies 10,000–10,000,000+"
lead_time: "Turret punching 1–5 business days; hard tooling 4–12 weeks, then production in hours per thousand"
url: https://manufacturingprocesses.org/processes/cutting/punching-and-blanking
---

# Punching and Blanking

Punching and blanking shear sheet metal between a punch and a die, producing holes and flat outlines at very high speed.

- **Category**: [Cutting](https://manufacturingprocesses.org/processes/cutting.md)
- **Family**: Mechanical
- **Materials**: Metal, Plastic
- **Typical tolerances**: ±0.003–0.010 in (±0.08–0.25 mm) on punched features; ±0.002 in (±0.05 mm) achievable in a well-maintained progressive die; features punched in one station relate to each other more tightly than features across stations
- **Surface finish**: A sheared edge shows four zones: rollover of 10–20% of thickness, a burnished band of about one third of thickness, a fracture zone, and a burr normally held under 10% of material thickness
- **Typical volumes**: Turret punching 1–5,000; progressive dies 10,000–10,000,000+
- **Lead time**: Turret punching 1–5 business days; hard tooling 4–12 weeks, then production in hours per thousand

## Overview

Punching and blanking shear sheet metal between a punch and a die. The difference is which piece you keep: in punching the slug that falls through is scrap and the sheet is the part; in blanking the slug is the part and the sheet becomes skeleton. Both are the fastest way to produce flat metal parts — a progressive die on a mechanical press runs hundreds of strokes per minute and holds **±0.003 in (±0.08 mm)**.

Press force comes straight out of the geometry: **tonnage = cut perimeter (in) × thickness (in) × shear strength (tons/in²)**, and the shear strength of mild steel is about 25 tons/in² (50,000 psi, 345 MPa).

CNC turret punching produces the same sheared edge with standard stocked tools and no dedicated die, which makes it the low-volume route — no tooling cost, but one hit at a time.

## How it works

1. **Clearance selection.** The gap between punch and die per side is the single most important variable: roughly **5–10% of material thickness** for mild steel, 3–6% for soft aluminum, and 10–12% for stainless and harder alloys. Everything about edge quality follows from getting this right.
2. **Rollover.** The punch contacts and first deforms the sheet plastically, drawing the top edge down into a rounded rollover of roughly 10–20% of the material thickness.
3. **Penetration and burnish.** The punch then shears into the material, producing a smooth burnished band that with correct clearance occupies about **one third of the thickness**. This band is the only part of the edge that is dimensionally reliable.
4. **Fracture.** Cracks initiate at both the punch and die edges. With correct clearance they meet cleanly and the slug separates. With too little or too much clearance they miss, producing a secondary shear step or a ragged, tapered fracture zone.
5. **Breakthrough and stripping.** The load releases suddenly at breakthrough — snap-through shock, which is why press and die design account for reverse tonnage. A stripper plate holds the sheet down as the punch withdraws.
6. **Progression.** In a progressive die the strip advances through stations located by pilot holes, gaining holes, forms and coined features until the final station blanks the part free.
7. **Burr formation.** The burr forms on the die side and grows as tooling dulls; **10% of material thickness** is a common maximum acceptable burr height, and it is the practical indicator that punches need regrinding.

## Design guidelines

### Minimum hole diameter
Keep punched holes at least **1× material thickness** in diameter in mild steel and **1.5×** in stainless and harder alloys. Below that the punch is loaded in buckling rather than shear and breaks, and the hole comes out poorly formed.

### Edge distance and hole spacing
Keep at least **2× material thickness** between a hole and a formed or sheared edge, and the same between adjacent holes. Less than that and the material between them bulges or tears as the punch enters.

### Slot proportions
Slot width should be at least **1× material thickness**, and slot length should stay within about 10× the width per punch — longer slots are made in multiple hits or with a dedicated tool.

### Blank corner radii
Give outside corners a radius of at least **0.5× material thickness**. A sharp corner in a blanking die is a stress riser in the tool, wears fastest, and is the first place a die cracks.

### Call out burr direction
The burr always forms on the die side of the cut. On a part where one face contacts a seal, a bearing or a user's hand, the drawing needs to state which face carries the burr — it is free to control and expensive to fix later.

### Design the part around the press
Tonnage scales with cut perimeter, so a lace-like part with a huge perimeter can exceed the press before it exceeds any dimensional limit. Use the formula in the summary, or a shear angle ground into the punch face to spread the cut over the stroke.

### Plan for what happens next
If the part will be bent, keep holes at least 1.5× thickness plus the bend radius from the bend line, and consider grain direction — bending across the rolling direction cracks less. See the [bend radius and K-factor chart](/charts/sheet-metal-bend-radius-k-factor) and specify stock from the [sheet metal gauge chart](/charts/sheet-metal-gauge-chart).

| Material | Punchability | Notes |
|---|---|---|
| Mild steel (CRS, HRPO) | Excellent | The reference material; ~25 tons/in² shear strength |
| Aluminum 5052, 3003 | Excellent | Lower tonnage; use tighter clearance, 3–6% per side |
| 304 / 316 stainless | Moderate | Higher tonnage, work hardens, wider clearance, faster tool wear |
| Galvanized steel | Good | Zinc pickup on tooling; needs frequent cleaning |
| Copper, brass | Excellent | Low tonnage, clean edges |
| Spring steel, hardened stock | Difficult | Cracks at the fracture zone; consider etching or wire EDM |
| Plastic sheet | Situational | Cracks when cold; larger clearance and warm stock help |

| Feature | Recommended | Limit | Why |
|---|---|---|---|
| Hole diameter | ≥ 1.5× thickness | 1× thickness (mild steel) | Punch buckles below this |
| Hole to edge | ≥ 2× thickness | 1.5× thickness | Material bulges or tears |
| Hole to hole | ≥ 2× thickness | 1.5× thickness | Web collapses between holes |
| Slot width | ≥ 1× thickness | 0.8× thickness | Punch strength |
| Blank corner radius | ≥ 0.5× thickness | Sharp | Die corners crack and wear |
| Punch-die clearance | 5–10% of t per side (steel) | 3–12% by material | Sets edge quality and burr |
| Burr height | ≤ 10% of thickness | — | Grows as tooling dulls |

## Cost drivers

The economics split cleanly between hard tooling and turret punching, and the crossover is driven entirely by volume.

**Tooling.** A progressive die is a substantial one-time investment with a lead time of weeks; after that, per-part cost is dominated by press time measured in strokes per minute. Turret punching has no tooling charge but hits one feature at a time.

**Part perimeter drives tonnage.** Tonnage sets which press the job runs on, and press size is a rate. A part that just exceeds a press's capacity moves to a bigger machine and a higher rate for no functional gain.

**Number of stations.** Each hole, form and coin in a progressive die is a station, and stations are die cost.

**Material utilization.** Strip layout, web width and part nesting decide how much of the coil becomes part rather than skeleton. On a high-volume part, a few percent of yield outweighs almost everything else.

**Tool maintenance.** Punches dull, burrs grow, and dies come out for regrinding on a schedule. Stainless and galvanized shorten that interval.

Four ways to take cost out:

1. Below about 5,000–10,000 parts, use turret punching or [laser cutting](/processes/cutting/laser-cutting) and skip the die entirely.
2. Design the strip layout with the toolmaker; web and pitch decisions set material yield for the life of the part.
3. Keep holes at 1× thickness or larger and 2× thickness from edges so standard punches survive.
4. Consolidate features so fewer die stations are needed.

## FAQ

### What is the difference between punching and blanking?

The tooling is the same; the difference is which piece you keep. In punching, the slug pushed through the die is scrap and the remaining sheet is the part. In blanking, the slug is the part and the sheet becomes skeleton scrap. Because the burnished band and burr form the same way in both, the edge quality rules are identical.

### How do I calculate press tonnage for punching?

Tonnage equals cut perimeter in inches multiplied by material thickness in inches multiplied by the material's shear strength in tons per square inch. Mild steel is about 25 tons/in² (50,000 psi, 345 MPa). A 4 in perimeter in 0.125 in mild steel therefore needs about 12.5 tons. Grinding a shear angle into the punch face spreads the cut over the stroke and reduces peak tonnage.

### What is the minimum hole size for punching?

About 1× material thickness in mild steel and 1.5× thickness in stainless and harder alloys. Below that the punch is loaded in buckling rather than shear and breaks quickly. Keep holes at least 2× thickness from any edge or from each other, or the material between them bulges and tears.

### What punch to die clearance should I use?

Roughly 5–10% of material thickness per side for mild steel, 3–6% for soft aluminum, and 10–12% for stainless and harder alloys. Correct clearance makes the cracks from the punch and die edges meet cleanly, producing a burnished band of about one third of the thickness. Wrong clearance produces a secondary shear step or a ragged, tapered fracture zone.

### Which side of a punched part has the burr?

The die side — the side the slug exits. Burr height grows as the punch dulls, and 10% of material thickness is a common maximum acceptable value and the usual trigger for regrinding tooling. If one face of the part must be burr-free for sealing, assembly or handling, state that on the drawing, since controlling burr direction is free at the die-design stage.

### At what volume does a progressive die pay for itself?

Typically somewhere above 5,000–10,000 parts, though the exact point depends on die complexity and part cycle time. Below that, CNC turret punching or laser cutting produces the same part with no tooling investment and no multi-week tooling lead time. Above it, press strokes per minute make the die the clear low-cost route.

## Alternative processes

- [Laser Cutting](https://manufacturingprocesses.org/processes/cutting/laser-cutting.md): Laser cutting melts, burns or vaporises a narrow kerf through flat stock with a focused beam, producing profiles from a 2D file with no tooling.
- [Water Jet Cutting](https://manufacturingprocesses.org/processes/cutting/water-jet-cutting.md): Water jet cutting drives a high-pressure jet of water and abrasive through the material, cutting almost anything with no heat-affected zone.
- [Photochemical Machining](https://manufacturingprocesses.org/processes/cutting/photochemical-machining.md): Photochemical machining masks sheet metal photographically and etches away the unmasked areas, cutting burr-free flat parts with no hard tooling.
- [Die Cutting](https://manufacturingprocesses.org/processes/cutting/die-cutting.md): Die cutting presses a shaped steel rule or rotary die through sheet stock, cutting flat outlines in paper, board, foam, film and textile.
- [Plasma Cutting](https://manufacturingprocesses.org/processes/cutting/plasma-cutting.md): Plasma cutting blows an ionised gas jet through conductive metal, melting and ejecting a kerf far faster than a laser on thick plate.

## Related processes

- [Metal Stamping](https://manufacturingprocesses.org/processes/forming/metal-stamping.md): Metal stamping shears and forms sheet metal between matched dies in a press, producing flat and shallow-formed parts at very high rates.
- [Press Braking](https://manufacturingprocesses.org/processes/forming/press-braking.md): Press braking bends sheet metal by driving a punch into a V-die, forming one straight-line bend at a time.
- [Sheet Metal Fabrication](https://manufacturingprocesses.org/processes/forming/sheet-metal-fabrication.md): Sheet metal fabrication cuts, bends and joins flat stock into a finished assembly, typically laser cutting followed by press braking, hardware and welding.
- [Deep Drawing](https://manufacturingprocesses.org/processes/forming/deep-drawing.md): Deep drawing pushes a sheet metal blank through a die with a punch so the metal flows into a seamless cup or box deeper than its own diameter.
- [Laser Cutting](https://manufacturingprocesses.org/processes/cutting/laser-cutting.md): Laser cutting melts, burns or vaporises a narrow kerf through flat stock with a focused beam, producing profiles from a 2D file with no tooling.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/cutting/punching-and-blanking)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Sawing"
category: "Cutting"
subcategory: "Mechanical"
materials: ["Metal", "Wood", "Plastic", "Composite"]
tolerances: "Band saw ±0.030 in (±0.75 mm) typical, ±0.010 in (±0.25 mm) with careful setup; cold saw ±0.005–0.010 in (±0.13–0.25 mm); abrasive saw ±0.020 in (±0.5 mm) or looser"
volumes: "1 to unlimited; bundle cutting makes high volumes efficient"
lead_time: "Minutes per cut; same-day service at any metal supplier or fabrication shop"
url: https://manufacturingprocesses.org/processes/cutting/sawing
---

# Sawing

Sawing separates stock with a toothed blade — band, circular or cold saw — and is the default first operation on bar, tube and plate.

- **Category**: [Cutting](https://manufacturingprocesses.org/processes/cutting.md)
- **Family**: Mechanical
- **Materials**: Metal, Wood, Plastic, Composite
- **Typical tolerances**: Band saw ±0.030 in (±0.75 mm) typical, ±0.010 in (±0.25 mm) with careful setup; cold saw ±0.005–0.010 in (±0.13–0.25 mm); abrasive saw ±0.020 in (±0.5 mm) or looser
- **Surface finish**: Roughly 125–500 µin Ra (3.2–12.5 µm) as-sawn, with cold saw cuts at the fine end and abrasive cuts at the coarse end
- **Typical volumes**: 1 to unlimited; bundle cutting makes high volumes efficient
- **Lead time**: Minutes per cut; same-day service at any metal supplier or fabrication shop

## Overview

Sawing separates stock with a toothed blade and is almost always the first operation performed on bar, tube, plate and structural section. Three machine types cover the field: the horizontal **band saw**, the general-purpose workhorse for metal; the **cold saw**, a slow-turning circular blade under flood coolant that gives the squarest and cleanest cut; and the **abrasive chop saw**, which is fast, crude and hot.

Blade selection follows one rule above all others: keep at least **3 teeth engaged in the cut** at all times and no more than about 24. Too few and the teeth strip out; too many and the chips cannot clear the gullets.

Expect **±0.030 in (±0.75 mm)** from a production band saw and **±0.005–0.010 in (±0.13–0.25 mm)** from a cold saw, and leave **0.030–0.125 in (0.8–3 mm)** of facing stock per end on anything that will be machined afterward.

## How it works

1. **Tooth pitch selection.** Pitch is chosen from the thickness of material the blade passes through, not from the overall size of the part — a 4 in tube with a 0.120 in wall needs a fine pitch, not a coarse one. Variable-pitch blades (4/6, 6/10, 10/14) are standard because the varying tooth spacing breaks up the harmonic that causes chatter.
2. **Blade speed.** Bimetal bands run roughly 200–300 SFM (60–90 m/min) in mild steel and 100–150 SFM (30–45 m/min) in stainless; carbide-tipped blades in aluminum run an order of magnitude faster. Cold saws turn HSS blades at far lower surface speeds under flood coolant, which is where their finish and squareness come from.
3. **Break-in.** A new blade is run at reduced feed for the first several cuts to hone a controlled radius onto each tooth tip. Skipping break-in strips teeth and is the most common cause of premature blade failure.
4. **Clamping.** The vise must grip on both sides of the blade. Unsupported tube collapses, and unsupported bar rolls into the blade and strips teeth.
5. **The cut.** Feed pressure is set so each tooth takes a real chip. Too light and the teeth rub and work-harden the material — particularly destructive in stainless.
6. **Blade drift.** A dull, mis-tensioned or wrongly guided band wanders, producing an out-of-square cut and a barrel-shaped face. Squareness, not length, is usually the first thing to go wrong.
7. **Deburring.** A saw cut leaves a burr at exit that has to be removed before the next operation.

## Design guidelines

### Leave facing stock
Specify cut length with enough extra material for a facing operation: **0.030–0.060 in (0.8–1.5 mm)** per end for cold-saw cuts and **0.060–0.125 in (1.5–3 mm)** per end for band-saw cuts. Asking a saw for a finished length in one operation is asking for a rework.

### Match the machine to the squareness requirement
Band saws are fast and cheap but hold squareness loosely; cold saws are slower but produce a face square and clean enough to weld or machine directly; abrasive saws are the fastest and the least accurate, and they heat the cut end. Choose deliberately rather than defaulting.

### Pick pitch from the thinnest section the blade crosses

| Section thickness at the cut | Typical variable pitch |
|---|---|
| Under 0.125 in (3 mm) — thin wall tube | 10/14 or 14/18 |
| 0.125–0.25 in (3–6 mm) | 8/12 or 10/14 |
| 0.25–1 in (6–25 mm) | 6/10 or 5/8 |
| 1–3 in (25–75 mm) solid | 4/6 or 3/4 |
| Over 3 in (75 mm) solid | 2/3 or 1.4/2.0 |

### Budget the kerf
Band saw kerf runs **0.035–0.065 in (0.9–1.65 mm)** depending on blade width, cold saw **0.060–0.125 in (1.5–3 mm)**, and abrasive wheels wider still. On short parts cut from bar, kerf plus facing stock can be a substantial share of the material.

### Bundle cutting is where the savings are
Multiple bars clamped and cut together divide machine time across every piece. It requires a machine and vise that can hold the bundle rigidly, and it trades some squareness for throughput.

| Material | Sawing notes |
|---|---|
| Mild and alloy steel | The reference case; bimetal band at 200–300 SFM (60–90 m/min) |
| Stainless steel | Work hardens if the feed is too light — maintain real chip load |
| Aluminum | Carbide-tipped blades, high speed, coarse pitch, chip clearance is everything |
| Titanium | Slow speeds, heavy coolant, sharp blades; heat is the enemy |
| Cast iron | Cuts dry readily; abrasive on blades |
| Plastics and acrylics | Melt and re-weld behind the blade; coarse pitch, high speed, no coolant |
| Wood and composites | Circular and band saws; carbide or diamond for abrasive laminates |

| Feature | Recommended | Limit | Why |
|---|---|---|---|
| Teeth engaged in cut | 6–12 | 3 minimum, ~24 maximum | Fewer strips teeth, more clogs gullets |
| Facing stock per end | 0.060 in (1.5 mm) | 0.030 in (0.8 mm) | Saw cuts are not finished faces |
| Length tolerance, band saw | ±0.030 in (±0.75 mm) | ±0.010 in (±0.25 mm) | Blade drift and clamping |
| Length tolerance, cold saw | ±0.010 in (±0.25 mm) | ±0.005 in (±0.13 mm) | Rigid circular blade, flood coolant |
| Kerf allowance, band saw | 0.050 in (1.3 mm) | 0.035–0.065 in (0.9–1.65 mm) | Blade width dependent |
| Minimum tube wall | 0.030 in (0.8 mm) | Thinner needs support | Wall collapses under vise pressure |

## Cost drivers

Sawing is cheap per cut and expensive in aggregate, because almost every metal part starts with one and the material it consumes never comes back.

**Cuts per bar.** Kerf plus facing stock is lost on every piece. On a 2 in long part cut from bar, that loss can approach 10% of the material.

**Machine choice.** A cold saw costs more per cut than a band saw and can save a facing operation downstream. If the sawn face is a finished face, that trade usually pays.

**Bundle capability.** Cutting a bundle divides machine time across every piece in it and is the largest single lever on cost for volume work.

**Blade life.** Blades are consumables. Correct pitch, correct speed, proper break-in and adequate chip load extend life several times over; the same blade misapplied in stainless can fail in a shift.

**Material.** Stainless and titanium cut slowly and consume blades quickly; aluminum cuts fast with carbide.

Three ways to take cost out:

1. Cut in bundles wherever the machine and the tolerance allow.
2. Use a cold saw when the sawn face is a finished face, and a band saw when it is going to be machined anyway.
3. Choose part lengths that divide cleanly into standard 12 or 20 ft (3.6 or 6 m) mill lengths so remnants are minimized.

## FAQ

### How do I choose the right saw blade pitch?

Keep at least 3 teeth engaged in the cut at all times and no more than about 24, and select from the thickness of material the blade actually passes through rather than the overall part size. A 4 in tube with a 0.120 in wall needs a fine 10/14 pitch, not the coarse pitch its diameter would suggest. Variable-pitch blades are standard because the varying spacing suppresses chatter.

### What tolerance can a saw cut hold?

A production band saw holds about ±0.030 in (±0.75 mm), tightening to ±0.010 in (±0.25 mm) with careful setup. A cold saw holds ±0.005–0.010 in (±0.13–0.25 mm) and produces a much squarer face. Abrasive chop saws are looser than either and heat the cut end. Squareness usually degrades before length does.

### How much extra length should I leave for facing?

0.030–0.060 in (0.8–1.5 mm) per end after a cold saw and 0.060–0.125 in (1.5–3 mm) per end after a band saw. A saw cut is a stock-preparation operation, not a finishing one, and expecting a finished length straight off the saw is the most common cause of rework at the lathe.

### Why do saw blades fail early?

Usually skipped break-in or insufficient chip load. A new blade should run at reduced feed for the first several cuts to hone a controlled radius onto each tooth tip. Feeding too lightly is equally damaging, especially in stainless, where rubbing rather than cutting work-hardens the surface ahead of the teeth.

### What is the kerf of a band saw?

About 0.035–0.065 in (0.9–1.65 mm) depending on blade width, compared with 0.060–0.125 in (1.5–3 mm) for a cold saw and wider still for abrasive wheels. On short parts cut from bar, kerf plus facing stock can consume close to a tenth of the material, which is worth calculating before ordering stock.

### Should I saw or laser cut tube?

Saw a plain length; laser cut anything with features. A saw cut is far cheaper per piece, but once the part needs holes, notches, a coped end or joint geometry, tube laser cutting produces all of it in one operation and eliminates the drilling, notching and fixturing that would otherwise follow the saw.

## Alternative processes

- [Laser Cutting](https://manufacturingprocesses.org/processes/cutting/laser-cutting.md): Laser cutting melts, burns or vaporises a narrow kerf through flat stock with a focused beam, producing profiles from a 2D file with no tooling.
- [Water Jet Cutting](https://manufacturingprocesses.org/processes/cutting/water-jet-cutting.md): Water jet cutting drives a high-pressure jet of water and abrasive through the material, cutting almost anything with no heat-affected zone.
- [Plasma Cutting](https://manufacturingprocesses.org/processes/cutting/plasma-cutting.md): Plasma cutting blows an ionised gas jet through conductive metal, melting and ejecting a kerf far faster than a laser on thick plate.
- [Tube and Profile Laser Cutting](https://manufacturingprocesses.org/processes/cutting/tube-and-profile-laser-cutting.md): Tube laser cutting rotates and feeds tube or profile through a laser head, cutting holes, slots and joint geometry along its length.

## Related processes

- [CNC Turning](https://manufacturingprocesses.org/processes/cutting/cnc-turning.md): CNC turning rotates the workpiece against a stationary tool, producing round parts with excellent concentricity and short cycle times.
- [CNC Milling](https://manufacturingprocesses.org/processes/cutting/cnc-milling.md): CNC milling removes material with a rotating multi-flute cutter moved along programmed paths, producing prismatic parts from solid stock.
- [Swiss Screw Machining](https://manufacturingprocesses.org/processes/cutting/swiss-screw-machining.md): Swiss screw machining feeds bar stock through a guide bushing so cutting happens right next to the support, holding tight tolerances on slender parts.
- [Tube and Profile Laser Cutting](https://manufacturingprocesses.org/processes/cutting/tube-and-profile-laser-cutting.md): Tube laser cutting rotates and feeds tube or profile through a laser head, cutting holes, slots and joint geometry along its length.
- [Joinery](https://manufacturingprocesses.org/processes/joining/joinery.md): Joinery cuts interlocking geometry into timber so the parts locate and hold each other, with or without adhesive or fasteners.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/cutting/sawing)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Swiss Screw Machining"
category: "Cutting"
subcategory: "Mechanical"
materials: ["Metal", "Plastic"]
tolerances: "±0.0005 in (±0.013 mm) routine; ±0.0002 in (±0.005 mm) achievable on diameters; ±0.001 in (±0.025 mm) on lengths"
volumes: "500–1,000,000+ parts"
lead_time: "2–4 weeks for first articles including setup; production releases in days once the setup is proven"
url: https://manufacturingprocesses.org/processes/cutting/swiss-screw-machining
---

# Swiss Screw Machining

Swiss screw machining feeds bar stock through a guide bushing so cutting happens right next to the support, holding tight tolerances on slender parts.

- **Category**: [Cutting](https://manufacturingprocesses.org/processes/cutting.md)
- **Family**: Mechanical
- **Materials**: Metal, Plastic
- **Typical tolerances**: ±0.0005 in (±0.013 mm) routine; ±0.0002 in (±0.005 mm) achievable on diameters; ±0.001 in (±0.025 mm) on lengths
- **Surface finish**: 32 µin Ra (0.8 µm) typical; 16 µin (0.4 µm) achievable with finishing tools
- **Typical volumes**: 500–1,000,000+ parts
- **Lead time**: 2–4 weeks for first articles including setup; production releases in days once the setup is proven

## Overview

Swiss screw machining — properly, Swiss-type or sliding-headstock turning — feeds bar stock through a carbide guide bushing and cuts immediately in front of it. Because the unsupported length between the support and the tool never changes, a slender part does not deflect no matter how long it becomes, and diameters hold **±0.0002 in (±0.005 mm)** on parts that a conventional lathe could not hold round.

Bar capacity runs roughly **0.020–1.5 in (0.5–38 mm)**, with 20 mm and 32 mm class machines covering most work. Several tool groups cut at once, a sub-spindle finishes the back end, and the machine runs unattended off a bar feeder, so cycle times fall to well under a minute and volumes of 500 to several million are normal.

It is the process behind bone screws, contact pins, needle hubs, fuel injector components, watch parts and connector shells — small, long, tight-tolerance turned parts made in quantity.

## How it works

1. **Bar preparation.** The guide bushing rides directly on the bar OD, so stock must be straight and held to a tight, consistent diameter — centerless-ground bar is normal for close-tolerance work. Poor bar quality shows up directly as diameter variation on the finished part.
2. **Guide bushing setup.** Bushing clearance is set to a few ten-thousandths of an inch over the bar diameter, roughly **0.0002–0.0005 in (5–13 µm)**. Too tight and the bar galls and seizes; too loose and the accuracy advantage disappears.
3. **Sliding headstock motion.** The headstock feeds the bar along Z while the tool holds a fixed Z position. The profile is generated by coordinating that feed with the tool's X motion, which is why a long slender diameter comes out straight instead of tapered.
4. **Simultaneous cutting.** Gang tools, back-working tools and live tools engage at the same time on different features, so the cycle is the longest single operation rather than the sum of them.
5. **Live-tool operations.** Cross-holes, flats, slots and thread milling are done in-cycle with driven tools; thread whirling cuts deep single-lead threads such as bone screw profiles in one pass.
6. **Sub-spindle pickoff.** The sub-spindle grips the finished front end, the part is cut off, and back-end facing, drilling, chamfering and tapping happen while the main spindle already starts the next part.
7. **Unattended running.** Straight cutting oil, often delivered through the tool, controls chips in small deep holes. Parts drop into a catcher or conveyor and the bar feeder reloads, allowing lights-out production.

## Design guidelines

### Play to the slenderness advantage
Swiss earns its rate on parts with a **length-to-diameter ratio above about 3:1**, and handles 20:1 and beyond. A short, stubby part with a large diameter is cheaper on a conventional chucker lathe.

### Respect the headstock stroke
Tooling reaches only a limited length past the bushing before the headstock must retract and re-grip. Usable stroke is commonly **4–12 in (100–300 mm)** depending on machine class; parts longer than that need a re-grip, which leaves a witness and introduces a tolerance step at that point in the length.

### Minimum diameters and features
Turned diameters down to about **0.010 in (0.25 mm)** and cross-holes of similar size are routine. Wall thickness on small turned tubes should stay at or above **0.010 in (0.25 mm)**. These numbers are an order of magnitude finer than conventional turning because the support is right at the cut.

### Use free-machining bar wherever you can
303 stainless, C360 brass, 12L14 steel and 2011 aluminum run far faster and give better finish and chip control than 304, 316 or titanium. Since the process is priced by the second, alloy choice moves the part price more than geometry does.

### Threads
Rolled and whirled threads are stronger and faster than single-point threading on small diameters. Use standard sizes from the [thread size chart](/charts/thread-size-chart) and [tap drill chart](/charts/tap-drill-chart); a custom pitch means a custom tool for a part that may run in the millions.

### Tolerance and finish
Call out **±0.0005 in (±0.013 mm)** as the general tolerance and reserve **±0.0002 in (±0.005 mm)** for the diameters that mate. Lengths hold looser than diameters, typically ±0.001 in (±0.025 mm). Finish is 32 µin Ra (0.8 µm) typical and 16 µin (0.4 µm) achievable ([surface finish chart](/charts/surface-finish-chart)).

| Material | Suitability | Watch for |
|---|---|---|
| 303 stainless | Excellent | The default Swiss stainless; best chip control |
| C360 brass | Excellent | Fastest cycles, best finish |
| 12L14 / 1215 steel | Excellent | Free-cutting; not for welded or plated-critical parts |
| 316 / 316L stainless | Good | Slower and gummier than 303; standard for implants |
| Ti-6Al-4V, nitinol | Difficult | Low speeds, sharp tools, high tool cost per part |
| 2011 / 6061 aluminum | Excellent | 2011 chips better; 6061 makes stringy chips |
| PEEK, acetal, PTFE | Good | Low cutting forces; support and coolant control matter |

| Feature | Recommended | Limit | Why |
|---|---|---|---|
| Length : diameter | 3:1 and up | 20:1+ | Below 3:1 a chucker is cheaper |
| Bar diameter | 0.062–1.25 in (1.6–32 mm) | 0.020–1.5 in (0.5–38 mm) | Machine bar capacity |
| Minimum turned diameter | 0.020 in (0.5 mm) | 0.010 in (0.25 mm) | Tool pressure vs part stiffness |
| Tube wall thickness | 0.020 in (0.5 mm) | 0.010 in (0.25 mm) | Collapse under tool pressure |
| Part length per grip | Within machine stroke | 4–12 in (100–300 mm) | Re-grip adds a witness line |
| General tolerance | ±0.0005 in (±0.013 mm) | ±0.0002 in (±0.005 mm) | Reserve the tight band for mating features |

## Cost drivers

Swiss work is priced per second of cycle plus a substantial one-time setup, so the economics invert compared with prototype machining: setup is large, per-part cost is small, and quantity decides everything.

**Setup.** A Swiss job may use a dozen or more tools, all of which must be set, offset and proven out. That setup is charged once, which is why first articles feel expensive and the tenth thousand part feels almost free.

**Cycle time and tool count.** Every feature that cannot be overlapped with another adds directly to the cycle. Features that can be cut simultaneously with existing tool groups are nearly free.

**Material.** Free-machining alloys cut faster and last longer on tooling; 316, titanium and nitinol multiply both cycle time and tool consumption. Bar stock must also be precision-ground for close work, which costs more per pound than mill-run bar.

**Secondary operations.** Deburring, passivation, plating and cleaning on very small parts can rival machining cost, and handling millions of tiny parts is not trivial.

Four ways to take cost out:

1. Get above a few thousand pieces before choosing Swiss — below that, [CNC turning](/processes/cutting/cnc-turning) absorbs the setup better.
2. Specify 303 rather than 304, and 12L14 rather than 1018, wherever corrosion and weldability allow.
3. Keep the part within one headstock grip so no re-grip witness or tolerance step is needed.
4. Apply ±0.0002 in (±0.005 mm) only to mating diameters and leave everything else at ±0.0005 in (±0.013 mm).

## FAQ

### What tolerance can Swiss screw machining hold?

±0.0005 in (±0.013 mm) is routine as a general tolerance and ±0.0002 in (±0.005 mm) is achievable on diameters. Lengths hold looser, typically ±0.001 in (±0.025 mm). The accuracy comes from the guide bushing supporting the bar immediately behind the cut, so the unsupported length never grows as the part gets longer.

### What is the difference between Swiss machining and CNC turning?

On a conventional lathe the tool moves along a stationary rotating part held only at the chuck, so a long slender part deflects. On a Swiss machine the headstock slides and feeds the bar through a guide bushing, and cutting happens right at that support. The practical result is that Swiss handles length-to-diameter ratios of 20:1 and beyond while chuck turning is limited to about 3:1 unsupported.

### What size parts can Swiss screw machines make?

Bar capacity is roughly 0.020–1.5 in (0.5–38 mm) in diameter, with 20 mm and 32 mm class machines covering the bulk of production. Turned diameters down to about 0.010 in (0.25 mm) are routine. Part length per grip is limited by headstock stroke, commonly 4–12 in (100–300 mm) depending on machine class.

### What is the minimum volume for Swiss screw machining?

There is no hard minimum, but the setup involves a dozen or more tools that must all be set and proven, and that cost is charged once. Below roughly a few thousand pieces the setup dominates and conventional CNC turning is usually cheaper; above it, cycle times measured in seconds make Swiss the low-cost option.

### Which materials run best on a Swiss machine?

Free-machining grades: 303 stainless, C360 brass, 12L14 steel and 2011 aluminum give the fastest cycles, best chip control and longest tool life. 316L, titanium and nitinol all run but multiply both cycle time and tool consumption, which matters because the part is priced per second of cycle.

### Why does bar stock quality matter so much on Swiss machines?

The guide bushing rides directly on the bar outside diameter with only about 0.0002–0.0005 in (5–13 µm) of clearance. Any variation in bar diameter or straightness translates into diameter variation on the finished part, or into galling and seizure if the bar is oversize. Centerless-ground bar is standard for close-tolerance work.

## Alternative processes

- [CNC Turning](https://manufacturingprocesses.org/processes/cutting/cnc-turning.md): CNC turning rotates the workpiece against a stationary tool, producing round parts with excellent concentricity and short cycle times.
- [Metal Injection Molding](https://manufacturingprocesses.org/processes/forming/metal-injection-molding.md): Metal injection molding molds a powder-and-binder feedstock exactly like a plastic, then debinds and sinters it into a dense small metal part.
- [Swaging](https://manufacturingprocesses.org/processes/forming/swaging.md): Swaging hammers or presses the end of a tube or bar inward against a die to reduce or taper its diameter without removing any material.
- [Powder Metallurgy (Press and Sinter)](https://manufacturingprocesses.org/processes/forming/powder-metallurgy-press-and-sinter.md): Press-and-sinter powder metallurgy compacts metal powder in a rigid die and sinters it below its melting point into a finished net-shape part.

## Related processes

- [CNC Turning](https://manufacturingprocesses.org/processes/cutting/cnc-turning.md): CNC turning rotates the workpiece against a stationary tool, producing round parts with excellent concentricity and short cycle times.
- [CNC Milling](https://manufacturingprocesses.org/processes/cutting/cnc-milling.md): CNC milling removes material with a rotating multi-flute cutter moved along programmed paths, producing prismatic parts from solid stock.
- [CNC Machining](https://manufacturingprocesses.org/processes/cutting/cnc-machining.md): CNC machining removes material from solid stock with a programmed cutting tool, producing accurate parts directly from a CAD model with no tooling.
- [5-Axis CNC Machining](https://manufacturingprocesses.org/processes/cutting/5-axis-cnc-machining.md): 5-axis CNC machining tilts and rotates the part or the head so the tool reaches every face and stays short and rigid on complex surfaces.
- [Electrical Discharge Machining](https://manufacturingprocesses.org/processes/cutting/electrical-discharge-machining.md): EDM erodes conductive metal with controlled electrical sparks across a dielectric gap, cutting hardened material and sharp internal corners.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/cutting/swiss-screw-machining)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Tube and Profile Laser Cutting"
category: "Cutting"
subcategory: "Thermal"
materials: ["Metal"]
tolerances: "±0.005 in (±0.13 mm) on feature position within a setup; roughly ±0.020 in (±0.5 mm) cumulative over multi-meter lengths; overall accuracy is also bounded by the mill tolerance of the incoming tube"
volumes: "10–50,000 parts; bundle loading makes production runs efficient"
lead_time: "3–10 business days; production runs quote by the bar"
url: https://manufacturingprocesses.org/processes/cutting/tube-and-profile-laser-cutting
---

# Tube and Profile Laser Cutting

Tube laser cutting rotates and feeds tube or profile through a laser head, cutting holes, slots and joint geometry along its length.

- **Category**: [Cutting](https://manufacturingprocesses.org/processes/cutting.md)
- **Family**: Thermal
- **Materials**: Metal
- **Typical tolerances**: ±0.005 in (±0.13 mm) on feature position within a setup; roughly ±0.020 in (±0.5 mm) cumulative over multi-meter lengths; overall accuracy is also bounded by the mill tolerance of the incoming tube
- **Surface finish**: Edge quality follows ISO 9013 thermal cut classification; nitrogen-cut stainless is bright and weld-ready, oxygen-cut mild steel carries an oxide layer
- **Typical volumes**: 10–50,000 parts; bundle loading makes production runs efficient
- **Lead time**: 3–10 business days; production runs quote by the bar

## Overview

Tube laser cutting feeds round, square, rectangular or structural profile through chucks that rotate and advance it under a laser head, cutting holes, slots, copes, miters and joint geometry anywhere along its length. One operation replaces the whole sequence of saw, drill, mill, notch and jig, working directly from a 3D model.

Common machines take **0.5–6 in (12–150 mm)** diameter or across flats, with large machines reaching 10–12 in (250–300 mm), fed from bars typically **20–27 ft (6–8 m)** long. Wall thickness capacity on typical machines runs to about **0.5 in (12 mm)** in steel.

The real payoff is tab-and-slot joinery. Features cut into mating tubes locate the assembly without a weld fixture, and the money saved in the weld shop usually exceeds the cost of the cutting.

## How it works

1. **Bulk loading.** A bundle loader feeds full-length bars into the machine automatically, which is what makes the process economic on production runs rather than one-offs.
2. **Chucking and rotation.** Self-centering chucks grip the profile and rotate it under the head while a feed system advances it along its axis. On square and rectangular section, the machine indexes face to face.
3. **Profile measurement.** Better machines measure the actual section before cutting to correct for the twist, seam position and dimensional variation that mill-supplied tube always carries. This matters because the laser cannot cut more accurately than the tube it is given.
4. **Programming from the 3D model.** CAM wraps 2D features onto the curved or faceted surface and generates the intersection curves for copes and miters directly from the assembly geometry.
5. **Cutting.** A 2D head cuts perpendicular to the surface; a 3D or 5-axis head tilts to cut bevels up to about 45° for weld preparation and to keep the beam normal to the wall around the corner radii of square section.
6. **Assist gas.** The same rules as flat sheet: oxygen on mild steel for speed with an oxide edge, nitrogen on stainless and aluminum for a bright, weld-ready edge.
7. **Slug management and unloading.** Slugs from internal cutouts drop inside the tube and must be able to escape, or they rattle around in the finished weldment. Finished parts are unloaded and the remnant is ejected.

## Design guidelines

### Design tab-and-slot joints
This is the reason to use the process. Cut a tab on one member and a matching slot in the other so the weldment self-locates and self-squares. Allow **0.004–0.010 in (0.1–0.25 mm)** of clearance per side — tight enough to locate, loose enough to assemble with mill tolerance on the tube.

### Understand where the tolerance actually comes from
Feature-to-feature position within one setup holds around **±0.005 in (±0.13 mm)**, but over a multi-meter part cumulative length is looser, realistically **±0.020 in (±0.5 mm)**. More importantly, the raw tube is only held to mill tolerance on OD, wall and straightness, and no laser can improve on that. If a dimension is critical, dimension it from a cut feature, not from the tube's own surface.

### Account for the weld seam
Welded (ERW) tube carries an internal bead that can foul close-fitting inserts and shifts wall thickness locally. If features must clear it, specify seam orientation on the print or call out DOM or seamless tube.

### Apply the flat-sheet feature rules to the wall
Minimum hole diameter about **1× wall thickness** (1.5× in stainless), minimum slot width 1× wall, and minimum web between features 1× wall. On square and rectangular section, keep features off the corner radii, which typically run 1.5–2.5× wall thickness on structural tube.

### Leave slugs a way out
Internal cutouts drop slugs inside the tube. Either provide an opening they can exit through, or accept that they will be tack-held by micro-tabs and removed manually.

### Use the bevel capability for weld prep
A 3D head cuts weld bevels up to about 45° in the same operation, eliminating a grinding step on structural joints.

| Profile / material | Suitability | Notes |
|---|---|---|
| Round tube, mild steel | Excellent | The core application; copes and miters cut directly |
| Square and rectangular tube | Excellent | Machine indexes face to face; avoid features on corner radii |
| Angle, channel, flat bar | Good | Open sections need support against twist |
| I-beam and structural section | Machine dependent | Large-format machines only; confirm capacity first |
| Stainless tube | Good | Nitrogen assist for a bright, weld-ready edge |
| Aluminum tube | Good | Slower than steel; dross on the inside wall |
| Very thin wall under 0.040 in (1 mm) | Marginal | Distortion and part collapse under chuck pressure |

| Feature | Recommended | Limit | Why |
|---|---|---|---|
| Tube diameter | 1–6 in (25–150 mm) | 0.5–12 in (12–300 mm) | Machine chuck capacity |
| Wall thickness | 0.06–0.25 in (1.5–6 mm) | ~0.5 in (12 mm) | Laser power and edge quality |
| Hole diameter | ≥ 1.5× wall | 1× wall | Pierce quality on a curved surface |
| Tab-to-slot clearance | 0.004–0.010 in (0.1–0.25 mm) | 0.002 in (0.05 mm) | Mill tolerance on the tube itself |
| Feature to tube end | ≥ 1× wall thickness | — | Edge distortion at the cut end |
| Bevel angle | ≤ 45° | Head dependent | 3D head tilt range |

## Cost drivers

Tube laser competes not against another cutting process but against an entire fabrication sequence, and that is how it should be priced.

**Operations eliminated.** A bracket that would have taken a saw cut, two drilling setups, a notching operation and a weld fixture becomes one machine cycle. Compare total delivered cost, not cost per cut.

**Cut length and pierce count.** As on flat sheet, every hole is a pierce plus a contour, and pierce time grows with wall thickness.

**Material utilization.** You buy full bars. A part length that divides evenly into a 20–24 ft bar leaves almost no remnant; one that divides badly wastes several feet per bar.

**Setup and profile changes.** Changing section size or shape means re-chucking and re-proving. Grouping parts of the same section into one run matters.

**Weld fixture avoided.** Tab-and-slot geometry can remove a welding fixture entirely, which on a low-to-mid volume weldment is often the single largest saving.

Four ways to take cost out:

1. Design self-locating tab-and-slot joints so no weld fixture is needed.
2. Choose a part length that nests cleanly into standard 20–24 ft bar.
3. Standardize on one or two tube sections across the assembly to avoid setup changes.
4. Cut weld bevels on the machine instead of grinding them afterward.

## FAQ

### What size tube can a tube laser cut?

Common machines handle 0.5–6 in (12–150 mm) diameter or across flats, with large-format machines reaching 10–12 in (250–300 mm). Wall thickness capacity is typically up to about 0.5 in (12 mm) in steel. Bars are usually fed at 20–27 ft (6–8 m) lengths from an automatic bundle loader.

### What tolerance does tube laser cutting hold?

About ±0.005 in (±0.13 mm) on feature position within one setup, opening to roughly ±0.020 in (±0.5 mm) cumulatively over a multi-meter part. The bigger constraint is that the raw tube is only held to mill tolerance on OD, wall and straightness, and the laser cannot improve on the section it is given.

### What are tab-and-slot joints and why do they matter?

They are interlocking features cut into mating tubes so the weldment locates and squares itself without a fixture. Allow 0.004–0.010 in (0.1–0.25 mm) of clearance per side. On low and mid volume weldments, eliminating the weld fixture is usually a larger saving than the cost of the laser cutting itself.

### Does the weld seam in ERW tube cause problems?

It can. Welded tube carries an internal bead that fouls close-fitting inserts and locally changes the effective wall. If features must clear it, specify seam orientation on the drawing or call out DOM or seamless tube instead. Better machines also measure the actual section before cutting to correct for seam position and twist.

### Can a tube laser cut bevels for welding?

Yes, on machines with a 3D or 5-axis head, typically up to about 45°. Cutting the weld preparation in the same operation removes a separate grinding step, which is a meaningful saving on structural joints where every member needs prep.

### When is sawing still the better choice?

When the part is a plain length with no features. A saw cut is far cheaper per piece than a laser cycle, and tube laser earns its cost through features, copes and joint geometry. Once a part needs holes, notches or a coped end, the laser usually wins on total cost.

## Alternative processes

- [Sawing](https://manufacturingprocesses.org/processes/cutting/sawing.md): Sawing separates stock with a toothed blade — band, circular or cold saw — and is the default first operation on bar, tube and plate.
- [Plasma Cutting](https://manufacturingprocesses.org/processes/cutting/plasma-cutting.md): Plasma cutting blows an ionised gas jet through conductive metal, melting and ejecting a kerf far faster than a laser on thick plate.
- [Water Jet Cutting](https://manufacturingprocesses.org/processes/cutting/water-jet-cutting.md): Water jet cutting drives a high-pressure jet of water and abrasive through the material, cutting almost anything with no heat-affected zone.
- [Punching and Blanking](https://manufacturingprocesses.org/processes/cutting/punching-and-blanking.md): Punching and blanking shear sheet metal between a punch and a die, producing holes and flat outlines at very high speed.

## Related processes

- [Laser Cutting](https://manufacturingprocesses.org/processes/cutting/laser-cutting.md): Laser cutting melts, burns or vaporises a narrow kerf through flat stock with a focused beam, producing profiles from a 2D file with no tooling.
- [Tube and Section Bending](https://manufacturingprocesses.org/processes/forming/tube-and-section-bending.md): Tube and section bending draws or presses metal tube and profile around a former, with an internal mandrel wherever the wall would otherwise collapse.
- [Sheet Metal Fabrication](https://manufacturingprocesses.org/processes/forming/sheet-metal-fabrication.md): Sheet metal fabrication cuts, bends and joins flat stock into a finished assembly, typically laser cutting followed by press braking, hardware and welding.
- [Arc Welding](https://manufacturingprocesses.org/processes/joining/arc-welding.md): Arc welding melts and fuses metal with an electric arc between an electrode and the work, shielded from the atmosphere by gas or flux.
- [Sawing](https://manufacturingprocesses.org/processes/cutting/sawing.md): Sawing separates stock with a toothed blade — band, circular or cold saw — and is the default first operation on bar, tube and plate.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/cutting/tube-and-profile-laser-cutting)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Water Jet Cutting"
category: "Cutting"
subcategory: "Mechanical"
materials: ["Metal", "Plastic", "Glass", "Ceramic", "Composite", "Wood"]
tolerances: "±0.005 in (±0.13 mm) typical on thin material; ±0.003 in (±0.076 mm) achievable; ±0.010–0.020 in (±0.25–0.5 mm) on thick plate; 1–3° of natural taper unless a compensating head is used"
volumes: "1–5,000 parts; no tooling, so single pieces are routine"
lead_time: "1–5 business days; often same-week for single parts"
url: https://manufacturingprocesses.org/processes/cutting/water-jet-cutting
---

# Water Jet Cutting

Water jet cutting drives a high-pressure jet of water and abrasive through the material, cutting almost anything with no heat-affected zone.

- **Category**: [Cutting](https://manufacturingprocesses.org/processes/cutting.md)
- **Family**: Mechanical
- **Materials**: Metal, Plastic, Glass, Ceramic, Composite, Wood
- **Typical tolerances**: ±0.005 in (±0.13 mm) typical on thin material; ±0.003 in (±0.076 mm) achievable; ±0.010–0.020 in (±0.25–0.5 mm) on thick plate; 1–3° of natural taper unless a compensating head is used
- **Surface finish**: Edge quality is specified on a Q1–Q5 scale, from a fast separation cut with visible striations to a slow finish cut with a nearly striation-free face; roughness increases toward the exit side of the cut
- **Typical volumes**: 1–5,000 parts; no tooling, so single pieces are routine
- **Lead time**: 1–5 business days; often same-week for single parts

## Overview

Water jet cutting drives water through a jeweled orifice at **60,000 psi (4,100 bar)** — 90,000 psi (6,200 bar) on high-pressure pumps — and, for anything harder than foam, entrains garnet abrasive into the stream in a mixing tube. The resulting jet erodes its way through material rather than melting it, so there is **no heat affected zone at all**: hardened steel keeps its temper, aluminum does not distort, and composites do not delaminate from heat.

It is the most material-agnostic cutting process available. The same machine cuts 6 in (150 mm) steel plate, glass, stone, titanium, carbon fiber, rubber gasket stock and food.

Kerf is **0.030–0.060 in (0.76–1.5 mm)**, wider than a laser and narrower than plasma. Tolerance is **±0.005 in (±0.13 mm)** on thin material with modern control, opening as thickness grows.

## How it works

1. **Pressure generation.** An intensifier or direct-drive pump raises water to 60,000 psi (4,100 bar). The high-pressure plumbing, seals and check valves are the machine's main maintenance burden.
2. **Orifice.** Water passes through a ruby, sapphire or diamond orifice of **0.010–0.014 in (0.25–0.35 mm)**, emerging as a coherent jet at roughly two to three times the speed of sound in air.
3. **Abrasive entrainment.** For metals, stone, glass and composites, garnet — most commonly **80 mesh** — is drawn into a mixing chamber and accelerated down a carbide mixing tube of **0.030–0.045 in (0.76–1.1 mm)** bore. That mixing tube bore sets the kerf. Soft materials such as foam, gasket, felt and food are cut with pure water and a much finer kerf.
4. **Piercing.** The jet pierces from a standing start, often with a lower-pressure ramp or a slight dwell on brittle material to avoid cracking or delamination.
5. **Cutting and quality selection.** Feed rate is chosen from a quality scale, conventionally **Q1 to Q5**: Q1 is a fast separation cut with pronounced striations at the bottom edge, Q5 is a slow finish cut with a nearly striation-free face. Moving from a separation cut to a finish cut can cut feed rate several times over.
6. **Taper control.** The jet loses energy with depth, so the kerf is naturally wider at the top — roughly 1–3° on a fixed head. Tilting taper-compensation heads swing the jet to bring the cut face back to near-perpendicular.
7. **Catcher tank and abrasive removal.** Spent garnet settles in the tank and is periodically removed. Abrasive consumption of roughly 0.5–1.0 lb/min (0.23–0.45 kg/min) is a direct, continuous consumable cost.

## Design guidelines

### Minimum hole and slot size
Keep holes at least **1× material thickness** in diameter and never smaller than about 1.5× the kerf. A hole narrower than the jet cannot be cut at all, and one only slightly larger comes out tapered and rough because the jet has no room to develop.

### Expect and control taper
A fixed head leaves **1–3° of taper**, wider at the entry side. On parts where the cut face is a mating or sealing surface, either specify a taper-compensating head or state which face is dimensionally controlled. On thin material taper is negligible; on 2 in (50 mm) plate it is a real dimension.

### Choose an edge quality deliberately
Striations increase with thickness and feed rate, so the bottom of a thick cut is always rougher than the top. Specify the quality level you need rather than the best available — a Q3 cut on a bracket outline and a Q5 cut only on the one edge that seals is far cheaper than Q5 everywhere.

### Use the absence of a heat affected zone
This is the reason to choose water jet over laser or plasma. Pre-hardened, tempered, plated or heat-treated stock can be cut without altering it, and there is no recast, no hardened cut edge to spoil a subsequent tapping operation, and no delamination on composites.

### Stack thin sheets, but only when the material allows
Multiple sheets of the same material can be clamped and cut in one pass, dividing cost per part — specify stock from the [sheet metal gauge chart](/charts/sheet-metal-gauge-chart) so every layer is the same nominal thickness. Sheets must be tightly clamped — any gap lets the jet spread between layers and taper the lower sheets.

### Watch for delamination and brittle fracture at pierce points
Laminates, composites and glass can crack at the pierce. Pierce in scrap and lead in, or use a low-pressure pierce, and expect the shop to ask for a lead-in allowance around every internal cutout.

| Material | Water jet suitability | Notes |
|---|---|---|
| Steel, stainless, tool steel | Excellent | To ~6 in (150 mm) practical; keeps temper, no recast |
| Aluminum | Excellent | No thermal distortion on thin or long parts |
| Titanium, Inconel | Excellent | Common in aerospace where HAZ is unacceptable |
| Glass, stone, tile | Excellent | Low-pressure pierce required to avoid cracking |
| Carbon fiber, G10, laminates | Excellent | No heat delamination; abrasive wear on tooling is nil |
| Foam, rubber, gasket, felt | Excellent | Pure-water cutting, no abrasive, very fine kerf |
| Copper, brass | Excellent | No reflectivity problem, unlike laser |
| Tempered glass | Not possible | Shatters; must be cut before tempering |
| Very thin sheet under 0.020 in | Marginal | Jet deflection makes laser cutting more accurate |

| Feature | Recommended | Limit | Why |
|---|---|---|---|
| Hole diameter | ≥ 1× thickness | ~1.5× kerf, 0.060 in (1.5 mm) | Jet needs room to cut a full circle |
| Material thickness | ≤ 2 in (50 mm) | ~6 in (150 mm) practical | Speed and taper degrade with depth |
| Internal corner radius | ≥ 0.5× kerf | Kerf radius | The jet is round |
| Edge taper allowance | 1–3° | Near zero with a tilting head | Jet energy falls with depth |
| Web between features | ≥ 1× thickness | 0.5× thickness | Jet spread and part stability |

## Cost drivers

Water jet is priced by the minute, and the minute is set by material, thickness and the edge quality you ask for.

**Thickness times cut length.** Feed rate falls steeply with thickness. A 2 in (50 mm) plate profile can take ten times as long as the same profile in 0.25 in (6 mm) stock.

**Edge quality.** The quality scale is a feed rate multiplier. Cutting an entire part at finish quality when only one edge seals is the single most common source of avoidable cost.

**Abrasive.** Garnet flows continuously at roughly 0.5–1.0 lb/min (0.23–0.45 kg/min) whenever the jet is on, and it is consumed, not recycled, on most machines.

**Pierce count.** Each internal cutout requires a pierce, which is slower than cutting and harder on the mixing tube.

**Consumables and maintenance.** Orifices, mixing tubes and high-pressure seals wear continuously and are built into the hourly rate.

Four ways to take cost out:

1. Specify finish quality only on the edges that need it and separation quality everywhere else.
2. Stack thin sheets of the same material so one cut produces several parts.
3. Reduce internal cutouts, or enlarge and consolidate them, to cut pierce count.
4. If the part is thin steel with no heat sensitivity, compare against [laser cutting](/processes/cutting/laser-cutting) — water jet earns its rate on thickness, hardness and heat-sensitive materials, not on thin mild steel.

## FAQ

### How thick can a water jet cut?

About 6 in (150 mm) in steel is the practical limit, with thicker cuts possible but slow enough to be uneconomic. Feed rate falls steeply with thickness, so a 2 in (50 mm) profile can take roughly ten times as long as the same profile in 0.25 in (6 mm) stock. Softer materials such as foam and gasket stock cut far thicker.

### What tolerance does water jet cutting hold?

±0.005 in (±0.13 mm) is typical on thin material with a modern controller and ±0.003 in (±0.076 mm) is achievable. On thick plate it opens to ±0.010–0.020 in (±0.25–0.5 mm), partly because the cut face carries 1–3° of natural taper unless a tilting taper-compensating head is used.

### Does water jet cutting produce a heat affected zone?

No. The jet erodes material mechanically rather than melting it, so there is no heat affected zone, no recast layer and no hardened cut edge. This is the reason to choose it for pre-hardened stock, heat-treated or plated material, composites that would delaminate, and long thin parts that would distort from thermal cutting.

### What is the minimum hole size for water jet cutting?

About 1× the material thickness in diameter, and never smaller than roughly 1.5× the kerf — around 0.060 in (1.5 mm) with a typical 0.040 in (1 mm) kerf. A hole narrower than the jet cannot be produced, and one only slightly larger comes out tapered and rough.

### What does water jet edge quality Q1 to Q5 mean?

It is a feed rate scale. Q1 is a fast separation cut that leaves pronounced striations toward the exit side of the material; Q5 is a slow finish cut with a nearly striation-free face. Because it is a speed multiplier, specifying Q5 across an entire part when only one edge seals is the most common avoidable cost on a water jet quote.

### What materials cannot be water jet cut?

Tempered glass shatters and must be cut before tempering. Very thin sheet below about 0.020 in (0.5 mm) is cut more accurately by laser because jet deflection dominates. Otherwise the process is close to material-agnostic: metals, stone, glass, ceramics, composites, rubber, foam and food are all routine.

## Alternative processes

- [Laser Cutting](https://manufacturingprocesses.org/processes/cutting/laser-cutting.md): Laser cutting melts, burns or vaporises a narrow kerf through flat stock with a focused beam, producing profiles from a 2D file with no tooling.
- [Plasma Cutting](https://manufacturingprocesses.org/processes/cutting/plasma-cutting.md): Plasma cutting blows an ionised gas jet through conductive metal, melting and ejecting a kerf far faster than a laser on thick plate.
- [Electrical Discharge Machining](https://manufacturingprocesses.org/processes/cutting/electrical-discharge-machining.md): EDM erodes conductive metal with controlled electrical sparks across a dielectric gap, cutting hardened material and sharp internal corners.
- [Punching and Blanking](https://manufacturingprocesses.org/processes/cutting/punching-and-blanking.md): Punching and blanking shear sheet metal between a punch and a die, producing holes and flat outlines at very high speed.
- [Sawing](https://manufacturingprocesses.org/processes/cutting/sawing.md): Sawing separates stock with a toothed blade — band, circular or cold saw — and is the default first operation on bar, tube and plate.

## Related processes

- [Laser Cutting](https://manufacturingprocesses.org/processes/cutting/laser-cutting.md): Laser cutting melts, burns or vaporises a narrow kerf through flat stock with a focused beam, producing profiles from a 2D file with no tooling.
- [Plasma Cutting](https://manufacturingprocesses.org/processes/cutting/plasma-cutting.md): Plasma cutting blows an ionised gas jet through conductive metal, melting and ejecting a kerf far faster than a laser on thick plate.
- [Electrical Discharge Machining](https://manufacturingprocesses.org/processes/cutting/electrical-discharge-machining.md): EDM erodes conductive metal with controlled electrical sparks across a dielectric gap, cutting hardened material and sharp internal corners.
- [CNC Milling](https://manufacturingprocesses.org/processes/cutting/cnc-milling.md): CNC milling removes material with a rotating multi-flute cutter moved along programmed paths, producing prismatic parts from solid stock.
- [Sheet Metal Fabrication](https://manufacturingprocesses.org/processes/forming/sheet-metal-fabrication.md): Sheet metal fabrication cuts, bends and joins flat stock into a finished assembly, typically laser cutting followed by press braking, hardware and welding.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/cutting/water-jet-cutting)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: category
name: "Joining"
processes: 18
url: https://manufacturingprocesses.org/processes/joining
---

# Joining

18 manufacturing processes in the joining family.

Joining processes combine two or more parts into an assembly, either
permanently (welding, brazing, adhesive bonding) or reversibly
(fasteners, snap fits).

The design question is rarely which joining process is strongest, but
which one tolerates the materials, thicknesses and access you already
committed to upstream.

## Processes

| Process | Tolerances | Typical volumes | Lead time |
| --- | --- | --- | --- |
| [Adhesive Bonding](https://manufacturingprocesses.org/processes/joining/adhesive-bonding.md) | Bond line 0.005–0.010 in (0.13–0.25 mm) held with glass beads or shims; assembly position set by the bond fixture, commonly ±0.010 in (±0.25 mm) | 1 to 1,000,000+ | Seconds to fixture with cyanoacrylate; 4–24 h to handling strength and about 7 days to full cure at room temperature; 30–60 min at 250–350 °F (120–175 °C) for heat-cured structural epoxy |
| [Arc Welding](https://manufacturingprocesses.org/processes/joining/arc-welding.md) | ±1/16 in (±1.6 mm) on welded assembly dimensions; ±1/32 in (±0.8 mm) with hard fixturing; tighter only by machining after welding | 1 to 1,000,000+ per year — manual for one-offs and short runs, robotic cells for repeat production | Same day to 1 week for manual one-off weldments; 4–8 weeks to bring a robotic cell online including fixture build and WPS/PQR qualification |
| [Friction Welding](https://manufacturingprocesses.org/processes/joining/friction-welding.md) | Rotary: finished length controlled to about ±0.010 in (±0.25 mm) through upset control; FSW: butt gap under 10% of thickness, plate flatness governed by clamping | 500 to 1,000,000+ per year for rotary; FSW is economic from low hundreds of panels upward | 1–30 s cycle for rotary friction welds; 4–12 weeks for machine tooling, clamping fixtures and parameter development |
| [Hot Plate Welding](https://manufacturingprocesses.org/processes/joining/hot-plate-welding.md) | Assembly height after welding ±0.010–0.020 in (±0.25–0.5 mm), controlled by a positive stop rather than by molded part height; meltdown 0.04–0.16 in (1–4 mm) | 100 to 1,000,000+ per year; field pipe fusion is one joint at a time | 15–120 s per cycle; 3–8 weeks for platen and nest tooling |
| [Joinery](https://manufacturingprocesses.org/processes/joining/joinery.md) | Glue line of 0.002–0.006 in (0.05–0.15 mm) for PVA; joints fitted to firm hand pressure rather than to a numeric tolerance | 1 to a few thousand — hand work for one-offs, jigged or CNC-cut joinery for batch production | Minutes to an hour per joint depending on method; 30–60 minutes clamp time and 24 hours to full adhesive cure |
| [Laser Plastic Welding](https://manufacturingprocesses.org/processes/joining/laser-plastic-welding.md) | Clamped joint gap ≤ 0.004 in (0.1 mm) required across the seam; quasi-simultaneous collapse controlled to about ±0.001 in (±0.025 mm) | 1,000 to 5,000,000+ per year; contour welding is viable from prototype quantities | Seconds to a few tens of seconds per part for contour welding, under 5 s for simultaneous; 4–10 weeks for transparent clamping tooling and part-specific optics |
| [Mechanical Fastening](https://manufacturingprocesses.org/processes/joining/mechanical-fastening.md) | Normal-fit clearance holes about 1/32 in (0.8 mm) over nominal per ASME B18.2.8 for screws up to 1/2 in; preload scatter ±25–35% with torque control, ±15% with angle control, ±5–10% with bolt elongation measurement | 1 to 10,000,000+ | Off-the-shelf hardware in hours to days; seconds per fastener at assembly |
| [PCB Assembly (SMT Reflow)](https://manufacturingprocesses.org/processes/joining/pcb-assembly-smt-reflow.md) | Placement accuracy ±0.001–0.002 in (±25–50 µm) at 3σ; stencil foil 0.004–0.006 in (100–150 µm) with aperture area ratio above 0.66; board bow and twist ≤ 0.75% per IPC-6012 | 5 to 1,000,000+ boards | 1–3 minutes of process time per board once running; 1–3 weeks for a prototype build including stencil, programming and component procurement |
| [Power Beam Welding](https://manufacturingprocesses.org/processes/joining/power-beam-welding.md) | Beam-to-seam alignment ±0.004 in (±0.1 mm); joint gap ≤ 0.004 in (0.1 mm) for autogenous welds; post-weld distortion roughly an order of magnitude below an equivalent arc weld | 100 to 1,000,000+ per year | Seconds of weld time per joint; 2–8 weeks for precision fixtures and schedule development; EBW adds minutes of chamber pump-down per load |
| [Resistance Welding](https://manufacturingprocesses.org/processes/joining/resistance-welding.md) | ±1/32 in (±0.8 mm) on fixtured sheet assemblies; nugget diameter held to 4√t–5√t (t in mm); electrode indentation ≤ 20–25% of sheet thickness | 5,000 to 10,000,000+ per year — the capital only pays back at production volume | 0.1–0.5 s of weld time and 1–2 s per spot on a robotic gun; 2–8 weeks for weld fixtures, guns and electrode tooling |
| [Riveting](https://manufacturingprocesses.org/processes/joining/riveting.md) | Hole 0.003–0.006 in (0.08–0.15 mm) over nominal shank; driven head 1.5x shank diameter wide by 0.5x shank diameter high; SPR interlock commonly ≥ 0.016 in (0.4 mm) with ≥ 0.008 in (0.2 mm) remaining bottom sheet | 1 to 10,000,000+ | Immediate with off-the-shelf hardware; 1–2 s per joint for automated self-piercing riveting; 4–10 weeks for SPR dies and C-frame tooling |
| [Soldering and Brazing](https://manufacturingprocesses.org/processes/joining/soldering-and-brazing.md) | Joint clearance 0.001–0.005 in (0.025–0.13 mm) for brazing, 0.003–0.005 in (0.075–0.13 mm) for soldering, specified at brazing temperature; assembly position set by fixture or self-locating features | 1 to 1,000,000+ — torch for one-offs and repair, continuous or vacuum furnace at production volume | Seconds to minutes for torch and induction work; 20–90 minute furnace cycles including ramp and controlled cool |
| [Staking](https://manufacturingprocesses.org/processes/joining/staking.md) | Formed head height controlled to about ±0.005 in (±0.13 mm) by tool stop; hole-to-boss clearance 0.002–0.010 in (0.05–0.25 mm) | 1,000 to 10,000,000+ per year | 0.2–0.8 s per ultrasonic stake, 4–15 s per hot air cycle; 2–5 weeks for multi-tip forming tooling |
| [Timber Frame Structures](https://manufacturingprocesses.org/processes/joining/timber-frame-structures.md) | CNC-cut joinery to about ±1/16 in (±1.5 mm) over member lengths of 40 ft (12 m) and more; erection tolerances are set by the project specification | — | Months from design freeze to erection, dominated by engineering, shop drawing approval and engineered-timber mill scheduling; CNC cutting of a frame itself takes days |
| [Ultrasonic Welding](https://manufacturingprocesses.org/processes/joining/ultrasonic-welding.md) | Weld collapse held to ±0.002 in (±0.05 mm) with distance-mode control; joint faces should be molded flat and parallel to the limit the tool allows | 10,000 to 10,000,000+ per year | 0.1–1.0 s weld, under 2 s total cycle; 2–4 weeks for horn and nest fixture tooling |
| [Upholstery](https://manufacturingprocesses.org/processes/joining/upholstery.md) | — | 1 to 100,000+ — hand work for one-offs and restoration, CNC cutting and sewing cells for production runs | Hours of hand labor per seat for traditional sprung upholstery; production cut-and-sew cells work in minutes per panel once patterns and cutting nests exist |
| [Vibration Welding](https://manufacturingprocesses.org/processes/joining/vibration-welding.md) | Meltdown controlled to about ±0.004 in (±0.1 mm); final assembly height set by a positive stop or machine collapse control rather than by molded part height | 5,000 to 5,000,000+ per year | 6–20 s cycle including 1–10 s of vibration and a 2–6 s hold; 3–8 weeks for matched nest tooling |
| [Weaving](https://manufacturingprocesses.org/processes/joining/weaving.md) | — | 1 to a few thousand — hand weaving for one-offs and restoration, pre-woven webbing for production | Hours per seat for hand weaving; pre-woven webbing is routed, splined and trimmed in well under an hour |

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/joining)*


<!-- ---------------- -->

---
type: process
name: "Adhesive Bonding"
category: "Joining"
subcategory: "Mechanical"
materials: ["Metal", "Plastic", "Wood", "Glass", "Ceramic", "Composite"]
tolerances: "Bond line 0.005–0.010 in (0.13–0.25 mm) held with glass beads or shims; assembly position set by the bond fixture, commonly ±0.010 in (±0.25 mm)"
volumes: "1 to 1,000,000+"
lead_time: "Seconds to fixture with cyanoacrylate; 4–24 h to handling strength and about 7 days to full cure at room temperature; 30–60 min at 250–350 °F (120–175 °C) for heat-cured structural epoxy"
url: https://manufacturingprocesses.org/processes/joining/adhesive-bonding
---

# Adhesive Bonding

Adhesive bonding joins parts with a cured polymer layer that spreads load over the whole bond area instead of concentrating it at fastener points.

- **Category**: [Joining](https://manufacturingprocesses.org/processes/joining.md)
- **Family**: Mechanical
- **Materials**: Metal, Plastic, Wood, Glass, Ceramic, Composite
- **Typical tolerances**: Bond line 0.005–0.010 in (0.13–0.25 mm) held with glass beads or shims; assembly position set by the bond fixture, commonly ±0.010 in (±0.25 mm)
- **Typical volumes**: 1 to 1,000,000+
- **Lead time**: Seconds to fixture with cyanoacrylate; 4–24 h to handling strength and about 7 days to full cure at room temperature; 30–60 min at 250–350 °F (120–175 °C) for heat-cured structural epoxy

## Overview

Adhesive bonding joins parts with a cured polymer layer that carries load across the entire overlap instead of concentrating it at fastener holes. That distribution is the reason to use it: a bonded lap joint has no drilled holes, no stress risers, no local bearing stress, and a fatigue life that a riveted joint in the same panels cannot match.

It also joins what nothing else will — metal to plastic, composite to metal, glass to steel — and the bond line electrically isolates dissimilar metals that would otherwise form a galvanic couple.

Structural epoxies reach 2,500–5,000 psi (17–35 MPa) in lap shear on properly prepared metal, toughened grades higher; structural acrylics 2,500–4,000 psi (17–28 MPa); polyurethanes 1,000–2,500 psi (7–17 MPa) with far more elongation. The catch is that adhesive joints are weak in peel and cleavage, and almost impossible to inspect non-destructively — which makes surface preparation and process control the real engineering work.

## How it works

1. **Surface preparation.** This determines whether the joint works. Degrease with solvent, abrade — 80–180 grit paper or grit blast — to break the weak oxide layer and increase surface area, then degrease again to remove the debris and bond promptly. Aerospace aluminum bonding goes further, to phosphoric acid anodizing or a sol-gel treatment, because a mechanically abraded aluminum oxide layer rehydrates and the bond degrades over years.
2. **Surface energy check.** The adhesive must wet the substrate, which requires the substrate's surface energy to exceed the adhesive's surface tension. Low-surface-energy plastics — PP, PE, PTFE, POM, silicone — sit below about 36 dyn/cm and will not bond without flame, corona or plasma treatment, or a dedicated primer.
3. **Apply and set the bond line.** Dispense a controlled bead or film. Bond line thickness is a design variable, not an accident: 0.005–0.010 in (0.13–0.25 mm) is the structural sweet spot, held with glass beads mixed into the adhesive, wire spacers or molded standoffs.
4. **Close and fixture.** Parts are clamped or fixtured so the bond line is uniform and the joint does not move while the adhesive gels.
5. **Cure.** Two-part epoxies reach handling strength in 4–24 h at room temperature and full properties in about 7 days; heat curing at 250–350 °F (120–175 °C) for 30–60 min gives higher glass transition temperature and strength. As a rule of thumb, every 10 °C increase roughly halves cure time. Cyanoacrylates fixture in seconds; anaerobics cure only when confined between metal surfaces with air excluded.

### Adhesive families at a glance

Epoxy for the highest structural strength and temperature resistance. Structural acrylic where the surfaces cannot be perfectly cleaned — it tolerates light oil far better than epoxy. Polyurethane where the joint must flex or absorb differential expansion. Silicone for sealing and gasketing from −65 to 400 °F (−55 to 205 °C) rather than for strength. Cyanoacrylate for fast fixturing of small parts, with poor peel strength and a temperature ceiling near 180 °F (82 °C). Anaerobic threadlockers and retaining compounds for fastener locking and cylindrical press-fit augmentation.

## Design guidelines

### Load the joint in shear, never in peel

Adhesives are strong in shear and compression and weak in peel and cleavage — often by an order of magnitude. Convert every peel-loaded joint into a shear-loaded one: use a lap, double lap, scarf or joggle rather than a butt or a right-angle T. Where peel cannot be designed out, add a mechanical arrest — a rivet, a screw or a formed lip at the end of the overlap — to stop a peel crack from propagating.

### Set the overlap and the bond line

Start at an overlap of 4–8x the thinner adherend's thickness. Beyond about 20x thickness, added overlap buys very little, because load transfers at the two ends of the joint and the middle carries almost nothing. Hold a bond line of 0.005–0.010 in (0.13–0.25 mm) with beads or shims: starved joints are weak, and thick bond lines lose strength and cure unevenly. Add a spew fillet at the overlap ends — a small radius of squeezed-out adhesive measurably reduces the peel stress peak there.

### Design for differential expansion

Bonding steel (CTE roughly 6.5 µin/in/°F, 11.7 µm/m/°C) to aluminum (13, 23) or to CFRP (near 1, 2) means the joint sees strain every thermal cycle. A more flexible adhesive and a thicker bond line both absorb it; a rigid epoxy in a thin bond line between mismatched materials will eventually fail at the interface. Coefficients of expansion and moduli for the substrates are on the [material properties chart](/charts/material-properties).

### Substrate preparation and compatibility

| Substrate | Prep | Notes |
| --- | --- | --- |
| Steel, stainless | Degrease, abrade 80–120 grit, degrease | Bond promptly — flash rust starts within minutes |
| Aluminum (structural / long life) | Phosphoric acid anodize or sol-gel | Abrasion alone gives a bond that degrades with humidity over years |
| PP, PE, POM, PTFE, silicone | Flame, corona or plasma, or a primer | Surface energy below ~36 dyn/cm — untreated joints fail at the interface |
| ABS, PC, PVC, acrylic | Solvent wipe, light abrade | Bond well; check adhesive compatibility for stress cracking |
| CFRP, GFRP | Peel ply then abrade, or grit blast | Never bond over release agent |
| Glass, ceramic | Clean, silane primer | Silane coupling agents dramatically improve durability |
| Galvanized, painted, plated surfaces | Bond to the coating, not the metal | The joint is only as strong as the coating's adhesion |

Abrasion targets a specific roughness rather than a polish — see the [surface finish chart](/charts/surface-finish-chart) for the Ra values that correspond to common grit grades.

### Anaerobics on threaded fasteners

Threadlockers cure in the confined, airless gap between engaged threads. During assembly they act as a lubricant, so torque to the **lubricated** values rather than the dry ones — see the [bolt torque chart](/charts/bolt-torque-chart). Retaining compounds do the same job on cylindrical press fits, and can let a looser, cheaper machined fit carry the same torque.

### Inspection

This is adhesive bonding's genuine weakness. Ultrasonic through-transmission and thermography reliably find voids, porosity and disbonds, but neither detects a **kissing bond** — surfaces in intimate contact with no chemical adhesion, usually caused by contamination or bad surface prep. Because the worst defect is invisible, quality is controlled by process rather than inspection: documented surface preparation, adhesive lot traceability, controlled time between prep and bonding, and witness coupons bonded alongside every batch and tested destructively.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Bond line thickness | 0.005–0.010 in (0.13–0.25 mm) | Held with beads or shims | Starved joints are weak, thick ones cure unevenly |
| Overlap length | 4–8x thinner adherend thickness | Gains plateau past ~20x | Load transfers at the overlap ends |
| Loading mode | Shear or compression | Never peel or cleavage | Peel strength is an order of magnitude lower |
| Spew fillet at overlap ends | Small radius, retained | Do not wipe off | Reduces the peel stress peak |
| Time from surface prep to bonding | Minutes to a few hours | Per adhesive datasheet | Prepared surfaces re-oxidize and re-contaminate |
| Low-surface-energy plastics | Flame, corona or plasma treat | Untreated will not bond | Substrate energy must exceed adhesive surface tension |
| Quality method | Witness coupons per batch | NDT alone is insufficient | Kissing bonds are not detectable |

## Cost drivers

Adhesive bonding has almost no tooling cost and a significant process cost. There is no die, no fastener tooling, no weld fixture beyond a simple bond jig — but there is surface preparation labor, cure time, and floor space consumed by parts waiting to cure.

Cure time is usually the real constraint. A room-temperature epoxy that needs 4–24 h to reach handling strength forces work-in-progress inventory and fixtures that are tied up for the duration; a heat cure buys speed at the cost of oven time and energy. That trade — fixture count against cure schedule — is where bonded assembly economics are won or lost.

Surface preparation is the other cost, and it is the one most often cut, always with the same result. Abrasion and solvent wipe are labor; anodizing or plasma treatment is a process step with its own equipment. Neither is optional if the joint is structural.

1. **Match cure chemistry to takt time.** A faster adhesive often costs less overall than a cheaper one that ties up ten fixtures.
2. **Design self-fixturing joints.** A lip, tongue or molded standoff that holds alignment and bond line during cure deletes a fixture entirely.
3. **Use beads in the adhesive** rather than shims or spacer parts to set the bond line.
4. **Combine bonding with a few mechanical fasteners** at the overlap ends — this arrests peel and provides fixturing during cure at the same time.
5. **Bond witness coupons with every batch.** Destructive testing on coupons is far cheaper than the field failures that unverified surface preparation produces.

## FAQ

### How strong is a structural adhesive bond?

Two-part structural epoxy typically reaches 2,500–5,000 psi (17–35 MPa) in lap shear on properly prepared metal, with toughened grades higher; structural acrylics 2,500–4,000 psi (17–28 MPa); polyurethanes 1,000–2,500 psi (7–17 MPa) with much greater elongation. Those are shear numbers — peel and cleavage strength is roughly an order of magnitude lower and must be designed out.

### What bond line thickness should I design for?

0.005–0.010 in (0.13–0.25 mm) for structural adhesives. Thinner than that and the joint is starved, with the adherends effectively in contact and the adhesive unable to distribute load; much thicker and strength falls, cure becomes uneven and shrinkage stress rises. Control it with glass beads mixed into the adhesive, wire spacers or molded standoffs rather than by clamp pressure.

### Why won't adhesive stick to polypropylene?

Polypropylene, polyethylene, PTFE, acetal and silicone have surface energies below about 36 dyn/cm — lower than the surface tension of most adhesives — so the adhesive beads up instead of wetting. Flame, corona or atmospheric plasma treatment raises the surface energy, and dedicated polyolefin primers do the same chemically. Untreated joints fail cleanly at the interface.

### Can bonded joints be inspected non-destructively?

Only partly. Ultrasonic through-transmission and thermography find voids, porosity and disbonds reliably, but neither detects a kissing bond — surfaces in intimate contact with no actual adhesion, which is what contamination or bad surface prep produces. Because the most dangerous defect is invisible, quality control rests on documented surface preparation and destructively tested witness coupons bonded with each batch.

### Do threadlockers change bolt torque values?

Yes. An anaerobic threadlocker wets the threads during assembly and acts as a lubricant, lowering the friction that torque has to overcome, so torquing to dry values will overload the fastener. Use the lubricated column on the bolt torque chart. The adhesive only develops its locking function after it cures in the airless gap between the engaged threads.

## Alternative processes

- [Ultrasonic Welding](https://manufacturingprocesses.org/processes/joining/ultrasonic-welding.md): Ultrasonic welding applies high-frequency vibration through a horn to melt a thermoplastic joint locally in well under a second.
- [Laser Plastic Welding](https://manufacturingprocesses.org/processes/joining/laser-plastic-welding.md): Laser plastic welding passes a beam through a transmissive part into an absorbing one, melting only the interface between them.
- [Soldering and Brazing](https://manufacturingprocesses.org/processes/joining/soldering-and-brazing.md): Soldering and brazing join metals with a molten filler that wets the joint by capillary action while the parent metal stays solid.
- [Arc Welding](https://manufacturingprocesses.org/processes/joining/arc-welding.md): Arc welding melts and fuses metal with an electric arc between an electrode and the work, shielded from the atmosphere by gas or flux.

## Related processes

- [Mechanical Fastening](https://manufacturingprocesses.org/processes/joining/mechanical-fastening.md): Mechanical fastening joins parts with screws, inserts, clinch hardware or snap fits so the joint can be assembled and taken apart again.
- [Riveting](https://manufacturingprocesses.org/processes/joining/riveting.md): Riveting sets a rivet through aligned holes and upsets its tail so the joint clamps permanently, with no heat and no access to both sides for blind types.
- [Composite Laminating](https://manufacturingprocesses.org/processes/forming/composite-laminating.md): Composite laminating lays reinforcing fiber into a mold and consolidates it with resin, curing into a stiff, light monocoque part.
- [Insert Molding](https://manufacturingprocesses.org/processes/forming/insert-molding.md): Insert molding loads metal inserts into the cavity and molds polymer around them, capturing threads, terminals or stampings in a single shot.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/joining/adhesive-bonding)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Arc Welding"
category: "Joining"
subcategory: "Thermal"
materials: ["Metal"]
tolerances: "±1/16 in (±1.6 mm) on welded assembly dimensions; ±1/32 in (±0.8 mm) with hard fixturing; tighter only by machining after welding"
volumes: "1 to 1,000,000+ per year — manual for one-offs and short runs, robotic cells for repeat production"
lead_time: "Same day to 1 week for manual one-off weldments; 4–8 weeks to bring a robotic cell online including fixture build and WPS/PQR qualification"
url: https://manufacturingprocesses.org/processes/joining/arc-welding
---

# Arc Welding

Arc welding melts and fuses metal with an electric arc between an electrode and the work, shielded from the atmosphere by gas or flux.

- **Category**: [Joining](https://manufacturingprocesses.org/processes/joining.md)
- **Family**: Thermal
- **Materials**: Metal
- **Typical tolerances**: ±1/16 in (±1.6 mm) on welded assembly dimensions; ±1/32 in (±0.8 mm) with hard fixturing; tighter only by machining after welding
- **Typical volumes**: 1 to 1,000,000+ per year — manual for one-offs and short runs, robotic cells for repeat production
- **Lead time**: Same day to 1 week for manual one-off weldments; 4–8 weeks to bring a robotic cell online including fixture build and WPS/PQR qualification

## Overview

Arc welding fuses metal by drawing an electric arc between an electrode and the workpiece, melting both faces and — in every variant except autogenous TIG — adding filler metal that solidifies into a continuous joint. The arc column runs above 10,000 °F (5,500 °C), so the process melts anything from 0.030 in (0.8 mm) sheet to plate of effectively unlimited thickness in multiple passes.

The family covers five industrial variants: manual metal arc (MMA/SMAW), metal inert gas (MIG/MAG, GMAW), tungsten inert gas (TIG/GTAW), plasma arc (PAW) and submerged arc (SAW). Between them they cover carbon and low-alloy steel, stainless, aluminum, nickel alloys, copper alloys and titanium.

The number that characterizes arc welding: a full-penetration groove weld made with matching or overmatching filler develops 100% of base-metal tensile strength — in a qualified tensile test the coupon fails in the parent plate, not the weld. Volumes run from a single fabricated frame to robotic lines producing hundreds of thousands of assemblies a year.

## How it works

1. **Joint preparation.** Plate over about 1/4 in (6 mm) is beveled — 30–37.5° per side for a single-V groove (60–75° included), 1/16 in (1.6 mm) root face, 1/16–1/8 in (1.6–3.2 mm) root opening. Scale, rust, paint, oil and moisture come off at least 1/2 in (13 mm) back from the joint.
2. **Shielding.** MIG uses argon, argon/CO2 blends (75/25 for steel) or straight CO2 at 20–50 CFH (10–24 L/min); TIG uses argon or argon/helium; SMAW and flux-cored wires make their own gas and slag from the flux; SAW buries the arc under granular flux.
3. **Heat input.** Heat input = (arc volts x amps x 60) / travel speed, landing between roughly 12 and 64 kJ/in (0.5–2.5 kJ/mm) on structural work. That one number governs bead size, penetration, HAZ width and distortion.
4. **Deposition.** Thick joints are filled in multiple passes with interpass temperature controlled — commonly under 350 °F (175 °C) on austenitic stainless to limit sensitization.
5. **Cooling.** Parent metal that got hot but never melted becomes the heat-affected zone, typically 1/16–1/4 in (1.5–6 mm) wide. Cooling rate sets HAZ hardness, so thick sections and high-carbon-equivalent steels are preheated to 150–300 °F (65–150 °C) to avoid hydrogen cracking.
6. **Post-weld.** Slag chipping, back-gouging for double-sided joints, and on heavy pressure-boundary steel a stress relief at 1,100–1,250 °F (595–675 °C).

### How do the five variants differ?

**MMA / SMAW (stick).** A flux-coated consumable electrode at 40–300 A. No gas bottle, so it works outdoors; deposition is only 1–5 lb/hr (0.5–2.3 kg/hr) and slag is chipped between passes.

**MIG / MAG (GMAW).** Continuously fed solid wire, 0.023–0.045 in (0.6–1.2 mm), with external gas, at 5–12 lb/hr (2.3–5.5 kg/hr). Spray transfer in argon-rich gas for thick plate flat; short-circuit transfer for thin sheet and out-of-position work.

**TIG / GTAW.** A non-consumable tungsten electrode from 5 to 300 A with heat and filler controlled independently. Well under 2 lb/hr, but the cleanest — the only practical arc route for titanium and the normal choice below 0.060 in (1.5 mm).

**Plasma (PAW).** A TIG arc constricted through a nozzle orifice, giving higher energy density and a stable arc down to 0.1 A for foil. In keyhole mode it makes a single-pass full-penetration weld to about 1/4 in (6 mm).

**Submerged arc (SAW).** Wire fed under granular flux at 300–1,500 A, reaching 10–45 lb/hr (5–20 kg/hr). The flux must sit on the joint, so it is limited to flat and horizontal seams — heavy plate, pipe mills, pressure vessels.

## Design guidelines

### How big should the fillet be?

AWS D1.1 sets minimum fillet size by the thicker part joined: 1/8 in (3 mm) up to 1/4 in (6 mm) base metal; 3/16 in (5 mm) over 1/4 through 1/2 in (6–12 mm); 1/4 in (6 mm) over 1/2 through 3/4 in (12–20 mm); 5/16 in (8 mm) above 3/4 in (20 mm) — an undersized fillet on thick plate cools too fast and cracks.

Fillets are then checked on the throat, not the leg: effective throat = 0.707 x leg, and nominal shear strength is 0.60 x the electrode classification strength on that throat — 42 ksi (290 MPa) for E70 filler. Throat scales with the leg but deposited volume scales with the leg squared, so a 3/8 in fillet buys 50% more strength than a 1/4 in for 125% more weld metal.

### Groove welds, access and distortion

Keep 60° included angle on single-V groove welds; below about 45° the electrode cannot reach the sidewall and lack-of-fusion follows. Design so welds fall in the flat (1F/1G) or horizontal (2F) position — vertical and overhead work roughly halves deposition rate.

Weld metal shrinks about 3–5% by volume as it solidifies, pulling the joint transversely and rotating it angularly. Balance welds about the neutral axis, alternate sides on double-sided joints, and specify the smallest weld that carries the load. Expect ±1/16 in (1.6 mm) over a 3 ft (1 m) weldment without hard fixturing; if a surface must be true, leave 0.030–0.060 in (0.8–1.5 mm) of stock and machine after welding.

### Which material pairs actually weld?

| Pair | Arc weldable? | Notes |
| --- | --- | --- |
| Carbon steel to carbon steel | Yes | E70XX / ER70S-6 matching filler; the default case |
| Carbon steel to 300-series stainless | Yes | ER309L filler; expect differential expansion |
| 304 to 304 / 316 to 316 | Yes | ER308L and ER316L; hold interpass under 350 °F (175 °C) |
| Aluminum 6xxx to 6xxx | Yes | ER4043 resists cracking, ER5356 is stronger and anodizes to a closer color match |
| Aluminum 2024 / 7075 | No, practically | Hot-cracks and loses temper; rivet, bond or friction weld instead |
| Aluminum to steel | No | Brittle Fe-Al intermetallics; use friction welding or a bimetallic transition insert |
| Titanium to titanium | TIG or plasma only | Full inert coverage plus trailing and backing gas; oxygen pickup embrittles |
| Galvanized steel | Yes, with care | Zinc boils below the pool and causes porosity; grind back 1/2 in (13 mm), extract fume |

Melting points are on the [metal melting points chart](/charts/metal-melting-points); strength and expansion figures for sizing welds and predicting distortion are on the [material properties chart](/charts/material-properties).

### How are arc welds inspected?

Visual inspection to AWS D1.1 is the baseline. Dye penetrant (PT) finds surface cracks on any metal; magnetic particle (MT) finds surface and near-surface flaws in ferromagnetic material only; ultrasonic (UT) is the normal volumetric method for groove welds above 5/16 in (8 mm); radiography (RT) is better for porosity and slag. Procedure and welder qualification (WPS/PQR) is what actually controls quality — inspection only samples it.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Fillet leg on 1/2 in (12 mm) plate | Size for the calculated load | 3/16 in (5 mm) minimum | Undersized fillets on thick plate quench and crack |
| Included groove angle | 60° | 45° | Electrode access to the sidewall |
| Root opening, open-root V | 1/8 in (3 mm) | ±1/32 in (0.8 mm) of nominal | Too wide burns through, too tight leaves lack of penetration |
| Weld reinforcement (cap height) | Flush to 1/16 in (1.6 mm) | 1/8 in (3 mm) | Toe angle drives the fatigue stress concentration |
| Torch access around joint | 1 in (25 mm) | 5/8 in (16 mm) | Gas cup and electrode extension |
| Distortion over 3 ft (1 m) | ±1/16 in (1.6 mm) | ±1/32 in (0.8 mm) fixtured | Transverse and angular shrinkage |

## Variants

- Manual Metal Arc Welding (MMA)
- Metal Inert Gas Welding (MIG)
- Tungsten Inert Gas Welding (TIG)
- Plasma Welding
- Submerged Arc Welding (SAW)

## Cost drivers

Arc welding is a labor process with a consumable bill attached, not a tooling process. On manual stick work, arc-on time is typically only 10–30% of the welder's shift; semi-automatic MIG reaches 30–50%, and a robotic cell 50–80%. That operator factor — not wire price — is what separates a cheap weldment from an expensive one.

Weld volume is the second driver, and it scales badly. A fillet's cross-sectional area goes with the square of the leg, so specifying 3/8 in where 1/4 in carries the load costs 2.25x the deposited metal and 2.25x the arc time, plus proportionally more distortion to straighten afterward. Consumables are usually a small fraction of joint cost next to labor. Inspection can dominate on critical work: blanket 100% radiography on non-critical welds is one of the most common overspecs in fabrication.

1. **Size fillets to the calculated load.** Use the 0.707 throat rule rather than a habitual leg size; the area penalty is quadratic.
2. **Use intermittent fillets** where the joint is not fatigue-loaded or leak-tight — a 2 in on / 6 in off pattern deposits a quarter of the metal.
3. **Design for the flat and horizontal positions.** Out-of-position welding roughly halves deposition rate and needs a more skilled welder.
4. **Delete joints upstream.** A single laser-cut and formed part often replaces three welded pieces and two joints.
5. **Specify NDT per joint, not per drawing.** Reserve UT and RT for full-penetration structural or pressure joints and use visual plus PT elsewhere.

## FAQ

### How strong is an arc weld compared to the base metal?

A full-penetration groove weld made with matching or overmatching filler develops 100% of base-metal tensile strength — in a qualified tensile test the coupon breaks in the parent plate. Fillet welds are different: they are sized on shear across the effective throat, which is 0.707 x the leg, at 0.60 x the filler classification strength (42 ksi / 290 MPa for E70 filler).

### What is the minimum fillet weld size for 1/2 inch plate?

3/16 in (5 mm) per AWS D1.1 for base metal over 1/4 in through 1/2 in (6–12 mm). Minimum sizes exist because a small fillet deposited onto a thick, cold section cools too quickly, hardens the heat-affected zone and cracks. Above 3/4 in (20 mm) the minimum rises to 5/16 in (8 mm).

### Can you arc weld aluminum to steel?

No. Iron and aluminum form brittle Fe-Al intermetallic layers at the fusion line that crack under almost any load. The production routes are friction welding (rotary or friction stir), an explosion-bonded bimetallic transition insert, adhesive bonding, or self-piercing rivets.

### MIG or TIG — which should I specify?

MIG where deposition rate matters: 5–12 lb/hr (2.3–5.5 kg/hr) versus under 2 lb/hr for TIG. TIG below about 0.060 in (1.5 mm), for titanium and other reactive metals, and where the bead is a visible design feature, because heat and filler are controlled independently.

### How much distortion should I expect from welding?

Weld metal shrinks 3–5% by volume as it solidifies, producing transverse shrinkage across the joint and angular rotation of the plates. Without hard fixturing, ±1/16 in (1.6 mm) over a 3 ft (1 m) weldment is normal. Balanced sequence, back-stepping and the smallest adequate weld are the practical controls.

### When does carbon steel need preheat before welding?

AWS D1.1 sets preheat from thickness and steel group, and typical structural carbon steel above 3/4 in (20 mm) calls for 150–300 °F (65–150 °C). Preheat slows cooling through the martensite range, lowering HAZ hardness and giving diffusible hydrogen time to escape before the joint is cold.

## Alternative processes

- [Soldering and Brazing](https://manufacturingprocesses.org/processes/joining/soldering-and-brazing.md): Soldering and brazing join metals with a molten filler that wets the joint by capillary action while the parent metal stays solid.
- [Riveting](https://manufacturingprocesses.org/processes/joining/riveting.md): Riveting sets a rivet through aligned holes and upsets its tail so the joint clamps permanently, with no heat and no access to both sides for blind types.
- [Adhesive Bonding](https://manufacturingprocesses.org/processes/joining/adhesive-bonding.md): Adhesive bonding joins parts with a cured polymer layer that spreads load over the whole bond area instead of concentrating it at fastener points.
- [Mechanical Fastening](https://manufacturingprocesses.org/processes/joining/mechanical-fastening.md): Mechanical fastening joins parts with screws, inserts, clinch hardware or snap fits so the joint can be assembled and taken apart again.

## Related processes

- [Resistance Welding](https://manufacturingprocesses.org/processes/joining/resistance-welding.md): Resistance welding passes a heavy current through the joint so its own electrical resistance melts a nugget between the sheets.
- [Power Beam Welding](https://manufacturingprocesses.org/processes/joining/power-beam-welding.md): Power beam welding fuses metal with a laser or electron beam focused to a very small spot, producing deep, narrow welds with little distortion.
- [Friction Welding](https://manufacturingprocesses.org/processes/joining/friction-welding.md): Friction welding rubs two parts together under load until the interface plasticises and forges together, with no melting and no filler metal.
- [Sheet Metal Fabrication](https://manufacturingprocesses.org/processes/forming/sheet-metal-fabrication.md): Sheet metal fabrication cuts, bends and joins flat stock into a finished assembly, typically laser cutting followed by press braking, hardware and welding.
- [Heat Treatment](https://manufacturingprocesses.org/processes/forming/heat-treatment.md): Heat treatment cycles metal through controlled heating and cooling to change its hardness, strength, ductility or internal stress state.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/joining/arc-welding)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Friction Welding"
category: "Joining"
subcategory: "Thermal"
materials: ["Metal", "Plastic"]
tolerances: "Rotary: finished length controlled to about ±0.010 in (±0.25 mm) through upset control; FSW: butt gap under 10% of thickness, plate flatness governed by clamping"
volumes: "500 to 1,000,000+ per year for rotary; FSW is economic from low hundreds of panels upward"
lead_time: "1–30 s cycle for rotary friction welds; 4–12 weeks for machine tooling, clamping fixtures and parameter development"
url: https://manufacturingprocesses.org/processes/joining/friction-welding
---

# Friction Welding

Friction welding rubs two parts together under load until the interface plasticises and forges together, with no melting and no filler metal.

- **Category**: [Joining](https://manufacturingprocesses.org/processes/joining.md)
- **Family**: Thermal
- **Materials**: Metal, Plastic
- **Typical tolerances**: Rotary: finished length controlled to about ±0.010 in (±0.25 mm) through upset control; FSW: butt gap under 10% of thickness, plate flatness governed by clamping
- **Typical volumes**: 500 to 1,000,000+ per year for rotary; FSW is economic from low hundreds of panels upward
- **Lead time**: 1–30 s cycle for rotary friction welds; 4–12 weeks for machine tooling, clamping fixtures and parameter development

## Overview

Friction welding rubs two parts together under load until the interface plasticizes, then forges them together. Nothing melts — peak temperature stays at roughly 0.6–0.9 of the absolute melting point — so the joint is made in the solid state, without filler, flux, shielding gas or an arc.

That has two consequences a design engineer cares about. First, the solidification defects that limit fusion welding (hot cracking, gas porosity, filler dilution) simply do not occur. Second, alloys that are effectively unweldable by arc become joinable: 2024 and 7075 aluminum, aluminum to copper, aluminum to steel.

Four variants cover the field: rotary friction welding (RFW) for round parts, linear (LFW) and orbital (OFW) for non-round sections, and friction stir welding (FSW) for seams in plate and extrusion. FSW joint efficiency in aluminum typically runs 70–90% of parent tensile strength in heat-treatable alloys and approaches 100% in non-heat-treatable 5xxx — against roughly 50–70% for the same alloys arc welded.

## How it works

### Rotary, linear and orbital friction welding

1. **Load and rub.** Rotary drives one part at a surface speed of roughly 3–10 ft/s (1–3 m/s) at the interface; linear reciprocates at 25–125 Hz with 0.04–0.12 in (1–3 mm) amplitude; orbital moves both parts in a small circle so every point on the interface sees the same relative velocity.
2. **Friction phase.** Friction pressure — commonly 4,000–13,000 psi (30–90 MPa) on steel — breaks up oxide and contamination and drives it into the flash while the interface heats and plasticizes. The parts shorten (upset) as material extrudes radially.
3. **Braking or stopping.** Relative motion stops in milliseconds. Direct-drive machines brake the spindle; inertia machines let a flywheel run down, so delivered energy is fixed by flywheel mass and speed.
4. **Forge phase.** Pressure is raised, often to 9,000–29,000 psi (60–200 MPa), and held while the joint consolidates and cools under load. Total cycle is typically 1–30 s.
5. **Flash removal.** Upset flash is machined or trimmed off. Upset length is the process's own quality signal, controlled to about ±0.010 in (±0.25 mm) on production machines.

Rotary welding needs one part round and free to spin, which is why it dominates axles, drive shafts, valves, drill pipe and bimetallic tooling. Linear friction welding removed that constraint and is the production route for integrally bladed rotors (blisks) in titanium and nickel superalloys. Orbital suits non-round sections where neither part can be spun.

### Friction stir welding

1. **Clamp.** FSW generates large lateral and downward forces, so plates are clamped hard against a rigid backing anvil.
2. **Plunge.** A non-consumable tool with a profiled pin and a wider shoulder rotates at 200–2,000 rpm and plunges until the shoulder contacts the surface. Friction under the shoulder supplies most of the heat.
3. **Traverse.** The tool travels along the seam at 2–120 in/min (50–3,000 mm/min), stirring plasticized material from the advancing side around to the retreating side. Peak temperature in aluminum sits around 750–930 °F (400–500 °C) — hot enough to plasticize, well below the melting range.
4. **Retract.** The pin leaves an exit hole at the end of the weld unless a retractable-pin tool or a run-off tab is used. This must be designed for, not discovered.

Production FSW welds aluminum up to roughly 1 in (25 mm) per side, and about 2 in (50 mm) double-sided. The weld nugget is fine-grained and recrystallized; the thermo-mechanically affected zone and HAZ on either side are where heat-treatable tempers soften.

## Design guidelines

### Rotary friction welding: geometry constraints

At least one part must be axisymmetric and free to rotate. Solid bar from about 0.2 to 4 in (5–100 mm) diameter covers most production; tube welds well because interface pressure is uniform. Allow for length loss: upset consumes 0.06–0.4 in (1.5–10 mm) of combined length depending on section, so dimension blanks with an upset allowance and control finished length from the weld. Provide clearance for the flash and a way to machine it off — a relief groove next to the joint is the usual answer.

### Friction stir welding: joint and fixture design

Design for a rigid backing anvil under the entire weld line: FSW pushes down with thousands of pounds of force and the plate cannot deflect. Butt joints need edges machined or sawn square with a gap under about 10% of thickness. Plan the run-on and run-off — either put the exit hole in a sacrificial tab that gets trimmed, or specify a retractable pin tool. Lap joints produce hooking defects at the sheet interface, which must be kept off the tension side.

### Strength and heat effects

FSW does not melt the metal, but it does put the joint through a thermal cycle. In 6061-T6 and 7075-T6 the HAZ over-ages and softens, so joint efficiency lands around 70–90% of parent tensile — still far better than the 50–70% typical of arc welding those alloys, and without hot cracking. In non-heat-treatable 5xxx alloys, where strength comes from work hardening rather than precipitates, joint efficiency is close to 100%.

### Which material pairs work?

| Pair | Process | Notes |
| --- | --- | --- |
| Steel to steel | RFW, LFW | Full base-metal strength; standard for shafts and axles |
| Tool steel to mild steel | RFW | The classic bimetallic economy — expensive alloy only where it cuts |
| Stainless to carbon steel | RFW | No filler dilution and no sensitized weld metal |
| Aluminum to aluminum, same alloy | FSW, RFW | The mainstream FSW case: panels, extrusion seams, battery trays |
| Aluminum 2xxx and 7xxx | FSW | Joinable, where arc welding hot-cracks |
| Aluminum to copper | FSW, RFW | Practical for busbars and battery tabs; intermetallic layer must be kept thin |
| Aluminum to steel | FSW, RFW | Achievable in the solid state; impossible by arc |
| Titanium to titanium | LFW, RFW | Blisk manufacture; no atmospheric contamination since nothing melts |
| Nickel superalloys | LFW | Avoids the strain-age cracking that plagues fusion welding |

Strength and thermal data are on the [material properties chart](/charts/material-properties); melting points that set the 0.6–0.9 Tm working window are on the [metal melting points chart](/charts/metal-melting-points).

### Inspection

Rotary welds are inspected by the flash itself — shape and symmetry are a direct process signature — plus upset length monitoring, ultrasonic testing through the bar, and destructive coupons. FSW is inspected visually for surface flash, galling and root lack-of-penetration, with phased-array ultrasonics for wormhole (tunnel) and root defects. The FSW "kissing bond" — surfaces in contact but not metallurgically joined — is notoriously hard to detect by any NDT method, which makes force and parameter monitoring the primary quality tool.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| RFW: upset allowance on blank length | 0.1–0.4 in (2.5–10 mm) | Section dependent | Material is consumed into flash |
| RFW: flash relief groove at the joint | Provide | — | Flash must be trimmable |
| FSW: joint gap | 0 | 10% of thickness | The tool cannot bridge a gap |
| FSW: backing anvil support | Full length, rigid | — | Downforce deflects unsupported plate and causes root defects |
| FSW: exit hole | Run-off tab or retractable pin | Never in the part | The pin leaves a hole where it retracts |
| FSW: single-pass thickness in aluminum | Up to 1 in (25 mm) per side | ~2 in (50 mm) double-sided | Tool strength and machine force limits |

## Variants

- Rotary Friction Welding (RFW)
- Linear Friction Welding (LFW)
- Orbital Friction Welding (OFW)
- Friction Stir Welding (FSW)

## Cost drivers

Friction welding is machine-dominated. There are no consumables beyond FSW tool wear, and cycle times are short — 1–30 s for rotary — so per-part cost at volume is low. The offsetting costs are a purpose-built machine, dedicated clamping fixtures, and a parameter development program, because each material pair and section needs its own schedule proved on coupons.

Rotary friction welding usually pays for itself through material substitution rather than through the weld: replacing a solid alloy-steel shaft with a mild-steel body friction-welded to a small alloy end removes both expensive stock and the machining that goes with it. FSW pays for itself by replacing riveted panel joints — deleting hundreds of holes, rivets, sealant and inspection steps. FSW machines are large and stiff because the process reacts high loads, which is the main capital barrier and the reason contract capacity is worth checking before designing around it.

1. **Use a bimetallic design.** Put the expensive alloy only where function needs it and friction weld it to a cheap body.
2. **Design flash relief into the part** so trimming is a single facing cut rather than a separate operation.
3. **Detail the FSW exit hole out of the part** with a run-off tab, or budget for a retractable-pin tool.
4. **Consolidate extrusions.** Friction stir welding two profiles is usually cheaper than tooling a single wide extrusion, and can beat casting for panel structures.
5. **Qualify on coupons.** Parameter development destroys parts; do it on representative sections, not finished assemblies.

## FAQ

### How strong is a friction stir weld compared to the base metal?

In heat-treatable aluminum such as 6061-T6 or 7075-T6, joint efficiency is typically 70–90% of parent tensile strength — the loss comes from over-aging in the heat-affected zone, not from the weld nugget. In non-heat-treatable 5xxx alloys the welded joint approaches annealed parent strength, so efficiency is close to 100%. Arc welding the same alloys gives roughly 50–70%.

### Can friction welding join aluminum to steel?

Yes — this is the main reason to use it. Because nothing melts, the thick brittle Fe-Al intermetallic layer that ruins a fusion weld never has time to form. Both rotary friction welding (for round transition pieces) and friction stir welding (for lap and butt seams) are used in production for aluminum-to-steel and aluminum-to-copper joints.

### Does friction stir welding leave a hole at the end of the weld?

Yes. When the tool retracts it leaves an exit hole the diameter of the pin. Production designs either place that hole in a sacrificial run-off tab that gets trimmed away, or use a retractable-pin tool that withdraws the pin gradually while the shoulder closes the surface. This has to be designed in from the start.

### How is friction welding inspected?

Rotary welds are judged first by flash shape and symmetry, which is a direct signature of the parameters, plus upset length monitoring, ultrasonic testing and destructive bend and tensile coupons. FSW uses visual inspection, phased-array ultrasonics for tunnel and root defects, and bend tests. The kissing bond — surfaces touching but not joined — is very hard to detect by NDT, so force and parameter monitoring carry the quality burden.

## Alternative processes

- [Power Beam Welding](https://manufacturingprocesses.org/processes/joining/power-beam-welding.md): Power beam welding fuses metal with a laser or electron beam focused to a very small spot, producing deep, narrow welds with little distortion.
- [Riveting](https://manufacturingprocesses.org/processes/joining/riveting.md): Riveting sets a rivet through aligned holes and upsets its tail so the joint clamps permanently, with no heat and no access to both sides for blind types.
- [Adhesive Bonding](https://manufacturingprocesses.org/processes/joining/adhesive-bonding.md): Adhesive bonding joins parts with a cured polymer layer that spreads load over the whole bond area instead of concentrating it at fastener points.
- [Soldering and Brazing](https://manufacturingprocesses.org/processes/joining/soldering-and-brazing.md): Soldering and brazing join metals with a molten filler that wets the joint by capillary action while the parent metal stays solid.

## Related processes

- [Arc Welding](https://manufacturingprocesses.org/processes/joining/arc-welding.md): Arc welding melts and fuses metal with an electric arc between an electrode and the work, shielded from the atmosphere by gas or flux.
- [Forging](https://manufacturingprocesses.org/processes/forming/forging.md): Forging deforms solid metal under compressive load in a hammer, press or dies, so the grain flow follows the part contour and fatigue strength rises.
- [Metal Extrusion](https://manufacturingprocesses.org/processes/forming/metal-extrusion.md): Metal extrusion pushes heated billet through a shaped die, producing continuous aluminum or copper profile of constant cross-section.
- [Resistance Welding](https://manufacturingprocesses.org/processes/joining/resistance-welding.md): Resistance welding passes a heavy current through the joint so its own electrical resistance melts a nugget between the sheets.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/joining/friction-welding)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Hot Plate Welding"
category: "Joining"
subcategory: "Thermal"
materials: ["Plastic"]
tolerances: "Assembly height after welding ±0.010–0.020 in (±0.25–0.5 mm), controlled by a positive stop rather than by molded part height; meltdown 0.04–0.16 in (1–4 mm)"
volumes: "100 to 1,000,000+ per year; field pipe fusion is one joint at a time"
lead_time: "15–120 s per cycle; 3–8 weeks for platen and nest tooling"
url: https://manufacturingprocesses.org/processes/joining/hot-plate-welding
---

# Hot Plate Welding

Hot plate welding presses two thermoplastic parts against a heated platen to melt the joint faces, then clamps them together to fuse.

- **Category**: [Joining](https://manufacturingprocesses.org/processes/joining.md)
- **Family**: Thermal
- **Materials**: Plastic
- **Typical tolerances**: Assembly height after welding ±0.010–0.020 in (±0.25–0.5 mm), controlled by a positive stop rather than by molded part height; meltdown 0.04–0.16 in (1–4 mm)
- **Typical volumes**: 100 to 1,000,000+ per year; field pipe fusion is one joint at a time
- **Lead time**: 15–120 s per cycle; 3–8 weeks for platen and nest tooling

## Overview

Hot plate welding presses two thermoplastic parts against a heated platen until the joint faces melt, withdraws the platen, and closes the parts together under pressure to fuse. The platen typically runs 100–200 °F (55–110 °C) above the resin's melt temperature — around 400–570 °F (200–300 °C) for polyolefins, up to 750 °F (400 °C) for high-temperature resins.

It is the slowest of the plastic welding processes at 15–120 s per cycle, and the most forgiving. Because the melt is created by conduction rather than by mechanical motion, the joint can be any shape in three dimensions, the parts can be very large, and dimensional variation in the moldings is absorbed by the melt phase.

That combination is why it dominates two very different applications: HDPE pipe butt fusion in the field, where ASTM F2620 specifies a 400–450 °F (204–232 °C) heater and 60–90 psi (0.41–0.62 MPa) interfacial pressure, and large molded assemblies like fluid reservoirs, battery cases and automotive lamps.

## How it works

1. **Fixture and align.** Both parts are clamped in nests. Because the melt phase absorbs a lot of variation, fit-up requirements are looser than for any other plastic welding process.
2. **Matching (melt) phase.** The heated platen is inserted between the parts and they are pressed against it at low pressure for 15–60 s. This first phase planarizes the joint faces — high spots melt away — and establishes an even melt layer.
3. **Heat soak.** Pressure is reduced or removed while the melt layer grows to the required depth by conduction. Depth of melt, not surface temperature, is what determines weld quality.
4. **Changeover.** The platen retracts and the parts close. This has to happen fast — typically under 1–2 s — because the exposed melt begins to cool and, in oxidizable resins, to degrade.
5. **Joining (forge) phase.** The parts are pressed together to a controlled meltdown, then held under pressure for 10–30 s while the joint cools and solidifies.

### Contact and non-contact platens

**Contact** platens are aluminum with a PTFE or PTFE-composite coating so PE and PP release cleanly. The coating limits platen temperature to roughly 500 °F (260 °C).

**Non-contact (radiant / infrared)** platens hold the part a small distance off a hot or IR-emitting surface and heat it by radiation. Nothing touches the melt, so nothing sticks and nothing contaminates — this is the route for nylon, POM and filled grades that stick to or degrade on a contact platen, and for high-temperature resins above the coating limit.

### Pipe butt fusion

The field version of the same process. Pipe ends are faced square, the heater plate is inserted, a bead is raised, the plate is removed and the ends are pressed together and held while the joint cools. A qualified HDPE butt fusion joint is not the weak point of the system — bend and tensile testing fails in the pipe wall, not the fusion line.

## Design guidelines

### The joint can be any shape — use that

Hot plate welding's real advantage is geometric freedom. Joint lines can step, curve in three dimensions and run around irregular perimeters, because a matching platen profile is machined to suit. Do not flatten a joint line for the process the way linear vibration welding demands.

### Design in the meltdown

Total meltdown is typically 0.04–0.16 in (1–4 mm) of combined height. Dimension the assembly so the finished height is set by a positive stop on the tooling rather than by the sum of the two molded parts, and put the tolerance-critical features away from the joint.

### Flash traps

Melt is squeezed outward on both sides of the joint. Specify flash traps — recessed channels sized for the displaced volume — on any surface where flash is cosmetically or functionally unacceptable. In pipe fusion, the equivalent is the external and internal bead, and its double-roll shape is the field inspector's primary quality signal.

### Joint land width and support

Make the joint land at least as wide as the nominal wall, wider where the joint seals or carries load. Support the joint plane against the forge pressure with ribs or a nesting fixture; an unsupported wall bows and the weld opens as it cools.

### Resin selection

| Resin | Hot plate weldable? | Notes |
| --- | --- | --- |
| HDPE, LDPE, PP | Excellent | The core application; PTFE-coated contact platen at 400–570 °F (200–300 °C) |
| PA (nylon) | Good, non-contact preferred | Sticks to contact platens and oxidizes at the melt face; dry the parts first |
| POM (acetal) | Difficult | Degrades and sticks; non-contact radiant heating only |
| PC, PC/ABS, ABS | Good | Higher platen temperature; watch for stress cracking at the weld line |
| PEI, PPS, PEEK | Non-contact only | Platen temperatures above the PTFE coating limit |
| 30–50% glass filled grades | Good | Better than ultrasonic; weld line still weaker than bulk material |
| PE to PP | Poor | Different polyolefins do not form a strong weld |
| Dissimilar resins generally | Case by case | Needs chemical compatibility and close melt temperature |

Joint lands, flash traps and support ribs are molded features — see the [injection molding design guidelines chart](/charts/injection-molding-design-guidelines).

### Inspection

Process data first: platen temperature, melt time, changeover time and meltdown distance are recorded per cycle and are what actually correlate with joint quality. On pipe fusion, the visible double-roll bead shape and size is the field acceptance criterion, backed by bend-back testing and high-speed tensile impact on qualification joints. On molded assemblies, burst pressure and pressure-decay leak testing, plus cross-sections through the weld to confirm melt depth across the joint.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Platen temperature above melt temp | 100–200 °F (55–110 °C) | Coating limit ~500 °F (260 °C) | Too cold gives no melt depth, too hot degrades the resin |
| Melt (matching) time | 15–60 s | Resin dependent | Melt *depth* is the controlled variable, not surface temperature |
| Changeover time | Under 1 s | 2 s | Exposed melt cools and oxidizes |
| Meltdown allowance | 0.08 in (2 mm) | 0.04–0.16 in (1–4 mm) | Assembly height must be dimensioned around it |
| Joint land width | 1–2x nominal wall | 1x wall | Sets weld area and seal reliability |
| Flash trap | Both sides where flash matters | — | Displaced melt must be accommodated |
| HDPE pipe interfacial pressure | Per ASTM F2620 | 60–90 psi (0.41–0.62 MPa) | Standardized for qualified field joints |

## Cost drivers

Machine capital scales with part size and platen area, and the platen itself is part-specific tooling: a machined, coated or IR-element plate profiled to the joint line, plus a pair of holding nests. That is more tooling than ultrasonic welding needs but less than a mold.

Cycle time is the operating cost, and it is long — 15–120 s against 2 s for ultrasonic. At high volume that difference dominates and pushes designs toward vibration or ultrasonic welding wherever the geometry allows. At moderate volume, or where the joint is large or three-dimensional, hot plate welding is often the only process that works at all.

Energy is a genuine line item because the platen is held at temperature continuously, and platen coatings are consumable — PTFE degrades and must be recoated on a schedule.

1. **Use non-contact heating for sticky resins** rather than fighting PTFE coating life on nylon, acetal or filled grades.
2. **Keep the joint line as simple as the part allows.** Platen cost tracks joint complexity even though the process tolerates complexity.
3. **Check vibration welding first** if the joint happens to be planar and the volume is high — the cycle is 3–10x faster.
4. **Design flash traps in** so no deflashing operation is needed downstream.
5. **Consolidate joints onto one platen.** Multiple joint lines welded in the same cycle amortize both the platen and the cycle time.

## FAQ

### What temperature should the hot plate be set to?

Typically 100–200 °F (55–110 °C) above the resin's melt temperature — roughly 400–570 °F (200–300 °C) for polyolefins and higher for engineering resins. PTFE-coated contact platens cap out near 500 °F (260 °C), so anything above that requires non-contact radiant heating. For HDPE pipe butt fusion, ASTM F2620 specifies a 400–450 °F (204–232 °C) heater.

### Why is hot plate welding used on large parts when it is so slow?

Because it is the only plastic welding process with no geometric constraint on the joint. Melt is created by conduction from a platen machined to match the joint profile, so the joint can step and curve in three dimensions and the parts can be large. It also absorbs molded-part variation in the melt phase, which vibration and ultrasonic welding cannot.

### Is a butt-fused HDPE pipe joint as strong as the pipe?

Yes, when made to a qualified procedure. In bend-back and tensile testing a properly fused joint fails in the pipe wall rather than at the fusion line, which is why butt fusion is accepted for buried gas and water mains. Field acceptance rests on the visible double-roll bead shape plus adherence to the specified heater temperature, pressure and cooling time.

### Why do some resins need non-contact hot plate welding?

Nylon, acetal and many filled grades stick to a coated platen or oxidize at the exposed melt face, leaving stringing and contaminated joints. Radiant or infrared platens heat the part across a small air gap so nothing touches the melt. Non-contact heating is also mandatory above about 500 °F (260 °C), where PTFE platen coatings break down.

### How much do the parts shorten during hot plate welding?

Total meltdown is typically 0.04–0.16 in (1–4 mm) of combined height — more than vibration or ultrasonic welding. Dimension the assembly so the finished height comes from a positive stop on the tooling rather than from the sum of the two molded parts, and keep tolerance-critical features away from the joint plane.

## Alternative processes

- [Laser Plastic Welding](https://manufacturingprocesses.org/processes/joining/laser-plastic-welding.md): Laser plastic welding passes a beam through a transmissive part into an absorbing one, melting only the interface between them.
- [Adhesive Bonding](https://manufacturingprocesses.org/processes/joining/adhesive-bonding.md): Adhesive bonding joins parts with a cured polymer layer that spreads load over the whole bond area instead of concentrating it at fastener points.
- [Mechanical Fastening](https://manufacturingprocesses.org/processes/joining/mechanical-fastening.md): Mechanical fastening joins parts with screws, inserts, clinch hardware or snap fits so the joint can be assembled and taken apart again.
- [Overmolding](https://manufacturingprocesses.org/processes/forming/overmolding.md): Overmolding molds a second material, usually a soft elastomer, directly onto an already-molded rigid substrate so the two bond permanently.

## Related processes

- [Vibration Welding](https://manufacturingprocesses.org/processes/joining/vibration-welding.md): Vibration welding rubs two thermoplastic parts together in-plane at low frequency under pressure until the interface melts and fuses.
- [Ultrasonic Welding](https://manufacturingprocesses.org/processes/joining/ultrasonic-welding.md): Ultrasonic welding applies high-frequency vibration through a horn to melt a thermoplastic joint locally in well under a second.
- [Staking](https://manufacturingprocesses.org/processes/joining/staking.md): Heat staking softens a molded plastic boss with hot air or ultrasonic energy and reforms it into a head that traps the mating part.
- [Plastic Extrusion](https://manufacturingprocesses.org/processes/forming/plastic-extrusion.md): Plastic extrusion pushes molten polymer through a shaped die and cools it, producing continuous profile, sheet, pipe or film of constant cross-section.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/joining/hot-plate-welding)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Joinery"
category: "Joining"
subcategory: "Mechanical"
materials: ["Wood"]
tolerances: "Glue line of 0.002–0.006 in (0.05–0.15 mm) for PVA; joints fitted to firm hand pressure rather than to a numeric tolerance"
volumes: "1 to a few thousand — hand work for one-offs, jigged or CNC-cut joinery for batch production"
lead_time: "Minutes to an hour per joint depending on method; 30–60 minutes clamp time and 24 hours to full adhesive cure"
url: https://manufacturingprocesses.org/processes/joining/joinery
---

# Joinery

Joinery cuts interlocking geometry into timber so the parts locate and hold each other, with or without adhesive or fasteners.

- **Category**: [Joining](https://manufacturingprocesses.org/processes/joining.md)
- **Family**: Mechanical
- **Materials**: Wood
- **Typical tolerances**: Glue line of 0.002–0.006 in (0.05–0.15 mm) for PVA; joints fitted to firm hand pressure rather than to a numeric tolerance
- **Typical volumes**: 1 to a few thousand — hand work for one-offs, jigged or CNC-cut joinery for batch production
- **Lead time**: Minutes to an hour per joint depending on method; 30–60 minutes clamp time and 24 hours to full adhesive cure

## Overview

Joinery cuts interlocking geometry into timber so the parts locate and hold each other, with adhesive and fasteners as reinforcement rather than as the joint itself. The reason it exists is a material fact: an end-grain glue joint develops only a fraction of the strength of a long-grain one, so a plain butt joint between two pieces of wood is structurally worthless. Joinery converts that end-grain butt into long-grain glue area plus a mechanical interlock.

Properly executed, the result is stronger than the wood: a long-grain glued joint tested in block shear fails in the timber rather than at the glue line, and acceptance is generally stated as a percentage of wood failure rather than a stress.

The dominant families are mortise and tenon (frames), dovetail and box joints (carcase corners), lap and bridle joints (crossing members), dado and rabbet (shelving and casework), and loose-tenon systems — dowels, biscuits and domino tenons — which trade some strength for repeatability and speed.

## How it works

1. **Mill the stock true.** Every joint dimension is referenced from flat, square, parallel faces. Timber that is not four-square first will not produce a tight joint no matter how carefully the joint is cut.
2. **Lay out from a face side and face edge.** All measurements come from the same two reference surfaces so cumulative error does not accumulate around a frame.
3. **Cut to a push fit.** A mortise and tenon should assemble under firm hand pressure. A joint that needs a mallet is too tight: it squeezes the adhesive out of the joint line and starves it, and it can split the mortise cheek.
4. **Glue and clamp.** PVA and aliphatic resin adhesives work best in a thin, close-fitting glue line of roughly 0.002–0.006 in (0.05–0.15 mm) and have poor gap-filling ability beyond that. Clamp at 100–150 psi for softwoods and 175–250 psi for hardwoods. Open assembly time is typically 5–10 minutes, clamp time 30–60 minutes, and full cure 24 hours. Where gaps are unavoidable, epoxy tolerates them and PVA does not.
5. **Peg or wedge where appropriate.** Draw-boring a pegged mortise and tenon — offsetting the peg hole in the tenon by 1/16–1/8 in (1.5–3 mm) toward the shoulder — pulls the joint tight mechanically and holds it without adhesive at all.

### Classic proportions

Tenon thickness is traditionally about one third of the stock thickness, rounded to the nearest available chisel or router bit size, with tenon length two thirds to three quarters of the mortised member's width. Dovetails are cut at a 1:6 slope for softwood and 1:8 for hardwood — roughly 9.5° and 7° — the shallower angle reflecting hardwood's greater resistance to short-grain failure at the pin tips.

## Design guidelines

### Design around wood movement, not against it

Wood moves across the grain and essentially not along it. In a typical interior environment a wide board will move roughly 1/8–1/4 in per 12 in (3–6 mm per 300 mm) of width between the humid and dry seasons, and no joint will stop it. The design rules follow directly: never glue a wide panel cross-grain to a rail, let tabletops float on slotted fasteners or buttons, size breadboard ends so only the center is pinned, and orient panels in frames so they can expand into their grooves.

### Maximize long-grain glue area

The joint's strength comes from long-grain-to-long-grain contact. A mortise and tenon works because the tenon cheeks present long grain to the mortise walls; a dovetail works because the interlocking geometry adds mechanical resistance where glue area is limited. When choosing between joints, count the long-grain area, not the total surface area.

### Fit is a strength variable

Aim for a joint that assembles under firm hand pressure. Too loose and PVA cannot bridge the gap; too tight and the adhesive is squeezed out and the joint is starved, with the additional risk of splitting the mortise. Where a gap is unavoidable — a repair, a hand-cut joint that opened up — switch adhesives rather than adding clamping force.

| Joint | Typical proportion | Use | Why |
| --- | --- | --- | --- |
| Mortise and tenon | Tenon 1/3 stock thickness, length 2/3–3/4 of member width | Frames, doors, chairs, tables | Maximum long-grain area in a right-angle frame joint |
| Through dovetail | 1:6 softwood, 1:8 hardwood | Carcase corners, drawers | Mechanically resists pull-out even without adhesive |
| Box / finger joint | Fingers roughly equal to stock thickness | Boxes, drawers, machine-cut casework | Large glue area, simple to jig |
| Half lap | Half thickness removed from each member | Crossing rails, frames | Fast; the weakest of the frame joints |
| Dado / housing | 1/3 stock thickness deep | Shelving, casework dividers | Supports load without relying on the glue line |
| Loose tenon (domino, dowel) | Tenon 1/3 stock thickness | Repeat production frames | Mortise and tenon strength with machine repeatability |
| Butt joint, end grain | — | Avoid | End-grain glue joints carry a fraction of long-grain strength |

Mortise widths and dowel holes are cut with standard bit sizes — see the [drill size chart](/charts/drill-size-chart).

### Inspection

Visual first: gap-free shoulders, continuous glue squeeze-out along the whole joint line, and shoulders that pull tight without clamp force fighting a badly cut joint. Beyond that, block shear testing to ASTM D905 on sample joints, where acceptance is generally stated as a minimum percentage of wood failure rather than a stress value, and moisture content verification with a meter — the timber and the shop should be at the same equilibrium moisture content before any joint is cut.

## Cost drivers

Joinery is labor. Hand-cut dovetails in a drawer take an experienced maker a substantial fraction of an hour per corner; the same corner cut on a jig takes minutes, and a machine-cut loose tenon takes seconds. The cost question is therefore always which joint, cut by what method, rather than which material.

Machine setup is the countervailing cost. Router jigs, dovetail templates, hollow-chisel mortisers and domino machines all pay for themselves over a batch and cost time on a one-off. Timber itself is a real input cost, and joinery consumes length — tenons, laps and dovetails all require stock beyond the finished dimension.

Clamp time occupies fixtures and floor space. A 30–60 minute clamp cycle at 24 hours to full cure sets how many assemblies can move through a bench per day, and it is usually the binding constraint in a small shop.

1. **Use loose tenons for repeat work.** Domino and dowel joinery gives most of the strength of a cut mortise and tenon at a fraction of the machine time.
2. **Jig anything you make more than twice.** The jig pays for itself on the second batch and improves consistency on the first.
3. **Batch the same joint across an assembly** so one setup cuts every mortise before the machine is reconfigured.
4. **Buy timber at the right moisture content** and let it acclimate — a joint cut in wet stock will open regardless of how well it was made.
5. **Reserve hand-cut joinery for what is seen.** Through dovetails on a visible drawer front, machine joints everywhere else.

## FAQ

### Why can't you just glue and butt two pieces of wood together?

Because an end-grain glue joint develops only a fraction of the strength of a long-grain one — the open cell ends absorb adhesive and there is nothing for it to key into. Joinery exists to convert an end-grain butt into long-grain glue area plus a mechanical interlock. A properly glued long-grain joint, by contrast, is stronger than the surrounding timber.

### How thick should a tenon be?

About one third of the stock thickness, rounded to the nearest available chisel or router bit size, with a length of two thirds to three quarters of the mortised member's width. Thinner and the tenon itself becomes the weak point; thicker and the mortise cheeks are left too thin and split. The one-third rule balances the two failure modes.

### What dovetail angle should I use?

1:6 for softwoods, about 9.5°, and 1:8 for hardwoods, about 7°. The steeper angle in softwood provides more mechanical interlock in a material that compresses easily; the shallower angle in hardwood avoids leaving fragile short grain at the pin tips, which would break out when the joint is assembled.

### How tight should a mortise and tenon fit?

It should assemble under firm hand pressure. A joint driven home with a mallet is too tight: it squeezes the adhesive out of the glue line and starves the joint, and it can split the mortise cheek. A joint that falls together is too loose for PVA, which has almost no gap-filling ability beyond about 0.006 in (0.15 mm) — use epoxy if a gap cannot be avoided.

### How much does wood move seasonally, and how do I design for it?

Across the grain, roughly 1/8–1/4 in per 12 in (3–6 mm per 300 mm) of width in a typical interior environment; along the grain, effectively nothing. Design so the movement is unrestrained: float tabletops on slotted fasteners or buttons, leave panels loose in their frame grooves, and never glue a wide panel cross-grain to a rail.

## Alternative processes

- [Mechanical Fastening](https://manufacturingprocesses.org/processes/joining/mechanical-fastening.md): Mechanical fastening joins parts with screws, inserts, clinch hardware or snap fits so the joint can be assembled and taken apart again.
- [Adhesive Bonding](https://manufacturingprocesses.org/processes/joining/adhesive-bonding.md): Adhesive bonding joins parts with a cured polymer layer that spreads load over the whole bond area instead of concentrating it at fastener points.
- [Wood Laminating](https://manufacturingprocesses.org/processes/forming/wood-laminating.md): Wood laminating glues thin veneers or kerfed strips over a form so the assembly holds a curve permanently once the adhesive has cured.

## Related processes

- [Timber Frame Structures](https://manufacturingprocesses.org/processes/joining/timber-frame-structures.md): Timber frame structures assemble large solid or engineered timber members into a load-bearing frame using cut joints and steel connectors.
- [Steam Bending](https://manufacturingprocesses.org/processes/forming/steam-bending.md): Steam bending plasticises solid timber with saturated steam so it can be bent around a former and held until it dries into shape.
- [CNC Milling](https://manufacturingprocesses.org/processes/cutting/cnc-milling.md): CNC milling removes material with a rotating multi-flute cutter moved along programmed paths, producing prismatic parts from solid stock.
- [Sawing](https://manufacturingprocesses.org/processes/cutting/sawing.md): Sawing separates stock with a toothed blade — band, circular or cold saw — and is the default first operation on bar, tube and plate.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/joining/joinery)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Laser Plastic Welding"
category: "Joining"
subcategory: "Thermal"
materials: ["Plastic"]
tolerances: "Clamped joint gap ≤ 0.004 in (0.1 mm) required across the seam; quasi-simultaneous collapse controlled to about ±0.001 in (±0.025 mm)"
volumes: "1,000 to 5,000,000+ per year; contour welding is viable from prototype quantities"
lead_time: "Seconds to a few tens of seconds per part for contour welding, under 5 s for simultaneous; 4–10 weeks for transparent clamping tooling and part-specific optics"
url: https://manufacturingprocesses.org/processes/joining/laser-plastic-welding
---

# Laser Plastic Welding

Laser plastic welding passes a beam through a transmissive part into an absorbing one, melting only the interface between them.

- **Category**: [Joining](https://manufacturingprocesses.org/processes/joining.md)
- **Family**: Thermal
- **Materials**: Plastic
- **Typical tolerances**: Clamped joint gap ≤ 0.004 in (0.1 mm) required across the seam; quasi-simultaneous collapse controlled to about ±0.001 in (±0.025 mm)
- **Typical volumes**: 1,000 to 5,000,000+ per year; contour welding is viable from prototype quantities
- **Lead time**: Seconds to a few tens of seconds per part for contour welding, under 5 s for simultaneous; 4–10 weeks for transparent clamping tooling and part-specific optics

## Overview

Laser plastic welding passes a near-infrared beam — 808–1,064 nm from a diode or fiber source — straight through a transmissive upper part and into an absorbing lower part, where it converts to heat. Only the interface melts. The upper part stays cold enough to hold a mirror finish, a printed graphic or a bonded film through the weld.

That selectivity is the whole value: no vibration, no particulate, no flash, no heat anywhere except a seam 0.02–0.08 in (0.5–2 mm) wide. It is the standard route for microfluidic cartridges, sensor housings, medical devices and electronics enclosures with parts that other welding processes would shake apart.

The controlling constraint is contact. The clamped gap between the two parts must stay under about 0.004 in (0.1 mm) across the whole seam, because the beam heats only what it reaches — there is no melt pool to bridge with.

## How it works

1. **Material pairing.** One part must transmit the beam and the other must absorb it. The absorber is usually pigmented with 0.1–0.5% carbon black, or with a near-IR absorbing additive that leaves the part visually clear or color-matched.
2. **Clamp.** A transparent fixture — glass or acrylic — presses the parts together through the beam path. Contact pressure closes the gap and, once melting starts, pushes the parts together as the joint collapses.
3. **Deliver the beam.** 10–500 W of diode or fiber laser power passes through the upper part with little absorption, then deposits energy in the first few thousandths of an inch of the absorber.
4. **Melt and conduct.** The absorber melts, heat conducts back into the transmissive part, and a melt layer forms on both sides of the interface. Collapse of 0.002–0.008 in (0.05–0.2 mm) confirms fusion.
5. **Cool under clamp.** Pressure is held while the seam solidifies.

### The four scanning strategies

**Contour welding** traces the joint line once with a moving spot at 0.4–20 in/s (10–500 mm/s). Simple, flexible, and suits long or one-off seams; no collapse is possible because the seam behind the spot has already frozen.

**Quasi-simultaneous welding** uses galvanometer mirrors to sweep the entire joint line many times per second, so the whole seam is molten at once. That allows genuine collapse — the parts close under clamp pressure, taking up molded-part variation — and gives the best hermetic sealing. It is the workhorse for small and medium parts.

**Simultaneous welding** illuminates the whole joint at once through a fixed array of fiber-coupled diodes or a mask. Cycle times drop under a couple of seconds, but the optics are part-specific tooling.

**Mask welding** projects the beam through a mask onto flat parts, producing seams down to a few hundredths of an inch — used for microfluidic and lab-on-chip devices where channel walls must be sharply defined.

## Design guidelines

### Close the gap — this is the design problem

Laser welding has no capacity to bridge a gap: any place the parts do not touch is a place that will not weld. Design the clamped gap to stay under 0.004 in (0.1 mm) along the entire seam. That means supporting the joint from both sides of the beam path, avoiding long unsupported spans, and designing the parts so clamp force closes them rather than bowing them.

### Give the beam a clean path

Everything the beam passes through attenuates it. Keep the transmissive wall as thin as function allows, keep it free of ribs, bosses, gate marks, sink and text within the beam footprint, and mold it with a smooth surface — a textured or frosted surface scatters the beam and widens the seam unpredictably. Approach the joint perpendicular where possible; steep angles cost transmission and smear the focus.

### Design the joint for collapse

With quasi-simultaneous or simultaneous welding, specify 0.004–0.010 in (0.1–0.25 mm) of collapse into the design, and set the finished assembly height off a positive stop or the machine's collapse control. Collapse is what absorbs molding variation and squeezes out any residual gap, and it is the single biggest reliability lever available.

### Material transmission

| Material at the joint | Role | Notes |
| --- | --- | --- |
| Natural / unpigmented PP, PE, PA, PC, ABS, PMMA | Transmitter | Good transmission at 808–1,064 nm |
| Carbon-black pigmented resin (0.1–0.5%) | Absorber | The standard, cheapest absorber |
| Near-IR absorbing additive | Absorber | Lets the absorber stay clear or color-matched to the transmitter |
| TiO2-pigmented white resin | Poor transmitter | Scatters the beam heavily; white-on-white is difficult |
| 30%+ glass filled grade | Poor transmitter | Fibers scatter; transmission falls sharply with fill and wall thickness |
| Carbon black in the **upper** part | Will not work | The beam never reaches the joint |
| Dissimilar resin pair | Case by case | Needs chemical compatibility and close melt temperature, same as any plastic weld |

Wall thickness, gate placement and rib rules that govern the beam path are on the [injection molding design guidelines chart](/charts/injection-molding-design-guidelines).

### Inspection

The transmissive part is an inspection window: the weld seam can be seen and measured optically through it, which no other plastic welding process allows. Beyond that, collapse distance is logged per cycle and is the primary process signal for quasi-simultaneous welding; pressure-decay or helium leak testing verifies hermetic parts; burst and pull testing on lot samples verifies strength; and infrared thermography during welding detects cold spots caused by local gaps.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Clamped joint gap | 0.001 in (0.025 mm) | 0.004 in (0.1 mm) | The beam cannot bridge what it does not reach |
| Transmissive wall in the beam path | As thin as function allows | — | Every millimeter of path attenuates |
| Surface finish in the beam path | Smooth, untextured | — | Texture scatters the beam and widens the seam |
| Designed collapse (quasi-simultaneous) | 0.006 in (0.15 mm) | 0.004–0.010 in (0.1–0.25 mm) | Absorbs molding variation and closes residual gaps |
| Weld seam width | 0.04 in (1 mm) | 0.02–0.08 in (0.5–2 mm) | Sets joint strength; wider needs more power or slower travel |
| Absorber pigment | 0.1–0.5% carbon black | — | Enough to absorb in the first few thousandths of an inch |
| Clamp support | Both sides of the seam | — | Unsupported spans bow open under clamp force |

## Cost drivers

Capital is the main cost. A laser source, scanner or fiber array, transparent clamping tooling and a Class 1 enclosure represent a larger investment than an ultrasonic or hot plate welder, and simultaneous welding adds part-specific optics on top. Against that, the cycle is short, there are no consumables, and the process has essentially no reject mechanism once the gap is under control.

The economics work when the alternative fails on quality rather than cost: parts that cannot tolerate vibration, assemblies where particulate is a contamination issue, seams that must be narrow and clean, or products where the visible upper surface must survive the joining operation. Comparing laser welding to ultrasonic welding on price per part alone will always favor ultrasonics — the comparison only makes sense against scrap and warranty.

Tooling cost is dominated by the clamping fixture, which must be optically transparent, dimensionally accurate and stiff enough to close the gap everywhere along the seam.

1. **Use contour welding for prototypes and low volume** — no part-specific optics, just a program change.
2. **Design one absorber and one transmitter across a product family** so the same material pair and parameters carry over.
3. **Spend on the clamping fixture, not the laser.** Gap control is what determines yield.
4. **Use an IR-absorbing additive** instead of carbon black only where the part must be clear or color-matched; it costs more.
5. **Check that quasi-simultaneous collapse is available** before tightening molded part tolerances — collapse is cheaper than tooling precision.

## FAQ

### How does laser welding melt only the joint and not the top part?

The upper part is transmissive at the laser wavelength (808–1,064 nm) and passes the beam with little absorption; the lower part contains carbon black or a near-IR absorbing additive that converts the beam to heat in the first few thousandths of an inch. Melting therefore starts exactly at the interface and conducts back into the upper part from there.

### How tight does the fit-up need to be for laser plastic welding?

The clamped gap must stay under about 0.004 in (0.1 mm) across the entire seam. There is no melt pool to bridge a gap — anywhere the parts do not touch simply does not weld. Quasi-simultaneous welding relaxes this somewhat because the whole seam is molten at once and clamp pressure can collapse the joint by 0.004–0.010 in (0.1–0.25 mm).

### Can you laser weld two clear plastic parts?

Yes, using a near-IR absorbing additive applied at the interface or compounded into the lower part instead of carbon black. The additive absorbs at the laser wavelength but is nearly invisible, so both parts stay clear or color-matched. It costs more than carbon black, which is why carbon black remains the default where appearance permits.

### Why would I choose laser welding over ultrasonic welding?

When the part cannot tolerate 20 kHz vibration — populated PCBs, sensors, crystals, filled microfluidic channels — or when particulate generation is a contamination problem, or when the visible surface must survive the joining step. On price per part alone ultrasonic welding usually wins; laser welding justifies itself against scrap, contamination and warranty.

### Why are white plastic parts hard to laser weld?

Titanium dioxide, the pigment that makes plastic white, scatters near-infrared light strongly, so the beam disperses inside the transmissive part instead of reaching the joint. Glass fill above about 30% causes the same problem. Both can sometimes be worked around with thinner walls, higher power or a different pigment package, but they should be flagged at design review rather than discovered at process validation.

## Alternative processes

- [Vibration Welding](https://manufacturingprocesses.org/processes/joining/vibration-welding.md): Vibration welding rubs two thermoplastic parts together in-plane at low frequency under pressure until the interface melts and fuses.
- [Adhesive Bonding](https://manufacturingprocesses.org/processes/joining/adhesive-bonding.md): Adhesive bonding joins parts with a cured polymer layer that spreads load over the whole bond area instead of concentrating it at fastener points.
- [Staking](https://manufacturingprocesses.org/processes/joining/staking.md): Heat staking softens a molded plastic boss with hot air or ultrasonic energy and reforms it into a head that traps the mating part.
- [Mechanical Fastening](https://manufacturingprocesses.org/processes/joining/mechanical-fastening.md): Mechanical fastening joins parts with screws, inserts, clinch hardware or snap fits so the joint can be assembled and taken apart again.

## Related processes

- [Ultrasonic Welding](https://manufacturingprocesses.org/processes/joining/ultrasonic-welding.md): Ultrasonic welding applies high-frequency vibration through a horn to melt a thermoplastic joint locally in well under a second.
- [Hot Plate Welding](https://manufacturingprocesses.org/processes/joining/hot-plate-welding.md): Hot plate welding presses two thermoplastic parts against a heated platen to melt the joint faces, then clamps them together to fuse.
- [Laser Marking and Engraving](https://manufacturingprocesses.org/processes/finishing/laser-marking-and-engraving.md): Laser marking alters the surface with a focused beam — annealing, etching or engraving it — to make permanent, contact-free marks.
- [Injection Molding](https://manufacturingprocesses.org/processes/forming/injection-molding.md): Injection molding forces molten thermoplastic into a steel mold under high pressure, producing high-volume identical plastic parts.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/joining/laser-plastic-welding)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Mechanical Fastening"
category: "Joining"
subcategory: "Mechanical"
materials: ["Metal", "Plastic", "Wood", "Composite"]
tolerances: "Normal-fit clearance holes about 1/32 in (0.8 mm) over nominal per ASME B18.2.8 for screws up to 1/2 in; preload scatter ±25–35% with torque control, ±15% with angle control, ±5–10% with bolt elongation measurement"
volumes: "1 to 10,000,000+"
lead_time: "Off-the-shelf hardware in hours to days; seconds per fastener at assembly"
url: https://manufacturingprocesses.org/processes/joining/mechanical-fastening
---

# Mechanical Fastening

Mechanical fastening joins parts with screws, inserts, clinch hardware or snap fits so the joint can be assembled and taken apart again.

- **Category**: [Joining](https://manufacturingprocesses.org/processes/joining.md)
- **Family**: Mechanical
- **Materials**: Metal, Plastic, Wood, Composite
- **Typical tolerances**: Normal-fit clearance holes about 1/32 in (0.8 mm) over nominal per ASME B18.2.8 for screws up to 1/2 in; preload scatter ±25–35% with torque control, ±15% with angle control, ±5–10% with bolt elongation measurement
- **Typical volumes**: 1 to 10,000,000+
- **Lead time**: Off-the-shelf hardware in hours to days; seconds per fastener at assembly

## Overview

Mechanical fastening joins parts with screws, bolts, threaded inserts, self-clinching hardware or molded snap fits, so the assembly can be taken apart again without destroying it. That serviceability is the reason it survives against faster and cheaper joining processes.

A bolted joint does not work the way most drawings imply. Clamp load, not the bolt shank, carries shear in a properly designed joint — friction between the clamped faces does the work, and the bolt only sees shear in a bearing-type connection. Preload is therefore the design variable, typically targeted at 65–75% of the fastener's proof load.

Getting that preload is the hard part. Torque control delivers it with ±25–35% scatter because most of the applied torque is consumed by friction; angle control tightens that to about ±15%, and direct bolt elongation measurement to ±5–10%. Volumes run from one-off to tens of millions of joints a year.

## How it works

1. **Select the fastener and the thread engagement.** Engagement length is set by the weaker material: about 1x nominal diameter into steel of similar strength, 1.5–2x into aluminum, 2x into magnesium, and 2–2.5x into thermoplastic. Too little engagement strips the female thread before the bolt reaches preload.
2. **Prepare the hole.** Tapped holes are drilled to the tap drill size for the required thread engagement percentage — see the [tap drill chart](/charts/tap-drill-chart) and the [drill size chart](/charts/drill-size-chart). Clearance holes follow ASME B18.2.8: a normal-fit clearance hole is about 1/32 in (0.8 mm) over nominal for screws up to 1/2 in. Thread dimensions and pitch are on the [thread size chart](/charts/thread-size-chart).
3. **Tighten and develop preload.** Torque relates to preload through T = K x D x F, where K is the nut factor — about 0.20 for plain steel, 0.15 for zinc-plated or lubricated, 0.10 with wax or MoS2. Because K varies with surface condition, so does the resulting clamp load: this is the source of the ±25–35% scatter. Torque values by size, grade and lubrication are on the [bolt torque chart](/charts/bolt-torque-chart).
4. **Verify.** Residual or breakaway torque audits, torque-angle signature monitoring on production tools, or ultrasonic bolt elongation measurement where the joint is critical.

### The other mechanical fastening routes

**Threaded inserts in plastic.** Heat-set or ultrasonic inserts give the highest pull-out and torque-out in thermoplastic because the melt flows into the knurl. Design the boss outside diameter at roughly 2x the insert outside diameter.

**Self-tapping screws in plastic.** Thread-forming screws for ductile thermoplastics (they displace rather than cut material, so there are no chips); thread-cutting screws for brittle and filled resins and for thermosets.

**Self-clinching hardware.** Nuts, studs and standoffs pressed into sheet metal, providing a load-bearing thread in material far too thin to tap. They require sheet softer than the fastener — typically HRB 80 or below for standard steel clinch hardware.

**Snap fits.** A molded cantilever or annular feature that deflects during assembly and returns to retain. No hardware at all, but every snap is a strain event in the resin.

## Design guidelines

### Design the joint around preload, not around torque

Target 65–75% of proof load for a joint that will be reused, and understand that torque control alone delivers that with ±25–35% scatter. If the joint's function depends on clamp load — a gasketed face, a friction-critical connection, a fatigue-loaded bolt — either specify angle control (±15%) or design so the scatter does not matter. Preload also has to survive: use a hardened washer to prevent embedment, and keep soft or gasket material out of the clamp path where it will creep and shed preload.

### Get the thread engagement right

Engagement is set by the weaker material, not by the bolt: 1x diameter into steel, 1.5–2x into aluminum, 2x into magnesium, 2–2.5x into thermoplastic. In a thin plate that cannot provide it, use a self-clinching nut, a threaded insert or a nut plate — not a longer screw into more of the same thin material.

### Bosses and inserts in plastic

Boss outside diameter should be about 2x the insert or screw major diameter. Add a 0.010–0.015 in (0.25–0.4 mm) counterbore at the top of the boss so the first thread does not blow it open. For self-tapping screws, the ratio of stripping torque to driving torque should be at least 3:1, and 5:1 is a better target for automated assembly — below 3:1 the driver will strip bosses routinely.

### Snap fits

Design a cantilever snap so maximum bending strain during assembly stays within the resin's published allowable — commonly around 2–4% for unfilled engineering thermoplastics on a one-time assembly, roughly 60% of that for a joint that cycles repeatedly, and 1–2% or less for glass-filled grades. Always take the allowable from the specific resin datasheet, and taper the beam so strain is distributed rather than concentrated at the root.

### Fastener and substrate compatibility

| Pair | Notes |
| --- | --- |
| Steel screw into steel | Baseline; 1x diameter engagement |
| Steel or stainless screw into aluminum | 1.5–2x engagement; galvanic couple — isolate or seal in wet environments |
| Stainless screw into stainless | Prone to galling under load; use a dissimilar grade, a lubricant or an anti-seize |
| Screw into magnesium | 2x engagement; strongly anodic — isolation is mandatory |
| Screw into thermoplastic | 2–2.5x engagement; use thread-forming screws or a heat-set insert |
| Screw into thermoset or 30%+ glass filled | Thread-cutting screws; thread-forming will crack the boss |
| Clinch nut into sheet | Sheet must be softer than the fastener, typically HRB 80 or below |
| Fastener into composite | Bearing failure governs — use a washer or a bushing to spread load |

For dissimilar-metal joints, keep the anodic index difference under about 0.25 V in harsh environments and 0.50 V in benign ones, per MIL-STD-889, or isolate with a washer, coating or sealant.

### Inspection

Torque audits — residual or breakaway — are the standard production check, though they measure torque rather than the preload that actually matters. Torque-angle signature monitoring on DC electric tools catches cross-threads, missing washers and soft joints in real time and is the practical high-volume method. For critical joints, ultrasonic bolt elongation measures preload directly. Threaded inserts and self-clinching hardware are verified by pull-out and torque-out testing on lot samples.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Preload target | 65–75% of proof load | Reusable joints | Leaves margin against yield and relaxation |
| Preload scatter, torque control | Expect ±25–35% | ±15% with angle control | Friction consumes most of the applied torque |
| Thread engagement into aluminum | 2x nominal diameter | 1.5x minimum | Female thread strips before the bolt reaches preload |
| Boss outside diameter for an insert | 2x insert diameter | — | Hoop strength around the knurl |
| Strip-to-drive torque ratio | 5:1 | 3:1 minimum | Below 3:1 assembly strips bosses routinely |
| Snap fit assembly strain | Per resin datasheet | ~2–4% unfilled, 1–2% filled | Exceeding it cracks the beam on first assembly |
| Dissimilar metal anodic index difference | Under 0.25 V harsh | 0.50 V benign | Galvanic corrosion of the anodic member |

## Cost drivers

Fastened assembly cost lives in three places: the purchased hardware, the features that receive it, and the labor to drive it. The hardware is usually the smallest of the three. Every fastener also implies a hole, a boss or a tapped feature in both parts, and a driving operation with its own station, tool and verification step.

Part count is therefore the lever. A design with twenty screws in five sizes needs five bit changes or five stations; the same design with twelve screws in one size needs one. Unique fastener part numbers carry their own cost in purchasing, inventory and line-side presentation.

Torque verification is a genuine recurring cost on anything safety-related — logged DC tools, audit sampling and the documentation behind them. That cost scales with the number of joints that need verifying, not with the number of fasteners.

1. **Standardize on one or two fastener sizes** per assembly and one drive type. Bit changes and tool stations cost more than the hardware.
2. **Delete fasteners with snap fits or staking** on non-serviceable joints — a snap fit costs nothing per unit once the mold exists.
3. **Design captive hardware in.** Self-clinching nuts, nut plates and captive screws remove the loose-part handling that dominates manual assembly time.
4. **Do not tap thin sheet.** A clinch nut is cheaper than the extra material or the second operation needed to get thread engagement.
5. **Reserve torque verification for joints that need it.** Blanket logged-torque requirements are a common and expensive overspec.

## FAQ

### What preload should a bolted joint be designed for?

65–75% of the fastener's proof load for a joint that will be taken apart and reused. That leaves margin against yield and against the relaxation that follows initial tightening. Remember that torque control delivers this with ±25–35% scatter, because most of the applied torque is spent overcoming thread and underhead friction rather than stretching the bolt.

### How much thread engagement do I need in aluminum?

1.5–2x the nominal fastener diameter — so a 1/4-20 screw needs roughly 0.375–0.500 in (9.5–12.7 mm) of engaged thread. Engagement is set by the weaker material: 1x diameter suffices in steel of comparable strength, 2x is needed in magnesium, and 2–2.5x in thermoplastic. If the part is too thin to provide it, use a clinch nut or a threaded insert.

### Why does torque control give such variable clamp load?

Because roughly 85–90% of applied torque is consumed by friction under the head and in the threads, and only the remainder actually stretches the bolt. Any variation in surface finish, plating, lubrication or debris changes the nut factor K in T = K x D x F, and with it the preload. That is why torque control scatters ±25–35% while angle control reaches about ±15%.

### How do I keep a self-tapping screw from stripping a plastic boss?

Design for a stripping-torque to driving-torque ratio of at least 3:1, and 5:1 for automated assembly. That comes from correct pilot hole diameter for the resin, a boss outside diameter of about 2x the screw major diameter, and a 0.010–0.015 in (0.25–0.4 mm) counterbore at the top of the boss so the first thread does not split it. For joints that will be opened repeatedly, use a heat-set insert instead.

### When should I use a self-clinching nut instead of a tapped hole?

Whenever the sheet is too thin to give the required thread engagement — which is most sheet metal below about 0.090 in (2.3 mm). A clinch nut presses permanently into the sheet and provides full-strength threads without a second operation or added material. The sheet must be softer than the fastener, typically HRB 80 or below for standard steel clinch hardware.

## Alternative processes

- [Riveting](https://manufacturingprocesses.org/processes/joining/riveting.md): Riveting sets a rivet through aligned holes and upsets its tail so the joint clamps permanently, with no heat and no access to both sides for blind types.
- [Adhesive Bonding](https://manufacturingprocesses.org/processes/joining/adhesive-bonding.md): Adhesive bonding joins parts with a cured polymer layer that spreads load over the whole bond area instead of concentrating it at fastener points.
- [Resistance Welding](https://manufacturingprocesses.org/processes/joining/resistance-welding.md): Resistance welding passes a heavy current through the joint so its own electrical resistance melts a nugget between the sheets.
- [Ultrasonic Welding](https://manufacturingprocesses.org/processes/joining/ultrasonic-welding.md): Ultrasonic welding applies high-frequency vibration through a horn to melt a thermoplastic joint locally in well under a second.

## Related processes

- [Staking](https://manufacturingprocesses.org/processes/joining/staking.md): Heat staking softens a molded plastic boss with hot air or ultrasonic energy and reforms it into a head that traps the mating part.
- [Sheet Metal Fabrication](https://manufacturingprocesses.org/processes/forming/sheet-metal-fabrication.md): Sheet metal fabrication cuts, bends and joins flat stock into a finished assembly, typically laser cutting followed by press braking, hardware and welding.
- [Insert Molding](https://manufacturingprocesses.org/processes/forming/insert-molding.md): Insert molding loads metal inserts into the cavity and molds polymer around them, capturing threads, terminals or stampings in a single shot.
- [CNC Machining](https://manufacturingprocesses.org/processes/cutting/cnc-machining.md): CNC machining removes material from solid stock with a programmed cutting tool, producing accurate parts directly from a CAD model with no tooling.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/joining/mechanical-fastening)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "PCB Assembly (SMT Reflow)"
category: "Joining"
subcategory: "Thermal"
materials: ["Metal", "Composite", "Plastic"]
tolerances: "Placement accuracy ±0.001–0.002 in (±25–50 µm) at 3σ; stencil foil 0.004–0.006 in (100–150 µm) with aperture area ratio above 0.66; board bow and twist ≤ 0.75% per IPC-6012"
volumes: "5 to 1,000,000+ boards"
lead_time: "1–3 minutes of process time per board once running; 1–3 weeks for a prototype build including stencil, programming and component procurement"
url: https://manufacturingprocesses.org/processes/joining/pcb-assembly-smt-reflow
---

# PCB Assembly (SMT Reflow)

PCB assembly prints solder paste, places components and reflows the board through a heated oven so every joint forms in one pass.

- **Category**: [Joining](https://manufacturingprocesses.org/processes/joining.md)
- **Family**: Thermal
- **Materials**: Metal, Composite, Plastic
- **Typical tolerances**: Placement accuracy ±0.001–0.002 in (±25–50 µm) at 3σ; stencil foil 0.004–0.006 in (100–150 µm) with aperture area ratio above 0.66; board bow and twist ≤ 0.75% per IPC-6012
- **Typical volumes**: 5 to 1,000,000+ boards
- **Lead time**: 1–3 minutes of process time per board once running; 1–3 weeks for a prototype build including stencil, programming and component procurement

## Overview

Surface-mount assembly prints solder paste through a laser-cut stencil, places components onto the wet paste, and carries the whole board through a reflow oven where every joint on the board forms in a single thermal pass. A modern line places 10,000–100,000 components per hour with a placement accuracy of ±0.001–0.002 in (±25–50 µm).

The process is defined by its thermal profile. Lead-free SAC305 solder melts at 217 °C (423 °F), and a standard profile ramps at 1–3 °C/s, soaks at 150–200 °C for 60–120 s, peaks at 235–250 °C, and holds 45–90 s above liquidus before cooling at up to 4 °C/s.

Solder joints are electrical connections first — SAC305 shears at roughly 5,000 psi (35 MPa), so mechanical loads belong in mounting hardware, not in the solder. Volumes span a five-board prototype to millions of units on the same equipment.

## How it works

1. **Stencil printing.** Solder paste is squeegeed through a laser-cut stainless foil, typically 0.004–0.006 in (100–150 µm) thick. Aperture design follows IPC-7525: the area ratio (aperture opening area divided by aperture wall area) must stay above 0.66 or the paste sticks in the aperture instead of releasing onto the pad. Paste type matters — Type 3 powder for general work, Type 4 or 5 for 0201 and finer.
2. **Solder paste inspection (SPI).** A 3D scanner measures deposited paste volume, height and offset on every pad. Most reflow defects are printing defects, so this is where they get caught.
3. **Placement.** Vision-centered pick-and-place heads set components from 01005 (0.016 x 0.008 in / 0.4 x 0.2 mm) up to large connectors, holding them in the tacky paste.
4. **Reflow.** The board passes through a multi-zone convection oven. Preheat drives off solvent and activates flux; the soak equalizes temperature between light and heavy thermal masses; the spike takes every joint above liquidus long enough to wet and form intermetallic; controlled cooling sets the grain structure. Peak temperature is bounded above by the component body limits in J-STD-020.
5. **Inspection and second side.** AOI checks placement, polarity and joint appearance; X-ray inspects hidden joints under BGAs and QFNs. Double-sided boards run the whole sequence again, with the first side's components held by their own solidified joints.
6. **Through-hole and finishing.** Any remaining leaded parts are wave or selective soldered to at least 75% hole fill per IPC-A-610, then the board is cleaned, conformal coated or potted as required.

## Design guidelines

### Design the panel, not just the board

Assembly equipment handles panels, not individual boards. Provide 0.2–0.5 in (5–12 mm) tooling rails on at least two edges, keep components clear of the rails, and add global fiducials — 0.040 in (1 mm) copper circles with 0.080 in (2 mm) of clear solder-mask-free area — at three corners, plus local fiducials next to fine-pitch and BGA footprints.

### Give the machines and the rework tech room

Keep at least 0.020 in (0.5 mm) between component bodies for placement accuracy and AOI discrimination, and 0.100 in (2.5 mm) of clear area around a BGA so a rework nozzle can seat. Tall components next to short ones create shadowing in both convection and inspection — group by height where the layout allows.

### Balance the thermal mass

A large ground plane, a heavy connector or a thick copper pour acts as a heat sink and pulls its joints below liquidus while neighboring joints overheat. Thermally relieve pads that connect to planes with spoke connections, and avoid putting a small chip resistor immediately beside a large thermal mass.

### Respect board flatness

IPC-6012 limits bow and twist to 0.75% for boards carrying surface-mount components. Beyond that, fine-pitch parts and BGAs cannot be co-planar with their pads. Board thickness (0.062 in / 1.6 mm is the default), copper balance across layers and panel routing all feed into this.

### Thermal and reliability limits

| Item | Value | Notes |
| --- | --- | --- |
| SAC305 liquidus | 217 °C (423 °F) | Lead-free default |
| Sn63/Pb37 liquidus | 183 °C (361 °F), eutectic | Lower process temperature, still used in high-reliability work |
| Peak reflow temperature | 235–250 °C for SAC305 | Bounded by component body limits in J-STD-020 |
| Time above liquidus | 45–90 s | Too short gives incomplete wetting, too long grows brittle intermetallic |
| Ramp rate | 1–3 °C/s | Faster cracks ceramic capacitors and lifts components |
| Cooling rate | Up to 4 °C/s | Faster refines grain but stresses joints |
| Stencil area ratio | Above 0.66 | Below this, paste stays in the aperture |
| BGA void area per ball | ≤ 25% commonly specified | IPC-7095 flags voiding above 30% |
| Plated through-hole fill | ≥ 75% | IPC-A-610 Class 2 and 3 |
| Board bow and twist | ≤ 0.75% | IPC-6012 for surface-mount assemblies |

### Inspection

Surface-mount assembly is the most heavily inspected joining process in manufacturing, and the inspection sequence is the design constraint. SPI verifies paste volume before anything is placed. AOI after reflow checks presence, polarity, alignment and fillet shape on every visible joint. Automated X-ray (2D or 3D) is the only way to see BGA, QFN and bottom-terminated joints, and is where voiding and head-in-pillow defects are found. In-circuit test or flying probe verifies electrical connectivity, and functional test closes the loop. Acceptance is defined by IPC-A-610 class: Class 1 general electronics, Class 2 dedicated service, Class 3 high-reliability.

## Cost drivers

Assembly cost splits into non-recurring setup and per-board processing. Setup means a stencil, a placement program, feeder loading and a first-article inspection — a fixed cost paid once per revision, which is why a five-board prototype run costs far more per board than a thousand-board run of the identical design.

Per-board cost is driven by placement count and by anything that breaks the single-pass flow: double-sided assembly doubles the line passes, through-hole parts add a wave or selective soldering operation and often hand labor, odd-form components need manual placement, and unique part numbers each consume a feeder slot.

Component procurement, not assembly, usually dominates total cost and lead time. Design decisions that constrain sourcing to a single supplier or an unusual package have far more cost impact than anything on the line.

1. **Put everything on one side** where the design allows — it halves the line passes and eliminates the adhesive or second-reflow constraints of double-sided assembly.
2. **Eliminate through-hole parts.** One remaining leaded connector adds a whole soldering operation to an otherwise fully surface-mount board.
3. **Reduce unique part numbers.** Consolidating resistor and capacitor values frees feeder slots and lowers setup time.
4. **Use standard package sizes.** 0402 and 0603 place faster and more reliably than 0201, and need only Type 3 paste.
5. **Panelize deliberately.** Board utilization within a panel directly sets the per-board share of bare board and processing cost.

## FAQ

### What is a standard lead-free reflow profile?

For SAC305: ramp at 1–3 °C/s to a 150–200 °C soak held 60–120 s, spike to a 235–250 °C peak, hold 45–90 s above the 217 °C liquidus, then cool at up to 4 °C/s. Peak temperature is bounded by the component body limits in J-STD-020, and the soak exists to equalize temperature between light chip parts and heavy thermal masses.

### What is the stencil area ratio rule?

Aperture opening area divided by aperture wall area must exceed 0.66, per IPC-7525. Below that, surface tension holds the paste against the aperture walls instead of releasing it onto the pad, and you get insufficient or missing deposits. It is the practical limit on how fine a pitch a given stencil thickness can print.

### How much voiding is acceptable in a BGA solder joint?

A void area of 25% or less per ball is the figure most commonly specified; IPC-7095 treats voiding above 30% as a concern requiring evaluation. Voids are found by X-ray inspection, since BGA joints are hidden under the package. Excessive voiding usually traces back to paste volume, flux activity or an overly aggressive reflow profile.

### Can solder joints carry mechanical load?

Not meaningfully. SAC305 shears at roughly 5,000 psi (35 MPa) and creeps at room temperature. Connectors that get plugged and unplugged, heavy components and anything subject to shock or vibration need mechanical retention — screws, standoffs, brackets or staking — with the solder carrying only the electrical connection.

### Why does a prototype PCB assembly cost so much per board?

Because setup cost is fixed per revision: a laser-cut stencil, a placement program, feeder loading and first-article inspection are paid once whether the run is five boards or five thousand. Component procurement adds its own minimum order quantities and lead time. Per-board processing itself is only 1–3 minutes of line time.

### What is the difference between IPC-A-610 Class 1, 2 and 3?

They are acceptance levels, not process differences. Class 1 covers general consumer electronics where function is all that matters; Class 2 covers dedicated service equipment where extended life is expected; Class 3 covers high-reliability applications where failure is not tolerable, and tightens criteria for fillet shape, voiding, cleanliness and hole fill. Specify the class on the assembly drawing.

## Alternative processes

- [Adhesive Bonding](https://manufacturingprocesses.org/processes/joining/adhesive-bonding.md): Adhesive bonding joins parts with a cured polymer layer that spreads load over the whole bond area instead of concentrating it at fastener points.
- [Mechanical Fastening](https://manufacturingprocesses.org/processes/joining/mechanical-fastening.md): Mechanical fastening joins parts with screws, inserts, clinch hardware or snap fits so the joint can be assembled and taken apart again.
- [Ultrasonic Welding](https://manufacturingprocesses.org/processes/joining/ultrasonic-welding.md): Ultrasonic welding applies high-frequency vibration through a horn to melt a thermoplastic joint locally in well under a second.

## Related processes

- [Photo Etching](https://manufacturingprocesses.org/processes/finishing/photo-etching.md): Photo etching masks a surface photographically and etches shallow detail into it for decoration, marking or texture.
- [Screen Printing](https://manufacturingprocesses.org/processes/finishing/screen-printing.md): Screen printing pushes ink through a stencilled mesh onto the surface below, laying down one opaque color per pass.
- [Soldering and Brazing](https://manufacturingprocesses.org/processes/joining/soldering-and-brazing.md): Soldering and brazing join metals with a molten filler that wets the joint by capillary action while the parent metal stays solid.
- [Laser Marking and Engraving](https://manufacturingprocesses.org/processes/finishing/laser-marking-and-engraving.md): Laser marking alters the surface with a focused beam — annealing, etching or engraving it — to make permanent, contact-free marks.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/joining/pcb-assembly-smt-reflow)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Power Beam Welding"
category: "Joining"
subcategory: "Thermal"
materials: ["Metal"]
tolerances: "Beam-to-seam alignment ±0.004 in (±0.1 mm); joint gap ≤ 0.004 in (0.1 mm) for autogenous welds; post-weld distortion roughly an order of magnitude below an equivalent arc weld"
volumes: "100 to 1,000,000+ per year"
lead_time: "Seconds of weld time per joint; 2–8 weeks for precision fixtures and schedule development; EBW adds minutes of chamber pump-down per load"
url: https://manufacturingprocesses.org/processes/joining/power-beam-welding
---

# Power Beam Welding

Power beam welding fuses metal with a laser or electron beam focused to a very small spot, producing deep, narrow welds with little distortion.

- **Category**: [Joining](https://manufacturingprocesses.org/processes/joining.md)
- **Family**: Thermal
- **Materials**: Metal
- **Typical tolerances**: Beam-to-seam alignment ±0.004 in (±0.1 mm); joint gap ≤ 0.004 in (0.1 mm) for autogenous welds; post-weld distortion roughly an order of magnitude below an equivalent arc weld
- **Typical volumes**: 100 to 1,000,000+ per year
- **Lead time**: Seconds of weld time per joint; 2–8 weeks for precision fixtures and schedule development; EBW adds minutes of chamber pump-down per load

## Overview

Power beam welding fuses metal with a laser or electron beam focused to a spot a few thousandths of an inch across. Power density reaches roughly 10⁶ W/cm² — about a hundred times an arc — vaporizing a narrow capillary (a keyhole) through the joint rather than melting a shallow puddle on top of it.

The result is the defining number: depth-to-width ratios of 10:1 for laser and up to 50:1 for electron beam, against roughly 1:1 for arc welding. Because so little metal melts, the heat-affected zone is typically 0.010–0.040 in (0.25–1 mm) wide and distortion is an order of magnitude below an equivalent arc weld — which is why the process is used on finish-machined assemblies.

Two variants: laser beam welding (LBW) with 1–20 kW fiber or disk lasers in air, and electron beam welding (EBW) at 30–200 kV inside a vacuum chamber, capable of single-pass penetration up to about 6 in (150 mm) in steel. Both are autogenous by default, which puts the burden squarely on joint fit-up.

## How it works

1. **Fit-up and fixturing.** Autogenous beam welding adds no metal, so any gap is a missing volume. Butt joints need the faying surfaces closed to under 0.004 in (0.1 mm), or roughly 10% of thickness, whichever is smaller. Parts are machined, not sheared, and held in rigid fixtures.
2. **Beam delivery.** LBW: a 1,064 nm fiber or disk laser routed to a focusing head and focused to a 0.008–0.024 in (0.2–0.6 mm) spot. EBW: electrons accelerated at 30–200 kV in a vacuum near 10⁻⁴ mbar, magnetically focused and steerable at kilohertz rates for beam oscillation.
3. **Keyhole formation.** Above roughly 10⁶ W/cm² the surface vaporizes and recoil pressure opens a vapor capillary that the beam penetrates. Molten metal flows around the keyhole and closes behind it as the beam travels.
4. **Travel.** Laser welds run at 20–400 in/min (0.5–10 m/min). A useful rule for steel is about 1 mm of penetration per kW at 1 m/min, so a 6 kW fiber laser gives roughly 0.25 in (6 mm) at that speed.
5. **Solidification.** Cooling rates are extreme, which refines weld metal but also hardens the HAZ of medium-carbon and low-alloy steels; those may need preheat or post-weld tempering.
6. **EBW chamber cycle.** Pump-down between loads is the throughput limit for electron beam work — minutes per load on production chambers — and the process generates X-rays, so the chamber is also a radiation shield.

### Laser versus electron beam

**Laser beam welding** works in air with an inert shield gas, needs no chamber, and can be delivered on a robot arm or a remote scanner head. It struggles with highly reflective metals at 1 µm wavelength — copper especially — and with aluminum alloys that boil off magnesium and go porous.

**Electron beam welding** has no reflectivity problem, and its vacuum is a perfect shield: it is the standard route for titanium, zirconium, niobium and other reactive and refractory metals, and for thick single-pass welds in steel that would take dozens of arc passes. The costs are the chamber, the pump-down cycle, X-ray shielding, and a part size capped by chamber size.

Both processes also run in conduction mode at lower power density, producing a smooth, wide, shallow weld with a cosmetic surface — used for hermetic seals on thin-wall enclosures.

## Design guidelines

### Fit-up is the whole design problem

An autogenous beam weld cannot bridge a gap. Specify butt joint gaps under 0.004 in (0.1 mm) — machined or laser-cut edges, not sheared. If the assembly cannot hold that, add filler wire (which tolerates roughly 0.010 in / 0.25 mm) or use hybrid laser-arc welding, where the arc fills the gap and the laser provides the penetration.

### Beam-to-seam alignment

The weld is only 0.02–0.06 in (0.5–1.5 mm) wide, so the beam must find the joint to within about ±0.004 in (±0.1 mm). Either hold that with fixturing and part tolerance or add seam tracking. In practice this drives designs toward self-locating joints — a machined step, spigot or shoulder that both centers the parts and presents a consistent seam.

### Joint geometry and root support

Favor square butt joints, with the beam entering perpendicular to the seam. For circumferential welds on shafts and housings, a machined lip that puts the seam on a cylindrical or flat face lets the part rotate under a stationary head. A keyhole weld that only just penetrates has an unstable root and produces spiking and porosity, so either overshoot into a backing or run-out feature, or deliberately design a partial-penetration weld with margin — and keep the unwelded root notch, which is a stress raiser, out of the load path.

### Materials

| Material | Laser | Electron beam | Notes |
| --- | --- | --- | --- |
| Carbon and low-alloy steel | Good | Excellent | Fast cooling hardens the HAZ above about 0.25% C — preheat or temper |
| Austenitic stainless | Excellent | Excellent | The workhorse combination for hermetic housings |
| Titanium and zirconium | Good with full shielding | Excellent | Vacuum is the ideal shield; oxygen pickup embrittles |
| Nickel superalloys | Good | Excellent | Low heat input limits strain-age cracking |
| Aluminum 5xxx / 6xxx | Fair | Good | Magnesium loss causes porosity; filler wire and beam oscillation help |
| Aluminum 2xxx / 7xxx | Poor | Fair | Hot cracking |
| Copper | Poor at 1 µm | Excellent | High reflectivity and conductivity; green or blue lasers change this |
| Refractory metals (Mo, Nb, Ta, W) | Poor | Good | Vacuum welding only |
| Carbon steel to stainless | Yes | Yes | The tiny fused volume limits dilution — an advantage over arc welding |
| Aluminum to steel | No | No | Brittle intermetallics form regardless of beam source |

Melting and boiling points that set the process window are on the [metal melting points chart](/charts/metal-melting-points); gauge conversions are on the [sheet metal gauge chart](/charts/sheet-metal-gauge-chart).

### Inspection

Visual inspection covers undercut, underfill and surface porosity. Radiography detects the porosity, spiking and root voids that keyhole welds are prone to; ultrasonic testing detects lack of penetration, though the narrow weld geometry makes probe positioning critical. Helium leak testing to 10⁻⁹ std cc/s is standard on hermetic enclosures. Because the welds are so small, destructive cross-sections on setup coupons are how penetration depth is actually qualified before a production run.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Butt joint gap, autogenous | 0.001 in (0.025 mm) | 0.004 in (0.1 mm) or 10% of t | No filler is available to bridge a gap |
| Beam-to-seam alignment | ±0.002 in (±0.05 mm) | ±0.004 in (±0.1 mm) | The weld is only 0.02–0.06 in (0.5–1.5 mm) wide |
| Edge preparation | Machined or laser cut | Not sheared | Sheared edges are neither flat nor square |
| Penetration margin | 20% beyond the joint | Full or clearly partial | Marginal penetration gives root spiking and porosity |
| Joint located by geometry | Machined spigot or step | — | Removes the alignment burden from the fixture |

## Variants

- Laser Beam Welding (LBW)
- Electron Beam Welding (EBW)

## Cost drivers

Beam welding is capital-intensive and consumable-light. There is no filler, no flux, and no shielding gas on EBW; cost sits in the machine, the fixture, and — for electron beam — the chamber cycle. That inverts the usual welding economics: the weld itself is fast and cheap, and everything upstream of it is where the money goes.

Fit-up is the dominant hidden cost. Because the joint has to close to within 0.004 in (0.1 mm), parts often need a machining operation they would not otherwise require, and the fixture has to be a precision tool rather than a weld jig. Throughput on EBW is governed by pump-down, so batch size in the chamber matters more than weld time; laser welding avoids that entirely but pays for its flexibility with seam tracking and a Class 1 safety enclosure.

1. **Machine the joint faces in the same setup** that establishes the mating features, so fit-up comes for free rather than as an added operation.
2. **Design a self-locating joint** — spigot, step or shoulder — and you can often delete seam tracking altogether.
3. **Load EBW chambers with a full fixture plate**; pump-down cost is per cycle, not per weld.
4. **Weld last, after finish machining,** and exploit the low distortion — that is the actual value proposition versus arc welding.
5. **Check whether hybrid laser-arc solves the fit-up problem** more cheaply than tightening part tolerance: the arc bridges gaps the laser cannot.

## FAQ

### How deep can a laser or electron beam weld penetrate in one pass?

A useful rule for laser welding steel is about 1 mm of penetration per kW at 1 m/min travel, so a 6 kW fiber laser gives roughly 0.25 in (6 mm) in a single pass. Electron beam welding in vacuum goes far deeper — up to about 6 in (150 mm) in steel in one pass, with depth-to-width ratios reaching 50:1.

### How tight does fit-up need to be for laser welding?

For an autogenous butt weld with no filler, the gap must stay under about 0.004 in (0.1 mm), or roughly 10% of material thickness, whichever is smaller. Adding filler wire raises that to roughly 0.010 in (0.25 mm). Hybrid laser-arc welding tolerates more still, because the arc supplies metal while the laser supplies penetration.

### Laser or electron beam — which should I choose?

Laser if the parts are large or awkward, volume is high, and the material is steel, stainless or nickel alloy — no chamber means no pump-down and easy robot delivery. Electron beam for reactive and refractory metals, for copper, and for thick single-pass welds, since the vacuum is a perfect shield and there is no reflectivity limit.

### Can you laser weld copper?

Not well with a conventional 1 µm fiber laser — copper reflects most of the beam until it melts, and the process is unstable. Green (515 nm) and blue diode lasers absorb far better and have made copper welding practical for battery and busbar work. Electron beam welding has no reflectivity problem at all.

### How are power beam welds inspected?

Radiography for the porosity, root spiking and voids that keyhole welds produce; ultrasonic testing for lack of penetration; visual for undercut and underfill; and helium leak testing down to 10⁻⁹ std cc/s on hermetic enclosures. Destructive cross-sections on setup coupons are how penetration depth is actually qualified before a run.

## Alternative processes

- [Arc Welding](https://manufacturingprocesses.org/processes/joining/arc-welding.md): Arc welding melts and fuses metal with an electric arc between an electrode and the work, shielded from the atmosphere by gas or flux.
- [Soldering and Brazing](https://manufacturingprocesses.org/processes/joining/soldering-and-brazing.md): Soldering and brazing join metals with a molten filler that wets the joint by capillary action while the parent metal stays solid.
- [Adhesive Bonding](https://manufacturingprocesses.org/processes/joining/adhesive-bonding.md): Adhesive bonding joins parts with a cured polymer layer that spreads load over the whole bond area instead of concentrating it at fastener points.
- [Riveting](https://manufacturingprocesses.org/processes/joining/riveting.md): Riveting sets a rivet through aligned holes and upsets its tail so the joint clamps permanently, with no heat and no access to both sides for blind types.

## Related processes

- [Laser Cutting](https://manufacturingprocesses.org/processes/cutting/laser-cutting.md): Laser cutting melts, burns or vaporises a narrow kerf through flat stock with a focused beam, producing profiles from a 2D file with no tooling.
- [Friction Welding](https://manufacturingprocesses.org/processes/joining/friction-welding.md): Friction welding rubs two parts together under load until the interface plasticises and forges together, with no melting and no filler metal.
- [Resistance Welding](https://manufacturingprocesses.org/processes/joining/resistance-welding.md): Resistance welding passes a heavy current through the joint so its own electrical resistance melts a nugget between the sheets.
- [Heat Treatment](https://manufacturingprocesses.org/processes/forming/heat-treatment.md): Heat treatment cycles metal through controlled heating and cooling to change its hardness, strength, ductility or internal stress state.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/joining/power-beam-welding)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Resistance Welding"
category: "Joining"
subcategory: "Thermal"
materials: ["Metal"]
tolerances: "±1/32 in (±0.8 mm) on fixtured sheet assemblies; nugget diameter held to 4√t–5√t (t in mm); electrode indentation ≤ 20–25% of sheet thickness"
volumes: "5,000 to 10,000,000+ per year — the capital only pays back at production volume"
lead_time: "0.1–0.5 s of weld time and 1–2 s per spot on a robotic gun; 2–8 weeks for weld fixtures, guns and electrode tooling"
url: https://manufacturingprocesses.org/processes/joining/resistance-welding
---

# Resistance Welding

Resistance welding passes a heavy current through the joint so its own electrical resistance melts a nugget between the sheets.

- **Category**: [Joining](https://manufacturingprocesses.org/processes/joining.md)
- **Family**: Thermal
- **Materials**: Metal
- **Typical tolerances**: ±1/32 in (±0.8 mm) on fixtured sheet assemblies; nugget diameter held to 4√t–5√t (t in mm); electrode indentation ≤ 20–25% of sheet thickness
- **Typical volumes**: 5,000 to 10,000,000+ per year — the capital only pays back at production volume
- **Lead time**: 0.1–0.5 s of weld time and 1–2 s per spot on a robotic gun; 2–8 weeks for weld fixtures, guns and electrode tooling

## Overview

Resistance welding passes a high current — 5,000 to 20,000 A at only 1–10 V — through overlapping sheet metal held between two copper alloy electrodes. The joint's own electrical resistance is the heat source (Q = I²Rt), so a molten nugget forms at the sheet interface and nowhere else, and the whole cycle is over in 0.1–0.5 s.

It is the dominant joining process for sheet steel assemblies: a typical car body contains 3,000–5,000 spot welds. Three variants cover most work — spot welding for discrete nuggets, seam welding for pressure-tight continuous joints, and projection welding for embossed features and weld nuts made several at a time in one press stroke.

Sheet thickness runs from about 0.020 to 0.125 in (0.5–3 mm) per sheet in normal practice. There is no filler, no shielding gas and no consumable except electrode tips, so per-joint cost is essentially electricity and cycle time — which is why the process only makes sense at volume.

## How it works

1. **Squeeze.** The electrodes close on the stack-up and build force before any current flows — typically 200–1,500 lbf (0.9–6.7 kN) for sheet steel. Force sets contact resistance and holds the sheets together; too little causes expulsion, too much shunts heat into the electrodes.
2. **Weld.** Current flows for 5–30 cycles of 60 Hz line frequency (0.08–0.5 s), or an equivalent millisecond schedule on a medium-frequency DC inverter. Heating follows Q = I²Rt, and the highest resistance in the stack is the faying surface between the sheets, so that is where melting starts.
3. **Nugget growth.** A lens of molten metal grows outward from the interface, contained by surrounding solid metal and electrode force — the nugget is forged as well as melted. Target diameter follows d = 4√t to 5√t with t in millimeters, so 1 mm sheet wants a 4–5 mm (0.16–0.20 in) nugget.
4. **Hold.** Current stops but the electrodes stay closed for a few cycles while the nugget solidifies under pressure. Releasing early gives shrinkage porosity and center cracks.
5. **Off and index.** Water-cooled electrodes recover and the part indexes. Robotic guns average 1–2 s per spot including travel.

### How do the three variants differ?

**Spot welding** uses two opposed copper alloy electrodes (RWMA Class 2 CuCrZr for steel) to make one discrete nugget at a time — the default for sheet assemblies and the process behind every C-frame and X-gun on a body line.

**Seam welding** replaces the tips with two rotating copper wheels that weld continuously as the part passes between them. Overlapping nuggets by roughly a quarter of their diameter produces a pressure-tight seam — fuel tanks, radiators, drums, can bodies.

**Projection welding** embosses a dimple, ring or rib into one part so current is forced through that small area. Because geometry rather than the electrode locates the heat, flat platens replace pointed tips and several welds are made in one press stroke. Standard weld nuts and studs carry three projections and are attached this way in a single hit.

Electrode life is the practical process limit: several thousand welds on bare steel between tip dressings, often under a thousand on hot-dip galvanized, because the zinc alloys with the copper face.

## Design guidelines

### Nugget size and sheet thickness

Size the nugget with d = 4√t to 5√t (t in mm), and let the thinner sheet govern. In a stack-up, the thickness ratio between outer sheets should stay under about 3:1 — beyond that the thin sheet burns through before the thick one melts. Three-sheet stacks are routine; four is specialist work.

### Flange width, edge distance and spacing

Edge distance from nugget center to sheet edge should be at least 2x the nugget diameter, or molten metal expels sideways. Spot spacing needs to be at least 10x sheet thickness for steel; closer than that and a share of the current shunts through the previous weld, starving the new joint. In practice that means a flange 0.5–0.75 in (13–19 mm) wide. Gauge conversions are on the [sheet metal gauge chart](/charts/sheet-metal-gauge-chart).

### Access and appearance

Both sides of the joint must be reachable, and the flange has to be presentable to a gun throat — design flanges that run in one plane and avoid deep re-entrant pockets. The electrodes also leave a visible depression; standard automotive acceptance practice limits indentation to 20–25% of sheet thickness, and deeper than that means the weld has thinned the sheet. On a Class A visible surface, put the spot weld on a hidden flange or switch to adhesive bonding or laser welding.

### Which material pairs actually weld?

| Pair | Resistance weldable? | Notes |
| --- | --- | --- |
| Low-carbon steel to low-carbon steel | Excellent | The reference case — high enough resistivity, moderate conductivity |
| HSLA and advanced high-strength steel | Yes | Narrower current window; often needs post-weld tempering pulses to avoid a brittle martensitic nugget |
| Galvanized / galvannealed steel | Yes | Needs roughly 20–50% more current and destroys electrode tips far faster; specify a tip dress interval |
| 300-series stainless to itself | Yes | Higher resistivity means lower current and shorter time than carbon steel |
| Aluminum to aluminum | Difficult | Roughly 3x the current of steel at a fraction of the time; oxide film must be broken; electrode life is poor |
| Copper or brass to itself | Poor | Conductivity is too high — heat goes into the electrodes, not the joint |
| Aluminum to steel | No | Brittle intermetallics; use self-piercing rivets, adhesive bonding or friction stir |
| Painted or oiled surfaces | No | Insulating layers block current entirely — weld before paint |

Conductivity and resistivity figures behind those rows are on the [material properties chart](/charts/material-properties).

### How are resistance welds inspected?

Nugget quality cannot be seen from outside, so inspection combines destructive sampling with in-process monitoring. Chisel and peel tests on production samples must produce a **button pull-out** — base metal tearing in a ring around the nugget — rather than interfacial fracture through the weld. Macro-sections confirm nugget diameter and penetration into each sheet. Phased-array ultrasonic spot inspection is the non-destructive option, and adaptive controllers watching dynamic resistance and electrode displacement are the primary line-side control.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Nugget diameter | 5√t (t in mm) | 4√t minimum | Below this the joint fails interfacially instead of pulling a button |
| Edge distance | 2.5x nugget diameter | 2x nugget diameter | Molten metal expels through the free edge |
| Spot spacing | 15x sheet thickness | 10x sheet thickness | Current shunts through the adjacent weld |
| Flange width | 0.75 in (19 mm) | 0.5 in (13 mm) | Edge distance plus electrode access |
| Thickness ratio in a stack | 2:1 | 3:1 | Heat balance shifts toward the thick sheet |
| Electrode indentation | 10% of sheet thickness | 20–25% | Deeper indentation means the sheet has been thinned |

## Variants

- Projection Welding
- Spot Welding
- Seam Welding

## Cost drivers

Per-joint consumable cost is nearly zero — no filler, no gas, only electricity and copper tips — so resistance welding's economics are dominated by equipment and cycle time. A pedestal spot welder is inexpensive; a robotic weld cell with servo guns, transformers, water cooling and adaptive controls is a capital project. That capital only amortizes across large quantities, which is why the process owns automotive body-in-white and almost nothing at prototype volumes.

The second driver is spot count. Each spot is 1–2 s of robot time on a production line, which is why body engineers fight over weld counts. Electrode maintenance is the hidden operating cost: tip dressing intervals collapse on galvanized steel, and unplanned dressing stops the line.

1. **Justify every spot.** Load-path analysis usually shows spacing can be opened up away from high-stress regions; going from 10x to 20x sheet thickness halves the spot count.
2. **Consider weld bonding.** A bead of structural adhesive plus a reduced spot count gives higher stiffness and fatigue life than spot welds alone.
3. **Design one-sided access out of the part.** Deep-throat and offset guns cost force accuracy and cycle time; a flange in an accessible plane is free.
4. **Use projection welding for hardware.** Attaching four weld nuts in one press stroke beats four separate spot welding operations.
5. **Budget tip dressing on coated steel.** If the part must be galvanized, plan the dress interval into cycle time rather than discovering it in production.

## FAQ

### What size should a spot weld nugget be?

The standard rule is d = 4√t to 5√t with t the thinner sheet thickness in millimeters — so 1 mm sheet needs a 4–5 mm (0.16–0.20 in) nugget and 2 mm sheet needs 5.7–7 mm (0.22–0.28 in). Below 4√t the joint tends to fail interfacially through the weld instead of pulling a button out of the base metal.

### How far apart should spot welds be spaced?

At least 10x the sheet thickness for steel, and 15x is a better default. Welds placed closer shunt part of the welding current through the previously made nugget, which starves the new joint and produces an undersized weld even though the machine settings never changed.

### Can you spot weld aluminum?

Yes, but it is much harder than steel. Aluminum's high electrical and thermal conductivity means roughly three times the current for a fraction of the time, the tenacious oxide film has to be broken down, and electrode life is poor because aluminum alloys with copper. Most volume aluminum body structures use self-piercing rivets or friction stir welding instead.

### How do you inspect a spot weld?

Destructively, by chisel or peel testing production samples — an acceptable weld pulls a button of base metal out of one sheet rather than fracturing through the nugget. Non-destructively, by phased-array ultrasonic inspection. On production lines the real control is adaptive monitoring of dynamic resistance and electrode displacement during each weld.

### Why does galvanized steel wear out spot welding electrodes so fast?

Zinc melts and alloys with the copper electrode face, so tip geometry and contact area change weld by weld. Bare steel runs several thousand welds between dressings; hot-dip galvanized often needs dressing under a thousand, and also requires roughly 20–50% more current for the same nugget.

### What is projection welding used for?

Attaching hardware and making multiple welds in one stroke. An embossed dimple or ring on one part concentrates current where you want the weld, so flat platens replace pointed electrodes. Standard weld nuts and weld studs carry three projections and are attached to sheet steel in a single press hit.

## Alternative processes

- [Arc Welding](https://manufacturingprocesses.org/processes/joining/arc-welding.md): Arc welding melts and fuses metal with an electric arc between an electrode and the work, shielded from the atmosphere by gas or flux.
- [Riveting](https://manufacturingprocesses.org/processes/joining/riveting.md): Riveting sets a rivet through aligned holes and upsets its tail so the joint clamps permanently, with no heat and no access to both sides for blind types.
- [Adhesive Bonding](https://manufacturingprocesses.org/processes/joining/adhesive-bonding.md): Adhesive bonding joins parts with a cured polymer layer that spreads load over the whole bond area instead of concentrating it at fastener points.
- [Mechanical Fastening](https://manufacturingprocesses.org/processes/joining/mechanical-fastening.md): Mechanical fastening joins parts with screws, inserts, clinch hardware or snap fits so the joint can be assembled and taken apart again.

## Related processes

- [Sheet Metal Fabrication](https://manufacturingprocesses.org/processes/forming/sheet-metal-fabrication.md): Sheet metal fabrication cuts, bends and joins flat stock into a finished assembly, typically laser cutting followed by press braking, hardware and welding.
- [Metal Stamping](https://manufacturingprocesses.org/processes/forming/metal-stamping.md): Metal stamping shears and forms sheet metal between matched dies in a press, producing flat and shallow-formed parts at very high rates.
- [Power Beam Welding](https://manufacturingprocesses.org/processes/joining/power-beam-welding.md): Power beam welding fuses metal with a laser or electron beam focused to a very small spot, producing deep, narrow welds with little distortion.
- [Punching and Blanking](https://manufacturingprocesses.org/processes/cutting/punching-and-blanking.md): Punching and blanking shear sheet metal between a punch and a die, producing holes and flat outlines at very high speed.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/joining/resistance-welding)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Riveting"
category: "Joining"
subcategory: "Mechanical"
materials: ["Metal", "Plastic", "Composite"]
tolerances: "Hole 0.003–0.006 in (0.08–0.15 mm) over nominal shank; driven head 1.5x shank diameter wide by 0.5x shank diameter high; SPR interlock commonly ≥ 0.016 in (0.4 mm) with ≥ 0.008 in (0.2 mm) remaining bottom sheet"
volumes: "1 to 10,000,000+"
lead_time: "Immediate with off-the-shelf hardware; 1–2 s per joint for automated self-piercing riveting; 4–10 weeks for SPR dies and C-frame tooling"
url: https://manufacturingprocesses.org/processes/joining/riveting
---

# Riveting

Riveting sets a rivet through aligned holes and upsets its tail so the joint clamps permanently, with no heat and no access to both sides for blind types.

- **Category**: [Joining](https://manufacturingprocesses.org/processes/joining.md)
- **Family**: Mechanical
- **Materials**: Metal, Plastic, Composite
- **Typical tolerances**: Hole 0.003–0.006 in (0.08–0.15 mm) over nominal shank; driven head 1.5x shank diameter wide by 0.5x shank diameter high; SPR interlock commonly ≥ 0.016 in (0.4 mm) with ≥ 0.008 in (0.2 mm) remaining bottom sheet
- **Typical volumes**: 1 to 10,000,000+
- **Lead time**: Immediate with off-the-shelf hardware; 1–2 s per joint for automated self-piercing riveting; 4–10 weeks for SPR dies and C-frame tooling

## Overview

Riveting sets a rivet through aligned holes and upsets its tail into a second head, clamping the joint permanently. There is no heat, so nothing is annealed, nothing distorts, and heat-treated alloys keep their temper — which is precisely why 2024 and 7075 aluminum aircraft structure is riveted rather than welded.

The joint is a shear joint. A driven solid rivet's allowable comes from the shear area of its shank at the rivet material's shear strength — 30 ksi (207 MPa) for the 2117-T4 alloy used in standard aircraft rivets — and the joint can equally fail by sheet bearing or edge tear-out, which is what edge distance and pitch rules exist to prevent.

The family covers solid rivets, blind rivets that need access to one side only, structural blind rivets, self-piercing rivets that make their own hole, and rivet nuts that leave a thread behind. Volumes run from a single repair to millions of joints a year on an automotive line.

## How it works

1. **Drill and deburr.** The hole is drilled 0.003–0.006 in (0.08–0.15 mm) over the rivet's nominal shank — a 1/8 in rivet takes a #30 drill at 0.1285 in. Both sides are deburred, and in aerospace practice the hole may be reamed and cold-worked. Drill sizes are on the [drill size chart](/charts/drill-size-chart).
2. **Insert.** Rivet length equals grip length plus about 1.5x diameter for a solid rivet — enough material to form the driven head without so much that it buckles.
3. **Upset.** A bucking bar backs the tail while a rivet gun or squeeze tool deforms it. The shank swells to fill the hole (an important part of the joint's shear performance) and the tail flows into a driven head.
4. **Verify.** The driven head is gauged: 1.5x shank diameter across and 0.5x shank diameter high is the standard target.

### The other rivet types

**Blind rivets** are set from one side. A mandrel is pulled through the rivet body, flaring the far end, and then snaps at a designed break point. Standard open-end pop rivets are for light-duty and non-structural work; structural blind rivets retain and lock the mandrel stem so it contributes shear area and the joint does not lose the pin in service.

**Self-piercing rivets (SPR)** need no hole at all. A punch drives a semi-tubular rivet through the top sheet and flares it into — but not through — the bottom sheet, which is held in a profiled die. Setting force is roughly 6,700–18,000 lbf (30–80 kN), the cycle is 1–2 s, and total stack thickness runs about 0.04–0.24 in (1–6 mm). SPR is how aluminum-intensive car bodies are assembled, because those alloys resist spot welding and cannot be arc welded without losing temper.

**Rivet nuts (blind threaded inserts)** are set like a blind rivet but leave a threaded barrel behind, providing a serviceable thread in thin sheet or tube with one-sided access.

## Design guidelines

### Edge distance and pitch are the strength rules

Edge distance from the rivet centerline to the sheet edge must be at least 2x the rivet diameter, and 2.5x is the working target — closer and the joint tears out through the edge instead of loading the rivet. Rivet pitch along a row is a minimum of 3x diameter and typically 4–6x; row spacing at least 2.5x diameter. These are the numbers that keep failure in the rivet, where it is predictable, rather than in the sheet.

### Hole and grip

Drill the hole 0.003–0.006 in (0.08–0.15 mm) over the nominal shank so the rivet enters without force but swells to fill on upsetting. Rivet length for a solid rivet is grip plus 1.5x diameter. Blind rivets are sold in grip ranges — a rivet whose grip range does not match the actual stack will either not clamp or will not form a proper blind head, and this is the single most common riveting error.

### Loading and access

Rivets carry shear well and tension poorly, and solid rivets in particular are not tension fasteners. Where a joint sees pull-through, use a structural blind rivet with a retained locking stem, a bolted joint, or a washer or backing plate to spread the bearing load; in composite structures, bearing failure usually governs before rivet shear does. On access: solid riveting needs both sides (gun on one face, bucking bar behind), blind riveting needs one, and self-piercing riveting needs a C-frame with the die behind the joint — the same throat-depth constraint that governs resistance welding.

### Material combinations

| Pair | Notes |
| --- | --- |
| Aluminum sheet to aluminum sheet | The classic case; 2117-T4 rivets in 2024/7075 structure retain full temper |
| Steel to steel | Solid or blind steel rivets; galvanized sheet rivets without any of the welding problems |
| Aluminum to steel | SPR and blind rivets both work; use a compatible rivet material and seal against galvanic corrosion |
| Composite to metal | Bearing failure governs — spread load with a washer or bushing, and never overtorque |
| Plastic to metal | Blind rivets with large flanges or backup washers; avoid clamping brittle resin directly |
| Dissimilar metals generally | Choose the rivet to be cathodic to the sheet, and wet-install with sealant in corrosive service |

Sheet thickness and gauge conversions are on the [sheet metal gauge chart](/charts/sheet-metal-gauge-chart).

### Inspection

Driven heads are gauged for diameter and height with go/no-go gauges — 1.5x diameter across and 0.5x diameter high on solid rivets. Blind rivets are checked for a properly formed blind-side head and, on structural types, for a retained and locked mandrel. Self-piercing rivets are qualified by cross-section: the accept criteria are a measured mechanical interlock (commonly 0.016 in / 0.4 mm or greater) and a minimum remaining bottom-sheet thickness (commonly 0.008 in / 0.2 mm), with setting force and displacement monitored on every joint in production. Eddy current and ultrasonic inspection are used to find cracked holes in aerospace structure.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Edge distance | 2.5x rivet diameter | 2x minimum | Prevents edge tear-out |
| Pitch along a row | 4–6x diameter | 3x minimum | Prevents sheet failure between rivets |
| Row spacing | 3x diameter | 2.5x minimum | Same reason, transverse |
| Hole diameter | Nominal + 0.004 in (0.1 mm) | +0.003 to +0.006 in (0.08–0.15 mm) | Rivet must enter freely and swell to fill |
| Solid rivet length | Grip + 1.5x diameter | — | Material to form the driven head |
| Driven head | 1.5x diameter wide, 0.5x diameter high | Gauge every joint | The only visible evidence the rivet was set correctly |
| Blind rivet grip range | Match the actual stack | Never exceed | Wrong grip gives no clamp or no blind head |
| Loading | Shear | Not primary tension | Solid rivets are not tension fasteners |

## Cost drivers

Riveting cost is hardware plus holes plus labor. The rivets themselves are inexpensive; the drilling, deburring and — in aerospace — hole quality control that precede them are not, and hand riveting a large panel is a substantial labor operation. This is why automated drill-and-rivet cells exist for aircraft panels and why any design decision that reduces hole count pays twice.

Self-piercing riveting inverts the structure: no drilling at all, a 1–2 s automated cycle, but a C-frame robot cell and part-specific dies as capital. It only makes sense at production volume, and it is chosen because the materials cannot be welded rather than because it is cheap.

The comparison that matters is usually against welding. Riveting adds hardware and holes but adds no heat, so heat-treated and dissimilar-material assemblies stay strong and stay flat — with no distortion correction and no post-weld heat treatment downstream.

1. **Reduce hole count by increasing rivet size** where edge distance and pitch rules allow; fewer, larger rivets beat many small ones on labor.
2. **Use blind rivets to delete backside access,** which often removes an entire fixture or an assembly sequence constraint.
3. **Standardize on one or two rivet diameters and grip ranges** to cut tool changes and part numbers.
4. **Consider riveting plus adhesive (rivet bonding).** The adhesive carries fatigue load and seals the joint while the rivets fixture it during cure — a common aerospace and automotive combination.
5. **Design flanges reachable by a C-frame** if self-piercing riveting is in the plan; throat depth constrains the geometry exactly as it does for spot welding.

## FAQ

### What edge distance do rivets need?

At least 2x the rivet diameter from the centerline to the sheet edge, with 2.5x as the working target. Closer than 2x and the joint fails by tearing out through the edge of the sheet rather than by shearing the rivet, which is both weaker and far less predictable. Pitch along a row should be at least 3x diameter and is typically 4–6x.

### Why rivet instead of weld?

Because riveting adds no heat. Heat-treated aluminum alloys such as 2024 and 7075 lose their temper and hot-crack when fusion welded, dissimilar metals form brittle intermetallics, and any welded assembly distorts. A riveted joint keeps the parent material at full strength, stays flat, and needs no post-weld heat treatment or straightening.

### What is a self-piercing rivet and when is it used?

A semi-tubular rivet driven straight through the top sheet and flared into — not through — the bottom sheet against a profiled die, with no pre-drilled hole. Setting force is roughly 6,700–18,000 lbf (30–80 kN) over a 1–2 s cycle for stacks of about 0.04–0.24 in (1–6 mm). It is the standard joining method for aluminum-intensive car bodies, which resist spot welding.

### How is a driven rivet inspected?

By gauging the driven head: 1.5x the shank diameter across and 0.5x the shank diameter high is the standard target, checked with go/no-go gauges. Blind rivets are checked for a properly formed blind-side head and a retained locked mandrel on structural types. Self-piercing rivets are qualified by sectioning, measuring the mechanical interlock and the remaining bottom-sheet thickness.

### Can rivets carry tension loads?

Poorly, and solid rivets are not designed for it at all — they are shear fasteners. Where a joint sees pull-through, use a structural blind rivet with a retained locking stem, switch to a bolted joint, or spread the bearing load with a washer or backing plate. In composite joints, bearing failure of the laminate usually governs before the rivet does.

## Alternative processes

- [Mechanical Fastening](https://manufacturingprocesses.org/processes/joining/mechanical-fastening.md): Mechanical fastening joins parts with screws, inserts, clinch hardware or snap fits so the joint can be assembled and taken apart again.
- [Adhesive Bonding](https://manufacturingprocesses.org/processes/joining/adhesive-bonding.md): Adhesive bonding joins parts with a cured polymer layer that spreads load over the whole bond area instead of concentrating it at fastener points.
- [Resistance Welding](https://manufacturingprocesses.org/processes/joining/resistance-welding.md): Resistance welding passes a heavy current through the joint so its own electrical resistance melts a nugget between the sheets.
- [Arc Welding](https://manufacturingprocesses.org/processes/joining/arc-welding.md): Arc welding melts and fuses metal with an electric arc between an electrode and the work, shielded from the atmosphere by gas or flux.

## Related processes

- [Sheet Metal Fabrication](https://manufacturingprocesses.org/processes/forming/sheet-metal-fabrication.md): Sheet metal fabrication cuts, bends and joins flat stock into a finished assembly, typically laser cutting followed by press braking, hardware and welding.
- [Punching and Blanking](https://manufacturingprocesses.org/processes/cutting/punching-and-blanking.md): Punching and blanking shear sheet metal between a punch and a die, producing holes and flat outlines at very high speed.
- [Metal Stamping](https://manufacturingprocesses.org/processes/forming/metal-stamping.md): Metal stamping shears and forms sheet metal between matched dies in a press, producing flat and shallow-formed parts at very high rates.
- [Staking](https://manufacturingprocesses.org/processes/joining/staking.md): Heat staking softens a molded plastic boss with hot air or ultrasonic energy and reforms it into a head that traps the mating part.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/joining/riveting)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Soldering and Brazing"
category: "Joining"
subcategory: "Thermal"
materials: ["Metal"]
tolerances: "Joint clearance 0.001–0.005 in (0.025–0.13 mm) for brazing, 0.003–0.005 in (0.075–0.13 mm) for soldering, specified at brazing temperature; assembly position set by fixture or self-locating features"
volumes: "1 to 1,000,000+ — torch for one-offs and repair, continuous or vacuum furnace at production volume"
lead_time: "Seconds to minutes for torch and induction work; 20–90 minute furnace cycles including ramp and controlled cool"
url: https://manufacturingprocesses.org/processes/joining/soldering-and-brazing
---

# Soldering and Brazing

Soldering and brazing join metals with a molten filler that wets the joint by capillary action while the parent metal stays solid.

- **Category**: [Joining](https://manufacturingprocesses.org/processes/joining.md)
- **Family**: Thermal
- **Materials**: Metal
- **Typical tolerances**: Joint clearance 0.001–0.005 in (0.025–0.13 mm) for brazing, 0.003–0.005 in (0.075–0.13 mm) for soldering, specified at brazing temperature; assembly position set by fixture or self-locating features
- **Typical volumes**: 1 to 1,000,000+ — torch for one-offs and repair, continuous or vacuum furnace at production volume
- **Lead time**: Seconds to minutes for torch and induction work; 20–90 minute furnace cycles including ramp and controlled cool

## Overview

Soldering and brazing join metals with a molten filler that wets both faces and is drawn into a close-fitting joint by capillary action, while the parent metal never melts. The dividing line is a temperature: filler that melts below 840 °F (450 °C) is solder, above it is braze.

Because the base metal stays solid, there is no fusion zone, no dilution and almost no distortion — the joint can be made on finish-machined parts, on thin-wall tube, on carbide inserts, and on assemblies with dozens of joints made simultaneously in one furnace pass.

The number to design around: a lap joint with an overlap of 3x the thinner member's thickness develops the full strength of that member, so brazed assemblies are designed by overlap area rather than by filler strength. Silver brazing filler melts at 1,125–1,145 °F (607–618 °C); eutectic Sn63/Pb37 solder at 361 °F (183 °C); lead-free SAC305 at 423–428 °F (217–220 °C).

## How it works

1. **Clean and fit.** Oxide, oil and oxide-forming residue must be gone — capillary filler will not wet a dirty surface. Parts are then assembled to a controlled clearance, typically 0.001–0.005 in (0.025–0.13 mm) for brazing and 0.003–0.005 in (0.075–0.13 mm) for soldering. Wider than that and capillary action fails; tighter and the filler cannot enter.
2. **Flux or atmosphere.** Flux dissolves the oxide film and protects the joint while it heats. Furnace brazing in vacuum or dissociated ammonia / hydrogen does the same job chemically and leaves nothing to clean off.
3. **Heat the joint, not the filler.** The assembly is brought above the filler's liquidus so the parent metal itself melts the filler. Preplaced rings, shims, paste or clad layers are common; hand-fed wire works for torch brazing.
4. **Capillary fill.** Molten filler is drawn through the clearance, flowing toward the hottest region. A complete fillet visible all the way around the joint is the signal that the gap filled.
5. **Cool and clean.** The joint solidifies as the filler drops below solidus. Residual halide-containing flux is corrosive and must be washed off. Furnace-brazed assemblies cool under controlled atmosphere and need no cleaning.

### The three heating methods

**Conduction.** A soldering iron, resistance-heated tip or hot plate conducts heat into the joint. Local, cheap and fully manual — electronics rework, sheet metal seams, jewelry.

**Torch.** Oxy-fuel or air-fuel flame heats the parts directly. The most flexible route, and the standard for one-off and repair work: HVAC copper tube, bicycle frames, tool tipping. Skill-dependent, because the operator controls the whole thermal cycle by eye.

**Furnace.** The whole assembly goes through a controlled thermal cycle — continuous belt with a reducing atmosphere, or a vacuum furnace for stainless and superalloys. Every joint in an assembly is made at once with identical, repeatable heat, which is why heat exchangers, honeycomb structures and multi-joint manifolds are furnace brazed. Cycles run 20–90 minutes including ramp and cool.

Induction and dip brazing sit between torch and furnace: fast, localized, and repeatable enough for production without a full furnace cycle.

## Design guidelines

### Joint clearance is the controlling dimension

Capillary flow needs a gap, and the optimum is narrow: 0.001–0.005 in (0.025–0.13 mm) for silver brazing with flux, essentially 0.000–0.002 in (0–0.05 mm) for nickel filler in a vacuum furnace, and 0.003–0.005 in (0.075–0.13 mm) for soldering. Design that clearance **at brazing temperature**, not at room temperature: a steel bushing in an aluminum housing closes up on heating, while a copper sleeve in a steel bore opens. Cylindrical joints are dimensioned as fits — see the [ISO 286 fits and tolerances chart](/charts/iso-286-fits-tolerances).

### Design a lap, never a butt

A butt joint has only the cross-section as bond area and no capillary path, so it is weak by definition. Use a lap, a sleeve or a scarf. The standard rule is an **overlap of 3x the thickness of the thinner member**, which makes the joint stronger than the member it joins — the part yields before the braze shears. Longer overlaps add little.

### Position the joint, and respect the thermal cycle

Capillary action can pull filler uphill, but do not rely on it: preplace filler where gravity assists flow and design a visible fillet exit so an operator or camera can confirm the joint filled through. Remember too that brazing temperature is a heat treatment — brazing 6061 aluminum at 1,070–1,080 °F (577–582 °C) puts it near solution temperature and it must be re-aged, brazing hardened steel above its tempering temperature softens it, and work-hardened copper anneals.

### Filler and material selection

| Filler | Melting range | Use on | Notes |
| --- | --- | --- | --- |
| Sn63/Pb37 solder | 361 °F (183 °C), eutectic | Cu, brass, steel, PCBs | Shear strength only ~3,000–5,000 psi (20–35 MPa) — never structural |
| SAC305 (Sn/Ag/Cu) | 423–428 °F (217–220 °C) | Electronics, RoHS work | Higher process temperature than SnPb; wets less readily |
| BAg silver filler | 1,125–1,145 °F (607–618 °C) for BAg-1 | Steel, stainless, copper, carbide, nickel | The general-purpose braze; needs flux in air |
| BCuP (copper-phosphorus) | 1,310–1,460 °F (710–795 °C) | Copper to copper only | Self-fluxing on copper; **never on ferrous** — brittle iron phosphides |
| BCuZn brass/bronze | ~1,600–1,650 °F (870–900 °C) | Steel, cast iron | Also used as bronze fillet-braze without capillary action |
| BNi nickel filler | 1,780–2,100 °F (970–1,150 °C) | Stainless, superalloys | Vacuum furnace; high-temperature and corrosion service |
| BAlSi (Al-12Si) | 1,070–1,080 °F (577–582 °C) | Aluminum to aluminum | Uncomfortably close to the parent's solidus — tight furnace control |

Base-metal melting points, which set the process window against the filler ranges above, are on the [metal melting points chart](/charts/metal-melting-points).

### Inspection

Visual inspection of the fillet all the way around the joint is the primary method and is genuinely informative — a continuous fillet at the far side means the capillary gap filled. Beyond that: helium leak testing to 10⁻⁹ std cc/s for pressure and vacuum joints, radiography for internal voids in critical brazements, ultrasonic testing for bond coverage, and destructive peel or shear coupons run alongside a production lot. Be careful with dye penetrant — it wicks into any porosity and can be impossible to remove.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Braze clearance, flux brazing | 0.002 in (0.05 mm) | 0.001–0.005 in (0.025–0.13 mm) | Outside this band capillary flow stops |
| Vacuum braze clearance | 0.001 in (0.025 mm) | 0.000–0.002 in (0–0.05 mm) | Nickel fillers flow in near-zero gaps |
| Solder joint clearance | 0.004 in (0.1 mm) | 0.003–0.005 in (0.075–0.13 mm) | Solder is more viscous than braze |
| Lap overlap | 3x thinner member thickness | 2x minimum | Makes the joint stronger than the member |
| Joint type | Lap, sleeve or scarf | Never a plain butt | No capillary path and minimal bond area |
| Fillet visibility | Both ends of the joint | — | The only cheap confirmation the gap filled |

## Variants

- Conduction Method
- Torch Method
- Furnace Method

## Cost drivers

The split is between manual and furnace routes. Torch brazing is labor and skill: no tooling, immediate start, and a cost per joint that never falls with volume. Furnace brazing inverts that — fixtures, filler preforms and a qualified cycle up front, then a cost per joint that collapses because a single cycle can make hundreds of joints across a full furnace load at once.

Filler metal is a real line item when it contains silver, so preform sizing matters: hand-fed wire routinely puts two or three times the necessary filler into a joint. Vacuum furnace time is the other significant cost, and it is charged by the cycle, so a partly loaded furnace is expensive per part.

1. **Use preplaced preforms** — rings, washers, shims or clad layers — instead of hand-feeding. Filler use drops and joints become repeatable.
2. **Braze every joint in one cycle.** Design the assembly so all joints are made in one furnace pass rather than sequenced with progressively lower-melting fillers.
3. **Hold the clearance with the parts, not the fixture.** A machined shoulder, knurl or dimple that self-centers the joint deletes fixture cost.
4. **Check whether a cadmium-free or lower-silver filler works.** Filler choice can move joint cost significantly with no design change.
5. **Do not over-specify the filler.** A BAg alloy on a joint that a copper-phosphorus filler would handle is a pure cost adder on copper-to-copper work.

## FAQ

### What is the difference between soldering and brazing?

Filler melting temperature, and nothing else about the mechanism. Filler that melts below 840 °F (450 °C) is solder; above that it is braze. Both rely on capillary action to fill a close-fitting joint while the parent metal stays solid. The practical difference is strength: brazed joints can develop the strength of the base member, while tin-lead solder shears at only about 3,000–5,000 psi (20–35 MPa).

### What joint clearance does brazing need?

0.001–0.005 in (0.025–0.13 mm) for silver filler with flux, and essentially 0.000–0.002 in (0–0.05 mm) for nickel filler in a vacuum furnace. Outside that band capillary action stops — too wide and the filler will not bridge, too tight and it cannot enter. Specify the clearance at brazing temperature, since differential expansion of dissimilar metals changes the gap on heating.

### How strong is a brazed joint?

Strong enough that the joint is not the weak point, if it is designed as a lap with an overlap of 3x the thinner member's thickness. At that overlap the parent member yields before the braze shears, so brazed assemblies are sized by overlap area rather than by filler strength. A plain butt joint, by contrast, has no capillary path and minimal bond area.

### Can you braze aluminum?

Yes, with an Al-12Si filler (BAlSi-4) melting at 1,070–1,080 °F (577–582 °C), typically in a controlled-atmosphere brazing furnace. The difficulty is that this is only a few degrees below the solidus of common aluminum alloys, so furnace temperature uniformity has to be tight. Brazing also puts 6061 near solution temperature, so the part needs re-aging afterward.

### How are brazed joints inspected?

Visual inspection of the fillet all the way around is the primary and surprisingly informative method: a continuous fillet at the far end of the joint means the capillary gap filled through. Beyond that, helium leak testing to 10⁻⁹ std cc/s for pressure joints, radiography for internal voids, and destructive shear coupons run with the lot. Avoid dye penetrant, which wicks into porosity and cannot be removed.

## Alternative processes

- [Adhesive Bonding](https://manufacturingprocesses.org/processes/joining/adhesive-bonding.md): Adhesive bonding joins parts with a cured polymer layer that spreads load over the whole bond area instead of concentrating it at fastener points.
- [Mechanical Fastening](https://manufacturingprocesses.org/processes/joining/mechanical-fastening.md): Mechanical fastening joins parts with screws, inserts, clinch hardware or snap fits so the joint can be assembled and taken apart again.
- [Friction Welding](https://manufacturingprocesses.org/processes/joining/friction-welding.md): Friction welding rubs two parts together under load until the interface plasticises and forges together, with no melting and no filler metal.
- [Riveting](https://manufacturingprocesses.org/processes/joining/riveting.md): Riveting sets a rivet through aligned holes and upsets its tail so the joint clamps permanently, with no heat and no access to both sides for blind types.

## Related processes

- [PCB Assembly (SMT Reflow)](https://manufacturingprocesses.org/processes/joining/pcb-assembly-smt-reflow.md): PCB assembly prints solder paste, places components and reflows the board through a heated oven so every joint forms in one pass.
- [Arc Welding](https://manufacturingprocesses.org/processes/joining/arc-welding.md): Arc welding melts and fuses metal with an electric arc between an electrode and the work, shielded from the atmosphere by gas or flux.
- [Power Beam Welding](https://manufacturingprocesses.org/processes/joining/power-beam-welding.md): Power beam welding fuses metal with a laser or electron beam focused to a very small spot, producing deep, narrow welds with little distortion.
- [Heat Treatment](https://manufacturingprocesses.org/processes/forming/heat-treatment.md): Heat treatment cycles metal through controlled heating and cooling to change its hardness, strength, ductility or internal stress state.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/joining/soldering-and-brazing)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Staking"
category: "Joining"
subcategory: "Thermal"
materials: ["Plastic"]
tolerances: "Formed head height controlled to about ±0.005 in (±0.13 mm) by tool stop; hole-to-boss clearance 0.002–0.010 in (0.05–0.25 mm)"
volumes: "1,000 to 10,000,000+ per year"
lead_time: "0.2–0.8 s per ultrasonic stake, 4–15 s per hot air cycle; 2–5 weeks for multi-tip forming tooling"
url: https://manufacturingprocesses.org/processes/joining/staking
---

# Staking

Heat staking softens a molded plastic boss with hot air or ultrasonic energy and reforms it into a head that traps the mating part.

- **Category**: [Joining](https://manufacturingprocesses.org/processes/joining.md)
- **Family**: Thermal
- **Materials**: Plastic
- **Typical tolerances**: Formed head height controlled to about ±0.005 in (±0.13 mm) by tool stop; hole-to-boss clearance 0.002–0.010 in (0.05–0.25 mm)
- **Typical volumes**: 1,000 to 10,000,000+ per year
- **Lead time**: 0.2–0.8 s per ultrasonic stake, 4–15 s per hot air cycle; 2–5 weeks for multi-tip forming tooling

## Overview

Heat staking softens the tip of a molded plastic boss and reforms it into a head that traps a mating part underneath — the plastic equivalent of riveting, using material the part already has. No fastener, no adhesive, no added part number.

Two heat sources dominate. **Hot air / cold stake** blows hot air at 500–900 °F (260–480 °C) onto the boss and then consolidates the softened material with an unheated forming tool, which produces a clean head with no sticking or stringing. **Ultrasonic staking** uses a contoured horn at 20 kHz and low amplitude to melt and form the head in under a second.

The characterizing dimension: a boss must protrude above the mating part by roughly 1.5–2x its own diameter to supply enough material for a standard dome head. Staking is a shear-and-retention joint, not a structural weld — it holds parts in place, resists rattle and locates assemblies, but it is not sized to carry significant tension.

## How it works

1. **Assemble.** The mating part — a PCB, a metal bracket, a decorative trim panel — drops over the molded boss through a clearance hole, typically 0.002–0.010 in (0.05–0.25 mm) larger than the boss.
2. **Soften.** Heat is applied only to the protruding boss tip. Hot air staking heats to a depth of a few thousandths of an inch in 2–8 s; ultrasonic staking melts the tip in 0.2–0.8 s through friction from a low-amplitude horn.
3. **Form.** A tool with the head profile machined into it descends and displaces the softened material sideways into a head. In hot air / cold stake the tool is deliberately cold so it chills and sets the head instantly.
4. **Hold and release.** Pressure is held for 1–3 s while the head solidifies, then the tool retracts. Multiple bosses are almost always staked simultaneously from one multi-tip head.

### Choosing the head profile

- **Dome** — the general-purpose profile, for bosses up to about 1/16 in (1.5 mm) diameter.
- **Flush / rosette** — sits within a counterbore, used when the head cannot stand proud of the mating surface.
- **Hollow** — for bosses over about 5/32 in (4 mm) diameter, where a solid head would need far too much material and would sink; the tool forms only the wall of the boss outward.
- **Knurled** — spreads a wide, textured head to distribute load on soft or thin mating parts.

### Hot air versus ultrasonic

Hot air staking is slower (4–15 s per cycle) but produces the best cosmetic head, generates no particulate, transmits no vibration into the assembly, and works well on glass-filled grades where the abrasive filler destroys ultrasonic horns. It is the standard for staking populated PCBs and visible trim.

Ultrasonic staking is faster (under 1 s) and uses cheaper tooling, but the horn contacts the melt, can leave marks and fines, and puts 20 kHz vibration into whatever is being retained — a real concern for crystals, relays and MEMS devices.

## Design guidelines

### Size the boss protrusion to the head volume

The head is formed from the material that stands above the mating part, so protrusion is the controlling dimension. For a standard dome head, allow a protrusion of roughly 1.5–2x the boss diameter; the exact figure follows from equating the protruding volume to the head volume, so confirm it against the specific head profile before cutting the mold. Too little protrusion gives a thin, weak head that has not filled the tool; too much squeezes out around the tool as flash.

### Head and hole proportions

Aim for a formed head diameter of roughly 1.5–2x the boss diameter and a head height of about 0.5x the boss diameter. Keep the clearance hole in the mating part 0.002–0.010 in (0.05–0.25 mm) over the boss diameter — enough for assembly, tight enough that the head has material to bear against all the way around.

### Use a hollow head above 5/32 in (4 mm)

A solid dome on a large boss requires so much displaced material that the head sinks, voids and takes far too long to cool. Above roughly 5/32 in (4 mm) boss diameter, switch to a hollow profile that forms only the boss wall outward, or to a knurled head.

### Support the boss and design for the load

Put a radius at the base of the boss (0.010–0.020 in / 0.25–0.5 mm) to avoid a stress riser where the retention load lands, and support the underside of the mating part so the stake is not the only thing carrying the assembly load. Staked joints resist shear and rattle well; they are poor in tension and should never be the sole retention for a safety-critical or heavily loaded part.

### Material suitability

| Resin | Hot air staking | Ultrasonic staking | Notes |
| --- | --- | --- | --- |
| ABS, PC, PC/ABS, PS | Excellent | Excellent | Amorphous resins form clean heads |
| PP, PE | Good | Good | Wide melt range, forgiving |
| PA (nylon) | Good | Fair | Dry before staking — absorbed moisture foams the head |
| POM (acetal) | Fair | Fair | Narrow melt window; head can crack on cooling |
| 20–50% glass filled | Good | Poor | Abrasive fill destroys horns; hot air is the practical route |
| Thermosets | No | No | Cannot be reformed once cured |
| Boss already stressed or knit-lined | Avoid | Avoid | Staking loads the boss base — a knit line there will split |

Boss diameter, wall thickness and base radius are molded features governed by the rules on the [injection molding design guidelines chart](/charts/injection-molding-design-guidelines).

### Inspection

Formed head height and diameter are gauged, often with a simple go/no-go or a vision check on a multi-stake assembly, and head height is controlled to about ±0.005 in (±0.13 mm) by tool stop. Pull-out and torque-out testing on lot samples confirms retention. Cross-sections through the head reveal voids and unfused material. Ultrasonic stakers log energy and collapse per cycle; hot air stations log air temperature, dwell and forming force.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Boss protrusion above the mating part | 1.5–2x boss diameter | Profile dependent | Supplies the volume the head is formed from |
| Formed head diameter | 1.5–2x boss diameter | — | Bearing area against the mating part |
| Formed head height | ~0.5x boss diameter | — | Enough section to carry the retention load |
| Hole clearance over boss | 0.004 in (0.1 mm) | 0.002–0.010 in (0.05–0.25 mm) | Assembly fit without losing bearing area |
| Boss diameter for a solid dome | Under 5/32 in (4 mm) | Use hollow head above | Solid heads sink and void on large bosses |
| Boss base radius | 0.010–0.020 in (0.25–0.5 mm) | — | The retention load reacts at the boss base |
| Joint loading | Shear and retention | Not primary tension | The head is a formed cap, not a fastener |

## Variants

- Hot Air Staking
- Ultrasonic Staking

## Cost drivers

Staking is one of the cheapest joining operations available to a molded assembly, because the fastener is already part of the molding. There is no purchased hardware, no adhesive, no consumable at all — only machine time and a forming tool.

Tooling cost tracks the number of stakes and how they are grouped: a multi-tip head that forms twelve stakes in one stroke costs little more than a single-tip head and turns twelve operations into one. That grouping decision, made at design time, is the dominant cost lever.

Hot air stations cost more than ultrasonic ones and run longer cycles, but avoid horn wear on filled resins and avoid vibration damage to what is being retained — which usually shows up as scrap avoided rather than cost saved.

1. **Group every stake in an assembly onto one multi-tip head.** One stroke instead of twelve is the single biggest saving.
2. **Standardize boss diameters** across a product family so the same tips and profiles are reused.
3. **Use staking to delete screws and their bosses.** Each deleted screw removes a purchased part, a driver station and a torque verification step.
4. **Choose hot air for glass-filled resin.** Horn replacement on abrasive fill quickly outweighs the cheaper ultrasonic machine.
5. **Do not stake what needs to come apart.** A staked head must be drilled out for rework; use a screw or a threaded insert where serviceability matters.

## FAQ

### How tall does the boss need to be for heat staking?

For a standard dome head, allow the boss to protrude above the mating part by roughly 1.5–2x its own diameter. The head is formed entirely from that protruding material, so the exact figure comes from matching protruding volume to head volume for the chosen profile. Too little gives a thin unfilled head; too much extrudes as flash around the tool.

### Hot air or ultrasonic staking — which should I specify?

Hot air / cold stake for glass-filled resins, for cosmetic heads, and whenever what is being retained cannot take 20 kHz vibration — populated PCBs, crystals, relays. Ultrasonic staking for speed (under 1 s versus 4–15 s) and cheaper tooling on unfilled resins where a small horn mark and some particulate are acceptable.

### How much load can a heat-staked joint carry?

Enough to retain and locate a part and resist rattle, but it is not a structural fastener. Staked joints work well in shear and poorly in tension, because the head is a formed cap of the same resin rather than a separate high-strength element. Qualify retention with pull-out testing on lot samples rather than calculating it.

### Why use a hollow head instead of a dome?

Above about 5/32 in (4 mm) boss diameter a solid dome needs so much displaced material that the head sinks, voids internally and takes a long time to cool. A hollow profile forms only the boss wall outward into an annular head, using far less material, cooling faster and giving a flatter, more uniform bearing surface.

### Can you heat stake glass-filled plastic?

Yes, with hot air staking. Glass-filled grades up to about 50% form good heads under a cold forming tool. Ultrasonic staking is a poor choice on filled resin because the abrasive fill erodes the horn rapidly, and because the fibers at the head surface can leave a rough, weak-looking formed head.

## Alternative processes

- [Mechanical Fastening](https://manufacturingprocesses.org/processes/joining/mechanical-fastening.md): Mechanical fastening joins parts with screws, inserts, clinch hardware or snap fits so the joint can be assembled and taken apart again.
- [Adhesive Bonding](https://manufacturingprocesses.org/processes/joining/adhesive-bonding.md): Adhesive bonding joins parts with a cured polymer layer that spreads load over the whole bond area instead of concentrating it at fastener points.
- [Riveting](https://manufacturingprocesses.org/processes/joining/riveting.md): Riveting sets a rivet through aligned holes and upsets its tail so the joint clamps permanently, with no heat and no access to both sides for blind types.
- [Laser Plastic Welding](https://manufacturingprocesses.org/processes/joining/laser-plastic-welding.md): Laser plastic welding passes a beam through a transmissive part into an absorbing one, melting only the interface between them.

## Related processes

- [Ultrasonic Welding](https://manufacturingprocesses.org/processes/joining/ultrasonic-welding.md): Ultrasonic welding applies high-frequency vibration through a horn to melt a thermoplastic joint locally in well under a second.
- [Injection Molding](https://manufacturingprocesses.org/processes/forming/injection-molding.md): Injection molding forces molten thermoplastic into a steel mold under high pressure, producing high-volume identical plastic parts.
- [Insert Molding](https://manufacturingprocesses.org/processes/forming/insert-molding.md): Insert molding loads metal inserts into the cavity and molds polymer around them, capturing threads, terminals or stampings in a single shot.
- [Overmolding](https://manufacturingprocesses.org/processes/forming/overmolding.md): Overmolding molds a second material, usually a soft elastomer, directly onto an already-molded rigid substrate so the two bond permanently.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/joining/staking)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Timber Frame Structures"
category: "Joining"
subcategory: "Mechanical"
materials: ["Wood"]
tolerances: "CNC-cut joinery to about ±1/16 in (±1.5 mm) over member lengths of 40 ft (12 m) and more; erection tolerances are set by the project specification"
lead_time: "Months from design freeze to erection, dominated by engineering, shop drawing approval and engineered-timber mill scheduling; CNC cutting of a frame itself takes days"
url: https://manufacturingprocesses.org/processes/joining/timber-frame-structures
---

# Timber Frame Structures

Timber frame structures assemble large solid or engineered timber members into a load-bearing frame using cut joints and steel connectors.

- **Category**: [Joining](https://manufacturingprocesses.org/processes/joining.md)
- **Family**: Mechanical
- **Materials**: Wood
- **Typical tolerances**: CNC-cut joinery to about ±1/16 in (±1.5 mm) over member lengths of 40 ft (12 m) and more; erection tolerances are set by the project specification
- **Lead time**: Months from design freeze to erection, dominated by engineering, shop drawing approval and engineered-timber mill scheduling; CNC cutting of a frame itself takes days

## Overview

Timber frame construction assembles large solid-sawn or engineered timber members into a load-bearing frame, joined either with cut joinery — pegged mortise and tenon, the traditional route — or with steel connectors such as knife plates, concealed hangers, bolts and self-tapping structural screws.

The modern version runs on engineered products. Glulam is built up from laminations of 1-3/8 in (35 mm) nominal thickness for straight members and 3/4 in (19 mm) for tight curves; cross-laminated timber (CLT) uses 3, 5 or 7 orthogonal layers to give panels 4–12 in (100–300 mm) thick that act as both floor and diaphragm.

Fire design is what makes heavy timber viable structurally. Wood chars at a nominal 1.5 in/hour (38 mm/hour) and the char layer insulates the sound wood beneath, so a member is fire-rated by calculating its residual section after the required exposure — an oversized beam is a fire-rated beam, with no applied protection.

## How it works

1. **Engineer the frame and the connections.** Member sizes and connection capacities come from the NDS, with CLT to ANSI/APA PRG 320 and glulam to ANSI A190.1. Traditional cut joinery is designed to TFEC 1. Connections, not members, usually govern.
2. **Model and produce shop drawings.** The frame is modelled in 3D and every joint is dimensioned, because cutting is done off the model rather than off site measurements.
3. **Machine the joinery.** CNC timber processing centers cut mortises, tenons, housings, bird's mouths and connector pockets in members up to 40 ft (12 m) or longer, holding roughly ±1/16 in (±1.5 mm). What was once weeks of layout and chisel work is now a machine program.
4. **Dry fit.** Frames are traditionally assembled flat in the shop or on site to confirm every joint closes before the raising.
5. **Raise and pin.** Bents are assembled on the deck and lifted into place. Traditional joints are secured with 1 in (25 mm) hardwood pegs, draw-bored with a 1/16–1/8 in (1.5–3 mm) offset so driving the peg pulls the shoulder tight. Modern frames use bolts, screws and steel plates, tightened snug rather than to a preload.
6. **Enclose.** Structural insulated panels or CLT panels wrap the frame, which stays exposed inside.

### Solid sawn versus engineered timber

Solid heavy timber is usually delivered green or partly seasoned at 12–19% moisture content and will check and shrink across the grain as it dries in service — that is expected and generally not a structural problem, but the connections must tolerate it. Glulam, LVL, PSL and CLT are manufactured at 8–12% moisture content, are dimensionally stable, and can span further for a given depth, which is why they dominate long-span and mass timber work.

## Design guidelines

### Design connections for shrinkage across the grain

Timber shrinks across the grain and barely at all along it. The classic detailing error is a bolted connection with several inches of cross-grain thickness between fasteners: as the timber dries and shrinks, the steel does not, and the wood splits between the bolts. Keep cross-grain dimensions between fasteners short, use slotted holes where movement is expected, and place fasteners so shrinkage relieves rather than builds stress.

### Do not torque timber bolts like steel bolts

Bolts in timber connections are snug-tightened, not preloaded. Wood creeps under sustained compression perpendicular to the grain, so any preload applied at installation is lost within months and excessive torque simply crushes the wood under the washer. The values on the [bolt torque chart](/charts/bolt-torque-chart) apply to steel-to-steel joints and must not be carried over to timber connections; follow the connection design and the connector manufacturer's installation instructions instead.

### Size for the char layer, not for a coating

At a nominal char rate of 1.5 in/hour (38 mm/hour), a one-hour rating consumes roughly 1.5 in (38 mm) from each exposed face, and the member is designed on the residual section with reduced strength factors. That is why exposed heavy timber achieves ratings that exposed steel cannot without protection — but it only works if the member is genuinely oversized from the outset. Retrofitting fire performance into an undersized timber member is not possible.

### Detail out end-grain exposure and standing water

Timber decays where it stays wet. Keep end grain off concrete and out of the weather, provide capillary breaks at bearing points, flash exposed beam ends, and design roof geometry so water never sits on a horizontal timber surface.

| Element | Typical figure | Notes |
| --- | --- | --- |
| Glulam lamination, straight members | 1-3/8 in (35 mm) nominal | Thinner laminations for tight radii |
| Glulam lamination, curved members | 3/4 in (19 mm) | Radius-dependent |
| CLT panel thickness | 4–12 in (100–300 mm) | 3, 5 or 7 orthogonal layers per ANSI/APA PRG 320 |
| Traditional peg | 1 in (25 mm) hardwood, draw-bored 1/16–1/8 in (1.5–3 mm) | The offset is what pulls the shoulder tight |
| CNC-cut joinery tolerance | About ±1/16 in (±1.5 mm) | Over member lengths of 40 ft (12 m) and more |
| Solid timber moisture content | 12–19% at erection | Expect checking and cross-grain shrinkage |
| Engineered timber moisture content | 8–12% | Dimensionally stable |
| Nominal char rate | 1.5 in/hr (38 mm/hr) | Per the NDS; design on the residual section |

### Inspection

Grade stamps on solid timber and mill certificates on glulam and CLT are the starting point. Moisture content is verified with a meter at delivery and before enclosure. Joint fit-up is checked at dry fit, connector installation is verified against the shop drawings and the manufacturer's instructions, and glulam bond lines are qualified by delamination testing to ASTM D2559 at the mill. Post-erection inspection is largely visual — plumb, level, bearing, and connector installation — with any splitting at connections recorded and evaluated.

## Cost drivers

Engineering and shop drawings come first and are a substantial fraction of the cost of a timber frame, because every joint is designed and dimensioned before anything is cut. Connection design in particular drives both cost and schedule; a frame with dozens of unique connector types is far more expensive than one with a repeating detail.

Material follows. Engineered timber is a manufactured product with mill lead times and minimum order quantities, and long-span glulam or wide CLT panels are made to order. Solid heavy timber in large sections is a specialty purchase.

Fabrication itself is comparatively fast once the model exists — CNC timber processing has collapsed what used to be the dominant labor cost. Erection is craned work with a small crew, usually measured in days for a house-sized frame.

1. **Repeat the connection detail.** One well-engineered joint used sixty times costs a fraction of sixty variations.
2. **Design bays on a regular module** so members repeat and the CNC program runs efficiently.
3. **Choose engineered timber where spans are long** — the added material cost is usually less than the cost of the deeper solid section it replaces.
4. **Fix the design before the mill order.** Changes after glulam or CLT is scheduled reset the lead time entirely.
5. **Detail for weather during construction.** Mass timber left open to rain is expensive to dry and to remediate.

## FAQ

### Why are timber bolts not torqued to steel bolt values?

Because wood creeps under sustained compression perpendicular to the grain. Any preload applied at installation relaxes within months, and high installation torque simply crushes the timber under the washer. Timber connection bolts are snug-tightened per the connection design and the connector manufacturer's instructions; steel-to-steel torque tables do not apply.

### How does heavy timber achieve a fire rating without protection?

Wood chars at a nominal 1.5 in/hour (38 mm/hour) and the char layer insulates the sound wood beneath it, so the member keeps a load-bearing core. Fire design under the NDS calculates the residual section after the required exposure and checks it with reduced strength factors. The consequence is that the member must be oversized from the start — the rating cannot be retrofitted.

### What is the difference between glulam and CLT?

Glulam is a beam or column built from laminations all running the same direction — 1-3/8 in (35 mm) nominal for straight members, 3/4 in (19 mm) for tight curves — and carries load along its length. CLT is a panel of 3, 5 or 7 layers laid at right angles to each other, 4–12 in (100–300 mm) thick, which spans in two directions and acts as a floor, wall or diaphragm.

### Why do timber connections split, and how is it avoided?

Because timber shrinks across the grain as it dries while steel connectors do not. A connection with several inches of cross-grain thickness between fasteners is restrained as the wood tries to shrink, and it splits. The fix is detailing: keep cross-grain distances between fasteners short, use slotted holes where movement is expected, and place fasteners so shrinkage relieves rather than builds stress.

### How accurate is CNC-cut timber joinery?

Roughly ±1/16 in (±1.5 mm), held over members 40 ft (12 m) and longer. That accuracy is what makes modern timber framing practical: joinery is cut directly from the 3D model, so a frame can be shipped and raised with confidence that the joints will close, rather than being fitted piece by piece on site.

## Alternative processes

- [Mechanical Fastening](https://manufacturingprocesses.org/processes/joining/mechanical-fastening.md): Mechanical fastening joins parts with screws, inserts, clinch hardware or snap fits so the joint can be assembled and taken apart again.
- [Adhesive Bonding](https://manufacturingprocesses.org/processes/joining/adhesive-bonding.md): Adhesive bonding joins parts with a cured polymer layer that spreads load over the whole bond area instead of concentrating it at fastener points.
- [Arc Welding](https://manufacturingprocesses.org/processes/joining/arc-welding.md): Arc welding melts and fuses metal with an electric arc between an electrode and the work, shielded from the atmosphere by gas or flux.

## Related processes

- [Joinery](https://manufacturingprocesses.org/processes/joining/joinery.md): Joinery cuts interlocking geometry into timber so the parts locate and hold each other, with or without adhesive or fasteners.
- [Wood Laminating](https://manufacturingprocesses.org/processes/forming/wood-laminating.md): Wood laminating glues thin veneers or kerfed strips over a form so the assembly holds a curve permanently once the adhesive has cured.
- [Sawing](https://manufacturingprocesses.org/processes/cutting/sawing.md): Sawing separates stock with a toothed blade — band, circular or cold saw — and is the default first operation on bar, tube and plate.
- [CNC Milling](https://manufacturingprocesses.org/processes/cutting/cnc-milling.md): CNC milling removes material with a rotating multi-flute cutter moved along programmed paths, producing prismatic parts from solid stock.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/joining/timber-frame-structures)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Ultrasonic Welding"
category: "Joining"
subcategory: "Thermal"
materials: ["Plastic", "Metal"]
tolerances: "Weld collapse held to ±0.002 in (±0.05 mm) with distance-mode control; joint faces should be molded flat and parallel to the limit the tool allows"
volumes: "10,000 to 10,000,000+ per year"
lead_time: "0.1–1.0 s weld, under 2 s total cycle; 2–4 weeks for horn and nest fixture tooling"
url: https://manufacturingprocesses.org/processes/joining/ultrasonic-welding
---

# Ultrasonic Welding

Ultrasonic welding applies high-frequency vibration through a horn to melt a thermoplastic joint locally in well under a second.

- **Category**: [Joining](https://manufacturingprocesses.org/processes/joining.md)
- **Family**: Thermal
- **Materials**: Plastic, Metal
- **Typical tolerances**: Weld collapse held to ±0.002 in (±0.05 mm) with distance-mode control; joint faces should be molded flat and parallel to the limit the tool allows
- **Typical volumes**: 10,000 to 10,000,000+ per year
- **Lead time**: 0.1–1.0 s weld, under 2 s total cycle; 2–4 weeks for horn and nest fixture tooling

## Overview

Ultrasonic welding drives a thermoplastic joint with high-frequency mechanical vibration — usually 20 kHz at 0.0008–0.005 in (20–125 µm) peak-to-peak amplitude — delivered through a tuned horn under clamping force. Friction at the joint interface generates heat exactly where the parts touch, and the weld is made in 0.1–1.0 s with a total cycle under 2 s.

Nothing else joins plastics that fast. A well-designed joint in an amorphous resin reaches 90–100% of parent material strength and can be hermetic, with no adhesive, no fastener and no added mass.

The same principle also welds metal foils and thin sheet — ultrasonic metal welding is a solid-state scrubbing process, not a melting one, and it joins copper to aluminum readily, which is why it dominates battery tab and wire-splice work. Volumes are high: the process only makes sense once horn and fixture tooling can be amortized, typically from tens of thousands of parts upward.

## How it works

1. **Clamp.** The horn descends and applies trigger force — commonly 50–500 lbf (0.2–2.2 kN) — pressing the parts together in a nest fixture that supports the joint directly beneath the weld.
2. **Vibrate.** A converter turns 20 kHz (or 15, 30, 35, 40 kHz) electrical energy into mechanical motion; a booster scales the amplitude and the horn delivers it into the part. Motion is perpendicular to the joint face.
3. **Localize the heat.** Cyclic surface friction and, once softening begins, viscoelastic hysteresis in the polymer concentrate heat at the interface. A molded **energy director** — a small triangular rib on one joint face — focuses that energy so melting starts at a defined point rather than randomly.
4. **Melt and collapse.** The director melts and flows across the interface; the parts close by a controlled collapse distance. Welders run in time, energy, peak-power or distance mode, and distance mode holds the collapse to about ±0.002 in (±0.05 mm).
5. **Hold.** Vibration stops and force is maintained for 0.2–1 s while the melt solidifies under pressure.

### Near field, far field and resin choice

If the joint sits within 0.25 in (6 mm) of where the horn contacts the part, the weld is **near field**; beyond that it is **far field** and the energy has to travel through the part to reach the joint.

Amorphous resins — ABS, PC, PS, acrylic, PSU — soften gradually over a broad range and transmit vibration well, so they weld reliably in both near and far field. Semicrystalline resins — PA, PP, PE, POM, PBT — absorb energy melting their crystalline phase and have a sharp melting point, so they need near-field welding and usually a shear joint rather than an energy director.

### Ultrasonic metal welding

The metal version applies the vibration **parallel** to the interface, scrubbing the oxide away and bringing clean metal into atomic contact without melting anything. It works on thin, ductile, conductive material — copper, aluminum, nickel foils, wire and tabs — and makes copper-to-aluminum joints that resistance welding and soldering both struggle with.

## Design guidelines

### Energy directors

For amorphous resins, mold a triangular rib on one joint face with a 90° included angle, 0.010–0.025 in (0.25–0.64 mm) high, base width roughly twice the height. One director per joint line, positioned on the part that is *not* in contact with the horn where possible. For hermetic seals, run the director as a continuous, unbroken ring — a single gap is a leak path.

### Shear joints for semicrystalline resins

A shear joint has a small interference between a boss and a socket so the parts weld progressively down the sidewall as they telescope together, keeping the melt away from air. Use 0.008–0.012 in (0.2–0.3 mm) of interference on parts under about 0.75 in (19 mm), increasing to roughly 0.016 in (0.4 mm) on larger parts, with a 0.020–0.030 in (0.5–0.75 mm) lead-in chamfer to align the parts before the interference engages.

### Support the joint and the horn contact area

The nest fixture must support the joint plane directly under the weld — unsupported walls flex and absorb energy instead of welding. The horn contact face should be flat, unbroken and perpendicular to the horn axis; textured or graphic surfaces mark under a horn. Keep sharp internal corners away from the energy path: they concentrate stress and crack under 20 kHz vibration.

### Resin compatibility

| Pair | Weldable? | Notes |
| --- | --- | --- |
| Same amorphous resin (ABS-ABS, PC-PC) | Excellent | Far-field capable; energy director joint |
| Same semicrystalline resin (PA-PA, PP-PP) | Good, near field only | Shear joint; keep the joint within 0.25 in (6 mm) of the horn |
| ABS to PC | Fair | Chemically compatible but melt temperatures differ — expect reduced strength |
| PC to acrylic | Fair | Compatible pair; validate on parts |
| PE to PP | No | Not chemically compatible despite both being polyolefins |
| Any resin to a filled grade over ~30% glass | Poor | Too little polymer at the interface to form a weld |
| Nylon after moisture pickup | Poor | Absorbed water foams at the joint — dry the parts or weld immediately after molding |
| Copper to aluminum (metal welding) | Excellent | Solid-state scrubbing; the standard battery-tab route |

Joint features are molded, so the wall, rib and radius rules that govern them are on the [injection molding design guidelines chart](/charts/injection-molding-design-guidelines).

### Inspection

Ultrasonic welds are validated in-process and by destructive sampling. Modern welders record energy, peak power, collapse distance and weld time for every cycle and alarm on out-of-window results — that data trail is the primary quality record. Destructively, pull and burst tests on lot samples, and cross-sections to confirm the melt has spread across the full joint width. For sealed parts, pressure decay or vacuum leak testing.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Energy director height | 0.015 in (0.4 mm) | 0.010–0.025 in (0.25–0.64 mm) | Too small starves the weld, too large leaves unmelted material |
| Energy director angle | 90° included | — | Concentrates energy at a repeatable point |
| Shear joint interference | 0.010 in (0.25 mm) | 0.008–0.016 in (0.2–0.4 mm) by part size | Sets the welded sidewall area |
| Joint-to-horn distance, semicrystalline | Under 0.25 in (6 mm) | Near field required | Crystalline phase absorbs the energy en route |
| Glass fill at the joint | Under 20% | 30% | Not enough polymer at the interface above this |
| Horn contact face | Flat, unbroken, perpendicular | — | Textured surfaces mark; angled faces lose coupling |
| Collapse control | Distance mode, ±0.002 in (±0.05 mm) | — | Time mode does not compensate for part variation |

## Cost drivers

Ultrasonic welding is a tooling-plus-cycle-time process. The press, converter and power supply are moderate capital; the real per-part cost is a tuned horn and a machined nest fixture, both specific to the part. Once those exist, the cycle is under 2 s and consumable cost is zero — no adhesive, no fastener, no filler — so unit cost is essentially machine time.

That structure means the process wins decisively at volume and loses at low volume, where adhesive or screws carry no tooling burden. Horns are also consumable on abrasive filled resins: titanium horns last far longer than aluminum on glass-filled material but cost more.

The hidden cost is joint development. Energy director and shear joint dimensions interact with resin, part stiffness and horn amplitude, and getting a hermetic seal usually takes an iteration on the mold — which is expensive after tooling is cut.

1. **Design the joint into the first mold revision.** Adding or resizing an energy director after tooling is cut is a steel change, not a process tweak.
2. **Use one horn for multiple parts** by keeping joint plane dimensions in a family across a product line.
3. **Specify unfilled or low-fill resin at the joint,** even if the body of the part is glass filled — a locally unfilled joint region welds far better.
4. **Buy distance-mode (servo) control** if hermeticity matters; it absorbs part-to-part molding variation that time mode cannot.
5. **Check whether staking or ultrasonic insertion** does the job instead — both use the same machine and cheaper tooling than a full hermetic weld.

## FAQ

### What is an energy director and how big should it be?

A small triangular rib molded onto one joint face that concentrates ultrasonic energy so melting starts at a defined point. Standard practice is a 90° included angle, 0.010–0.025 in (0.25–0.64 mm) high, with a base width about twice the height. For a hermetic seal, run it as a continuous unbroken ring around the joint — a single gap is a leak path.

### How strong is an ultrasonic weld?

A well-designed joint in an amorphous resin such as ABS or PC reaches 90–100% of parent material strength and can be hermetic. Strength falls off with glass content — above roughly 30% fill there is not enough polymer at the interface to form a weld — and with semicrystalline resins welded in the far field.

### Why won't polypropylene weld as easily as ABS?

Polypropylene is semicrystalline: it has a sharp melting point and absorbs energy melting its crystalline phase, so vibration is attenuated on the way to the joint. It needs near-field welding, with the joint within 0.25 in (6 mm) of the horn contact, and usually a shear joint (0.008–0.016 in / 0.2–0.4 mm interference) rather than an energy director.

### Can ultrasonic welding join dissimilar plastics?

Only if the resins are chemically compatible and their melt temperatures are close. ABS to PC and PC to acrylic work with reduced strength; PE to PP does not weld despite both being polyolefins. As a rule, plan on welding a resin to itself and treat any dissimilar pair as something to validate on real parts before committing tooling.

### Can you ultrasonically weld metal?

Yes, but it is a different process. Ultrasonic metal welding vibrates parallel to the interface, scrubbing away oxide and bringing clean metal into atomic contact with no melting at all. It suits thin, ductile, conductive material — copper and aluminum foils, wires and battery tabs — and makes copper-to-aluminum joints that resistance welding and soldering both handle poorly.

### How are ultrasonic welds inspected?

Primarily in-process: modern welders log energy, peak power, collapse distance and weld time for every cycle and alarm on out-of-window results. That trail is the quality record. Destructive pull and burst tests on lot samples, cross-sections to confirm melt across the full joint width, and pressure-decay leak testing on sealed parts back it up.

## Alternative processes

- [Vibration Welding](https://manufacturingprocesses.org/processes/joining/vibration-welding.md): Vibration welding rubs two thermoplastic parts together in-plane at low frequency under pressure until the interface melts and fuses.
- [Hot Plate Welding](https://manufacturingprocesses.org/processes/joining/hot-plate-welding.md): Hot plate welding presses two thermoplastic parts against a heated platen to melt the joint faces, then clamps them together to fuse.
- [Adhesive Bonding](https://manufacturingprocesses.org/processes/joining/adhesive-bonding.md): Adhesive bonding joins parts with a cured polymer layer that spreads load over the whole bond area instead of concentrating it at fastener points.
- [Mechanical Fastening](https://manufacturingprocesses.org/processes/joining/mechanical-fastening.md): Mechanical fastening joins parts with screws, inserts, clinch hardware or snap fits so the joint can be assembled and taken apart again.

## Related processes

- [Staking](https://manufacturingprocesses.org/processes/joining/staking.md): Heat staking softens a molded plastic boss with hot air or ultrasonic energy and reforms it into a head that traps the mating part.
- [Laser Plastic Welding](https://manufacturingprocesses.org/processes/joining/laser-plastic-welding.md): Laser plastic welding passes a beam through a transmissive part into an absorbing one, melting only the interface between them.
- [Injection Molding](https://manufacturingprocesses.org/processes/forming/injection-molding.md): Injection molding forces molten thermoplastic into a steel mold under high pressure, producing high-volume identical plastic parts.
- [Insert Molding](https://manufacturingprocesses.org/processes/forming/insert-molding.md): Insert molding loads metal inserts into the cavity and molds polymer around them, capturing threads, terminals or stampings in a single shot.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/joining/ultrasonic-welding)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Upholstery"
category: "Joining"
subcategory: "Mechanical"
materials: ["Composite"]
volumes: "1 to 100,000+ — hand work for one-offs and restoration, CNC cutting and sewing cells for production runs"
lead_time: "Hours of hand labor per seat for traditional sprung upholstery; production cut-and-sew cells work in minutes per panel once patterns and cutting nests exist"
url: https://manufacturingprocesses.org/processes/joining/upholstery
---

# Upholstery

Upholstery builds a padded surface from suspension, foam, wadding and textile over a frame, fixed with staples, ties and stitching.

- **Category**: [Joining](https://manufacturingprocesses.org/processes/joining.md)
- **Family**: Mechanical
- **Materials**: Composite
- **Typical volumes**: 1 to 100,000+ — hand work for one-offs and restoration, CNC cutting and sewing cells for production runs
- **Lead time**: Hours of hand labor per seat for traditional sprung upholstery; production cut-and-sew cells work in minutes per panel once patterns and cutting nests exist

## Overview

Upholstery builds a padded seating surface in layers over a frame: suspension, foundation, foam, wadding, cover and trim, fixed with staples, ties and stitching. Each layer does a distinct job, and the failure modes of a seat almost always trace back to one of them being specified for appearance rather than function.

The two numbers that define a cushion are foam density and IFD. Seat foam is typically 1.8–2.8 lb/ft³ (29–45 kg/m³) with an indentation force deflection of roughly 25–50 lbf at 25% deflection; density predicts how long the foam lasts, IFD predicts how firm it feels, and they are independent.

Cover fabric is specified by abrasion resistance — 15,000 double rubs on the Wyzenbeek test or 20,000 Martindale cycles is the usual minimum for contract upholstery, with heavy-duty commercial fabrics rated far higher. Flammability is governed by regulation, not preference: California TB 117-2013 in the US, TB 133 for public occupancy seating.

## How it works

1. **Frame.** Kiln-dried hardwood at 7–9% moisture content, or steel or molded plastic in production furniture. Corner blocks are glued and screwed, not just nailed — a racking frame destroys everything built on top of it.
2. **Suspension.** Interwoven jute webbing roughly 3-1/2 in (90 mm) wide stretched with a web stretcher, sinuous ("no-sag") springs at about 4–5 in (100–125 mm) centers with a crown of about 1 in per foot of span, or coil springs hand-tied eight ways in traditional work. This layer, not the foam, determines how the seat feels over years of use.
3. **Foundation.** Burlap or a decking fabric over the suspension to stop the padding from being pushed down between the springs or webbing.
4. **Padding.** Polyurethane foam cut to shape, often with a firmer core and softer topper, wrapped in polyester wadding to soften the edges and stop the cover from abrading directly against the foam.
5. **Cover.** Panels are cut — allowing extra yardage for pattern matching — sewn with bonded nylon or polyester thread at roughly 6–8 stitches per inch, then pulled over the padding and fixed with fine-wire staples at 1–1-1/2 in (25–38 mm) spacing into the frame.
6. **Trim and finish.** Welt, piping, buttoning, gimp or a dust cover closes out the visible edges.

Cut-and-sew is the production version of the same sequence: CNC fabric cutting from nested patterns, sewing cells, and foam pouring or cutting to a fixed profile, with the assembly still done by hand.

## Design guidelines

### Specify foam by density and IFD separately

Density (lb/ft³ or kg/m³) predicts durability — how many cycles the foam survives before it loses height. IFD (lbf at 25% deflection) predicts firmness. They are independent properties, and specifying only "firm foam" produces a cushion that feels right in the showroom and collapses in eighteen months. Seat cushions typically use 1.8–2.8 lb/ft³ (29–45 kg/m³) at 25–50 lbf IFD; back cushions use lower values of both.

### Match fabric abrasion rating to use

Specify by test result, not by appearance. Roughly 15,000 Wyzenbeek double rubs or 20,000 Martindale cycles is the accepted minimum for general contract upholstery; residential light-use fabrics sit below that and heavy commercial fabrics well above. Loose weaves and long floats also snag regardless of their abrasion number, which the test does not capture.

### Design the suspension for the load path

The suspension carries the load; the foam only distributes it. Sinuous springs at 4–5 in (100–125 mm) centers with a crown of about 1 in per foot of span, or properly stretched webbing, are what keep a seat from bottoming out. Adding foam to fix a sagging seat treats the symptom.

### Allow for the cover

Cut-and-sew consumes more material than the finished area suggests: a 1/2 in (13 mm) seam allowance on every seam, plus 15–30% extra yardage where a pattern must be matched across panels, plus enough pull-in at the staple line to tension the cover.

### Verify against the regulation that applies

Flammability requirements are jurisdictional and non-negotiable: California TB 117-2013 covers smolder resistance of cover fabrics and barrier materials in the US, TB 133 applies to public-occupancy seating, and the UK Furniture and Furnishings (Fire) (Safety) Regulations govern in Britain. Confirm which one applies before specifying either fabric or foam, since a compliant construction usually depends on a barrier layer as well as the cover.

### Inspection

Seam strength and stitch quality on samples, foam density and IFD verified against the supplier's certification, and cushion fatigue testing to ASTM D3574 constant-force pounding for foam durability. For commercial and institutional seating, BIFMA X5.4 covers lounge and public seating with a defined sequence of drop, cyclic load and stability tests, and EN 1728 is the European equivalent. Visual inspection covers seam alignment, pattern match, cover tension and staple spacing.

## Cost drivers

Upholstery is labor-dominated at every volume. Traditional hand upholstery over a sprung frame takes hours per seat and cannot be meaningfully mechanized; even in production cut-and-sew, the assembly step remains hand work. That is why upholstered furniture costs what it does.

Material cost is dominated by cover fabric, and it is driven as much by yield as by price per yard. A patterned fabric that must be matched across panels can consume 15–30% more than a plain one for the same seat, and railroaded versus up-the-roll orientation changes the nest entirely.

Foam and suspension are comparatively inexpensive and are exactly where specification cuts get made — usually to the long-term detriment of the product, since foam density and spring quality determine how the piece performs after a few years rather than on delivery day.

1. **Design panels to nest.** Pattern layout that yields well from a standard roll width is worth more than a small discount on the fabric.
2. **Specify plain or non-directional fabrics** where the design permits and recover the 15–30% pattern-matching allowance.
3. **Make cushions removable and re-coverable.** Serviceability extends product life and turns a failure into a repair.
4. **Standardize foam profiles** across a range so cutting or molding is shared between models.
5. **Do not economize on suspension or foam density.** Both are small parts of the bill of materials and large parts of how the product is judged after two years.

## FAQ

### How do I specify upholstery foam?

By density and IFD separately. Density — typically 1.8–2.8 lb/ft³ (29–45 kg/m³) for seat cushions — predicts how long the foam holds its height. IFD, roughly 25–50 lbf at 25% deflection for seating, predicts how firm it feels. They are independent, so specifying only firmness gives a cushion that feels correct at delivery and collapses within a couple of years.

### What abrasion rating does upholstery fabric need?

Around 15,000 double rubs on the Wyzenbeek test or 20,000 cycles on Martindale is the accepted minimum for general contract upholstery, with heavy commercial fabrics rated substantially higher and residential light-duty fabrics below. Note that abrasion testing does not capture snagging, so loose weaves and long floats can fail in service despite a good rating.

### Why does a seat sag even with good foam?

Because the suspension carries the load and the foam only distributes it. Stretched webbing, sinuous springs at 4–5 in (100–125 mm) centers with about 1 in of crown per foot of span, or hand-tied coils are what stop a seat bottoming out. Adding or upgrading foam to correct a sagging seat treats the symptom rather than the cause.

### How much extra fabric does pattern matching cost?

Typically 15–30% more yardage than an unmatched plain fabric for the same piece, because panels must be cut on repeat rather than nested for yield. Seam allowances of about 1/2 in (13 mm) and the pull-in needed to tension the cover at the staple line add further to the difference between finished area and purchased yardage.

### Which flammability standard applies to upholstered furniture?

It depends on jurisdiction and use. In the US, California TB 117-2013 covers smolder resistance of cover fabric and barrier materials and is widely adopted, while TB 133 applies to public-occupancy seating. In the UK the Furniture and Furnishings (Fire) (Safety) Regulations apply. Compliance is usually achieved by a combination of cover and barrier layer, so confirm the requirement before specifying either.

## Alternative processes

- [Weaving](https://manufacturingprocesses.org/processes/joining/weaving.md): Weaving interlaces cane, rattan or synthetic fiber over a frame to build a structural surface out of flexible material.
- [Reaction Injection Molding](https://manufacturingprocesses.org/processes/forming/reaction-injection-molding.md): Reaction injection molding mixes two liquid reactants that polymerize inside a low-pressure mold, producing large tough polyurethane parts on light tooling.
- [Injection Molding](https://manufacturingprocesses.org/processes/forming/injection-molding.md): Injection molding forces molten thermoplastic into a steel mold under high pressure, producing high-volume identical plastic parts.

## Related processes

- [Joinery](https://manufacturingprocesses.org/processes/joining/joinery.md): Joinery cuts interlocking geometry into timber so the parts locate and hold each other, with or without adhesive or fasteners.
- [Steam Bending](https://manufacturingprocesses.org/processes/forming/steam-bending.md): Steam bending plasticises solid timber with saturated steam so it can be bent around a former and held until it dries into shape.
- [Wood Laminating](https://manufacturingprocesses.org/processes/forming/wood-laminating.md): Wood laminating glues thin veneers or kerfed strips over a form so the assembly holds a curve permanently once the adhesive has cured.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/joining/upholstery)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Vibration Welding"
category: "Joining"
subcategory: "Thermal"
materials: ["Plastic"]
tolerances: "Meltdown controlled to about ±0.004 in (±0.1 mm); final assembly height set by a positive stop or machine collapse control rather than by molded part height"
volumes: "5,000 to 5,000,000+ per year"
lead_time: "6–20 s cycle including 1–10 s of vibration and a 2–6 s hold; 3–8 weeks for matched nest tooling"
url: https://manufacturingprocesses.org/processes/joining/vibration-welding
---

# Vibration Welding

Vibration welding rubs two thermoplastic parts together in-plane at low frequency under pressure until the interface melts and fuses.

- **Category**: [Joining](https://manufacturingprocesses.org/processes/joining.md)
- **Family**: Thermal
- **Materials**: Plastic
- **Typical tolerances**: Meltdown controlled to about ±0.004 in (±0.1 mm); final assembly height set by a positive stop or machine collapse control rather than by molded part height
- **Typical volumes**: 5,000 to 5,000,000+ per year
- **Lead time**: 6–20 s cycle including 1–10 s of vibration and a 2–6 s hold; 3–8 weeks for matched nest tooling

## Overview

Vibration welding rubs two thermoplastic parts against each other in-plane, at 100–250 Hz with 0.03–0.16 in (0.7–4 mm) amplitude, under 100–300 psi (0.7–2 MPa) of joint pressure, until the interface melts and the parts are clamped together to fuse.

It exists to do what ultrasonic welding cannot: weld large parts, semicrystalline resins and heavily glass-filled grades, all reliably. Machines handle joints up to roughly 24 x 40 in (600 x 1,000 mm), which is why intake manifolds, instrument panels, fluid reservoirs, bumpers and door modules are welded this way.

The joint is hermetic and, in unfilled resins, approaches base-resin strength. Cycle time is 6–20 s including 1–10 s of actual vibration and a 2–6 s hold — slower than ultrasonic, but the process is far more tolerant of part variation, contamination and joint geometry.

## How it works

1. **Clamp.** The lower part is fixtured rigidly; the upper part is held in a vibrating tool head. Joint pressure of 100–300 psi (0.7–2 MPa) is applied across the joint area — total force scales with joint area, which is why large parts need large machines.
2. **Vibrate.** Electromagnetic drivers oscillate the upper part in-plane at 100–250 Hz. Lower frequencies run larger amplitudes (up to about 0.16 in / 4 mm) and higher frequencies smaller ones (0.03 in / 0.7 mm). Friction at the interface generates the melt.
3. **Melt and steady-state flow.** After an initial dry-friction phase the interface reaches a steady melt film that is continuously extruded outward as flash while fresh polymer melts behind it. Total meltdown (collapse) is typically 0.04–0.10 in (1–2.5 mm) and is the primary process variable.
4. **Stop and hold.** Vibration halts, the parts align to their final position, and clamp pressure is held for 2–6 s while the joint solidifies. Solidifying under pressure is what makes the joint hermetic.

### Linear versus orbital

**Linear vibration welding** moves the part back and forth along one axis. The joint has to lie in a single plane, or on a shallow curve, so the relative motion stays tangential to the joint face everywhere.

**Orbital vibration welding** moves the upper part in a small circular orbit, so every point on the joint sees the same relative velocity in every direction. That removes the single-plane constraint and suits irregular and non-planar joint lines, at the cost of more complex machinery.

### Why it tolerates what ultrasonics will not

The heat comes from bulk relative motion, not from transmitting vibration *through* the part, so part stiffness, resin crystallinity and glass content barely matter. Glass-filled grades up to about 50% weld successfully, though the weld line itself is weaker than the base material because the fibers reorient parallel to the flow and no longer bridge the joint.

## Design guidelines

### The joint must lie in one plane

This is the fundamental constraint for linear welding. Design the parting line of the assembly as a single flat plane, or a curve shallow enough that motion stays tangential everywhere along it. If the joint has to be three-dimensional, specify orbital welding instead — do not try to force a linear machine onto a saddle-shaped joint.

### Joint width, flash traps and clearance

Make the joint land at least as wide as the nominal wall, and 1.5–2x wall where the joint is structural or must seal. Molten polymer is extruded outward during the weld, so design **flash traps** on both sides of the joint — a recessed channel roughly the volume of the displaced melt — unless flash on the finished part is acceptable. Provide clearance around the joint so the moving part can travel its full amplitude without touching adjacent features: at least the peak-to-peak amplitude plus 0.04 in (1 mm).

### Design in the meltdown

The parts get shorter by the meltdown distance, typically 0.04–0.10 in (1–2.5 mm). Dimension the assembly so its critical height is controlled by a positive stop or by machine collapse control, not by the nominal molded height of the two halves.

### Support the joint against clamp force

Joint pressure of 100–300 psi (0.7–2 MPa) over a large joint is a large total force, and unsupported walls will bow. Design internal ribs or a supporting nest so the joint plane stays flat under load; a joint that bows open mid-weld will not seal.

### Resin and fill compatibility

| Pair | Weldable? | Notes |
| --- | --- | --- |
| Same resin, unfilled | Excellent | Joint approaches base-resin strength, hermetic |
| Same resin, up to 30% glass | Good | Weld line weaker than base material — fibers reorient and stop bridging |
| Same resin, 30–50% glass | Fair | Welds, but size the joint on the weld-line strength, not the datasheet value |
| PA (nylon), PP, PE, PBT, POM | Good | Semicrystalline resins that ultrasonic welding struggles with |
| PC, ABS, PC/ABS | Good | Also weld well ultrasonically — choose on part size |
| Dissimilar resins | Generally no | Needs chemical compatibility and close melt temperature; validate before committing |
| Nylon with absorbed moisture | Poor | Water foams at the melt line; dry the parts before welding |

Joint lands, flash traps and rib supports are molded features — wall thickness and rib rules are on the [injection molding design guidelines chart](/charts/injection-molding-design-guidelines).

### Inspection

Meltdown distance is recorded for every cycle and is the primary process signal; a shallow meltdown means the joint did not reach steady-state melt. Beyond that: burst pressure and pressure-decay leak testing on sealed parts, destructive cross-sections through the weld line to confirm melt across the full joint width, and tensile coupons cut from welded parts for strength qualification.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Joint plane | One flat plane (linear) | Shallow curve | Motion must stay tangential to the joint face |
| Joint land width | 1.5–2x nominal wall | 1x wall | Sets weld area and seal reliability |
| Meltdown allowance | 0.06 in (1.5 mm) | 0.04–0.10 in (1–2.5 mm) | Assembly height must be dimensioned around it |
| Flash trap | Both sides of the joint | — | Displaced melt has to go somewhere |
| Motion clearance around joint | Amplitude + 0.04 in (1 mm) | — | The moving half must not strike adjacent features |
| Glass fill at the joint | Under 30% | 50% | Weld line strength falls as fiber content rises |
| Joint support under clamp | Ribbed or nested | — | 100–300 psi (0.7–2 MPa) will bow an unsupported wall |

## Cost drivers

Vibration welders are substantially larger and more expensive machines than ultrasonic welders, because they must react the vibrating mass and the full clamp force across a large joint. Machine size scales with joint area, and so does capital cost — this is the dominant economic factor and it is set by part size, not by volume.

Tooling is a pair of matched nests, typically machined aluminum, which cost more than ultrasonic nests because they carry the clamp load. Cycle time of 6–20 s is longer than ultrasonic but still short next to adhesive cure, and there are no consumables at all.

The process wins where the alternative is bonding a large assembly (cure time, fixturing, adhesive cost) or fastening it (leak paths, part count, assembly labor). It loses to ultrasonic welding on small parts, where a smaller machine and a 2 s cycle do the same job.

1. **Keep the joint in one plane** — an orbital machine costs more than a linear one and there is rarely a functional reason to need it.
2. **Size the joint to the load, not to the perimeter.** Total clamp force scales with joint area, and machine cost scales with force.
3. **Design flash traps in** rather than adding a deflashing operation after welding.
4. **Consolidate parts into the welded assembly.** The economics improve every time a weld deletes a bracket, gasket or fastener stack.
5. **Specify unfilled resin at the joint** where the part allows it; weld-line strength in a 50% glass grade is a fraction of the datasheet number.

## FAQ

### When should I use vibration welding instead of ultrasonic welding?

When the part is large, the resin is semicrystalline, or the grade is heavily glass filled. Vibration welding handles joints up to roughly 24 x 40 in (600 x 1,000 mm) and welds PP, PA, PBT and 30–50% glass grades reliably, because the heat comes from bulk relative motion rather than from transmitting vibration through the part. Ultrasonic wins on small parts with a 2 s cycle.

### How much do the parts shrink during a vibration weld?

Meltdown, or collapse, is typically 0.04–0.10 in (1–2.5 mm) of combined height. That has to be designed into the assembly: dimension the critical height off a positive stop or the machine's collapse control rather than off the nominal molded heights of the two halves.

### Does the joint have to be flat?

For linear vibration welding, effectively yes — the joint must lie in one plane or on a curve shallow enough that the relative motion stays tangential everywhere along it. Orbital vibration welding moves the part in a small circle so every point sees the same relative velocity, which removes the constraint and suits three-dimensional joint lines, at higher machine cost.

### Can vibration welding produce a hermetic seal?

Yes — hermetic sealing is one of its main uses, in fluid reservoirs, filter housings and intake manifolds. The seal comes from holding clamp pressure for 2–6 s after vibration stops so the melt solidifies under load. Verify with burst pressure or pressure-decay leak testing rather than assuming it.

### How does glass fill affect vibration weld strength?

It reduces it, and the reduction grows with fill level. During the weld the fibers reorient parallel to the melt flow and stop bridging the joint, so the weld line is weaker than the bulk material even though the process runs fine up to about 50% glass. Size structural joints on measured weld-line strength, not the resin datasheet.

## Alternative processes

- [Hot Plate Welding](https://manufacturingprocesses.org/processes/joining/hot-plate-welding.md): Hot plate welding presses two thermoplastic parts against a heated platen to melt the joint faces, then clamps them together to fuse.
- [Adhesive Bonding](https://manufacturingprocesses.org/processes/joining/adhesive-bonding.md): Adhesive bonding joins parts with a cured polymer layer that spreads load over the whole bond area instead of concentrating it at fastener points.
- [Mechanical Fastening](https://manufacturingprocesses.org/processes/joining/mechanical-fastening.md): Mechanical fastening joins parts with screws, inserts, clinch hardware or snap fits so the joint can be assembled and taken apart again.
- [Riveting](https://manufacturingprocesses.org/processes/joining/riveting.md): Riveting sets a rivet through aligned holes and upsets its tail so the joint clamps permanently, with no heat and no access to both sides for blind types.

## Related processes

- [Ultrasonic Welding](https://manufacturingprocesses.org/processes/joining/ultrasonic-welding.md): Ultrasonic welding applies high-frequency vibration through a horn to melt a thermoplastic joint locally in well under a second.
- [Injection Molding](https://manufacturingprocesses.org/processes/forming/injection-molding.md): Injection molding forces molten thermoplastic into a steel mold under high pressure, producing high-volume identical plastic parts.
- [Staking](https://manufacturingprocesses.org/processes/joining/staking.md): Heat staking softens a molded plastic boss with hot air or ultrasonic energy and reforms it into a head that traps the mating part.
- [Laser Plastic Welding](https://manufacturingprocesses.org/processes/joining/laser-plastic-welding.md): Laser plastic welding passes a beam through a transmissive part into an absorbing one, melting only the interface between them.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/joining/vibration-welding)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Weaving"
category: "Joining"
subcategory: "Mechanical"
materials: ["Wood", "Composite"]
volumes: "1 to a few thousand — hand weaving for one-offs and restoration, pre-woven webbing for production"
lead_time: "Hours per seat for hand weaving; pre-woven webbing is routed, splined and trimmed in well under an hour"
url: https://manufacturingprocesses.org/processes/joining/weaving
---

# Weaving

Weaving interlaces cane, rattan or synthetic fiber over a frame to build a structural surface out of flexible material.

- **Category**: [Joining](https://manufacturingprocesses.org/processes/joining.md)
- **Family**: Mechanical
- **Materials**: Wood, Composite
- **Typical volumes**: 1 to a few thousand — hand weaving for one-offs and restoration, pre-woven webbing for production
- **Lead time**: Hours per seat for hand weaving; pre-woven webbing is routed, splined and trimmed in well under an hour

## Overview

In furniture and structural work, weaving interlaces cane, rattan, willow, rush, paper cord or synthetic strap over or through a frame to build a surface out of flexible material. The woven surface carries load as a tensioned membrane, which means the frame has to resist the inward pull the weave applies to its rails — a detail that is frequently missed and shows up as a bowed seat rail years later.

Two production routes exist. Pre-woven cane webbing is machine woven in rolls, cut to shape, soaked, pressed into a routed groove around the opening and locked with a reed spline and adhesive. Hand caning threads individual strands through a ring of holes drilled around the seat opening — typically 1/8–3/16 in (3–5 mm) diameter at 1/2–5/8 in (13–16 mm) centers — in the traditional six-way pattern.

Natural cane is soaked 10–30 minutes before working and tightens as it dries. Synthetic polyethylene strap, woven over powder-coated aluminum frames, gives the same construction UV and moisture resistance for outdoor use.

## How it works

1. **Prepare the frame.** For pre-woven webbing, rout a continuous groove around the opening sized to the spline. For hand caning, drill the hole pattern around the opening at even spacing. For strap and cord seats, fit the rails with L-nails, tacks or slots to anchor the material.
2. **Condition the material.** Natural cane, rattan and rush are soaked for 10–30 minutes so they become pliable. Cane worked dry cracks; cane worked wet is easy to bend and tightens as it dries, which is what puts tension into the finished seat.
3. **Weave.** Hand caning follows a defined sequence of passes — front to back, side to side, a second layer of each, then two diagonals — that produces the familiar octagonal pattern and locks each strand against its neighbors. Danish paper cord is wound in a continuous figure over L-nails set into the inner rails. Synthetic strap is woven over and under in a simple plain weave.
4. **Tension and lock.** Pre-woven webbing is wedged into its groove with a reed spline over adhesive. Hand-caned seats are pegged at each hole. Cord and strap are terminated by wrapping and tucking under previous passes.
5. **Dry.** The seat is left to dry fully, during which natural materials shrink and the weave pulls taut.

## Design guidelines

### Design the frame to resist the weave

A woven seat is a tensioned membrane, and it pulls its supporting rails inward continuously for the life of the piece. Size the rails for that sustained load, brace or corner-block the frame, and avoid long unsupported spans between the anchor points — a rail that bows inward slackens the weave, which then accelerates the sagging.

### Match the material to the environment

Natural cane, rush and rattan are interior materials. They embrittle in sunlight, absorb moisture, and stain. Where a woven seat has to live outdoors, specify extruded polyethylene or polypropylene strap over an aluminum or coated steel frame, which weaves the same way and holds up under UV and rain.

### Space the anchors evenly

For hand caning, hole diameter and spacing must suit the cane width — commonly 1/8–3/16 in (3–5 mm) holes at 1/2–5/8 in (13–16 mm) centers. Uneven spacing produces a visibly irregular weave that no amount of tension will correct, and holes drilled too close to the inner edge of the rail break out.

### Design for replacement

A woven seat is a wear item. Pre-woven webbing in a spline groove can be cut out and replaced in an afternoon; a hand-caned seat can be re-caned through the same holes; a strap seat can be rewoven. That serviceability is a genuine advantage over a molded or bonded seat and is worth protecting in the detailing — do not glue a woven surface into a joint that has to be destroyed to open it.

### Inspection

Visual assessment of weave regularity, tension and termination quality is the primary method, since the construction is fundamentally a hand process. Moisture content of natural material at the time of weaving matters, since it determines the final tension. For commercial and contract seating, cyclic load testing to BIFMA X5.4 or EN 1728 applies to the finished chair regardless of how the seat surface is made, and the woven surface is usually what governs the result.

## Cost drivers

Hand weaving is skilled labor measured in hours per seat, and it does not scale — the cost per unit is essentially flat from the first piece to the thousandth. That is the single dominant fact in the economics.

Pre-woven cane webbing changes the equation completely. The weaving itself is done on a machine and sold by the roll; the shop operation reduces to routing a groove, cutting the webbing, soaking, splining and trimming. It looks similar at a glance, costs a fraction, and is the reason almost all production cane-seat furniture uses webbing rather than hand caning.

Material cost is modest in both cases. The genuine cost risk is rework: an irregular or under-tensioned weave usually has to be cut out and redone rather than corrected.

1. **Use pre-woven webbing** wherever the design does not specifically require hand caning through drilled holes.
2. **Standardize the opening geometry** across a product range so one webbing width and one spline size serve every model.
3. **Design for replacement** so a worn seat is a service item rather than a scrapped chair.
4. **Specify synthetic strap for outdoor products** — the labor is the same and the field life is far longer.
5. **Control the moisture content at weaving.** Material worked too dry cracks, and material worked too wet slackens as it shrinks past the intended tension.

## FAQ

### What is the difference between hand caning and pre-woven cane webbing?

Hand caning threads individual strands through a ring of holes drilled around the seat opening — typically 1/8–3/16 in (3–5 mm) at 1/2–5/8 in (13–16 mm) centers — in a six-way pattern, and takes hours per seat. Pre-woven webbing is machine woven by the roll, then soaked and pressed into a routed groove and locked with a reed spline. Almost all production furniture uses webbing.

### Why is cane soaked before weaving?

Soaking for 10–30 minutes makes the cane pliable enough to bend and pull tight without cracking. It also sets the final tension: cane is woven damp and shrinks as it dries, so the seat tightens itself. Working cane dry breaks strands, and working it excessively wet leaves the finished weave slack once it has shrunk past the intended tension.

### Does a woven seat need a stronger frame?

Yes. The woven surface behaves as a tensioned membrane and pulls the supporting rails inward for the life of the piece. Size the rails for that sustained load, corner-block or brace the frame, and avoid long unsupported spans between anchor points — a rail that bows inward slackens the weave, which then sags faster.

### Can woven seating be used outdoors?

Natural cane, rush and rattan should not be — they embrittle under UV, absorb moisture and stain. Extruded polyethylene or polypropylene strap woven over a powder-coated aluminum or steel frame gives the same construction and appearance with genuine weather resistance, and it is what outdoor woven furniture actually uses.

### Can a woven seat be repaired?

Yes, and that is one of its real advantages. Pre-woven webbing is cut out of its groove and replaced with new webbing and spline; a hand-caned seat is re-caned through the existing holes; a strap or cord seat is rewoven over the same anchors. Detail the frame so the woven surface can be removed without destroying a glued joint.

## Alternative processes

- [Composite Laminating](https://manufacturingprocesses.org/processes/forming/composite-laminating.md): Composite laminating lays reinforcing fiber into a mold and consolidates it with resin, curing into a stiff, light monocoque part.
- [Injection Molding](https://manufacturingprocesses.org/processes/forming/injection-molding.md): Injection molding forces molten thermoplastic into a steel mold under high pressure, producing high-volume identical plastic parts.
- [Filament Winding](https://manufacturingprocesses.org/processes/forming/filament-winding.md): Filament winding wraps resin-impregnated continuous fiber onto a rotating mandrel in a controlled pattern, producing tubes and pressure vessels.

## Related processes

- [Steam Bending](https://manufacturingprocesses.org/processes/forming/steam-bending.md): Steam bending plasticises solid timber with saturated steam so it can be bent around a former and held until it dries into shape.
- [Joinery](https://manufacturingprocesses.org/processes/joining/joinery.md): Joinery cuts interlocking geometry into timber so the parts locate and hold each other, with or without adhesive or fasteners.
- [Wood Laminating](https://manufacturingprocesses.org/processes/forming/wood-laminating.md): Wood laminating glues thin veneers or kerfed strips over a form so the assembly holds a curve permanently once the adhesive has cured.
- [Upholstery](https://manufacturingprocesses.org/processes/joining/upholstery.md): Upholstery builds a padded surface from suspension, foam, wadding and textile over a frame, fixed with staples, ties and stitching.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/joining/weaving)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: category
name: "Finishing"
processes: 25
url: https://manufacturingprocesses.org/processes/finishing
---

# Finishing

25 manufacturing processes in the finishing family.

Finishing processes change the surface rather than the bulk: appearance,
corrosion resistance, hardness, friction, or dimensional fine-tuning.

Finishing is where tolerances get consumed. Plating adds thickness,
anodizing both adds and eats into the substrate, and blasting rounds
edges, so the finish has to be chosen before the drawing is dimensioned.

## Processes

| Process | Tolerances | Typical volumes | Lead time |
| --- | --- | --- | --- |
| [Abrasive Blasting](https://manufacturingprocesses.org/processes/finishing/abrasive-blasting.md) | Cleaning passes remove very little, but aggressive or repeated blasting removes measurable stock; mask any toleranced feature. Anchor profile is specified at 1.5–4.0 mils (38–100 µm) for coating work | 1 part to continuous automated wheel-blast lines | Same day to 3 business days at a job shop; on-site structural work is scheduled by area and containment requirements |
| [Anodizing](https://manufacturingprocesses.org/processes/finishing/anodizing.md) | Type II adds 0.0002–0.001 in (5–25 µm) total, roughly half of it outward; Type III adds 0.0005–0.004 in (13–100 µm). Budget the full coating thickness on a diameter and 4× the radial buildup on thread pitch diameter | 1 to millions of parts; racked in batches and priced by area and rack space | 3–7 business days at a job shop; 1–2 days expedited; add a week for a first-article custom color match |
| [Bead Blasting](https://manufacturingprocesses.org/processes/finishing/bead-blasting.md) | Removes under 0.0005 in (13 µm) — generally treated as dimensionally neutral, but toleranced bores, threads and sealing faces should still be masked | 1 to millions of parts; automated cabinets are used above a few hundred pieces for cosmetic consistency | Same day to 3 business days at a job shop; usually run in line with anodizing |
| [Black Oxide](https://manufacturingprocesses.org/processes/finishing/black-oxide.md) | Adds roughly 0.00003–0.00005 in (0.75–1.25 µm) — treated as dimensionally neutral; no allowance needed on threads, gage surfaces or press fits | 1 to millions of parts; small hardware runs in bulk baskets | 1–5 business days at a job shop; often same or next day for small hardware lots |
| [CNC Engraving](https://manufacturingprocesses.org/processes/finishing/cnc-engraving.md) | Depth control of ±0.001–0.002 in (±0.025–0.05 mm) is routine; stroke width with a conical tool varies directly with depth, so both must be specified together | 1 to tens of thousands of parts; cost per part is dominated by cycle time above a few hundred | 1–5 business days; no tooling lead time beyond stock cutters |
| [Cerakote Ceramic Coating](https://manufacturingprocesses.org/processes/finishing/cerakote-ceramic-coating.md) | Adds about 0.001 in (25 µm) per coated surface; budget 2× on a bore diameter and roughly 4× on thread pitch diameter, or mask | 1 part to a few thousand; hand-sprayed, so it does not scale like powder or e-coat | 1–3 weeks at a specialist applicator; oven cure itself is about 2 hours |
| [Chemical Conversion Coating](https://manufacturingprocesses.org/processes/finishing/chemical-conversion-coating.md) | Chem film on aluminum adds 0.00001–0.00004 in (0.25–1 µm) and needs no allowance; manganese phosphate on steel can reach 0.0002–0.0004 in (5–10 µm) and should be checked on close fits | 1 to millions of parts; immersion lines handle bulk baskets and racked work alike | 1–5 business days at a job shop; commonly same-day when run in line with cleaning |
| [E-Coating (Electrophoretic Deposition)](https://manufacturingprocesses.org/processes/finishing/e-coating-electrophoretic-deposition.md) | Adds 0.6–1.2 mils (15–30 µm) per surface, uniform over the whole part including recesses; budget roughly 4× the film thickness on thread pitch diameter | Economical from a few thousand parts per run upward; automotive lines run millions | 3–10 business days at a job coater; production lines run continuously once qualified |
| [Electroplating](https://manufacturingprocesses.org/processes/finishing/electroplating.md) | Zinc 0.0002–0.001 in (5–25 µm), electroless nickel 0.0005–0.002 in (13–50 µm), hard chrome 0.0002–0.010 in (5–250 µm) per surface. Electrolytic deposits vary 2:1 or more between high- and low-current areas; electroless nickel holds about ±10% | Barrel plating is economical from a few pounds of hardware to millions of pieces; rack plating from 1 part upward | 3–10 business days at a job shop; add 1 day for an embrittlement-relief bake |
| [Electropolishing](https://manufacturingprocesses.org/processes/finishing/electropolishing.md) | Removes 0.0002–0.001 in (5–25 µm) per surface, so bores grow and shafts shrink by twice that on diameter; sharp external edges break by 0.002–0.003 in (50–75 µm) | 1 part to high-volume barrel work; small parts can be run in bulk baskets | 3–7 business days at a job shop; 1–2 days expedited |
| [Foil Blocking and Embossing](https://manufacturingprocesses.org/processes/finishing/foil-blocking-and-embossing.md) | Transferred foil layer is sub-micron and dimensionally negligible; image registration typically ±0.010–0.020 in (±0.25–0.5 mm), and exact between foil and emboss when a combination die is used | A few hundred to millions of impressions; die cost amortizes quickly at medium volume | 1–3 weeks including die manufacture; 3–7 business days for a repeat run on existing tooling |
| [Galvanizing](https://manufacturingprocesses.org/processes/finishing/galvanizing.md) | Hot dip adds 1.8–3.9 mils (45–100 µm) per surface per ASTM A123, so holes close by 3.6–7.8 mils on diameter; electrogalvanizing adds 0.0002–0.001 in (5–25 µm) and is far more predictable | One-off fabrications to continuous coil; priced by weight, so heavy structural work is the most economical | 2–10 business days at a commercial galvanizer, driven by kettle scheduling rather than processing time |
| [Grinding, Sanding and Polishing](https://manufacturingprocesses.org/processes/finishing/grinding-sanding-and-polishing.md) | Surface and cylindrical grinding ±0.0001–0.0005 in (±0.0025–0.013 mm); honing corrects bore geometry within a few tenths; lapping holds flatness to a few millionths of an inch | 1 to millions of parts; centerless and production grinding lines run continuously | 1–10 business days at a job shop depending on setup complexity; production grinding runs in seconds to minutes per part |
| [Hydro Transfer Printing](https://manufacturingprocesses.org/processes/finishing/hydro-transfer-printing.md) | The complete system adds 0.003–0.006 in (75–150 µm) per surface, dominated by the base and clear coats; the printed layer itself is negligible | 1 to tens of thousands of parts; manual lines for low volume, automated dipping for automotive trim | 1–3 weeks at a job shop, driven by the spray and cure cycles rather than the dip |
| [Laser Marking and Engraving](https://manufacturingprocesses.org/processes/finishing/laser-marking-and-engraving.md) | Annealed marks remove no measurable material; etching removes 0.0001–0.001 in (2.5–25 µm); engraving removes 0.001–0.020 in (0.025–0.5 mm). Beam positioning within the marking field is typically within a few thousandths of an inch | 1 to millions of parts; variable data such as serial numbers costs the same as static artwork | Same day to 5 business days; artwork is a file, so there is no tooling lead time |
| [PVD Coating](https://manufacturingprocesses.org/processes/finishing/pvd-coating.md) | Adds 0.5–5 µm (0.00002–0.0002 in) per surface — dimensionally negligible for most fits; thicker tooling coatings at 4–5 µm should be checked on close-fitting punch and die clearances | 1 part to high volume; cost per part is governed by how densely the chamber can be loaded | 3–10 business days at a coating service; cycle time itself is 2–6 hours |
| [Pad Printing](https://manufacturingprocesses.org/processes/finishing/pad-printing.md) | Image position typically within ±0.005 in (±0.13 mm); the transferred ink film is a few microns and adds nothing dimensionally | A few hundred to millions of parts; tooling cost is low enough for short runs | 1–2 weeks including cliché and fixture; 2–5 business days for a repeat order |
| [Passivation](https://manufacturingprocesses.org/processes/finishing/passivation.md) | No measurable dimensional change; the passive film is a few nanometers thick | 1 part to millions; small parts run in bulk baskets at very low cost per piece | 1–5 business days at a job shop; often same-day when combined with an existing cleaning operation |
| [Photo Etching](https://manufacturingprocesses.org/processes/finishing/photo-etching.md) | ±10–20% of material thickness is standard; approximately ±0.001 in (±0.025 mm) achievable on thin foil. Half-etch depth is controlled to a residual thickness rather than to a depth dimension | 1 to several hundred thousand parts; the same tooling serves prototype and production | 3–10 business days including phototool; prototypes in 1–3 days at some suppliers |
| [Powder Coating](https://manufacturingprocesses.org/processes/finishing/powder-coating.md) | Adds 2–4 mils (50–100 µm) per surface; budget 2× film thickness on a bore diameter and roughly 4× on thread pitch diameter, or mask | 1 to millions of parts; conveyorized lines are most economical above a few hundred parts per run | 3–7 business days at a job shop; 24–48 hours expedited on stock colors; 2–3 weeks for a custom-matched powder |
| [Screen Printing](https://manufacturingprocesses.org/processes/finishing/screen-printing.md) | Registration typically ±0.005–0.010 in (±0.13–0.25 mm); dried ink film 0.0004–0.002 in (10–50 µm), which is enough to matter on a stacked assembly | A hundred to millions of impressions; setup per color makes very short runs uneconomical | 1–2 weeks including screens and setup; 3–7 business days for repeat orders |
| [Spray Painting](https://manufacturingprocesses.org/processes/finishing/spray-painting.md) | A complete system adds 4–6 mils (100–150 µm) per coated surface; budget 2× on a bore diameter and roughly 4× on thread pitch diameter, or mask | 1 part to continuous automotive lines; the most viable option at volume one | 2–10 business days at a job shop depending on the number of coats and cure schedule; field work is scheduled by weather window |
| [UV Inkjet Printing](https://manufacturingprocesses.org/processes/finishing/uv-inkjet-printing.md) | Image placement typically within ±0.010–0.020 in (±0.25–0.5 mm) relative to the substrate datum; printed film 0.0004–0.0008 in (10–20 µm) for a standard CMYK build | 1 to a few thousand impressions; variable data and one-offs cost the same as repeats | Same day to 5 business days; no tooling, so there is no plate or screen lead time |
| [Vacuum Metalizing](https://manufacturingprocesses.org/processes/finishing/vacuum-metalizing.md) | Metal film 0.05–0.15 µm (2–6 µin) is dimensionally negligible; the base and topcoats add roughly 0.5–1.5 mils (13–38 µm) total per coated surface | Hundreds to millions; batch chamber cycles favor high-volume decorative parts | 2–4 weeks for first articles including basecoat qualification; days per lot in production |
| [Vibratory Tumbling and Mass Finishing](https://manufacturingprocesses.org/processes/finishing/vibratory-tumbling-and-mass-finishing.md) | Removes 0.0001–0.001 in (2.5–25 µm) per surface in a typical cycle and breaks exposed edges 0.002–0.010 in (0.05–0.25 mm); the edge break is not controllable to a tolerance | Dozens to millions of parts; cost per part is essentially independent of count once the machine is full | Same day to 5 business days at a job shop; cycle time itself is 30 minutes to 4 hours for vibratory work |

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/finishing)*


<!-- ---------------- -->

---
type: process
name: "Abrasive Blasting"
category: "Finishing"
subcategory: "Subtractive"
materials: ["Metal", "Glass", "Plastic", "Ceramic"]
tolerances: "Cleaning passes remove very little, but aggressive or repeated blasting removes measurable stock; mask any toleranced feature. Anchor profile is specified at 1.5–4.0 mils (38–100 µm) for coating work"
volumes: "1 part to continuous automated wheel-blast lines"
lead_time: "Same day to 3 business days at a job shop; on-site structural work is scheduled by area and containment requirements"
url: https://manufacturingprocesses.org/processes/finishing/abrasive-blasting
---

# Abrasive Blasting

Abrasive blasting propels media at a surface with compressed air to clean it, strip scale, or produce a uniform matte texture.

- **Category**: [Finishing](https://manufacturingprocesses.org/processes/finishing.md)
- **Family**: Subtractive
- **Materials**: Metal, Glass, Plastic, Ceramic
- **Typical tolerances**: Cleaning passes remove very little, but aggressive or repeated blasting removes measurable stock; mask any toleranced feature. Anchor profile is specified at 1.5–4.0 mils (38–100 µm) for coating work
- **Surface finish**: Typically 125–500 µin (3.2–12.5 µm) Ra depending on media and pressure; specified for coating work as an anchor profile of 1.5–4.0 mils (38–100 µm)
- **Typical volumes**: 1 part to continuous automated wheel-blast lines
- **Lead time**: Same day to 3 business days at a job shop; on-site structural work is scheduled by area and containment requirements

## Overview

Abrasive blasting propels hard media at a surface with compressed air or a centrifugal wheel to strip scale, rust, old coatings and casting sand, and to leave a clean, uniformly roughened surface. It is the dominant pre-treatment for paint, powder coating and thermal spray, because coating adhesion depends on the mechanical anchor profile that blasting creates.

The specification that matters is not Ra but anchor profile — the peak-to-valley depth of the blasted surface — with 1.5–4.0 mils (38–100 µm) covering most coating requirements, and cleanliness graded against SSPC/NACE standards from brush-off (SP 7) through commercial (SP 6) and near-white (SP 10) to white metal (SP 5). Nozzle pressure typically runs 60–100 psi (4–7 bar) for production cleaning and lower for delicate work. Media choice — aluminum oxide, steel grit or shot, garnet, glass bead, ceramic, plastic, walnut shell or soda — sets both the aggressiveness and whether the process cuts or peens.

## How it works

1. **Contain and mask.** Blasting is done in a cabinet, a blast room or on site with containment. Threads, bearing bores, sealing faces, machined datums and any surface that must not lose dimension are masked with plugs, caps, rubber or steel shielding.

2. **Select the media.** Angular media (aluminum oxide, steel grit, garnet) cut and produce a sharp, deep anchor profile. Round media (steel shot, glass bead, ceramic) peen and produce a dimpled, work-hardened surface with a shallower profile. Media size controls profile depth: coarser media, deeper profile.

3. **Blast.** Media is accelerated through a nozzle by compressed air, typically 60–100 psi (4–7 bar) at the nozzle for structural cleaning, or thrown by a centrifugal wheel on automated equipment. Standoff distance and angle control cutting rate — around 45–60° for cleaning, closer to 90° for maximum profile.

4. **Verify.** Cleanliness is compared against the SSPC/NACE visual standards. Profile is measured with replica tape or a depth micrometer. Both are recorded, because a coating warranty usually depends on them.

5. **Coat promptly.** A freshly blasted steel surface is chemically active and flash-rusts within hours in humid conditions. Coating normally follows within the same shift.

### Wet and vapor blasting

Introducing water into the stream suppresses dust, eliminates the silica inhalation hazard, and produces a softer, more uniform satin because the water cushions media impact. Vapor blasting is widely used on aluminum castings and engine components where a dry blast would be too aggressive and would embed media.

### Cutting versus peening

Angular media removes material. Round media at controlled intensity is shot peening — a distinct process, verified with Almen strips, that deliberately induces compressive residual stress to improve fatigue life. Ordinary blast cleaning also imparts some compressive stress, which is why thin sheet can bow after one-sided blasting.

## Design guidelines

### Specify profile and cleanliness, not "sandblast"

A usable drawing note names the cleanliness grade (for example SSPC-SP 10 near-white), the anchor profile range (for example 2.0–3.5 mils / 50–90 µm), and the media type. Without those, two shops will produce two different surfaces and the coating supplier cannot warrant either.

### Expect Ra to land in a coarse band

Blasted surfaces are rough. Typical blast finishes fall in the 125–500 µin (3.2–12.5 µm) Ra range depending on media and pressure, with heavy grit work coarser still and fine media finer. Ra correlates only loosely with anchor profile, so specify profile for coating work and Ra only for appearance. See the [surface finish chart](/charts/surface-finish-chart) for how that band compares with machined and ground surfaces.

### Mask every toleranced or sealing feature

Blasting removes material — little on a cleaning pass, but measurable on repeated or aggressive work — and it destroys a fine finish. Threads, O-ring grooves, bearing seats, dowel holes and gasket faces must be masked or produced after blasting.

### Watch distortion on thin sections

One-sided blasting induces compressive stress in the blasted face and bows thin panels. Sheet below roughly 0.030–0.060 in (0.8–1.5 mm) is at risk. Blast both sides, reduce pressure, or use a less aggressive medium.

### Match media to substrate

Never blast stainless steel with media that has been used on carbon steel, and never with steel grit or shot, unless the part will be passivated afterward — embedded free iron causes rust spots on an otherwise sound part. Aluminum, brass and plastics embed hard media easily, so glass bead, ceramic, plastic or walnut are preferred. Cast iron and steel take aluminum oxide or steel grit.

### Plan for media breakdown and contamination

Media fractures with use and gets finer and sharper, which changes the profile over a run. Production work uses classifiers and controlled media replacement. On critical jobs, specify new or classified media.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Anchor profile for coating | 1.5–4.0 mils (38–100 µm) | Match to the coating's data sheet | Too shallow and adhesion fails; too deep and peaks poke through |
| Cleanliness grade | SSPC-SP 6 general, SP 10 for demanding service | SP 5 white metal for immersion | Residual scale and rust undercut the coating |
| Nozzle pressure | 60–100 psi (4–7 bar) | Reduce for thin or soft parts | Pressure drives profile and distortion |
| Minimum sheet thickness | 0.060 in (1.5 mm) for one-sided blasting | 0.030 in (0.8 mm) | Compressive stress bows thin panels |
| Media on stainless | Stainless, ceramic or glass only | Never carbon steel grit | Embedded free iron rusts |
| Time to coating | Same shift | Before visible flash rust | Blasted steel re-oxidizes within hours |

## Cost drivers

Blasting is priced by area, by time, or per part depending on whether the work is cabinet, automated wheel, or a blast room. It is inexpensive per square foot but scales badly with handling and masking.

- **Handling and setup.** A part that goes into a cabinet and comes out in two minutes costs almost nothing. A structural weldment that needs rigging, a blast room and containment costs by the hour.
- **Masking.** Plugs, caps and shielding are manual work, applied and removed once per part, and they are the largest adder on machined components.
- **Media choice and consumption.** Aluminum oxide and steel grit are recyclable through classifiers; single-use media such as garnet and soda is consumed per job.
- **Containment and disposal.** On-site work, lead-containing old coatings and dust control drive the cost far above the blasting itself.
- **Verification.** Replica tape profile readings and documented cleanliness grades on every lot add inspection time.

Cost-reduction tactics:

1. Specify the loosest cleanliness grade the coating system actually requires — SP 10 costs materially more than SP 6.
2. Design so that toleranced features can be machined after blasting rather than masked during it.
3. Batch parts of the same media type together to avoid media changeovers.
4. Use recyclable media on repeat production work and reserve single-use media for jobs that require it.
5. Where the requirement is only cosmetic uniformity on small parts, price vibratory finishing or bead blasting against a manual blast operation.

## FAQ

### What surface finish does abrasive blasting produce?

Typically 125–500 µin (3.2–12.5 µm) Ra depending on media and pressure, with heavy grit work coarser and fine media finer. For coating work the meaningful specification is anchor profile — the peak-to-valley depth — usually 1.5–4.0 mils (38–100 µm), measured with replica tape rather than a profilometer.

### What is the difference between SSPC-SP 6, SP 10 and SP 5?

They are cleanliness grades. SP 6 (commercial blast) allows light staining over a limited portion of the surface, SP 10 (near-white) allows less, and SP 5 (white metal) requires the surface be free of all visible residue. Cost rises steeply from SP 6 to SP 5, so specify the grade the coating system actually needs.

### Which blast media should I use?

Angular media — aluminum oxide, steel grit, garnet — cuts and creates a sharp, deep anchor profile for coating. Round media — steel shot, glass bead, ceramic — peens and gives a shallower, satin surface. Media size controls profile depth, and soft substrates such as aluminum and plastics need non-embedding media such as glass bead, plastic or walnut shell.

### Can stainless steel be abrasive blasted?

Yes, but only with media that has never touched carbon steel, and never with steel grit or shot unless the part will be passivated afterward. Embedded free iron rusts and produces surface corrosion on an otherwise sound stainless part.

### Will blasting distort thin parts?

It can. Blasting induces compressive stress in the treated face, so one-sided blasting bows thin panels — sheet below roughly 0.030–0.060 in (0.8–1.5 mm) is at risk. Blast both sides, drop the pressure, or switch to a less aggressive medium.

### How soon after blasting must a part be coated?

Within the same shift in most conditions. A freshly blasted steel surface is chemically active and flash-rusts within hours in humid air, and any visible rust bloom means the surface has to be reblasted before coating.

## Alternative processes

- [Bead Blasting](https://manufacturingprocesses.org/processes/finishing/bead-blasting.md): Bead blasting propels glass or ceramic beads at a surface to produce an even satin matte texture without cutting into it aggressively.
- [Vibratory Tumbling and Mass Finishing](https://manufacturingprocesses.org/processes/finishing/vibratory-tumbling-and-mass-finishing.md): Vibratory tumbling agitates parts with abrasive media in a bowl so edges are broken and surfaces smoothed in bulk, without hand work.
- [Grinding, Sanding and Polishing](https://manufacturingprocesses.org/processes/finishing/grinding-sanding-and-polishing.md): Grinding, sanding and polishing remove a thin layer of material with bonded or coated abrasive to correct form or improve surface finish.
- [Electropolishing](https://manufacturingprocesses.org/processes/finishing/electropolishing.md): Electropolishing reverses plating: current dissolves the surface fastest at the peaks, leaving a bright, deburred, passive finish.
- [Chemical Conversion Coating](https://manufacturingprocesses.org/processes/finishing/chemical-conversion-coating.md): Chemical conversion coating reacts the metal surface with a chemical bath to form a thin protective, conductive or paint-adhering film.

## Related processes

- [Bead Blasting](https://manufacturingprocesses.org/processes/finishing/bead-blasting.md): Bead blasting propels glass or ceramic beads at a surface to produce an even satin matte texture without cutting into it aggressively.
- [Powder Coating](https://manufacturingprocesses.org/processes/finishing/powder-coating.md): Powder coating sprays electrostatically charged dry polymer powder onto an earthed part, then bakes it into a tough continuous film.
- [Spray Painting](https://manufacturingprocesses.org/processes/finishing/spray-painting.md): Spray painting atomises liquid coating onto a prepared surface, building a film that cures into a decorative and protective layer.
- [Vibratory Tumbling and Mass Finishing](https://manufacturingprocesses.org/processes/finishing/vibratory-tumbling-and-mass-finishing.md): Vibratory tumbling agitates parts with abrasive media in a bowl so edges are broken and surfaces smoothed in bulk, without hand work.
- [Grinding, Sanding and Polishing](https://manufacturingprocesses.org/processes/finishing/grinding-sanding-and-polishing.md): Grinding, sanding and polishing remove a thin layer of material with bonded or coated abrasive to correct form or improve surface finish.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/finishing/abrasive-blasting)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Anodizing"
category: "Finishing"
subcategory: "Additive"
materials: ["Metal"]
tolerances: "Type II adds 0.0002–0.001 in (5–25 µm) total, roughly half of it outward; Type III adds 0.0005–0.004 in (13–100 µm). Budget the full coating thickness on a diameter and 4× the radial buildup on thread pitch diameter"
volumes: "1 to millions of parts; racked in batches and priced by area and rack space"
lead_time: "3–7 business days at a job shop; 1–2 days expedited; add a week for a first-article custom color match"
url: https://manufacturingprocesses.org/processes/finishing/anodizing
---

# Anodizing

Anodizing grows a hard porous oxide layer on aluminum electrochemically, which can then be dyed and sealed for wear and corrosion resistance.

- **Category**: [Finishing](https://manufacturingprocesses.org/processes/finishing.md)
- **Family**: Additive
- **Materials**: Metal
- **Typical tolerances**: Type II adds 0.0002–0.001 in (5–25 µm) total, roughly half of it outward; Type III adds 0.0005–0.004 in (13–100 µm). Budget the full coating thickness on a diameter and 4× the radial buildup on thread pitch diameter
- **Surface finish**: Reproduces the substrate. Bead-blast plus Type II typically lands at 32–125 µin (0.8–3.2 µm) Ra; Type III raises Ra above the as-machined value
- **Typical volumes**: 1 to millions of parts; racked in batches and priced by area and rack space
- **Lead time**: 3–7 business days at a job shop; 1–2 days expedited; add a week for a first-article custom color match

## Overview

Anodizing is an electrochemical process that converts the surface of an aluminum part into a hard, porous aluminum oxide grown out of the base metal rather than deposited onto it. Because the oxide is part of the substrate, it cannot chip or peel the way paint can. Three coating types dominate, all defined by MIL-A-8625: Type I (chromic acid, 0.00002–0.0001 in / 0.5–2.5 µm), Type II (sulfuric acid, 0.0002–0.001 in / 5–25 µm), and Type III hardcoat (sulfuric acid run cold, 0.0005–0.004 in / 13–100 µm). Type II is the cosmetic workhorse — it accepts dye and is what most consumer aluminum enclosures wear. Type III is the wear coating, commonly quoted at 60–70 HRC equivalent surface hardness.

The number that governs every design decision: roughly half the coating grows into the part and half builds outward, so a 0.002 in (50 µm) hardcoat adds about 0.001 in (25 µm) per surface. Anodizing applies to aluminum, titanium and magnesium — never to steel.

## How it works

1. **Clean and degrease.** An alkaline soak cleaner at 140–160 °F (60–70 °C) removes oil, coolant and fingerprints. Anything left behind prints through as a visible defect, because the finish is transparent.

2. **Etch and desmut.** A sodium hydroxide bath at 120–150 °F (50–65 °C) for 1–3 minutes removes roughly 0.0001–0.0005 in (2.5–13 µm) and creates the uniform satin base that "matte anodized" parts show; a bright dip instead of an etch preserves gloss. A nitric acid dip then strips the dark smut of copper, silicon and iron that the etch leaves on alloyed and cast material.

3. **Anodize.** The racked part becomes the anode in a sulfuric acid electrolyte of roughly 15–20% concentration. Type II runs at 65–75 °F (18–24 °C) and about 12–18 A/ft²; Type III runs the same chemistry chilled to roughly 28–45 °F (−2 to 7 °C) at 24–36 A/ft², with the voltage climbing as the insulating film thickens. A colder bath and higher current density produce a denser, harder, thicker oxide. Time on the rectifier sets thickness — Type II clear typically takes 20–40 minutes, Type III 30–90 minutes.

4. **Color (optional).** The as-formed oxide is porous. Organic dye at 120–140 °F (50–60 °C) penetrates the pores and gives the full color range but fades fastest outdoors. Two-step electrolytic coloring deposits tin or nickel salts at the base of the pore for lightfast bronzes and blacks. Type III dyes black reliably; other colors come out muted because the natural hardcoat is already gray-bronze.

5. **Seal.** Hydrating the oxide swells it and closes the pores — boiling deionized water at 200–212 °F (93–100 °C) for 20–30 minutes, or a nickel-acetate seal at 180–190 °F (82–88 °C), or a cold seal. Sealing is what delivers corrosion resistance: MIL-A-8625 requires sealed coatings to withstand 336 hours of ASTM B117 neutral salt spray. Hardcoat destined for PTFE or oil impregnation is deliberately left unsealed.

## Design guidelines

### Subtract the buildup before you machine

Type II clear at 0.0003 in (7.6 µm) puts about 0.00015 in (4 µm) on each surface, which disappears inside a ±0.005 in tolerance. Type III does not: a 0.002 in (50 µm) hardcoat closes a bore by about 0.002 in on diameter and grows a shaft by the same amount. Machine toleranced features undersize or oversize by the full coating thickness on diameter, and state on the drawing whether the dimension applies before or after anodize.

### Allow four times the radial buildup on threads

A 60° thread multiplies radial coating thickness by roughly 4 on pitch diameter. Type II at 0.0002 in per surface consumes about 0.0008 in of pitch-diameter clearance — usually acceptable on a Class 2B thread. Type III at 0.001 in per surface consumes about 0.004 in and will bind small threads outright. Mask the thread, tap after anodizing, or specify an oversize tap.

### Mask everything that must stay conductive or dimensioned

Anodize is a dielectric. Ground pads, connector mating faces, press-fit bores, bearing journals and dowel holes need plugs, tape or lacquer masking, or a secondary machining pass after the tank. Where a whole face must stay conductive, a chromate conversion coating to MIL-DTL-5541 Class 3 is the usual substitute on that surface.

### Break outside corners before hardcoating

Hardcoat grows perpendicular to every surface it touches, so at a sharp external corner the two advancing films collide and the coating cracks or forms a brittle ridge that chips in service. Break outside corners to at least 0.020 in (0.5 mm) radius before Type III.

### Alloy choice decides the color

6061, 6063 and 5052 anodize clear and consistent. 7075 runs slightly darker. High-copper alloys such as 2024 produce thinner, less corrosion-resistant coatings and are difficult to hardcoat. High-silicon casting alloys (A380, A383) come out dark gray and mottled and cannot be color-matched to wrought parts. Never mix alloys inside a color-matched assembly.

### Designate the rack point and control the prep

Every part carries at least one contact mark where the rack gripped it; call out an acceptable location on the drawing. The coating is thin and transparent, so machining marks, extrusion die lines and weld heat tint read through — a bead blast beforehand hides tool marks and yields a uniform satin. See the [surface finish chart](/charts/surface-finish-chart) for the Ra band each prep leaves.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Type II thickness | 0.0002–0.0005 in (5–13 µm) | 0.001 in (25 µm) | Thicker Type II coatings soften and craze |
| Type III thickness | 0.002 in (50 µm) | 0.004 in (100 µm) | Very thick hardcoat becomes brittle and micro-cracks |
| Bore/shaft allowance | Full coating thickness on diameter | — | Half the coating builds outward from each wall |
| Thread pitch-diameter allowance | 4 × radial buildup | — | 60° thread geometry multiplies radial coating by 4 |
| Outside corner radius before Type III | 0.020 in (0.5 mm) | 0.010 in (0.25 mm) | Films from adjacent faces collide and crack |

## Cost drivers

Anodizing is a tank process billed by surface area and rack space over a lot minimum. The variables that move the price:

- **Rack density.** Cost tracks how many parts fit per tank cycle. Small flat parts nest tightly; a large weldment occupies a full rack for the same cycle and carries the whole cost.
- **Coating type.** Type III needs a chilled bath, more current and a longer cycle than Type II, and is priced accordingly.
- **Color.** Clear and black are stock. A custom match adds bath setup, a first-article approval loop and usually a dedicated lot.
- **Masking.** Every plug, cap and tape mask is manual touch time applied and removed once per part. On a precision machined part it frequently costs more than the anodizing.
- **Cosmetic grade.** Forbidding rack marks in a visible zone, water staining or lot-to-lot color drift raises reject and inspection cost far more than it raises processing cost.

Cost-reduction tactics:

1. Standardize on one alloy and one color across an assembly so everything runs in a single lot.
2. Design an intentional rack point instead of accepting whatever the shop picks.
3. Use Type II unless a wear, dielectric or salt-exposure requirement genuinely demands Type III.
4. Replace a masked feature with a post-anodize machining operation — reaming a bushing bore is often cheaper than plugging it.
5. Release full racks; cost per part drops steeply with lot size because the tank cycle is fixed.

## FAQ

### How much thickness does anodizing add to a part?

Roughly half the oxide grows into the base metal and half builds outward. Type II at 0.0002–0.001 in (5–25 µm) total adds about 0.0001–0.0005 in (2.5–13 µm) per surface. Type III hardcoat at 0.002 in (50 µm) adds about 0.001 in (25 µm) per surface, which closes a bore by 0.002 in on diameter.

### What is the difference between Type II and Type III anodizing?

Both use a sulfuric acid electrolyte. Type II runs at 65–75 °F (18–24 °C) and produces a 0.0002–0.001 in (5–25 µm) decorative coating that dyes in any color. Type III runs the bath chilled to roughly 28–45 °F (−2 to 7 °C) at higher current density and produces a 0.0005–0.004 in (13–100 µm) wear coating, commonly quoted at 60–70 HRC equivalent hardness but with a limited color range.

### Can you anodize steel or stainless steel?

No. Anodizing requires a metal that forms a coherent, adherent oxide under anodic current — aluminum, titanium and magnesium. Steel and stainless steel are finished by black oxide, passivation, electroplating, phosphating or paint instead.

### Is anodized aluminum electrically conductive?

No — the oxide is a dielectric. Ground paths, connector faces and bonding pads must be masked or reworked after anodizing. Where a surface must be both protected and conductive, a chromate conversion coating to MIL-DTL-5541 Class 3 is specified instead.

### Do threads need to be masked before anodizing?

For Type II, usually not: a 0.0002 in per-surface buildup consumes about 0.0008 in of pitch-diameter clearance, which normally fits within a Class 2B thread. For Type III, yes — 0.001 in per surface consumes roughly 0.004 in of pitch diameter and will bind small threads. Mask the thread or tap after anodizing.

### Which aluminum alloys anodize best?

6061, 6063 and 5052 give clear, consistent results and are the default choices for anodized parts. 7075 anodizes slightly darker. High-copper alloys like 2024 form thinner, less protective coatings, and high-silicon die-casting alloys such as A380 come out dark gray and mottled with no reliable color match.

### How corrosion resistant is anodizing?

It depends almost entirely on sealing. MIL-A-8625 requires sealed coatings to withstand 336 hours of ASTM B117 neutral salt spray. Unsealed coatings, used when the pores will be impregnated with PTFE or oil, offer far less corrosion protection.

## Alternative processes

- [Powder Coating](https://manufacturingprocesses.org/processes/finishing/powder-coating.md): Powder coating sprays electrostatically charged dry polymer powder onto an earthed part, then bakes it into a tough continuous film.
- [Chemical Conversion Coating](https://manufacturingprocesses.org/processes/finishing/chemical-conversion-coating.md): Chemical conversion coating reacts the metal surface with a chemical bath to form a thin protective, conductive or paint-adhering film.
- [Electroplating](https://manufacturingprocesses.org/processes/finishing/electroplating.md): Electroplating deposits a metal coating from solution onto a conductive part by passing current through an electrolyte bath.
- [PVD Coating](https://manufacturingprocesses.org/processes/finishing/pvd-coating.md): PVD coating vaporises a solid target inside a vacuum chamber so it condenses on the part as a very thin, extremely hard film.
- [Cerakote Ceramic Coating](https://manufacturingprocesses.org/processes/finishing/cerakote-ceramic-coating.md): Cerakote sprays a thin ceramic-filled polymer coating onto a prepared part and cures it, giving wear and corrosion resistance in a very thin film.

## Related processes

- [Chemical Conversion Coating](https://manufacturingprocesses.org/processes/finishing/chemical-conversion-coating.md): Chemical conversion coating reacts the metal surface with a chemical bath to form a thin protective, conductive or paint-adhering film.
- [Bead Blasting](https://manufacturingprocesses.org/processes/finishing/bead-blasting.md): Bead blasting propels glass or ceramic beads at a surface to produce an even satin matte texture without cutting into it aggressively.
- [Abrasive Blasting](https://manufacturingprocesses.org/processes/finishing/abrasive-blasting.md): Abrasive blasting propels media at a surface with compressed air to clean it, strip scale, or produce a uniform matte texture.
- [Electropolishing](https://manufacturingprocesses.org/processes/finishing/electropolishing.md): Electropolishing reverses plating: current dissolves the surface fastest at the peaks, leaving a bright, deburred, passive finish.
- [Laser Marking and Engraving](https://manufacturingprocesses.org/processes/finishing/laser-marking-and-engraving.md): Laser marking alters the surface with a focused beam — annealing, etching or engraving it — to make permanent, contact-free marks.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/finishing/anodizing)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Bead Blasting"
category: "Finishing"
subcategory: "Subtractive"
materials: ["Metal", "Plastic"]
tolerances: "Removes under 0.0005 in (13 µm) — generally treated as dimensionally neutral, but toleranced bores, threads and sealing faces should still be masked"
volumes: "1 to millions of parts; automated cabinets are used above a few hundred pieces for cosmetic consistency"
lead_time: "Same day to 3 business days at a job shop; usually run in line with anodizing"
url: https://manufacturingprocesses.org/processes/finishing/bead-blasting
---

# Bead Blasting

Bead blasting propels glass or ceramic beads at a surface to produce an even satin matte texture without cutting into it aggressively.

- **Category**: [Finishing](https://manufacturingprocesses.org/processes/finishing.md)
- **Family**: Subtractive
- **Materials**: Metal, Plastic
- **Typical tolerances**: Removes under 0.0005 in (13 µm) — generally treated as dimensionally neutral, but toleranced bores, threads and sealing faces should still be masked
- **Surface finish**: 32–125 µin (0.8–3.2 µm) Ra, non-directional, depending on bead size and pressure
- **Typical volumes**: 1 to millions of parts; automated cabinets are used above a few hundred pieces for cosmetic consistency
- **Lead time**: Same day to 3 business days at a job shop; usually run in line with anodizing

## Overview

Bead blasting propels spherical glass or ceramic beads at a surface to produce an even, low-gloss satin texture. Because the media is round rather than angular, it peens the surface rather than cutting it — it removes very little stock, work-hardens the surface slightly, and leaves a dimpled finish with no directional lay.

It is the standard cosmetic prep for machined aluminum, and bead blast followed by Type II anodize is the finish on most consumer electronics enclosures, camera bodies and instrument housings. Typical results land in the 32–125 µin (0.8–3.2 µm) Ra range depending on bead size and pressure, and working pressure is usually 20–60 psi (1.4–4 bar) — well below what an aluminum oxide blast would use. The trade-off is that bead blasting is a texturing operation, not a cleaning operation: it will not strip heavy scale, rust or old coating, and it will not remove a deep tool mark.

## How it works

1. **Deburr and clean first.** Bead blasting hides fine tool marks but not burrs, chatter or scratches. Machining marks deeper than the blast texture stay visible, and burrs simply become rounded burrs. Deburr and degrease before the cabinet.

2. **Mask.** Threads, reamed bores, bearing seats, sealing lands and any surface with a specified finish get plugs, caps or tape.

3. **Select bead size.** Finer beads produce a smoother, tighter satin; coarser beads produce a deeper, more matte texture. Media is specified by size number or mesh range, and the choice is the single biggest determinant of appearance.

4. **Blast.** Beads are propelled at 20–60 psi (1.4–4 bar) through a hand or automated nozzle at a consistent standoff and angle. Uniformity is everything on a cosmetic part: overlapping passes, dwell in one spot, or a changing standoff distance all show as banding or blotching under raking light.

5. **Blow off and inspect.** Residual media is removed with clean, dry, oil-free air. Trapped beads in blind holes and threads become contamination later.

6. **Finish promptly.** Blasted aluminum has a high-energy, active surface. It should move to anodize, chem film or paint quickly, and be handled with gloves in the meantime — fingerprints on freshly blasted aluminum are visible after anodizing.

### Media control

Beads fracture in use. A broken bead has sharp edges and cuts instead of peening, so a cabinet with degraded media produces a progressively different finish. Production cosmetic work uses classifiers to remove fines and broken media, and specifies media replacement intervals. This is why the same drawing can produce visibly different parts from two suppliers.

## Design guidelines

### Specify the finish by sample, not by adjective

"Bead blast" on a drawing is not a specification. Cosmetic work needs the bead size or mesh range, the nozzle pressure, and — most usefully — an approved physical master sample that the shop matches against under a defined lighting condition. Ra alone does not capture the appearance difference between two blast setups that measure the same.

### Expect 32–125 µin Ra and no directional lay

Bead blasting produces a non-directional, uniform texture in the 32–125 µin (0.8–3.2 µm) Ra band, which is why it hides the lay of a milled or turned surface so effectively. See the [surface finish chart](/charts/surface-finish-chart) for where that sits relative to as-machined, ground and polished surfaces.

### Do not use it to remove material or fix defects

Stock removal is negligible — typically well under 0.0005 in (13 µm) — so bead blasting cannot correct a dimension, remove a scratch, or clean heavy oxide. It is a texture operation applied to an already-correct surface.

### Mask every toleranced and sealing feature

Even light peening degrades a reamed bore, a lapped seal face or a thread. Plug or cap them. Blind holes also trap beads, so provide drainage or plan for a wash step.

### Watch soft materials and thin sections

Aluminum, copper, brass and plastics can retain embedded media, and repeated blasting of a thin panel will bow it through induced compressive stress. Reduce pressure and bead size on thin or soft parts.

### Sequence it correctly with anodizing

Bead blast, then anodize — the anodic film is thin and transparent and reproduces the blast texture. Blasting after anodizing destroys the coating. If a part needs both a blasted cosmetic face and a masked conductive pad, plan the masking for both operations at once.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Working pressure | 20–60 psi (1.4–4 bar) | Lower on thin or soft parts | Higher pressure fractures beads and embeds media |
| Achievable Ra | 32–125 µin (0.8–3.2 µm) | — | Set by bead size and pressure |
| Stock removal | < 0.0005 in (13 µm) | Not a material-removal process | Round media peens rather than cuts |
| Appearance control | Approved master sample | Ra alone is insufficient | Two setups at the same Ra can look different |
| Media condition | Classified, fines removed | — | Broken beads cut and change the finish |
| Sequence | Deburr, blast, then anodize | Never blast after anodizing | Blasting destroys the anodic film |

## Cost drivers

Bead blasting is inexpensive per part in a cabinet and expensive per part when the appearance requirement is tight, because the cost is almost entirely operator time and reject rate.

- **Cosmetic tolerance.** A functional matte finish costs a fraction of a Class A cosmetic finish with no banding, blotching or overlap witness under raking light.
- **Masking.** Plugs and caps on threads and bores are hand labor per part.
- **Part size and handling.** Cabinet work on small parts is fast; large housings need a room, a manipulator, or robotic blasting.
- **Media management.** Classified media and scheduled replacement cost more than running media to destruction, but they are what makes lot-to-lot consistency possible.
- **Consistency across suppliers.** Qualifying a second source to match an existing master sample is a real engineering cost, and one reason cosmetic blast specs are hard to dual-source.

Cost-reduction tactics:

1. Approve a physical master sample early and control it — this prevents the most expensive failure mode, which is a rejected cosmetic lot.
2. Robotic or automated blasting pays for itself on cosmetic parts above a few hundred pieces because it removes operator variation.
3. Reduce masked features by moving fitted surfaces onto a separate component.
4. Specify bead size and pressure on the drawing so the process is reproducible rather than negotiated per lot.
5. Where the surface is not visible, use a functional blast callout instead of a cosmetic one.

## FAQ

### What Ra does bead blasting produce?

Typically 32–125 µin (0.8–3.2 µm) Ra, with a non-directional texture, depending on bead size and nozzle pressure. Finer beads and lower pressure give a tighter satin; coarser beads give a deeper matte. Ra alone does not fully define the appearance, so cosmetic work should be controlled with an approved master sample.

### How much material does bead blasting remove?

Very little — typically well under 0.0005 in (13 µm). Round media peens rather than cuts, so bead blasting cannot correct a dimension, remove a scratch, or strip heavy scale or old coating. It is a texturing operation applied to an already-correct surface.

### Should I bead blast before or after anodizing?

Before. The anodic film is thin and transparent and reproduces the blast texture faithfully, which is why bead blast plus Type II anodize is the standard finish on machined aluminum enclosures. Blasting after anodizing destroys the coating.

### Why do bead-blasted parts from two suppliers look different?

Bead size, nozzle pressure, standoff distance and — most often — media condition. Beads fracture with use, and broken beads cut instead of peening, so a cabinet running degraded media produces a progressively different finish. Specify bead size, pressure and an approved master sample to make the process reproducible.

### Bead blasting or abrasive blasting with aluminum oxide?

Bead blasting for a uniform cosmetic satin with negligible stock removal, at 20–60 psi. Aluminum oxide blasting for cleaning, stripping and creating a deep anchor profile for coating, at 60–100 psi, with a much coarser resulting surface in the 125–500 µin (3.2–12.5 µm) Ra range.

### Do I need to mask threads for bead blasting?

Yes for anything toleranced or sealing. Even light peening degrades a reamed bore, a lapped seal face or a thread flank, and blind holes trap beads that become contamination later. Plug or cap those features, or produce them after blasting.

## Alternative processes

- [Abrasive Blasting](https://manufacturingprocesses.org/processes/finishing/abrasive-blasting.md): Abrasive blasting propels media at a surface with compressed air to clean it, strip scale, or produce a uniform matte texture.
- [Vibratory Tumbling and Mass Finishing](https://manufacturingprocesses.org/processes/finishing/vibratory-tumbling-and-mass-finishing.md): Vibratory tumbling agitates parts with abrasive media in a bowl so edges are broken and surfaces smoothed in bulk, without hand work.
- [Electropolishing](https://manufacturingprocesses.org/processes/finishing/electropolishing.md): Electropolishing reverses plating: current dissolves the surface fastest at the peaks, leaving a bright, deburred, passive finish.
- [Grinding, Sanding and Polishing](https://manufacturingprocesses.org/processes/finishing/grinding-sanding-and-polishing.md): Grinding, sanding and polishing remove a thin layer of material with bonded or coated abrasive to correct form or improve surface finish.
- [Photo Etching](https://manufacturingprocesses.org/processes/finishing/photo-etching.md): Photo etching masks a surface photographically and etches shallow detail into it for decoration, marking or texture.

## Related processes

- [Abrasive Blasting](https://manufacturingprocesses.org/processes/finishing/abrasive-blasting.md): Abrasive blasting propels media at a surface with compressed air to clean it, strip scale, or produce a uniform matte texture.
- [Anodizing](https://manufacturingprocesses.org/processes/finishing/anodizing.md): Anodizing grows a hard porous oxide layer on aluminum electrochemically, which can then be dyed and sealed for wear and corrosion resistance.
- [Vibratory Tumbling and Mass Finishing](https://manufacturingprocesses.org/processes/finishing/vibratory-tumbling-and-mass-finishing.md): Vibratory tumbling agitates parts with abrasive media in a bowl so edges are broken and surfaces smoothed in bulk, without hand work.
- [Grinding, Sanding and Polishing](https://manufacturingprocesses.org/processes/finishing/grinding-sanding-and-polishing.md): Grinding, sanding and polishing remove a thin layer of material with bonded or coated abrasive to correct form or improve surface finish.
- [Chemical Conversion Coating](https://manufacturingprocesses.org/processes/finishing/chemical-conversion-coating.md): Chemical conversion coating reacts the metal surface with a chemical bath to form a thin protective, conductive or paint-adhering film.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/finishing/bead-blasting)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Black Oxide"
category: "Finishing"
subcategory: "Additive"
materials: ["Metal"]
tolerances: "Adds roughly 0.00003–0.00005 in (0.75–1.25 µm) — treated as dimensionally neutral; no allowance needed on threads, gage surfaces or press fits"
volumes: "1 to millions of parts; small hardware runs in bulk baskets"
lead_time: "1–5 business days at a job shop; often same or next day for small hardware lots"
url: https://manufacturingprocesses.org/processes/finishing/black-oxide
---

# Black Oxide

Black oxide converts the steel surface to magnetite in a hot alkaline bath, giving a black finish that adds essentially no thickness.

- **Category**: [Finishing](https://manufacturingprocesses.org/processes/finishing.md)
- **Family**: Additive
- **Materials**: Metal
- **Typical tolerances**: Adds roughly 0.00003–0.00005 in (0.75–1.25 µm) — treated as dimensionally neutral; no allowance needed on threads, gage surfaces or press fits
- **Surface finish**: Reproduces the substrate exactly — ground parts finish satin black, blasted parts matte black, polished parts gloss black
- **Typical volumes**: 1 to millions of parts; small hardware runs in bulk baskets
- **Lead time**: 1–5 business days at a job shop; often same or next day for small hardware lots

## Overview

Black oxide is a conversion coating that reacts the surface of a steel part with a hot alkaline oxidizing bath to form magnetite (Fe₃O₄), a black iron oxide integral to the substrate. The coating is roughly 0.00003–0.00005 in (0.75–1.25 µm) thick — small enough that it is treated as dimensionally neutral, which is the entire reason the process exists. Gages, precision tooling, bearing components, gears and firearms parts get a black finish without a plating allowance.

Hot black oxide runs at 285–305 °F (141–152 °C) in a caustic soda and nitrate/nitrite bath, and the specification is MIL-DTL-13924 for steel. Black oxide by itself provides very little corrosion protection; the post-treatment does the work. A displacing oil, wax or dry-film sealer soaks into the porous oxide, and it is that oil retention — not the oxide — that delivers the rust resistance and the mild lubricity. Mid-temperature and room-temperature variants exist, with different chemistry and generally lower durability.

## How it works

1. **Clean and derust.** Alkaline soak clean, then acid pickle if scale or rust is present. Black oxide is a conversion reaction with bare steel; anything on the surface blocks it and shows up as a bare or brown patch.

2. **Convert.** The part is immersed in a boiling caustic solution — sodium hydroxide with nitrate and nitrite oxidizers — at 285–305 °F (141–152 °C) for roughly 5–20 minutes. Iron at the surface is oxidized directly to magnetite. Because it is a conversion rather than a deposition, no thickness is added by plating action: the oxide grows into and out of the existing surface with negligible net dimensional change.

3. **Rinse.** Multi-stage rinsing removes the caustic. The bath is alkaline, so the process does not involve the acid hydrogen charging that makes electroplating an embrittlement risk on hardened steel.

4. **Post-treat.** The finished oxide is porous and must be sealed. Water-displacing oil is the traditional choice and gives the deepest black; wax gives a drier feel; proprietary dry-film and polymer sealers give the best corrosion performance and are used where oil is unacceptable. Corrosion resistance in service is essentially a function of which sealer was used and whether it stays on the part.

### Temperature variants

**Hot black oxide** at 285–305 °F (141–152 °C) is the true magnetite conversion, and the finish is the most durable and abrasion-resistant of the three. **Mid-temperature** processes run around 200–250 °F (93–121 °C) and avoid the boiling caustic hazard while producing a similar black oxide. **Room-temperature** blackening is not the same chemistry — it deposits a selenium-based copper conversion film that is thinner, softer and much less abrasion resistant, but requires no heated tank and is common for touch-up and low-volume work.

### Non-steel variants

Separate baths blacken stainless steel (MIL-DTL-13924 Class 4), copper and copper alloys, and zinc. Stainless blackening produces a true black oxide but with a different bath chemistry and a longer cycle.

## Design guidelines

### Use it when dimension is the constraint

At roughly 0.00005 in (1.25 µm), black oxide fits inside almost any tolerance. Threads, gage surfaces, bearing seats, ground diameters, press fits and match-ground assemblies can be blackened after final machining without allowance and without masking. No other black finish on steel does this — powder coating adds 2–4 mils, plating 0.0002 in or more, and even a chemical conversion coating on aluminum adds more than black oxide does.

### Do not specify it where corrosion is the primary requirement

The oxide itself offers minimal protection. Sealed with oil it will pass a modest humidity exposure and resist handling rust indoors, but it is not a substitute for zinc plating, phosphate-and-oil, or paint in outdoor or wet service. If the part will see salt or standing water, choose a different system or accept a scheduled re-oiling regime.

### The finish reproduces the substrate exactly

Black oxide is transparent in the sense that it follows the surface underneath: a ground part comes out satin black, a blasted part comes out matte black, a polished part comes out gloss black. It hides nothing — grinding chatter, tool marks and decarburized skin all remain visible. Set the appearance with the mechanical prep, using the [surface finish chart](/charts/surface-finish-chart) to pick the incoming Ra.

### Watch the base material

Low-carbon, medium-carbon, alloy and tool steels all blacken well. Cast iron blackens but the graphite gives a mottled result. High-chromium and stainless grades need the dedicated stainless bath. Nitrided, heavily decarburized, or previously plated surfaces convert unevenly. Mixed-material assemblies must be disassembled — aluminum is attacked by the caustic bath.

### Sequence it after heat treat and grinding

Black oxide is a low-temperature process, so it does not affect temper or hardness, but any subsequent grinding, lapping or machining removes it locally. Put it at the end of the routing, immediately before the oil seal.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Coating thickness | 0.00003–0.00005 in (0.75–1.25 µm) | — | Conversion, not deposition — treat as dimensionally neutral |
| Bath temperature (hot process) | 285–305 °F (141–152 °C) | — | Below the boiling range the magnetite does not form properly |
| Post-treatment | Oil, wax or dry-film sealer, always | Never leave unsealed | Bare oxide is porous and offers little corrosion protection |
| Base materials | Carbon, alloy and tool steels | Aluminum is attacked by the bath | Caustic dissolves aluminum |
| Position in routing | After all machining and heat treat | — | Any subsequent cutting removes the coating locally |
| Appearance control | Set by incoming surface finish | — | The oxide reproduces the substrate exactly |

## Cost drivers

Black oxide is one of the least expensive finishes available for steel. Parts run in bulk baskets or on racks, the cycle is short, and there is no rectifier, no masking allowance and no dimensional inspection to perform afterward.

- **Handling method.** Basket work on small hardware costs a fraction of racked work on large or cosmetic parts.
- **Pre-cleaning.** Heat-treat scale, rust, weld discoloration or heavy oil require pickling or blasting first, which is a separate operation and usually the largest cost adder.
- **Sealer choice.** Standard water-displacing oil is inexpensive; dry-film and proprietary polymer sealers cost more but are required where oil transfer is unacceptable.
- **Cosmetic requirements.** Uniform, deep, streak-free black on a visible part means racking, careful rinsing and higher reject rates than a functional finish on hardware.
- **Lot minimums.** As with any tank process, the tank cycle costs the same whether the basket is full or half full.

Cost-reduction tactics:

1. Batch parts to fill baskets — per-part cost falls steeply with lot size.
2. Control heat-treat atmosphere so parts arrive clean and skip the pickling step.
3. Specify a functional rather than cosmetic acceptance standard where the part is not visible.
4. Use black oxide instead of black plating or black powder wherever the requirement is appearance plus dimensional stability, not corrosion resistance.
5. Keep aluminum and mixed-material subassemblies out of the routing so parts do not need to be disassembled and reassembled.

## FAQ

### How much thickness does black oxide add?

Roughly 0.00003–0.00005 in (0.75–1.25 µm), which is small enough to be treated as dimensionally neutral. Threads, gage surfaces, bearing seats and press fits can be blackened after final machining with no allowance and no masking — the reason black oxide is specified on precision tooling.

### Does black oxide prevent rust?

Only in combination with its post-treatment. The magnetite layer itself offers minimal corrosion protection; the oil, wax or dry-film sealer absorbed into the porous oxide does the work. Black oxide is suitable for indoor and handling protection, not as a substitute for zinc plating or paint in wet or salt service.

### What is the difference between hot, mid-temperature and room-temperature black oxide?

Hot black oxide runs at 285–305 °F (141–152 °C) in caustic soda and produces true magnetite — the most durable version. Mid-temperature processes run around 200–250 °F (93–121 °C) with similar results and less hazard. Room-temperature blackening is a different chemistry entirely, depositing a selenium-based film that is thinner, softer and much less abrasion resistant.

### Does black oxide cause hydrogen embrittlement?

The blackening bath itself is alkaline and does not charge hydrogen into the steel the way acid plating does. However, any acid pickling used to remove scale beforehand can, so hardened high-strength parts that are pickled should still follow the applicable embrittlement-relief practice.

### Can stainless steel or aluminum be black oxided?

Stainless steel yes, in a dedicated bath with different chemistry and a longer cycle, covered as MIL-DTL-13924 Class 4. Aluminum no — the caustic bath attacks it. Aluminum parts are anodized black or given a black chemical conversion coating instead.

### Why does black oxide look different on different parts?

The coating is only about a micron thick and follows the surface underneath exactly. A ground part comes out satin black, a blasted part matte black, a polished part gloss black. Appearance is controlled by the mechanical prep before the tank, not by the tank.

## Alternative processes

- [Chemical Conversion Coating](https://manufacturingprocesses.org/processes/finishing/chemical-conversion-coating.md): Chemical conversion coating reacts the metal surface with a chemical bath to form a thin protective, conductive or paint-adhering film.
- [Electroplating](https://manufacturingprocesses.org/processes/finishing/electroplating.md): Electroplating deposits a metal coating from solution onto a conductive part by passing current through an electrolyte bath.
- [Cerakote Ceramic Coating](https://manufacturingprocesses.org/processes/finishing/cerakote-ceramic-coating.md): Cerakote sprays a thin ceramic-filled polymer coating onto a prepared part and cures it, giving wear and corrosion resistance in a very thin film.
- [PVD Coating](https://manufacturingprocesses.org/processes/finishing/pvd-coating.md): PVD coating vaporises a solid target inside a vacuum chamber so it condenses on the part as a very thin, extremely hard film.
- [Powder Coating](https://manufacturingprocesses.org/processes/finishing/powder-coating.md): Powder coating sprays electrostatically charged dry polymer powder onto an earthed part, then bakes it into a tough continuous film.

## Related processes

- [Chemical Conversion Coating](https://manufacturingprocesses.org/processes/finishing/chemical-conversion-coating.md): Chemical conversion coating reacts the metal surface with a chemical bath to form a thin protective, conductive or paint-adhering film.
- [Passivation](https://manufacturingprocesses.org/processes/finishing/passivation.md): Passivation removes free iron from a stainless steel surface with an acid bath so the chromium oxide layer can reform unbroken.
- [Abrasive Blasting](https://manufacturingprocesses.org/processes/finishing/abrasive-blasting.md): Abrasive blasting propels media at a surface with compressed air to clean it, strip scale, or produce a uniform matte texture.
- [Electroplating](https://manufacturingprocesses.org/processes/finishing/electroplating.md): Electroplating deposits a metal coating from solution onto a conductive part by passing current through an electrolyte bath.
- [Vibratory Tumbling and Mass Finishing](https://manufacturingprocesses.org/processes/finishing/vibratory-tumbling-and-mass-finishing.md): Vibratory tumbling agitates parts with abrasive media in a bowl so edges are broken and surfaces smoothed in bulk, without hand work.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/finishing/black-oxide)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "CNC Engraving"
category: "Finishing"
subcategory: "Subtractive"
materials: ["Metal", "Plastic", "Wood", "Glass"]
tolerances: "Depth control of ±0.001–0.002 in (±0.025–0.05 mm) is routine; stroke width with a conical tool varies directly with depth, so both must be specified together"
volumes: "1 to tens of thousands of parts; cost per part is dominated by cycle time above a few hundred"
lead_time: "1–5 business days; no tooling lead time beyond stock cutters"
url: https://manufacturingprocesses.org/processes/finishing/cnc-engraving
---

# CNC Engraving

CNC engraving cuts text and artwork into a surface with a small pointed or ball-nose tool following a programmed path.

- **Category**: [Finishing](https://manufacturingprocesses.org/processes/finishing.md)
- **Family**: Subtractive
- **Materials**: Metal, Plastic, Wood, Glass
- **Typical tolerances**: Depth control of ±0.001–0.002 in (±0.025–0.05 mm) is routine; stroke width with a conical tool varies directly with depth, so both must be specified together
- **Surface finish**: Groove walls come out as-machined; ductile metals raise a fine burr that must be removed before any paint fill
- **Typical volumes**: 1 to tens of thousands of parts; cost per part is dominated by cycle time above a few hundred
- **Lead time**: 1–5 business days; no tooling lead time beyond stock cutters

## Overview

CNC engraving cuts text, artwork and serial data into a surface with a small rotating tool following a programmed path. Unlike laser marking, it is a mechanical cutting process, so it produces genuine depth with a defined cross-section in any machinable material, and the resulting groove can be filled with paint or lacquer for contrast.

Typical engraved depth for text is 0.005–0.020 in (0.13–0.5 mm), cut with a conical engraving tool whose flat tip runs from about 0.005 in to 0.030 in (0.13–0.75 mm), or with a small ball-nose end mill. Because the conical tool gets wider as it goes deeper, stroke width is a function of depth — a designer controls line weight by controlling Z. Spindle speeds are high, commonly 10,000–24,000 rpm on small cutters. Diamond drag engraving, in which a non-rotating stylus scratches the surface at 0.001–0.003 in (25–75 µm), is the low-force alternative for thin, hardened or delicate parts.

## How it works

1. **Prepare the artwork.** Engraving uses single-line (stick) fonts wherever possible, so each character is one toolpath rather than an outlined shape that must be pocketed. Outline fonts and logos require the enclosed area to be cleared out, which multiplies cycle time.

2. **Choose the tool.** A conical engraving tool with a defined tip flat and an included angle of roughly 15–90° is the standard choice: the shallow angle gives a strong tip and a stroke width that grows predictably with depth. Ball-nose end mills give a constant-width, round-bottom groove. Diamond drag styli scratch rather than cut and need almost no spindle power or rigidity.

3. **Cut.** The tool follows the toolpath at high spindle speed and modest feed. Depth is usually taken in one pass for text under about 0.010 in (0.25 mm) and in multiple passes below that. Chip evacuation and heat matter: acrylic and many thermoplastics melt and re-weld unless a single-flute O-flute cutter and high feed are used.

4. **Deburr.** Engraving raises a fine burr at the groove edges on ductile metals. A light hand deburr, bead blast or vibratory pass removes it — and must be done before any paint fill.

5. **Fill (optional).** A lacquer stick or paint is wiped into the groove and the excess removed from the surface. Fill depth and groove width determine how well the fill holds; shallow, narrow grooves shed fill in service.

### Where it beats laser marking

Engraving produces real depth in any material regardless of optical absorption, so it works on bare aluminum, copper, brass and clear plastics where a fiber laser struggles. It gives a groove that holds paint fill, and it produces a tactile result for control panels and safety markings. It is also the practical route to cutting through a two-color engraving laminate, where the top layer is removed to expose a contrasting core.

## Design guidelines

### Set line weight through depth, not through the drawing

With a conical tool, stroke width grows as the tool goes deeper. A 0.010 in wide stroke and a 0.020 in wide stroke on the same part require two different depths or two different tools. Specify the depth and the tool tip, or specify the stroke width and let the shop compute depth — but do not assume a drawn line weight will be reproduced.

### Respect the minimum internal radius

Every internal corner in the artwork is limited by the tool's effective radius at cutting depth. Sharp corners in a logo come out radiused. Keep detail features above roughly 0.010 in (0.25 mm) and expect finer detail to soften.

### Size characters for legibility and for fill

Practical minimum character height is around 0.060 in (1.5 mm) for a clean engraved result, and around 0.100 in (2.5 mm) if the groove must hold paint fill reliably. Below that, the fill will not stay in and the characters lose definition.

### Keep engraving off thin and unsupported walls

Cutting force deflects thin sections, and a deflecting wall produces an inconsistent depth and therefore an inconsistent stroke width. Support the area under the engraving, or use diamond drag engraving, which applies almost no force.

### Sequence it with the surface finish

Engraving before anodizing gives a groove that anodizes along with the rest of the part — subtle, uniform, corrosion-protected. Engraving after anodizing cuts through the colored layer and exposes bright metal, giving high contrast but leaving bare aluminum in the groove. Choose deliberately, and note that engraving after bead blasting will leave a bright groove against a matte field.

### Match the tool and speed to the material

Aluminum and brass engrave cleanly at high speed. Stainless and tool steels need slower feeds and a stronger tip angle. Acrylic and polycarbonate melt unless chip evacuation is aggressive. Wood and engraving laminates cut easily but chip at the edges of the stroke.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Engraved depth (text) | 0.005–0.020 in (0.13–0.5 mm) | — | Deeper needs multiple passes and widens the stroke |
| Diamond drag depth | 0.001–0.003 in (25–75 µm) | Scratch only, no fill | Non-rotating stylus, minimal force |
| Minimum character height | 0.060 in (1.5 mm) | 0.100 in (2.5 mm) if paint filled | Fill will not stay in a small groove |
| Minimum internal detail | 0.010 in (0.25 mm) | Set by tool radius at depth | Corners radius to the tool |
| Spindle speed | 10,000–24,000 rpm | — | Small-diameter tools need high surface speed |
| Sequence with anodize | Before for uniform color, after for contrast | — | Engraving after anodize exposes bare metal |

## Cost drivers

CNC engraving is priced as machining time plus setup, so it behaves like any other milling operation: cheap per part at volume, dominated by setup at low volume.

- **Engraved area and path length.** Cycle time follows total toolpath. Outlined or filled artwork that must be pocketed can take ten times as long as the same design in a single-line font.
- **Setup and fixturing.** Locating the part repeatably under the spindle is the fixed cost. Flat plates are trivial; contoured or cylindrical parts need dedicated work holding or an additional axis.
- **Tooling.** Engraving tips are small and wear quickly in hard materials; tip replacement frequency on stainless is a real consumable cost.
- **Secondary operations.** Deburring and paint filling are hand operations, and paint fill in particular is priced per part.
- **Variable data.** Serial numbers require either a program per part or macro programming, and add handling time per piece.

Cost-reduction tactics:

1. Use single-line engraving fonts rather than outlined text — this is the single largest cycle-time lever.
2. Engrave in the same setup as the machining operation that produced the surface, avoiding a second fixture and load.
3. Keep engraved artwork simple and above the minimum detail size so it cuts in one pass with one tool.
4. Skip paint fill unless contrast is functionally required; an engraved groove on a blasted surface often reads well without it.
5. For variable data such as serial numbers on many parts, compare against laser marking — the laser has no tool wear and no per-part handling penalty.

## FAQ

### How deep should CNC engraving be?

Typical engraved text runs 0.005–0.020 in (0.13–0.5 mm) deep. Diamond drag engraving, which scratches with a non-rotating stylus, runs 0.001–0.003 in (25–75 µm). If the groove will be paint filled, stay toward the deeper end so the fill has something to hold onto.

### Why does my engraved line width change with depth?

Standard engraving tools are conical with a small tip flat, so the cutting width grows as the tool descends. Line weight is therefore a function of Z depth, not of the drawn artwork. Specify tip size and depth together, or specify the required stroke width and let the shop derive the depth.

### How small can engraved characters be?

Around 0.060 in (1.5 mm) character height for a clean result, and around 0.100 in (2.5 mm) if the groove has to hold paint fill. Internal detail is limited by the tool radius at depth, so features below roughly 0.010 in (0.25 mm) will soften and sharp corners will come out radiused.

### CNC engraving or laser marking?

Engraving for genuine depth, for a groove that holds paint fill, for a tactile result, and for materials that laser sources handle poorly such as bare aluminum, copper and clear plastics. Laser marking for speed, for variable data at no extra cost, for no tool wear, and for marks on surfaces that must not be cut into.

### Should I engrave before or after anodizing?

Before, if you want the groove anodized along with the rest of the part for a uniform, corrosion-protected result. After, if you want maximum contrast — cutting through the colored anodic layer exposes bright metal, but leaves that groove unprotected.

### What font should I use for CNC engraving?

A single-line or stick font, where each character is one toolpath. Outlined fonts require the enclosed area to be cleared out, which can multiply cycle time by an order of magnitude for no visual benefit at small sizes.

## Alternative processes

- [Laser Marking and Engraving](https://manufacturingprocesses.org/processes/finishing/laser-marking-and-engraving.md): Laser marking alters the surface with a focused beam — annealing, etching or engraving it — to make permanent, contact-free marks.
- [Photo Etching](https://manufacturingprocesses.org/processes/finishing/photo-etching.md): Photo etching masks a surface photographically and etches shallow detail into it for decoration, marking or texture.
- [Pad Printing](https://manufacturingprocesses.org/processes/finishing/pad-printing.md): Pad printing lifts ink out of an etched plate with a silicone pad and transfers it onto a curved or recessed surface.
- [Screen Printing](https://manufacturingprocesses.org/processes/finishing/screen-printing.md): Screen printing pushes ink through a stencilled mesh onto the surface below, laying down one opaque color per pass.
- [Foil Blocking and Embossing](https://manufacturingprocesses.org/processes/finishing/foil-blocking-and-embossing.md): Foil blocking and embossing press a heated die into the surface to transfer metallic foil, raise a relief, or do both at once.

## Related processes

- [Laser Marking and Engraving](https://manufacturingprocesses.org/processes/finishing/laser-marking-and-engraving.md): Laser marking alters the surface with a focused beam — annealing, etching or engraving it — to make permanent, contact-free marks.
- [CNC Milling](https://manufacturingprocesses.org/processes/cutting/cnc-milling.md): CNC milling removes material with a rotating multi-flute cutter moved along programmed paths, producing prismatic parts from solid stock.
- [Photo Etching](https://manufacturingprocesses.org/processes/finishing/photo-etching.md): Photo etching masks a surface photographically and etches shallow detail into it for decoration, marking or texture.
- [Anodizing](https://manufacturingprocesses.org/processes/finishing/anodizing.md): Anodizing grows a hard porous oxide layer on aluminum electrochemically, which can then be dyed and sealed for wear and corrosion resistance.
- [Screen Printing](https://manufacturingprocesses.org/processes/finishing/screen-printing.md): Screen printing pushes ink through a stencilled mesh onto the surface below, laying down one opaque color per pass.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/finishing/cnc-engraving)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Cerakote Ceramic Coating"
category: "Finishing"
subcategory: "Additive"
materials: ["Metal", "Plastic"]
tolerances: "Adds about 0.001 in (25 µm) per coated surface; budget 2× on a bore diameter and roughly 4× on thread pitch diameter, or mask"
volumes: "1 part to a few thousand; hand-sprayed, so it does not scale like powder or e-coat"
lead_time: "1–3 weeks at a specialist applicator; oven cure itself is about 2 hours"
url: https://manufacturingprocesses.org/processes/finishing/cerakote-ceramic-coating
---

# Cerakote Ceramic Coating

Cerakote sprays a thin ceramic-filled polymer coating onto a prepared part and cures it, giving wear and corrosion resistance in a very thin film.

- **Category**: [Finishing](https://manufacturingprocesses.org/processes/finishing.md)
- **Family**: Additive
- **Materials**: Metal, Plastic
- **Typical tolerances**: Adds about 0.001 in (25 µm) per coated surface; budget 2× on a bore diameter and roughly 4× on thread pitch diameter, or mask
- **Surface finish**: Even matte to satin, set primarily by the blast profile beneath the film
- **Typical volumes**: 1 part to a few thousand; hand-sprayed, so it does not scale like powder or e-coat
- **Lead time**: 1–3 weeks at a specialist applicator; oven cure itself is about 2 hours

## Overview

Cerakote is a family of proprietary thin-film ceramic-filled polymer coatings sprayed onto a blasted, degreased part and then cured. The defining number is film thickness: a typical application is around 0.001 in (25 µm), roughly a quarter of a powder coat and an order of magnitude thicker than PVD. That range makes it usable on parts with real fits and moving surfaces where a 2–4 mil powder film would not clear.

The H-Series oven-cure products cure at roughly 250 °F (121 °C), low enough not to affect the temper of hardened steel or the properties of most aluminum. Elite and other air-cure series remove the oven requirement and can be applied to polymers and assembled parts. The C-Series high-temperature products are rated for continuous service far above what an organic coating survives, and are used on firearm suppressors and exhaust components. Substrates include steel, stainless, aluminum, titanium, and — with the air-cure chemistries — many polymers.

## How it works

1. **Degrease completely.** The coating is thin and its adhesion depends entirely on a clean, high-energy surface. Parts are solvent-cleaned, then often baked to drive out oil from pores and threads. Any residual oil, silicone or fingerprint produces a fisheye or a delamination.

2. **Blast.** Abrasive blasting with aluminum oxide, typically in the 100–120 grit range, at a controlled pressure creates the mechanical anchor profile. Too coarse a profile telegraphs through a 1 mil film; too fine and adhesion suffers. Blasting also removes any residual oxide or old coating.

3. **Mask.** Threads, bores, bearing surfaces, sealing faces and any lettering that must stay crisp are masked with plugs, caps and high-temperature tape before spraying.

4. **Mix and spray.** Most products are two-component, mixed with a hardener at a specified ratio and reduced to spray viscosity. An HVLP gun with a small tip — commonly around 0.8 mm — is used at low pressure, and the coating is applied in light passes to a target of roughly 0.001 in (25 µm). Because the film is so thin, film build is controlled by pass count and gun distance, and inspection is by measured dry film thickness rather than by eye.

5. **Flash and cure.** Solvent flashes off, then the part is oven cured. H-Series schedules are typically around 2 hours at 250 °F (121 °C). Air-cure products develop full properties over days at room temperature; handling strength comes much sooner.

6. **Demask and inspect.** Dry film thickness by magnetic or eddy-current gauge, adhesion by cross-hatch tape test to ASTM D3359.

### Where it sits between other finishes

Thicker and more forgiving than PVD, thinner and lower-cure-temperature than powder coating, far more wear resistant than ordinary paint, and available in a much wider color range than anodizing. The trade is that it is a sprayed organic-ceramic hybrid: line-of-sight application, operator-dependent, and not a metallurgical bond.

## Design guidelines

### Budget about 1 mil per surface and mask the rest

At roughly 0.001 in (25 µm) per coated surface, a bore closes about 0.002 in on diameter and a thread's pitch diameter grows by roughly 0.004 in — enough to interfere on a Class 3 fit. Mask threads, dowel holes, bearing journals and sealing lands, or machine them after coating.

### Set the appearance with the blast profile

The film is thin enough to follow the substrate. A 100–120 grit aluminum oxide blast gives the characteristic even matte; a coarser profile shows through as visible texture, and a polished substrate under a thin film reads glossier. Use the [surface finish chart](/charts/surface-finish-chart) to choose a prep that matches the intended look, and specify the blast media and grit on the drawing rather than leaving it to the shop.

### Design for line-of-sight spray

Like any sprayed coating, deep bores, blind pockets, internal channels and sharp inside corners receive less film than exposed faces. Where an internal surface must be protected, either provide gun access or select an immersion process.

### Check the cure temperature against the assembly

H-Series at 250 °F (121 °C) for 2 hours is well below the tempering temperature of hardened steel and does not affect most aluminum tempers, which is the main reason Cerakote is chosen over powder coating for finished, heat-treated parts. It is still too hot for many elastomers, plastics and adhesives — use an air-cure series or coat before assembly.

### Break sharp edges

As with any sprayed film, edges receive less coating and wear through first. Break outside edges to at least 0.010–0.020 in (0.25–0.5 mm) where the part will be handled or holstered.

### Verify performance claims against the specific product

Corrosion, abrasion and chemical resistance vary substantially across the product series, and published test data is per product and per film thickness. Specify the exact product code, the target dry film thickness and the acceptance test rather than the brand name alone.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Dry film thickness | ~0.001 in (25 µm) | Excess film reduces wear performance | Thin-film chemistry is engineered for a specific build |
| Blast profile | 100–120 grit aluminum oxide | Coarser profiles telegraph through | The film is too thin to fill a heavy profile |
| Bore allowance | 2 × film thickness on diameter | Mask instead | Film builds on both walls |
| Thread allowance | ~4 × film on pitch diameter | Mask instead | Thread geometry multiplies radial build |
| Oven cure (H-Series) | ~2 hours at 250 °F (121 °C) | Check elastomers and plastics | Low enough for hardened steel, not for all polymers |
| Outside edge break | 0.010–0.020 in (0.25–0.5 mm) | — | Sprayed films thin at sharp edges |

## Cost drivers

Cerakote is applied by hand at specialist shops, so it is priced by labor and part complexity rather than by area, over a lot minimum.

- **Masking.** This is the dominant cost on any part with threads, bores or fitted surfaces. Every plug and every tape edge is manual work applied and removed once per part.
- **Prep.** Degreasing, oil bake-out and blasting are separate operations. Parts that arrive with oil in porous castings or old coatings in place cost substantially more.
- **Color count and multi-color work.** Each additional color means another mask, spray and cure cycle. Patterned and graphic work is priced by the hour.
- **Batch size.** The oven cycle is fixed; a full oven costs little more than a single part.
- **Rework.** A defect requires stripping — usually blasting the coating off — and a complete recoat.

Cost-reduction tactics:

1. Reduce masked features. A design that puts fitted surfaces on a separate component removes the largest cost driver.
2. Specify a single color unless the second color has a functional purpose.
3. Batch parts to fill the oven cycle.
4. Give the shop clean, oil-free parts; bake-out is chargeable time.
5. Where the part is not heat sensitive and has no tight fits, price powder coating against Cerakote — powder is generally cheaper per part at volume, and the thin film is only worth paying for when the fits demand it.

## FAQ

### How thick is Cerakote?

A typical application is around 0.001 in (25 µm) per surface — roughly a quarter of a powder coat. That thinness is the point: it lets a part with real fits and moving surfaces be coated where a 2–4 mil powder film would interfere.

### What temperature does Cerakote cure at?

The H-Series oven-cure products cure at roughly 250 °F (121 °C) for about two hours, which is below the tempering temperature of hardened steel and does not affect most aluminum tempers. Air-cure series remove the oven entirely and can be used on polymers and assembled parts.

### Does Cerakote need masking on threads?

Usually yes. A 0.001 in film grows a thread's pitch diameter by roughly 0.004 in, which will interfere on a Class 3 fit, and a bore closes about 0.002 in on diameter. Mask threads, dowel holes, bearing journals and sealing lands, or machine them after coating.

### What surface preparation does Cerakote require?

Complete degreasing — often with an oil bake-out for porous or threaded parts — followed by abrasive blasting, typically with aluminum oxide in the 100–120 grit range. The film is too thin to fill a coarse profile, so blast grit and pressure directly control both adhesion and final appearance.

### Cerakote or powder coating?

Cerakote for thin film, low cure temperature and parts with tight fits or heat-sensitive tempers. Powder coating for a thicker, tougher film at lower cost per part in volume. Powder cures at 350–400 °F (177–204 °C) and lands at 2–4 mils, which rules it out where fits or temper are the constraint.

### What substrates can be Cerakoted?

Steel, stainless, aluminum and titanium with the oven-cure products, plus many polymers and assembled items with the air-cure series. Any substrate must tolerate abrasive blasting and, for H-Series, a 250 °F (121 °C) bake.

## Alternative processes

- [Powder Coating](https://manufacturingprocesses.org/processes/finishing/powder-coating.md): Powder coating sprays electrostatically charged dry polymer powder onto an earthed part, then bakes it into a tough continuous film.
- [PVD Coating](https://manufacturingprocesses.org/processes/finishing/pvd-coating.md): PVD coating vaporises a solid target inside a vacuum chamber so it condenses on the part as a very thin, extremely hard film.
- [Spray Painting](https://manufacturingprocesses.org/processes/finishing/spray-painting.md): Spray painting atomises liquid coating onto a prepared surface, building a film that cures into a decorative and protective layer.
- [Black Oxide](https://manufacturingprocesses.org/processes/finishing/black-oxide.md): Black oxide converts the steel surface to magnetite in a hot alkaline bath, giving a black finish that adds essentially no thickness.
- [Anodizing](https://manufacturingprocesses.org/processes/finishing/anodizing.md): Anodizing grows a hard porous oxide layer on aluminum electrochemically, which can then be dyed and sealed for wear and corrosion resistance.

## Related processes

- [Abrasive Blasting](https://manufacturingprocesses.org/processes/finishing/abrasive-blasting.md): Abrasive blasting propels media at a surface with compressed air to clean it, strip scale, or produce a uniform matte texture.
- [Spray Painting](https://manufacturingprocesses.org/processes/finishing/spray-painting.md): Spray painting atomises liquid coating onto a prepared surface, building a film that cures into a decorative and protective layer.
- [Powder Coating](https://manufacturingprocesses.org/processes/finishing/powder-coating.md): Powder coating sprays electrostatically charged dry polymer powder onto an earthed part, then bakes it into a tough continuous film.
- [PVD Coating](https://manufacturingprocesses.org/processes/finishing/pvd-coating.md): PVD coating vaporises a solid target inside a vacuum chamber so it condenses on the part as a very thin, extremely hard film.
- [Black Oxide](https://manufacturingprocesses.org/processes/finishing/black-oxide.md): Black oxide converts the steel surface to magnetite in a hot alkaline bath, giving a black finish that adds essentially no thickness.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/finishing/cerakote-ceramic-coating)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Chemical Conversion Coating"
category: "Finishing"
subcategory: "Additive"
materials: ["Metal"]
tolerances: "Chem film on aluminum adds 0.00001–0.00004 in (0.25–1 µm) and needs no allowance; manganese phosphate on steel can reach 0.0002–0.0004 in (5–10 µm) and should be checked on close fits"
volumes: "1 to millions of parts; immersion lines handle bulk baskets and racked work alike"
lead_time: "1–5 business days at a job shop; commonly same-day when run in line with cleaning"
url: https://manufacturingprocesses.org/processes/finishing/chemical-conversion-coating
---

# Chemical Conversion Coating

Chemical conversion coating reacts the metal surface with a chemical bath to form a thin protective, conductive or paint-adhering film.

- **Category**: [Finishing](https://manufacturingprocesses.org/processes/finishing.md)
- **Family**: Additive
- **Materials**: Metal
- **Typical tolerances**: Chem film on aluminum adds 0.00001–0.00004 in (0.25–1 µm) and needs no allowance; manganese phosphate on steel can reach 0.0002–0.0004 in (5–10 µm) and should be checked on close fits
- **Surface finish**: Conformal and submicron — reproduces the substrate finish exactly, with color from clear through gold iridescent to dark gray
- **Typical volumes**: 1 to millions of parts; immersion lines handle bulk baskets and racked work alike
- **Lead time**: 1–5 business days at a job shop; commonly same-day when run in line with cleaning

## Overview

A chemical conversion coating is formed by reacting the metal surface with a chemical bath so that part of the substrate is converted into an adherent, insoluble inorganic film. Nothing is plated on; the coating grows out of the metal itself, which is why these films are thin, tightly bonded, and add almost nothing to a dimension.

The two families that matter industrially are chromate conversion on aluminum — chem film, Alodine, Iridite, specified by MIL-DTL-5541 — and phosphate conversion on steel, covered by TT-C-490 and MIL-DTL-16232. Chem film on aluminum runs roughly 0.00001–0.00004 in (0.25–1 µm) thick. MIL-DTL-5541 defines two classes with very different purposes: Class 1A for maximum corrosion protection, which must survive 168 hours of ASTM B117 neutral salt spray, and Class 3 for electrical contact, which must hold contact resistance at or below 5000 microhms per square inch. Conversion coatings are used both as a standalone protective finish and as the adhesion base under paint and powder.

## How it works

1. **Clean.** Alkaline soak clean removes oil and shop soil. Every conversion coating failure traces back to cleaning.

2. **Deoxidize or etch.** Aluminum carries a natural oxide that blocks the reaction, so a mild acid deoxidizer strips it and leaves an active surface. Steel is pickled or activated depending on the phosphate type.

3. **React.** The part is immersed, sprayed or brushed with the conversion chemistry. The acid attacks the metal at the surface; the local pH rise at the interface precipitates an insoluble film of chromate, oxide or metal phosphate that is chemically bonded to the substrate. Time and temperature control film weight — typically 1–5 minutes for chem film on aluminum.

4. **Rinse and dry.** Rinsing must be thorough but not aggressive; a fresh chromate film is soft and gel-like until dried. Drying above roughly 140 °F (60 °C) can dehydrate and crack a chromate film, so dry-off temperature is controlled.

### The main chemistries

**Hexavalent chromate (MIL-DTL-5541 Type I)** on aluminum gives the well-known gold-iridescent film and the best self-healing corrosion protection, but hexavalent chromium is restricted under RoHS and REACH. **Trivalent chromium (Type II, and TCP processes)** is the compliant replacement, typically clear to pale blue, and is now the default on new designs.

**Iron phosphate** on steel is a light coating, typically 30–70 mg/ft², applied as a low-cost paint and powder base. **Zinc phosphate** is heavier, typically 150–400 mg/ft² as a paint base, and gives substantially better under-film corrosion performance — it is what separates a powder-coated part that lasts from one that blisters. **Manganese phosphate** is heavier still, dark gray, and is used with oil for anti-galling and break-in on gears, camshafts and firearms components.

## Design guidelines

### Use chem film where anodize would insulate

Anodize is a dielectric. A chromate conversion coating to MIL-DTL-5541 Class 3 protects aluminum while holding contact resistance at or below 5000 microhms/in², which is why it is specified on chassis grounding pads, RF enclosure flanges, connector mounting faces and bonding straps. A common arrangement is Type II anodize on the visible exterior with Class 3 chem film on the internal grounding surfaces, which requires masking and two operations.

### Treat it as dimensionally free

At 0.00001–0.00004 in (0.25–1 µm), chem film changes no fits and needs no allowance on threads, bores or gage surfaces. Phosphate coatings are heavier — a manganese phosphate can reach 0.0002–0.0004 in (5–10 µm) — and on close-fitting parts that build should be checked.

### Specify class and type, not just "chem film"

The drawing note must state the specification, type (hexavalent or trivalent) and class (1A or 3). A part called out only as "Alodine" leaves the shop to choose, and Class 1A and Class 3 are not interchangeable: the corrosion-optimized film is a poorer conductor and the conductive film is a weaker corrosion barrier.

### It is a base coat as much as a finish

Under paint or powder, the conversion layer is what the organic film mechanically keys into and what limits under-film creep from a scratch. Choosing iron phosphate over zinc phosphate to save money at the pretreatment stage is the most common reason a powder-coated steel part fails salt spray.

### Do not use it to hide surface defects

The film is submicron and completely conformal. Machining marks, weld discoloration, casting texture and blast profile all remain. Where appearance matters, set it with the mechanical prep before the bath.

### Handle it gently until sealed or painted

A fresh chromate film is soft and can be wiped off by handling before it dries and hardens. It is also degraded by drying above roughly 140 °F (60 °C). Parts should be racked, not stacked, and coated late in the routing.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Chem film thickness (aluminum) | 0.00001–0.00004 in (0.25–1 µm) | — | Conversion film, dimensionally negligible |
| MIL-DTL-5541 Class 1A | 168 h ASTM B117 minimum | — | Corrosion-optimized film |
| MIL-DTL-5541 Class 3 | ≤ 5000 µΩ/in² contact resistance | — | Electrical bonding surfaces |
| Iron phosphate coating weight | 30–70 mg/ft² | Paint base only | Light film; limited corrosion benefit |
| Zinc phosphate coating weight | 150–400 mg/ft² | — | Substantially better under-paint corrosion resistance |
| Drying temperature (chromate) | ≤ 140 °F (60 °C) | — | Higher temperatures dehydrate and crack the film |

## Cost drivers

Conversion coating is a low-cost, short-cycle tank or spray operation, usually the cheapest line on a finishing routing. Cost is driven by handling and by what else has to happen around it.

- **Application method.** Immersion in a line tank is cheapest. Spray application on large fabrications, and brush touch-up on individual machined faces, are hand operations priced by the hour.
- **Chemistry.** Trivalent and non-chrome systems generally cost more per gallon than hexavalent but avoid the hazardous-waste and compliance burden, which usually nets out in their favor.
- **Masking and dual finishes.** Anodizing the exterior and chem filming an internal ground pad on the same part means masking, two setups and two inspections — often several times the cost of either finish alone.
- **Testing.** Salt spray to 168 hours for Class 1A and contact resistance measurement for Class 3 are real costs on qualification lots.
- **Pretreatment level for paint base.** Upgrading from iron to zinc phosphate adds bath stages and sludge handling, and is the main reason one coater's price differs from another's.

Cost-reduction tactics:

1. Specify one class across the whole part where the requirements allow, and avoid dual-finish masking.
2. Use conversion coating rather than anodize where the part is painted afterward — it is cheaper and a better paint base.
3. Batch to fill the line; the tank cycle cost is fixed.
4. Where the design permits, put the ground path on a fastener or a machined boss that can be masked simply rather than on a large interrupted face.
5. Do not over-specify zinc phosphate on interior parts that will never see moisture; iron phosphate is materially cheaper.

## FAQ

### What is chem film and how thick is it?

Chem film — also called Alodine or Iridite — is a chromate conversion coating on aluminum specified by MIL-DTL-5541. It is roughly 0.00001–0.00004 in (0.25–1 µm) thick, thin enough to need no dimensional allowance on threads, bores or gage surfaces.

### What is the difference between MIL-DTL-5541 Class 1A and Class 3?

Class 1A is optimized for corrosion protection and must survive 168 hours of ASTM B117 neutral salt spray. Class 3 is optimized for electrical contact and must hold contact resistance at or below 5000 microhms per square inch. They are not interchangeable, so the drawing must state which one applies.

### Should I specify chem film or anodizing on an aluminum enclosure?

Anodize where you want durability, wear resistance and color; chem film where the surface must stay electrically conductive, where the part will be painted afterward, or where no dimensional change is acceptable. Many enclosures use both — anodize outside, Class 3 chem film on internal grounding pads — which requires masking and two operations.

### Is hexavalent chromate still allowed?

It remains specified in some aerospace and defense work but is restricted under RoHS and REACH, so trivalent chromium (MIL-DTL-5541 Type II and TCP processes) is the default for new designs. Trivalent films are typically clear to pale blue rather than gold iridescent and offer somewhat less self-healing behavior.

### Iron phosphate or zinc phosphate under powder coating?

Iron phosphate at roughly 30–70 mg/ft² is cheaper and adequate for interior parts. Zinc phosphate at roughly 150–400 mg/ft² gives substantially better under-film corrosion resistance and much less creep from a scratch. Choosing iron phosphate to save money is the most common reason a powder-coated steel part fails salt spray.

### What is manganese phosphate used for?

It is a heavy, dark gray phosphate typically 0.0002–0.0004 in (5–10 µm) thick, used with oil as a break-in and anti-galling surface on gears, camshafts, fasteners and firearms components. The porous coating holds oil, which is what provides the lubricity and the corrosion resistance.

## Alternative processes

- [Anodizing](https://manufacturingprocesses.org/processes/finishing/anodizing.md): Anodizing grows a hard porous oxide layer on aluminum electrochemically, which can then be dyed and sealed for wear and corrosion resistance.
- [Black Oxide](https://manufacturingprocesses.org/processes/finishing/black-oxide.md): Black oxide converts the steel surface to magnetite in a hot alkaline bath, giving a black finish that adds essentially no thickness.
- [Passivation](https://manufacturingprocesses.org/processes/finishing/passivation.md): Passivation removes free iron from a stainless steel surface with an acid bath so the chromium oxide layer can reform unbroken.
- [Electroplating](https://manufacturingprocesses.org/processes/finishing/electroplating.md): Electroplating deposits a metal coating from solution onto a conductive part by passing current through an electrolyte bath.
- [E-Coating (Electrophoretic Deposition)](https://manufacturingprocesses.org/processes/finishing/e-coating-electrophoretic-deposition.md): E-coating deposits paint electrically from a bath so it plates evenly into every recess, then cures to a uniform film.

## Related processes

- [Anodizing](https://manufacturingprocesses.org/processes/finishing/anodizing.md): Anodizing grows a hard porous oxide layer on aluminum electrochemically, which can then be dyed and sealed for wear and corrosion resistance.
- [Powder Coating](https://manufacturingprocesses.org/processes/finishing/powder-coating.md): Powder coating sprays electrostatically charged dry polymer powder onto an earthed part, then bakes it into a tough continuous film.
- [Spray Painting](https://manufacturingprocesses.org/processes/finishing/spray-painting.md): Spray painting atomises liquid coating onto a prepared surface, building a film that cures into a decorative and protective layer.
- [E-Coating (Electrophoretic Deposition)](https://manufacturingprocesses.org/processes/finishing/e-coating-electrophoretic-deposition.md): E-coating deposits paint electrically from a bath so it plates evenly into every recess, then cures to a uniform film.
- [Black Oxide](https://manufacturingprocesses.org/processes/finishing/black-oxide.md): Black oxide converts the steel surface to magnetite in a hot alkaline bath, giving a black finish that adds essentially no thickness.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/finishing/chemical-conversion-coating)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "E-Coating (Electrophoretic Deposition)"
category: "Finishing"
subcategory: "Additive"
materials: ["Metal"]
tolerances: "Adds 0.6–1.2 mils (15–30 µm) per surface, uniform over the whole part including recesses; budget roughly 4× the film thickness on thread pitch diameter"
volumes: "Economical from a few thousand parts per run upward; automotive lines run millions"
lead_time: "3–10 business days at a job coater; production lines run continuously once qualified"
url: https://manufacturingprocesses.org/processes/finishing/e-coating-electrophoretic-deposition
---

# E-Coating (Electrophoretic Deposition)

E-coating deposits paint electrically from a bath so it plates evenly into every recess, then cures to a uniform film.

- **Category**: [Finishing](https://manufacturingprocesses.org/processes/finishing.md)
- **Family**: Additive
- **Materials**: Metal
- **Typical tolerances**: Adds 0.6–1.2 mils (15–30 µm) per surface, uniform over the whole part including recesses; budget roughly 4× the film thickness on thread pitch diameter
- **Surface finish**: Uniform matte to semi-gloss; reproduces the substrate and does not fill weld spatter, scale or porosity
- **Typical volumes**: Economical from a few thousand parts per run upward; automotive lines run millions
- **Lead time**: 3–10 business days at a job coater; production lines run continuously once qualified

## Overview

E-coating deposits paint electrically. The part is immersed in a waterborne bath of charged resin particles and made an electrode; DC voltage drives the particles onto every wetted surface until the growing film insulates the substrate and deposition stops on its own. That self-limiting behavior is what makes the process unique — film thickness converges to a uniform 0.6–1.2 mils (15–30 µm) over the entire part, including inside box sections, weld seams, threads and blind recesses that no spray gun can reach.

Cathodic epoxy e-coat is the standard automotive body primer and the reference for corrosion performance on complex steel weldments. Anodic acrylic systems are cheaper and used for less demanding indoor work. Deposition takes 2–3 minutes at 50–400 V, followed by an ultrafiltrate rinse that recovers undeposited paint and a bake at roughly 350–375 °F (177–191 °C). Material utilization exceeds 95%, and because the process is fully automated and immersion-based, unit cost at volume is very low.

## How it works

1. **Pretreat.** A multi-stage washer cleans, then applies a phosphate or zirconium conversion coating. As with any paint system, this stage — not the e-coat — sets corrosion performance. Cathodic epoxy over zinc phosphate is the combination that delivers the long salt-spray results the process is known for.

2. **Deposit.** The part is immersed in the e-coat bath, typically 15–22% solids in deionized water, and connected as cathode (for cathodic epoxy) against counter-electrodes in the tank. At 50–400 V DC, charged resin micelles migrate to the part and coagulate on it. As the film builds, its electrical resistance rises and deposition slows, then stops — so low-current areas keep coating after high-current areas have finished. Dwell is typically 2–3 minutes.

3. **Rinse.** Ultrafiltration permeate rinses undeposited bath solution off the part and returns it to the tank. This closed loop is why utilization exceeds 95%.

4. **Bake.** A cure oven at roughly 350–375 °F (177–191 °C) for about 20 minutes crosslinks the film and flows it out to a smooth, hard finish.

5. **Topcoat if required.** Epoxy e-coat chalks quickly under UV, so exterior parts get a powder or liquid topcoat over the e-coat. The e-coat is doing the corrosion work; the topcoat is doing the appearance and weathering work.

### Throwpower

The practical measure of an e-coat line is throwpower — how far into a recess or box section the coating will deposit at a given voltage. It is what separates e-coat from every spray process. A tubular frame, a hem flange, a spot-welded overlap and the inside of a threaded hole all receive film, which is precisely where spray-coated assemblies begin to corrode.

## Design guidelines

### Provide drain and flow-out holes

The part is immersed and withdrawn. Any cup, pocket or enclosed volume carries bath solution out of the tank, which then drains onto the part during bake and leaves runs, or is trapped and cures as a puddle. Put drain holes at the low point in the dip orientation and vent holes at the high point, and design the hang angle so liquid sheets off rather than pooling.

### Design in an electrical contact point

Every part needs a hard, repeatable ground contact through the rack. That contact area is uncoated. Designate it — a hanging hole, a boss, a face that is later hidden — and make sure it is metal-to-metal clean, because a resistive contact produces a thin film everywhere.

### Budget 0.6–1.2 mils on every surface

The film is thin compared to powder but it is genuinely everywhere, including in threads. On a 1/4-20 thread, 1 mil radial builds roughly 0.004 in on pitch diameter. That is usually tolerable on a coarse thread but not on a fine or Class 3 fit; mask or chase where it matters.

### The part must conduct and must survive the bake

E-coat needs a conductive substrate — steel, aluminum, zinc die castings, galvanized steel. Plastics can only be coated if metallized first. The bake at 350–375 °F (177–191 °C) rules out most elastomers, thermoplastics and electronics, so parts are coated before those items are installed.

### Do not expect a Class A appearance or a color range

E-coat is typically black or gray, matte to semi-gloss, and it reproduces the substrate — mill scale, weld spatter and porosity all show. It is a primer and a functional coating. Where appearance is the requirement, e-coat then topcoat.

### Size the part to the tank

The whole part must be immersed, so the tank dimensions are a hard limit, and unlike a spray booth there is no partial-coverage workaround.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Film thickness | 0.6–1.2 mils (15–30 µm) | ~1.5 mils (38 µm) | The film is self-limiting; more voltage does not add much |
| Drain hole | At the low point in dip orientation | — | Trapped bath solution runs and puddles during bake |
| Vent hole | At the high point | — | Air pockets block deposition entirely |
| Ground contact | 1 designated location, bare metal | — | Contact area is uncoated and resistance thins the whole film |
| Thread allowance | ~4 × film thickness on pitch dia. | Mask fine threads | Thread geometry multiplies radial build |
| Component temperature rating | Above 375 °F (191 °C) | — | The cure oven destroys most plastics and elastomers |

## Cost drivers

E-coating is a high-fixed-cost, low-variable-cost process. A line represents a large capital investment in tanks, rectifiers, ultrafiltration, a washer and an oven, and the economics only work when that line runs full.

- **Volume.** Cost per part falls dramatically with volume. Below a few hundred parts a run, spray or powder is almost always cheaper; above a few thousand, e-coat is difficult to beat on a per-square-foot basis.
- **Rack density.** As with any immersion process, price tracks how many parts fit on a rack and through the line per hour.
- **Pretreatment stages.** A three-stage washer is cheaper than a five-stage zinc phosphate system, and the difference shows up directly in corrosion performance.
- **Color and topcoat.** E-coat comes in a narrow range, usually black. Any color requirement means a second coating operation on top, doubling the handling.
- **Part size.** Tank dimensions cap part size; oversized parts are simply not quotable at that line.

Cost-reduction tactics:

1. Consolidate parts onto a single e-coat run to fill racks and amortize line time.
2. Design drain and vent holes in from the start; rework for runs and puddles is the most common quality cost.
3. Use e-coat as the corrosion primer and powder only as the topcoat where color is needed — the combination usually outperforms either alone at a lower total film build.
4. Accept the standard black where appearance does not matter and skip the topcoat entirely.
5. Keep parts within the tank envelope; splitting a weldment into two coatable pieces is often cheaper than finding an oversized line.

## FAQ

### How thick is e-coating?

Typically 0.6–1.2 mils (15–30 µm), and unusually uniform. The film is self-limiting: as it builds, its electrical resistance rises and deposition slows and stops, so recesses keep coating after exposed faces have finished. Increasing voltage does not produce a much thicker film.

### Why does e-coating reach inside recesses when spraying cannot?

Deposition is driven by an electric field in an immersion bath rather than by line-of-sight delivery, and the growing film insulates the areas that have already coated. That property — throwpower — lets the coating reach box sections, hem flanges, spot-welded overlaps and threaded holes, which is exactly where spray-coated assemblies start to corrode.

### Does e-coat need a topcoat?

For exterior use, yes. Epoxy e-coat chalks quickly under UV, so it is used as a corrosion primer with a powder or liquid topcoat over it for appearance and weathering. Indoors, or on hidden structure, the e-coat is frequently the only coating.

### What substrates can be e-coated?

Anything conductive that survives the bake — steel, aluminum, zinc die castings and galvanized steel. Plastics must be metallized first. The cure oven at roughly 350–375 °F (177–191 °C) rules out most elastomers, thermoplastics and electronic components, so those are installed after coating.

### E-coat or powder coat?

E-coat wins on uniformity, recess coverage and corrosion protection of complex weldments, at 0.6–1.2 mils. Powder wins on film toughness, color range and appearance, at 2–4 mils, and is viable at low volumes. Many production parts use both: e-coat as primer, powder as topcoat.

### At what volume does e-coating make economic sense?

The process carries high fixed cost in tanks, rectifiers, ultrafiltration and an oven, so it needs volume to amortize. Below a few hundred parts per run, spray or powder is usually cheaper; above a few thousand, e-coat is hard to beat on cost per square foot of coverage.

## Alternative processes

- [Powder Coating](https://manufacturingprocesses.org/processes/finishing/powder-coating.md): Powder coating sprays electrostatically charged dry polymer powder onto an earthed part, then bakes it into a tough continuous film.
- [Spray Painting](https://manufacturingprocesses.org/processes/finishing/spray-painting.md): Spray painting atomises liquid coating onto a prepared surface, building a film that cures into a decorative and protective layer.
- [Electroplating](https://manufacturingprocesses.org/processes/finishing/electroplating.md): Electroplating deposits a metal coating from solution onto a conductive part by passing current through an electrolyte bath.
- [Galvanizing](https://manufacturingprocesses.org/processes/finishing/galvanizing.md): Galvanizing coats steel with zinc, usually by hot dipping, so the zinc corrodes sacrificially and protects the steel beneath it.
- [Chemical Conversion Coating](https://manufacturingprocesses.org/processes/finishing/chemical-conversion-coating.md): Chemical conversion coating reacts the metal surface with a chemical bath to form a thin protective, conductive or paint-adhering film.

## Related processes

- [Powder Coating](https://manufacturingprocesses.org/processes/finishing/powder-coating.md): Powder coating sprays electrostatically charged dry polymer powder onto an earthed part, then bakes it into a tough continuous film.
- [Spray Painting](https://manufacturingprocesses.org/processes/finishing/spray-painting.md): Spray painting atomises liquid coating onto a prepared surface, building a film that cures into a decorative and protective layer.
- [Chemical Conversion Coating](https://manufacturingprocesses.org/processes/finishing/chemical-conversion-coating.md): Chemical conversion coating reacts the metal surface with a chemical bath to form a thin protective, conductive or paint-adhering film.
- [Electroplating](https://manufacturingprocesses.org/processes/finishing/electroplating.md): Electroplating deposits a metal coating from solution onto a conductive part by passing current through an electrolyte bath.
- [Galvanizing](https://manufacturingprocesses.org/processes/finishing/galvanizing.md): Galvanizing coats steel with zinc, usually by hot dipping, so the zinc corrodes sacrificially and protects the steel beneath it.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/finishing/e-coating-electrophoretic-deposition)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Electroplating"
category: "Finishing"
subcategory: "Additive"
materials: ["Metal", "Plastic"]
tolerances: "Zinc 0.0002–0.001 in (5–25 µm), electroless nickel 0.0005–0.002 in (13–50 µm), hard chrome 0.0002–0.010 in (5–250 µm) per surface. Electrolytic deposits vary 2:1 or more between high- and low-current areas; electroless nickel holds about ±10%"
volumes: "Barrel plating is economical from a few pounds of hardware to millions of pieces; rack plating from 1 part upward"
lead_time: "3–10 business days at a job shop; add 1 day for an embrittlement-relief bake"
url: https://manufacturingprocesses.org/processes/finishing/electroplating
---

# Electroplating

Electroplating deposits a metal coating from solution onto a conductive part by passing current through an electrolyte bath.

- **Category**: [Finishing](https://manufacturingprocesses.org/processes/finishing.md)
- **Family**: Additive
- **Materials**: Metal, Plastic
- **Typical tolerances**: Zinc 0.0002–0.001 in (5–25 µm), electroless nickel 0.0005–0.002 in (13–50 µm), hard chrome 0.0002–0.010 in (5–250 µm) per surface. Electrolytic deposits vary 2:1 or more between high- and low-current areas; electroless nickel holds about ±10%
- **Surface finish**: Reproduces or slightly amplifies the substrate finish; bright baths add gloss but do not level deep tool marks
- **Typical volumes**: Barrel plating is economical from a few pounds of hardware to millions of pieces; rack plating from 1 part upward
- **Lead time**: 3–10 business days at a job shop; add 1 day for an embrittlement-relief bake

## Overview

Electroplating deposits a metal coating onto a conductive part by making it the cathode in an electrolyte containing dissolved ions of the plating metal. Current drives ions out of solution onto the surface, and Faraday's law makes thickness a direct function of current and time — which is why plating specifications are written in thickness, not in dip time.

Common deposits and their working thicknesses: zinc for sacrificial corrosion protection at 0.0002–0.001 in (5–25 µm), electroless nickel for uniform hardness and corrosion resistance at 0.0005–0.002 in (13–50 µm), hard chrome for wear surfaces at 0.0002–0.010 in (5–250 µm), and decorative chrome at 0.00001–0.00002 in (0.25–0.5 µm) over a nickel underlayer that does the actual protecting. ASTM B633 codifies zinc thickness by service condition: 5 µm for mild, 8 µm moderate, 13 µm severe and 25 µm very severe exposure. Any base metal that conducts can be plated; plastics require an electroless metallization step first.

## How it works

1. **Clean.** Alkaline soak and electroclean remove oil and shop soil. Plating adhesion failures are almost always cleaning failures, not bath failures.

2. **Activate.** An acid dip strips oxide and leaves an electrochemically active surface. Stainless steel, aluminum and titanium re-oxidize within seconds, so they need a special strike: a Wood's nickel strike for stainless, a double zincate for aluminum.

3. **Strike.** A thin, high-adhesion first layer from a low-efficiency bath anchors the main deposit. Copper strike under nickel on steel, nickel strike under chrome.

4. **Plate.** The part is racked or loaded into a barrel and made the cathode. Thickness follows current density × time, so the rectifier setting and the dwell define the deposit. Rack plating gives control and a cosmetic surface; barrel plating tumbles thousands of small parts at once for a fraction of the cost, with more thickness scatter and part-to-part contact marks.

5. **Post-treat.** Zinc deposits are chromate-conversion passivated immediately — trivalent clear, blue, black or iridescent — because bare zinc white-rusts within days. Electroless nickel is often heat treated at 750 °F (400 °C) for 1 hour to raise as-plated hardness from roughly 500–550 HV to about 900–1000 HV.

6. **Bake for hydrogen embrittlement.** Acid cleaning and plating charge atomic hydrogen into high-strength steel. Parts at or above roughly 40 HRC must be baked, typically at 375 ± 25 °F (190 ± 14 °C), starting within 4 hours of plating; durations run from 3 hours for moderate-strength parts to 23 hours or more for the highest-strength grades. This step is mandatory, not optional, and skipping it produces delayed brittle fracture days or weeks into service.

### Throwing power

Current density is not uniform. Edges, corners and outside features see high current and plate thick; recesses, bore interiors and the inside of a tube see low current and plate thin — 2:1 or worse across one part is normal for bright acid baths. Electroless nickel is the exception: it is an autocatalytic chemical reaction with no current, so it deposits within about ±10% of nominal on every wetted surface including deep bores and blind holes.

## Design guidelines

### Specify thickness at the worst-case location

A drawing that says "zinc plate per ASTM B633 SC3" is asking for 13 µm minimum, and the inspector measures at the low-current-density point. Call out where thickness is measured, or the shop will plate to hit the number on an edge and leave the recess bare.

### Allow 4 × the radial deposit on threads

Coating on a 60° thread changes pitch diameter by roughly four times the radial thickness. A 0.0002 in (5 µm) zinc deposit consumes about 0.0008 in of pitch-diameter clearance, which is why fastener specifications limit plating thickness rather than maximizing it. On tapped holes, either plate before tapping or specify an oversize tap.

### Break sharp edges and avoid deep blind recesses

Sharp corners build a thick, rough, sometimes nodular deposit ("treeing"); deep recesses starve. Break edges to 0.010–0.030 in (0.25–0.75 mm) and open up narrow slots. Where geometry cannot change, the shop must add auxiliary anodes or current thieves, which is setup labor.

### Design in a rack or contact point

Racked parts carry a contact mark; barrel parts carry random contact marks and can nest. Add a hanging hole or specify an acceptable contact area. Parts that nest — thin discs, cupped stampings, springs — should be rack plated even though it costs more.

### Choose the deposit for the job, not the look

Zinc protects steel sacrificially and is the default for fasteners and brackets. Zinc-nickel gives substantially longer salt-spray life than plain zinc and is the automotive under-hood standard. Electroless nickel gives uniform thickness on complex geometry plus hardness. Hard chrome gives roughly 850–1100 HV for wear surfaces but is deposited with a micro-crack network and does not by itself protect steel from corrosion. Decorative chrome is a flash over nickel — the nickel provides the corrosion barrier.

### Plan for aluminum and plastic separately

Aluminum needs a zincate pretreatment because its oxide reforms instantly. Plastics need an etched, catalyzed surface and an electroless copper or nickel layer before electroplating; ABS and ABS/PC are the standard platable resins, and glass-filled or highly crystalline resins generally are not.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Zinc thickness (ASTM B633) | 8–13 µm (SC2–SC3) | 5 µm (SC1) | Below SC1 the coating cannot protect the substrate |
| Thread pitch-dia. allowance | 4 × radial thickness | Plate before tapping | Thread geometry multiplies radial build |
| Edge break before plating | 0.010–0.030 in (0.25–0.75 mm) | — | Sharp edges grow thick, rough, nodular deposits |
| Recess depth vs. width | ≤ 1 : 1 for electrolytic baths | Use electroless nickel | Low current density starves deep recesses |
| Embrittlement relief bake | 375 ± 25 °F (190 ± 14 °C) | Start within 4 h of plating | Absorbed hydrogen causes delayed fracture above ~40 HRC |
| Electroless nickel uniformity | ±10% of nominal | — | Chemical, not electrical, deposition

## Cost drivers

Plating is quoted by weight or area for barrel work and per rack for rack work, over a lot minimum.

- **Rack vs. barrel.** Barrel plating is the cheapest metal finish available for small hardware because thousands of parts share one cycle. Rack plating costs many times more per part and is required for anything cosmetic, fragile, or prone to nesting.
- **Deposit metal.** Zinc is inexpensive. Electroless nickel is expensive because the bath is consumed as it plates and has a finite number of metal turnovers. Gold and silver are priced on the metal.
- **Thickness.** Time on the rectifier scales with thickness, so an SC4 (25 µm) zinc call-out costs materially more than SC1 (5 µm) — specify the service condition the part actually sees.
- **Masking and secondary operations.** Masking, embrittlement-relief baking, and post-plate thread chasing are separate labor lines.
- **Specification burden.** Certifications, thickness reports, salt-spray testing and adhesion testing per lot add cost that has nothing to do with the plating itself.

Cost-reduction tactics:

1. Design small parts so they can be barrel plated — no nesting shapes, no fragile projections, no deep cups.
2. Specify the ASTM B633 service condition your application needs rather than defaulting to the thickest option.
3. Use zinc-nickel or a better chromate rather than more zinc when salt-spray life is the requirement.
4. Keep base-metal hardness below about 40 HRC where the design allows, and the embrittlement bake disappears.
5. Where geometry is deeply recessed, price electroless nickel against electrolytic plating plus auxiliary anodes — it is often cheaper overall.

## FAQ

### How thick is electroplating?

It depends on the deposit and the specification. ASTM B633 sets zinc at 5 µm (0.0002 in) for mild service up to 25 µm (0.001 in) for very severe service. Electroless nickel typically runs 0.0005–0.002 in (13–50 µm), hard chrome 0.0002–0.010 in (5–250 µm), and decorative chrome only 0.00001–0.00002 in (0.25–0.5 µm) over nickel.

### What is hydrogen embrittlement relief baking and when is it required?

Acid cleaning and plating drive atomic hydrogen into high-strength steel, which later causes delayed brittle fracture. Parts at or above roughly 40 HRC must be baked, typically at 375 ± 25 °F (190 ± 14 °C), beginning within 4 hours of plating. Bake times run from about 3 hours for moderate-strength parts to 23 hours or more for the highest-strength grades.

### Why does plating come out thicker on edges than in recesses?

Current density concentrates at edges and corners and starves in recesses, so a typical bright acid bath deposits 2:1 or more thickness between the two. Auxiliary anodes and current thieves help. Electroless nickel avoids the problem entirely because it plates chemically, holding about ±10% on every wetted surface including blind holes.

### Can you electroplate aluminum or plastic?

Yes, but neither goes straight into the bath. Aluminum needs a double zincate pretreatment because its oxide reforms in seconds. Plastic needs a chemical etch, a catalyst, and an electroless copper or nickel layer to make it conductive — ABS and ABS/PC are the standard platable resins.

### How much does plating add to a thread?

Thread geometry multiplies radial coating thickness by roughly four on pitch diameter, so a 0.0002 in (5 µm) zinc deposit consumes about 0.0008 in of pitch-diameter clearance. That is why fastener standards cap plating thickness. For tapped holes, plate before tapping or use an oversize tap.

### What is the difference between hard chrome and decorative chrome?

Hard chrome is a functional wear coating deposited 0.0002–0.010 in (5–250 µm) thick at roughly 850–1100 HV, plated directly for hydraulic rods and molds. Decorative chrome is a 0.00001–0.00002 in (0.25–0.5 µm) flash over bright nickel — the nickel underneath provides the corrosion protection, and the chrome only supplies color and tarnish resistance.

## Alternative processes

- [Galvanizing](https://manufacturingprocesses.org/processes/finishing/galvanizing.md): Galvanizing coats steel with zinc, usually by hot dipping, so the zinc corrodes sacrificially and protects the steel beneath it.
- [PVD Coating](https://manufacturingprocesses.org/processes/finishing/pvd-coating.md): PVD coating vaporises a solid target inside a vacuum chamber so it condenses on the part as a very thin, extremely hard film.
- [Powder Coating](https://manufacturingprocesses.org/processes/finishing/powder-coating.md): Powder coating sprays electrostatically charged dry polymer powder onto an earthed part, then bakes it into a tough continuous film.
- [Chemical Conversion Coating](https://manufacturingprocesses.org/processes/finishing/chemical-conversion-coating.md): Chemical conversion coating reacts the metal surface with a chemical bath to form a thin protective, conductive or paint-adhering film.
- [E-Coating (Electrophoretic Deposition)](https://manufacturingprocesses.org/processes/finishing/e-coating-electrophoretic-deposition.md): E-coating deposits paint electrically from a bath so it plates evenly into every recess, then cures to a uniform film.

## Related processes

- [Galvanizing](https://manufacturingprocesses.org/processes/finishing/galvanizing.md): Galvanizing coats steel with zinc, usually by hot dipping, so the zinc corrodes sacrificially and protects the steel beneath it.
- [Electroforming](https://manufacturingprocesses.org/processes/forming/electroforming.md): Electroforming grows a metal shell by electrodeposition onto a mandrel that is afterwards removed, producing thin parts with sub-micron detail.
- [Chemical Conversion Coating](https://manufacturingprocesses.org/processes/finishing/chemical-conversion-coating.md): Chemical conversion coating reacts the metal surface with a chemical bath to form a thin protective, conductive or paint-adhering film.
- [Electropolishing](https://manufacturingprocesses.org/processes/finishing/electropolishing.md): Electropolishing reverses plating: current dissolves the surface fastest at the peaks, leaving a bright, deburred, passive finish.
- [Passivation](https://manufacturingprocesses.org/processes/finishing/passivation.md): Passivation removes free iron from a stainless steel surface with an acid bath so the chromium oxide layer can reform unbroken.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/finishing/electroplating)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Electropolishing"
category: "Finishing"
subcategory: "Subtractive"
materials: ["Metal"]
tolerances: "Removes 0.0002–0.001 in (5–25 µm) per surface, so bores grow and shafts shrink by twice that on diameter; sharp external edges break by 0.002–0.003 in (50–75 µm)"
volumes: "1 part to high-volume barrel work; small parts can be run in bulk baskets"
lead_time: "3–7 business days at a job shop; 1–2 days expedited"
url: https://manufacturingprocesses.org/processes/finishing/electropolishing
---

# Electropolishing

Electropolishing reverses plating: current dissolves the surface fastest at the peaks, leaving a bright, deburred, passive finish.

- **Category**: [Finishing](https://manufacturingprocesses.org/processes/finishing.md)
- **Family**: Subtractive
- **Materials**: Metal
- **Typical tolerances**: Removes 0.0002–0.001 in (5–25 µm) per surface, so bores grow and shafts shrink by twice that on diameter; sharp external edges break by 0.002–0.003 in (50–75 µm)
- **Surface finish**: Roughly halves incoming Ra — 32 µin (0.8 µm) in typically yields about 16 µin (0.4 µm) out; mirror results require mechanical polishing first
- **Typical volumes**: 1 part to high-volume barrel work; small parts can be run in bulk baskets
- **Lead time**: 3–7 business days at a job shop; 1–2 days expedited

## Overview

Electropolishing is electroplating run backwards. The part is the anode in a viscous acid electrolyte, and current dissolves metal preferentially from the high points of the surface profile, leaving a bright, micro-smooth, deburred and chemically passive finish. It is a removal process, not a coating: expect 0.0002–0.001 in (5–25 µm) off each surface in a typical cycle, and up to 0.0015 in (38 µm) on heavier work.

The characteristic result is a roughly 50% reduction in Ra — a 32 µin (0.8 µm) machined surface commonly finishes near 16 µin (0.4 µm) — combined with removal of the mechanically disturbed surface layer and enrichment of chromium at the surface of stainless steel. ASTM B912 recognizes electropolishing as a method of passivating stainless. It is the standard finish for pharmaceutical, semiconductor, food and vacuum components, where cleanability and low particle entrapment matter more than appearance.

## How it works

1. **Clean and rack.** Oil, drawing compound and shop soil must be gone before the part enters the bath. Racking is critical: current has to enter the part somewhere, and that contact point does not polish.

2. **Immerse and energize.** The part hangs as the anode in a phosphoric/sulfuric acid electrolyte typically run at 100–190 °F (38–88 °C), with cathodes positioned around it. Current density is usually in the 50–500 A/ft² range depending on alloy and bath.

3. **Anodic dissolution.** A viscous, ion-rich boundary layer forms at the surface. It is thinner over peaks than over valleys, so peaks see lower resistance, higher local current density and faster dissolution. The profile flattens from the top down. Removal rate is roughly proportional to current density and time, so the operator controls stock loss by the amp-minutes applied.

4. **Rinse and neutralize.** Multi-stage rinsing followed by a neutralizing dip removes entrained acid, especially from blind features and threads.

5. **Passivate (inherent).** On austenitic stainless, the iron dissolves preferentially, leaving a surface enriched in chromium that immediately forms a passive chromium oxide film. This is why an electropolished part usually needs no separate passivation step.

### What it does and does not fix

Electropolishing levels micro-roughness — the fine peaks left by machining, grinding or blasting. It does not level waviness or long-wavelength form error, and it will not remove a deep scratch, a tool witness line, a weld undercut or a pit. Those defects come out of the tank brighter and often more visible, because the surrounding surface is now mirror-like. Grinding, sanding or polishing to a uniform starting Ra is the prerequisite; see the [surface finish chart](/charts/surface-finish-chart) for the bands each prep leaves.

## Design guidelines

### Start from a uniform surface

The finish is only as good as what goes in. Electropolishing roughly halves Ra, so a 63 µin (1.6 µm) as-machined surface comes out near 32 µin (0.8 µm), and a 16 µin ground surface comes out near 8 µin. If a mirror finish is the requirement, mechanical polishing precedes the tank; the electropolish then brightens and passivates it.

### Expect edges and corners to round

Current density is highest at edges, external corners and thin sections, so those areas dissolve fastest. A sharp corner will visibly break, and thin knife edges can lose 0.002–0.003 in (50–75 µm). Where a sharp edge is functional — a sealing land, a cutting edge, a mating datum — mask it or move the operation earlier in the routing.

### Budget the stock loss on toleranced features

Removal is 0.0002–0.001 in (5–25 µm) per surface, so a bore grows and a shaft shrinks by twice that on diameter. Machine to the pre-polish dimension and state on the drawing which condition the tolerance applies to.

### Threads and blind features are problem areas

External thread crests polish aggressively while roots polish little, and internal threads may barely polish at all because current cannot reach them. Blind holes, deep bores and internal passages trap electrolyte and need designed drainage — a cross-drilled vent or a through-hole. Anything that traps acid becomes a corrosion site later.

### Choose the alloy deliberately

300-series austenitic stainless electropolishes exceptionally well and is the default. 400-series martensitic and precipitation-hardening grades polish acceptably but less brightly. Free-machining grades with sulfur (303) leave sulfide inclusions that pit and streak. Cast and sintered parts reveal their porosity. Copper alloys, nickel alloys, aluminum and titanium can be electropolished in specialized baths.

### Provide a rack contact location

The contact point does not polish and often carries a burn mark. Designate an acceptable location — an internal thread, a mounting hole, an end face that gets machined later.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Stock removal per surface | 0.0002–0.001 in (5–25 µm) | 0.0015 in (38 µm) | Removal scales with amp-minutes |
| Ra improvement | About 50% of the incoming value | — | Micro-peaks dissolve first; waviness does not |
| Incoming Ra for a bright result | ≤ 32 µin (0.8 µm) | — | The process brightens, it does not level |
| Sharp external edges | Mask, or accept 0.002–0.003 in break | — | Current density peaks at edges |
| Blind features | Provide a drain or vent path | — | Trapped electrolyte causes later corrosion |
| Best alloys | 304, 316, 316L | Avoid 303 and free-machining grades | Sulfide inclusions pit and streak |

## Cost drivers

Electropolishing is quoted per part or per rack over a lot minimum, and is materially more expensive per unit area than passivation because it consumes electricity, bath chemistry and rack time.

- **Rack density and fixturing.** Each part needs its own electrical contact and a clear path to a cathode. Complex internal geometry may need dedicated internal cathodes, which is tooling.
- **Incoming surface condition.** A part that must reach a specified Ra requires mechanical prep first, and that prep usually costs more than the electropolishing.
- **Alloy and thickness of removal.** Heavier removal means more amp-minutes and more bath consumption.
- **Verification.** Ra measurement, passivation testing to ASTM A967 practices, and documentation for pharmaceutical or semiconductor work are real line items.
- **Masking.** Sealing lands, threads and mating datums that must not be attacked all require masking labor.

Cost-reduction tactics:

1. Specify the Ra you need, not the brightest available — every step down the finish scale multiplies the prep cost.
2. Design a rack contact point in a non-critical area so the shop does not choose one for you.
3. Give internal passages a through path so they can be flushed rather than individually cathode-fixtured.
4. Avoid 303 and other free-machining grades on parts that will be electropolished; 304 and 316 cost less to finish.
5. Where the requirement is only corrosion resistance and cleanliness, price passivation against electropolishing — passivation is far cheaper and adds no dimensional change.

## FAQ

### How much material does electropolishing remove?

A typical cycle removes 0.0002–0.001 in (5–25 µm) per surface, and heavy work can reach 0.0015 in (38 µm). A bore therefore grows and a shaft shrinks by twice that amount on diameter, so toleranced features must be machined to a pre-polish dimension.

### How much does electropolishing improve surface finish?

It reduces Ra by roughly 50%. A 32 µin (0.8 µm) machined surface commonly finishes near 16 µin (0.4 µm). It levels micro-roughness only — waviness, deep scratches, tool witness lines and weld undercut survive and often become more visible against the brightened background.

### Does electropolishing passivate stainless steel?

Yes. Iron dissolves preferentially, leaving a chromium-enriched surface that immediately forms a passive oxide film. ASTM B912 recognizes electropolishing as a method of passivating stainless steel, which is why an electropolished part normally needs no separate passivation step.

### Why do sharp edges round off during electropolishing?

Current density concentrates at edges, external corners and thin sections, so metal dissolves fastest there. Sharp corners visibly break and knife edges can lose 0.002–0.003 in (50–75 µm). Mask any edge that has to stay sharp, such as a sealing land or a mating datum.

### Which stainless grades electropolish best?

304, 316 and 316L give the brightest, most consistent results and are the default choices. 400-series and precipitation-hardening grades polish acceptably but duller. Free-machining 303 contains sulfide inclusions that pit and streak, and cast or sintered parts reveal their porosity.

### Is electropolishing the same as passivation?

No, though electropolishing accomplishes passivation as a side effect. Passivation is a chemical dip that removes free iron and adds nothing and removes essentially nothing dimensionally. Electropolishing removes 0.0002–0.001 in of metal, brightens the surface, deburrs it, and costs considerably more.

## Alternative processes

- [Passivation](https://manufacturingprocesses.org/processes/finishing/passivation.md): Passivation removes free iron from a stainless steel surface with an acid bath so the chromium oxide layer can reform unbroken.
- [Grinding, Sanding and Polishing](https://manufacturingprocesses.org/processes/finishing/grinding-sanding-and-polishing.md): Grinding, sanding and polishing remove a thin layer of material with bonded or coated abrasive to correct form or improve surface finish.
- [Vibratory Tumbling and Mass Finishing](https://manufacturingprocesses.org/processes/finishing/vibratory-tumbling-and-mass-finishing.md): Vibratory tumbling agitates parts with abrasive media in a bowl so edges are broken and surfaces smoothed in bulk, without hand work.
- [Abrasive Blasting](https://manufacturingprocesses.org/processes/finishing/abrasive-blasting.md): Abrasive blasting propels media at a surface with compressed air to clean it, strip scale, or produce a uniform matte texture.
- [Bead Blasting](https://manufacturingprocesses.org/processes/finishing/bead-blasting.md): Bead blasting propels glass or ceramic beads at a surface to produce an even satin matte texture without cutting into it aggressively.

## Related processes

- [Passivation](https://manufacturingprocesses.org/processes/finishing/passivation.md): Passivation removes free iron from a stainless steel surface with an acid bath so the chromium oxide layer can reform unbroken.
- [Electroplating](https://manufacturingprocesses.org/processes/finishing/electroplating.md): Electroplating deposits a metal coating from solution onto a conductive part by passing current through an electrolyte bath.
- [Grinding, Sanding and Polishing](https://manufacturingprocesses.org/processes/finishing/grinding-sanding-and-polishing.md): Grinding, sanding and polishing remove a thin layer of material with bonded or coated abrasive to correct form or improve surface finish.
- [Vibratory Tumbling and Mass Finishing](https://manufacturingprocesses.org/processes/finishing/vibratory-tumbling-and-mass-finishing.md): Vibratory tumbling agitates parts with abrasive media in a bowl so edges are broken and surfaces smoothed in bulk, without hand work.
- [Photo Etching](https://manufacturingprocesses.org/processes/finishing/photo-etching.md): Photo etching masks a surface photographically and etches shallow detail into it for decoration, marking or texture.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/finishing/electropolishing)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Foil Blocking and Embossing"
category: "Finishing"
subcategory: "Printing"
materials: ["Plastic", "Wood"]
tolerances: "Transferred foil layer is sub-micron and dimensionally negligible; image registration typically ±0.010–0.020 in (±0.25–0.5 mm), and exact between foil and emboss when a combination die is used"
volumes: "A few hundred to millions of impressions; die cost amortizes quickly at medium volume"
lead_time: "1–3 weeks including die manufacture; 3–7 business days for a repeat run on existing tooling"
url: https://manufacturingprocesses.org/processes/finishing/foil-blocking-and-embossing
---

# Foil Blocking and Embossing

Foil blocking and embossing press a heated die into the surface to transfer metallic foil, raise a relief, or do both at once.

- **Category**: [Finishing](https://manufacturingprocesses.org/processes/finishing.md)
- **Family**: Printing
- **Materials**: Plastic, Wood
- **Typical tolerances**: Transferred foil layer is sub-micron and dimensionally negligible; image registration typically ±0.010–0.020 in (±0.25–0.5 mm), and exact between foil and emboss when a combination die is used
- **Surface finish**: Mirror metallic, satin or matte depending on foil grade; embossing adds relief bounded by substrate caliper
- **Typical volumes**: A few hundred to millions of impressions; die cost amortizes quickly at medium volume
- **Lead time**: 1–3 weeks including die manufacture; 3–7 business days for a repeat run on existing tooling

## Overview

Foil blocking — also called hot stamping or hot foil stamping — presses a heated metal die against a carrier foil to transfer a thin decorative layer onto a substrate. Embossing uses the same press and a matched die pair to deform the substrate into raised relief, and the two operations are routinely combined in a single hit with a combination die.

What transfers in foil blocking is a sub-micron stack: a release layer, a colored lacquer, a vacuum-deposited aluminum layer where a metallic look is wanted, and a heat-activated size. The polyester carrier peels away and is discarded. The result adds essentially no thickness, cannot be scratched off like a printed ink, and produces a genuine mirror metallic that no ink can match. Die temperature typically runs 200–400 °F (95–205 °C) depending on the foil and substrate, with dwell measured in fractions of a second to a couple of seconds. Substrates include coated board, plastics, painted metal, leather and wood.

## How it works

1. **Make the die.** The image is etched or engraved into a metal die. Magnesium is the cheapest and is etched, suiting short runs and simple shapes. Copper holds finer detail and runs longer. Brass is engraved or CNC machined, holds the finest detail, retains heat best, and is the choice for long production runs and for sculpted multi-level embossing.

2. **Mount and heat.** The die is mounted in a heated chase and brought to temperature. Foil transfer is a thermal process: temperature, pressure and dwell are the three variables, and they trade against each other. Too cool or too brief and the foil does not release; too hot or too long and it spreads, fills fine detail, and can scorch the substrate.

3. **Feed the foil.** Foil is fed from a roll across the substrate. Only the die's raised image contacts the foil, so only that area transfers. Foil advance per cycle is set so unused foil is not wasted, which matters because foil is a significant consumable cost.

4. **Stamp.** The press closes. Heat activates the size layer and softens the release layer, and pressure bonds the transferred stack to the substrate. On opening, the carrier lifts away carrying everything except the stamped image.

5. **Emboss (same hit or separate).** A male die pressed into a matched female counter — or into a hard counter built up on the platen — deforms the substrate into relief. Combination dies do foil and emboss simultaneously, which guarantees perfect registration between the two because they are the same tool.

### Foil types

Metallic foils carry a vacuum-deposited aluminum layer tinted to gold, silver, copper or color. Pigment foils are opaque colors with no metal and behave like a very thin dry paint film. Holographic foils carry an embossed diffraction pattern. Specialty foils are formulated for specific substrates — a foil that adheres to coated board will not necessarily adhere to polypropylene, and foil selection is substrate-specific rather than universal.

## Design guidelines

### Keep strokes above the minimum width

Fine hairlines and small serifs fill in or drop out. A practical minimum stroke width is around 0.004 in (0.1 mm), and small text with delicate serifs should be redrawn with heavier strokes or set in a face designed to survive the process. Reversed-out detail inside a foiled area closes up first.

### Break up large solid areas

A large uninterrupted solid is the hardest thing to foil cleanly — the foil can pinhole, and air trapped under the die shows as speckle. Either specify a foil grade formulated for solid coverage, break the area with a pattern, or accept a lightly textured result. Solids over roughly 1 in (25 mm) across warrant a discussion with the converter before the artwork is finalized.

### Never foil or emboss across a fold or score

The transferred layer and the deformed substrate both crack when the material is folded afterward. Keep foil and emboss areas clear of scores, creases and fold lines by a comfortable margin, and keep them away from the trimmed edge — 1/8 in (3 mm) is a reasonable minimum on board.

### Match emboss depth to the substrate

Embossing works by deforming the material, so the achievable relief is bounded by the caliper and ductility of the substrate. Deep single-level embossing on thin board tears or burnishes the surface; multi-level sculpted embossing needs a heavier stock and a brass die. On plastics, emboss depth is limited by wall thickness and by how much the material can be locally deformed without whitening.

### Use a combination die when foil and emboss must register

Foiling and embossing in separate operations means separate registration, typically to ±0.010–0.020 in (±0.25–0.5 mm). A combination die produces both features from the same tool, so registration between them is exact by construction. This is nearly always worth the higher die cost.

### Verify foil-to-substrate adhesion, not just appearance

Foil selection is substrate specific. Coated board, uncoated board, painted metal, ABS, polypropylene and leather all require different foil chemistries. Specify a cross-hatch tape adhesion test to ASTM D3359 on the actual production substrate, including any coating or release agent it carries.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Minimum stroke width | 0.008 in (0.2 mm) | 0.004 in (0.1 mm) | Fine strokes fill in or drop out |
| Solid foil area | Break up or specify a solids foil | ~1 in (25 mm) across | Large solids pinhole and speckle |
| Distance from fold or trim | 1/8 in (3 mm) | — | Foil and embossed relief crack when folded |
| Die material | Brass for long runs and sculpting | Magnesium for short runs | Heat retention and detail retention differ |
| Die temperature | 200–400 °F (95–205 °C) | Substrate-dependent | Too hot spreads foil and scorches |
| Foil plus emboss registration | Use a combination die | ±0.010–0.020 in if separate | One tool guarantees registration |

## Cost drivers

Foil blocking is a tooling-plus-consumable process. The die is the fixed cost, the foil is a genuine per-impression material cost, and press time is fast, so the economics favor medium and long runs.

- **Die material and complexity.** A flat etched magnesium die is inexpensive; a sculpted, multi-level, CNC-engraved brass die is a substantial tooling investment that pays back over long runs.
- **Foil consumption.** Foil is consumed by area of web advanced, not by area transferred, so poor image layout across the web wastes material on every cycle. Multi-up layouts and correct foil advance settings matter.
- **Number of hits.** Each color of foil and each separate emboss is another pass through the press with its own makeready.
- **Makeready.** Setting temperature, pressure, dwell and registration is skilled setup time and is the dominant cost on short runs.
- **Substrate variability.** Coating, moisture content and surface treatment change foil adhesion, so a substrate change usually means requalification.

Cost-reduction tactics:

1. Consolidate to one foil color and one hit; multi-color foil work multiplies press passes.
2. Use a combination die so foiling and embossing happen in a single pass with guaranteed registration.
3. Lay artwork out so foil web advance is minimized — this is often the largest recoverable material cost.
4. Choose magnesium dies for prototypes and short runs and reserve brass for production.
5. Where a metallic look is needed on a three-dimensional plastic part rather than a flat panel, price vacuum metalizing instead — foil blocking needs a surface the die can press flat against.

## FAQ

### How thick is a hot-stamped foil layer?

Sub-micron and dimensionally negligible. What transfers is a release layer, a colored lacquer, a vacuum-deposited aluminum layer where a metallic look is required, and a heat-activated size. The polyester carrier peels away and is discarded, so the finished part gains a true mirror metallic with essentially no added thickness.

### What temperature does foil blocking run at?

Typically 200–400 °F (95–205 °C) at the die, with dwell from a fraction of a second to a couple of seconds. Temperature, pressure and dwell trade against each other: too cool or too brief and the foil will not release, too hot or too long and it spreads, fills fine detail and can scorch the substrate.

### Which die material should I specify?

Magnesium is etched, cheap and suits prototypes and short runs. Copper holds finer detail and lasts longer. Brass is engraved or CNC machined, retains heat best, holds the finest detail, and is the choice for long production runs and for sculpted multi-level embossing.

### Can foil and embossing be done in one operation?

Yes, with a combination die, and it is usually worth the higher tooling cost. Doing them separately means two press passes with registration typically ±0.010–0.020 in (±0.25–0.5 mm) between the foil and the relief. A combination die produces both from the same tool, so registration is exact by construction.

### Why did my large foiled area come out speckled?

Large uninterrupted solids are the hardest thing to foil cleanly — air trapped under the die shows as speckle and standard foils can pinhole. Specify a foil grade formulated for solid coverage, break the area up with a pattern, or keep solids under roughly 1 in (25 mm) across.

### Can you foil stamp plastic parts?

Yes — hot stamping is widely used on ABS, painted plastic trim, pens and appliance components. Foil chemistry is substrate specific, so a foil formulated for coated board will not necessarily adhere to polypropylene. Qualify the foil against the actual production substrate with a cross-hatch tape adhesion test.

## Alternative processes

- [Screen Printing](https://manufacturingprocesses.org/processes/finishing/screen-printing.md): Screen printing pushes ink through a stencilled mesh onto the surface below, laying down one opaque color per pass.
- [Pad Printing](https://manufacturingprocesses.org/processes/finishing/pad-printing.md): Pad printing lifts ink out of an etched plate with a silicone pad and transfers it onto a curved or recessed surface.
- [UV Inkjet Printing](https://manufacturingprocesses.org/processes/finishing/uv-inkjet-printing.md): UV inkjet printing jets UV-curable ink directly onto a flat substrate and cures it instantly under UV light.
- [Vacuum Metalizing](https://manufacturingprocesses.org/processes/finishing/vacuum-metalizing.md): Vacuum metalizing evaporates aluminum inside a vacuum chamber so it condenses as a mirror-bright film on the part surface.
- [Laser Marking and Engraving](https://manufacturingprocesses.org/processes/finishing/laser-marking-and-engraving.md): Laser marking alters the surface with a focused beam — annealing, etching or engraving it — to make permanent, contact-free marks.

## Related processes

- [Screen Printing](https://manufacturingprocesses.org/processes/finishing/screen-printing.md): Screen printing pushes ink through a stencilled mesh onto the surface below, laying down one opaque color per pass.
- [Pad Printing](https://manufacturingprocesses.org/processes/finishing/pad-printing.md): Pad printing lifts ink out of an etched plate with a silicone pad and transfers it onto a curved or recessed surface.
- [Die Cutting](https://manufacturingprocesses.org/processes/cutting/die-cutting.md): Die cutting presses a shaped steel rule or rotary die through sheet stock, cutting flat outlines in paper, board, foam, film and textile.
- [UV Inkjet Printing](https://manufacturingprocesses.org/processes/finishing/uv-inkjet-printing.md): UV inkjet printing jets UV-curable ink directly onto a flat substrate and cures it instantly under UV light.
- [Vacuum Metalizing](https://manufacturingprocesses.org/processes/finishing/vacuum-metalizing.md): Vacuum metalizing evaporates aluminum inside a vacuum chamber so it condenses as a mirror-bright film on the part surface.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/finishing/foil-blocking-and-embossing)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Galvanizing"
category: "Finishing"
subcategory: "Additive"
materials: ["Metal"]
tolerances: "Hot dip adds 1.8–3.9 mils (45–100 µm) per surface per ASTM A123, so holes close by 3.6–7.8 mils on diameter; electrogalvanizing adds 0.0002–0.001 in (5–25 µm) and is far more predictable"
volumes: "One-off fabrications to continuous coil; priced by weight, so heavy structural work is the most economical"
lead_time: "2–10 business days at a commercial galvanizer, driven by kettle scheduling rather than processing time"
url: https://manufacturingprocesses.org/processes/finishing/galvanizing
---

# Galvanizing

Galvanizing coats steel with zinc, usually by hot dipping, so the zinc corrodes sacrificially and protects the steel beneath it.

- **Category**: [Finishing](https://manufacturingprocesses.org/processes/finishing.md)
- **Family**: Additive
- **Materials**: Metal
- **Typical tolerances**: Hot dip adds 1.8–3.9 mils (45–100 µm) per surface per ASTM A123, so holes close by 3.6–7.8 mils on diameter; electrogalvanizing adds 0.0002–0.001 in (5–25 µm) and is far more predictable
- **Surface finish**: Spangled to matte gray depending on steel chemistry and cooling; visibly rougher than paint, with runs and drips at drain points
- **Typical volumes**: One-off fabrications to continuous coil; priced by weight, so heavy structural work is the most economical
- **Lead time**: 2–10 business days at a commercial galvanizer, driven by kettle scheduling rather than processing time

## Overview

Galvanizing coats steel with zinc so the zinc corrodes preferentially and protects the steel underneath. Hot-dip galvanizing — total immersion in molten zinc at roughly 840–850 °F (449–454 °C) — is the dominant form and is what "galvanized" means on a structural drawing. The zinc reacts with the iron to form a series of zinc-iron intermetallic layers metallurgically bonded to the substrate, topped by a layer of pure zinc.

ASTM A123 specifies coating thickness by steel section thickness, from about 1.8 mils (45 µm) on material under 1/16 in up to 3.9 mils (100 µm) on material 1/4 in and thicker. That coating is applied to every surface the zinc touches, inside and out, including the interior of hollow sections. Two properties set galvanizing apart from paint: it is cathodic, so it protects exposed cut edges and scratches, and the coating is bonded well enough to survive handling and erection.

## How it works

1. **Degrease.** A hot alkaline or acidic bath removes oil, grease and paint. Zinc will not wet a contaminated surface, and the result is a bare "miss".

2. **Pickle.** Dilute hydrochloric or sulfuric acid strips mill scale and rust. Scale that survives pickling reappears as an uncoated patch. Weld slag, paint and marking crayon do not pickle off and must be removed mechanically beforehand.

3. **Flux.** A zinc ammonium chloride bath removes the last oxide and leaves a film that prevents re-oxidation between the rinse and the kettle.

4. **Dip.** The part is lowered into molten zinc at 840–850 °F (449–454 °C) and held until it reaches bath temperature — seconds for thin sheet, several minutes for heavy structural sections. During immersion, iron and zinc interdiffuse to form the gamma, delta and zeta intermetallic layers; on withdrawal, a layer of free zinc (eta) freezes on the outside.

5. **Withdraw and drain.** Rate and angle of withdrawal control runs, drips and spikes. Excess zinc is spun off centrifugally for threaded fasteners and small hardware under ASTM A153.

6. **Quench or air cool, then inspect.** Coating thickness is measured magnetically per ASTM A123, and adhesion by stout knife test.

### Variants

**Electrogalvanizing** plates zinc electrolytically at 0.0002–0.001 in (5–25 µm) — thinner, smoother, dimensionally predictable, and applied cold, which suits sheet that will be formed and painted afterward. **Continuous hot-dip sheet** to ASTM A653 runs coil through a zinc bath in line and is designated by total coating weight on both surfaces, G60 and G90 being the common commercial grades. **Zinc thermal spray (metallizing)** applies zinc from a wire-fed arc gun on site, with no size limit and no heat distortion.

## Design guidelines

### Vent and drain every enclosed section

This is a safety requirement, not a finish requirement. Any sealed hollow section — tube, box, gusseted joint, overlapped plate — traps moisture that flashes to steam in a 850 °F kettle and can burst the part or eject zinc. Provide vent and drain holes at diagonally opposite ends of every hollow member, sized to the section, and never smaller than about 3/8 in (10 mm). Undrained internals also fill with zinc, adding dead weight.

### Control the steel chemistry

Silicon and phosphorus in the steel drive the reaction rate. Silicon in the roughly 0.04–0.14% band, or above about 0.22%, produces thick, dull gray, coarse and comparatively brittle coatings (the Sandelin effect). Silicon below 0.04% or in the 0.15–0.22% window gives bright, well-controlled coatings. Where appearance matters, specify the steel chemistry and keep every part of a visible assembly from a single heat.

### Expect distortion on thin or asymmetric weldments

Full immersion at 840–850 °F relieves residual stress from rolling, cutting and welding. Thin, long, asymmetric fabrications warp. Use symmetric weld sequences, balanced sections, and avoid combining thin and heavy plate in the same member. Design for temporary bracing where flatness matters.

### Budget the coating on fits and threads

At 1.8–3.9 mils (45–100 µm) per surface, a hole loses 3.6–7.8 mils on diameter. Standard practice is to tap internal threads oversize after galvanizing, per ASTM A563, and to galvanize external threads and leave them alone. Bolted connections need clearance holes drilled 1/16 in (1.6 mm) larger than usual, and faying surfaces in slip-critical connections need a specified surface condition because galvanizing changes the slip coefficient.

### Do not paint over new galvanizing without preparation

Fresh zinc is too smooth and too reactive for paint to grip, and zinc reacts with alkyd binders to form soap. A duplex system requires sweep blasting, a wash primer or acrylic-based tie coat, or weathered zinc. Done correctly, the paint and zinc together outlast the sum of the two applied separately.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Coating thickness (ASTM A123) | 3.9 mils (100 µm) on ≥ 1/4 in steel | 1.8 mils (45 µm) on thin sheet | Thickness is set by section thickness, not by choice |
| Vent/drain hole | Diagonally opposite, both ends | 3/8 in (10 mm) minimum | Trapped moisture flashes to steam in the kettle |
| Hole diameter allowance | 2 × coating thickness | — | Zinc builds on both walls |
| Internal threads | Tap oversize after galvanizing | — | Coating will not clear a standard tapped hole |
| Silicon content for appearance | < 0.04% or 0.15–0.22% | Avoid 0.04–0.14% | Sandelin range produces thick, dull, brittle coatings |
| Overlap / faying surfaces | Seal-weld or vent | — | Unvented overlaps trap acid and moisture |

## Cost drivers

Hot-dip galvanizing is priced by weight, typically per hundredweight or per ton, with a lot minimum. It is one of the lowest-cost-per-square-foot corrosion systems available for structural steel, and the per-unit cost falls as parts get heavier.

- **Weight and handling.** Price follows tonnage, but awkward parts that need special jigs, multiple dips, or single-piece handling carry a surcharge.
- **Kettle size.** Anything longer than the kettle must be double-dipped, which doubles handling and leaves a visible line where the two dips meet. Check the galvanizer's kettle dimensions before finalizing member lengths.
- **Preparation exceptions.** Weld slag, paint, marking crayon, oil-based inks and previously coated surfaces do not pickle off and become a hand-labor line item.
- **Rework for drainage.** Parts arriving without vent holes are either returned or drilled by the galvanizer at cost, and zinc that fills an undrained cavity is billed as consumed weight.
- **Post-galvanizing work.** Oversize tapping, thread chasing, straightening and touch-up with zinc-rich paint are all separate operations.

Cost-reduction tactics:

1. Design members to fit the galvanizer's kettle in a single dip.
2. Put vent and drain holes on the drawing rather than leaving the galvanizer to add them.
3. Specify weld filler and steel chemistry consistently across an assembly so the coating looks uniform.
4. Use continuous galvanized sheet (ASTM A653, G90) for formed sheet parts instead of fabricating in bare steel and dipping afterward.
5. Where the part is small, threaded or dimensionally critical, price electrogalvanizing or zinc plating against hot dip — thinner coating, no distortion, no oversize tapping.

## FAQ

### How thick is hot-dip galvanizing?

ASTM A123 sets minimum coating thickness by steel section thickness — roughly 1.8 mils (45 µm) on material under 1/16 in, rising to 3.9 mils (100 µm) on material 1/4 in and thicker. The coating goes on every wetted surface, so a hole closes by twice the coating thickness on diameter.

### Do galvanized parts need vent holes?

Yes, and it is a safety issue rather than a finish issue. Any sealed hollow section traps moisture that flashes to steam in an 840 °F (449 °C) zinc kettle, which can burst the part. Provide vent and drain holes at diagonally opposite ends of every hollow member, never smaller than about 3/8 in (10 mm).

### Can you galvanize threaded fasteners?

Yes — hardware is galvanized to ASTM A153 and centrifuged to spin off excess zinc. Because the coating will not clear a standard tapped hole, mating nuts are tapped oversize after galvanizing per ASTM A563 while external threads are left coated.

### Why does galvanizing sometimes come out dull gray instead of shiny?

Steel chemistry. Silicon in the roughly 0.04–0.14% range, or above about 0.22%, accelerates the zinc-iron reaction and produces a thick, dull gray, coarser coating — the Sandelin effect. It is still a compliant coating and is often thicker than required, but it will not match a bright spangled part from a different heat.

### Does galvanizing warp steel?

It can. Immersing a fabrication in 840–850 °F (449–454 °C) zinc relieves residual stress from rolling, cutting and welding, so thin, long or asymmetric weldments distort. Balance sections, use symmetric weld sequences, and avoid mixing thin sheet with heavy plate in the same member.

### Can galvanized steel be painted or powder coated?

Yes, and the combination (a duplex system) outlasts either coating alone, but the zinc must be prepared first. Fresh zinc is too smooth and too reactive to hold paint, so it needs sweep blasting, a wash primer or a tie coat. Zinc also reacts with alkyd binders, so the paint chemistry must be selected for galvanized substrates.

### What is the difference between hot-dip galvanizing and zinc plating?

Hot-dip immersion in molten zinc produces a 1.8–3.9 mil (45–100 µm) metallurgically bonded coating with excellent life but visible texture and thermal distortion. Electrogalvanizing (zinc plating) deposits 0.0002–0.001 in (5–25 µm) electrolytically at room temperature — thinner, smoother and dimensionally predictable, but with far less corrosion life.

## Alternative processes

- [Electroplating](https://manufacturingprocesses.org/processes/finishing/electroplating.md): Electroplating deposits a metal coating from solution onto a conductive part by passing current through an electrolyte bath.
- [Powder Coating](https://manufacturingprocesses.org/processes/finishing/powder-coating.md): Powder coating sprays electrostatically charged dry polymer powder onto an earthed part, then bakes it into a tough continuous film.
- [E-Coating (Electrophoretic Deposition)](https://manufacturingprocesses.org/processes/finishing/e-coating-electrophoretic-deposition.md): E-coating deposits paint electrically from a bath so it plates evenly into every recess, then cures to a uniform film.
- [Spray Painting](https://manufacturingprocesses.org/processes/finishing/spray-painting.md): Spray painting atomises liquid coating onto a prepared surface, building a film that cures into a decorative and protective layer.
- [Chemical Conversion Coating](https://manufacturingprocesses.org/processes/finishing/chemical-conversion-coating.md): Chemical conversion coating reacts the metal surface with a chemical bath to form a thin protective, conductive or paint-adhering film.

## Related processes

- [Electroplating](https://manufacturingprocesses.org/processes/finishing/electroplating.md): Electroplating deposits a metal coating from solution onto a conductive part by passing current through an electrolyte bath.
- [Powder Coating](https://manufacturingprocesses.org/processes/finishing/powder-coating.md): Powder coating sprays electrostatically charged dry polymer powder onto an earthed part, then bakes it into a tough continuous film.
- [Abrasive Blasting](https://manufacturingprocesses.org/processes/finishing/abrasive-blasting.md): Abrasive blasting propels media at a surface with compressed air to clean it, strip scale, or produce a uniform matte texture.
- [Chemical Conversion Coating](https://manufacturingprocesses.org/processes/finishing/chemical-conversion-coating.md): Chemical conversion coating reacts the metal surface with a chemical bath to form a thin protective, conductive or paint-adhering film.
- [Spray Painting](https://manufacturingprocesses.org/processes/finishing/spray-painting.md): Spray painting atomises liquid coating onto a prepared surface, building a film that cures into a decorative and protective layer.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/finishing/galvanizing)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Grinding, Sanding and Polishing"
category: "Finishing"
subcategory: "Subtractive"
materials: ["Metal", "Plastic", "Wood", "Glass", "Ceramic", "Composite"]
tolerances: "Surface and cylindrical grinding ±0.0001–0.0005 in (±0.0025–0.013 mm); honing corrects bore geometry within a few tenths; lapping holds flatness to a few millionths of an inch"
volumes: "1 to millions of parts; centerless and production grinding lines run continuously"
lead_time: "1–10 business days at a job shop depending on setup complexity; production grinding runs in seconds to minutes per part"
url: https://manufacturingprocesses.org/processes/finishing/grinding-sanding-and-polishing
---

# Grinding, Sanding and Polishing

Grinding, sanding and polishing remove a thin layer of material with bonded or coated abrasive to correct form or improve surface finish.

- **Category**: [Finishing](https://manufacturingprocesses.org/processes/finishing.md)
- **Family**: Subtractive
- **Materials**: Metal, Plastic, Wood, Glass, Ceramic, Composite
- **Typical tolerances**: Surface and cylindrical grinding ±0.0001–0.0005 in (±0.0025–0.013 mm); honing corrects bore geometry within a few tenths; lapping holds flatness to a few millionths of an inch
- **Surface finish**: Grinding 8–63 µin (0.2–1.6 µm) Ra; honing 4–32 µin (0.1–0.8 µm); lapping 1–16 µin (0.025–0.4 µm); superfinishing and polishing 1–8 µin (0.025–0.2 µm)
- **Typical volumes**: 1 to millions of parts; centerless and production grinding lines run continuously
- **Lead time**: 1–10 business days at a job shop depending on setup complexity; production grinding runs in seconds to minutes per part

## Overview

Grinding, sanding and polishing remove material with bonded or coated abrasive to correct form, hold tight size, and reach surface finishes that cutting tools cannot. The family spans a wide range of removal rates and results: surface and cylindrical grinding hold ±0.0001–0.0005 in (±0.0025–0.013 mm) and produce 8–63 µin (0.2–1.6 µm) Ra; honing corrects bore geometry at 4–32 µin (0.1–0.8 µm); lapping reaches 1–16 µin (0.025–0.4 µm) with flatness measurable in light bands; polishing and superfinishing go below 8 µin but correct no geometry at all.

Grinding is also the only practical way to machine material above roughly 60 HRC, which is why hardened parts are ground after heat treatment rather than cut. The distinguishing constraint for a designer is that abrasive processes need access, relief and stock: a wheel has to reach the surface, run out somewhere, and have 0.005–0.020 in (0.13–0.5 mm) of material to remove.

## How it works

1. **Leave grind stock.** Parts are machined oversize before heat treatment. Typical allowance is 0.010–0.020 in (0.25–0.5 mm) per surface where distortion and scale must be cleaned up, and 0.005 in (0.13 mm) or less where only a finish pass is needed. Too little stock leaves black spots; too much wastes cycle time and risks thermal damage.

2. **Select the abrasive.** Aluminum oxide for steels, silicon carbide for cast iron and non-ferrous metals, CBN for hardened steels and superalloys, diamond for carbide, ceramics and glass. Wheel specification also covers grit size (roughly 36–600), grade (bond hardness) and structure.

3. **Grind.** Conventional vitrified wheels run at roughly 5,500–6,500 sfpm (28–33 m/s); superabrasive wheels run considerably faster. Depth of cut is typically 0.001–0.003 in (0.025–0.075 mm) for roughing and 0.0002–0.0005 in (0.005–0.013 mm) for finishing, followed by spark-out passes at zero infeed to let the machine settle to size.

4. **Manage heat.** Grinding puts nearly all its energy into the workpiece as heat. Excessive rates cause grinding burn — local retempering, softening, tensile residual stress and in severe cases surface cracks. Flood coolant, correct wheel dressing and conservative infeed are the controls; nital etch and Barkhausen noise inspection are how burn is detected.

5. **Dress the wheel.** A loaded or glazed wheel cuts by rubbing, which generates heat instead of chips. Regular dressing with a diamond restores sharp cutting points and the wheel's form.

### The specialized operations

**Honing** uses expanding abrasive stones in a reciprocating, rotating tool to correct bore roundness, straightness and taper — removing 0.0005–0.010 in (0.013–0.25 mm) — and leaves the cross-hatch pattern that retains oil in cylinder bores. **Lapping** rubs the part against a soft plate charged with loose abrasive, removing 0.0001–0.0005 in (0.0025–0.013 mm) and producing exceptional flatness on seal faces and gage surfaces. **Belt sanding and polishing** conform to contours and produce appearance finishes, but they follow the existing form rather than correcting it — a polished part is a smoother version of whatever shape it already was.

## Design guidelines

### Provide wheel relief at every shoulder

A grinding wheel cannot cut into a sharp internal corner; its own edge breaks down and the corner ends up radiused and out of tolerance. Design an undercut or relief groove at the shoulder — typically 0.030–0.060 in (0.75–1.5 mm) wide and at least as deep as the grind stock — wherever a ground diameter runs into a face.

### Prefer through features to blind ones

Through-grinding a bore or a slot is straightforward; blind bores require the wheel to reverse inside the feature, which limits the achievable finish and size control near the bottom. The same applies to honing.

### Specify the loosest finish that works

Each step down the finish scale costs disproportionately. Finishes tighter than about 63 µin (1.6 µm) Ra usually require a dedicated grinding operation, and below about 32 µin (0.8 µm) a secondary lapping or polishing operation. Use the [surface finish chart](/charts/surface-finish-chart) to match the callout to what the process actually delivers rather than defaulting to a tight number.

### Do not ask polishing to fix geometry

Polishing, buffing and belt sanding remove the peaks of the existing surface. They cannot correct flatness, roundness, parallelism or a dimension, and on a machined part they will round edges and blur detail. Where both form and finish are required, grind for form and then polish for appearance.

### Specify Rz as well as Ra on critical surfaces

Ra averages the whole profile and is insensitive to isolated deep scratches; Rz is peak-to-valley and is not. A ground sealing face can pass an Ra spec and still leak because of one deep scratch. For seal lands, fatigue-critical surfaces and bearing journals, specify both.

### Allow for distortion and support

Thin, long or unsupported parts deflect under wheel pressure and spring back out of tolerance. Provide clamping surfaces, center holes, or steady-rest locations, and expect that a thin part may need multiple spring passes.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Grind stock allowance | 0.010–0.020 in (0.25–0.5 mm) per surface | 0.005 in (0.13 mm) finish only | Too little leaves unground low spots |
| Shoulder relief groove | 0.030–0.060 in (0.75–1.5 mm) wide | Required at every ground shoulder | Wheels cannot cut a sharp internal corner |
| Grinding tolerance | ±0.0005 in (±0.013 mm) | ±0.0001 in (±0.0025 mm) | Tighter needs temperature control and slow finishing |
| Ra by operation | Grinding 8–63 µin; honing 4–32 µin; lapping 1–16 µin | Polishing < 8 µin, no form correction | Each stage trades removal rate for finish |
| Finishing depth of cut | 0.0002–0.0005 in (0.005–0.013 mm) | — | Heavier cuts risk grinding burn |
| Critical surfaces | Specify Ra and Rz | — | Ra alone hides isolated deep scratches |

## Variants

- Wheel Cutting
- Belt Sanding
- Honing
- Lapping

## Cost drivers

Abrasive finishing is priced by machine time, and machine time is driven by how much material has to come off and how tight the result must be.

- **Stock to remove.** Grinding removes material slowly compared to milling or turning. Every extra thousandth of allowance is cycle time, so leave what the process needs and no more.
- **Finish requirement.** Cycle time rises sharply below about 63 µin (1.6 µm) Ra, and a callout below 32 µin (0.8 µm) usually adds an entire secondary operation.
- **Setup and fixturing.** Grinding setups are precise and slow to establish. Cost per part falls dramatically with lot size; one-off grinding is dominated by setup.
- **Wheel and dressing consumption.** Superabrasive CBN and diamond wheels are expensive but last far longer and hold form, so they win on volume and lose on short runs.
- **Inspection.** Tight tolerances require temperature-stabilized inspection, and burn-sensitive parts require nital etch or Barkhausen testing per lot.

Cost-reduction tactics:

1. Grind only the surfaces that need it. A drawing that calls out a ground finish on every face doubles or triples cost for no function.
2. Specify Ra to the function, not to the shop's best capability.
3. Design relief grooves and through features so the operation is a single straightforward pass.
4. Consider hard turning instead of cylindrical grinding on hardened parts where ±0.0005 in and 32 µin are adequate — it is often faster.
5. Where the requirement is only deburring and edge break rather than form correction, price vibratory finishing against hand or machine grinding.

## FAQ

### What surface finish can grinding achieve?

Surface and cylindrical grinding typically produce 8–63 µin (0.2–1.6 µm) Ra. Honing reaches 4–32 µin (0.1–0.8 µm), lapping 1–16 µin (0.025–0.4 µm), and superfinishing or polishing below 8 µin (0.2 µm). Each step down trades removal rate and cost for finish.

### How much stock should I leave for grinding?

Typically 0.010–0.020 in (0.25–0.5 mm) per surface where heat-treat distortion and scale must be cleaned up, and 0.005 in (0.13 mm) or less for a finish pass only. Too little leaves unground low spots; too much is wasted cycle time and raises the risk of grinding burn.

### Why does a ground shoulder need a relief groove?

A grinding wheel cannot cut into a sharp internal corner — its edge breaks down and the corner ends up radiused and out of tolerance. An undercut 0.030–0.060 in (0.75–1.5 mm) wide and at least as deep as the grind stock gives the wheel somewhere to run out.

### What tolerance can grinding hold?

±0.0005 in (±0.013 mm) is routine for surface and cylindrical grinding, and ±0.0001 in (±0.0025 mm) is achievable with temperature control, careful spark-out and slow finishing passes. Jig grinding and lapping go tighter still, at correspondingly higher cost.

### Can polishing fix a flatness or roundness problem?

No. Polishing, buffing and belt sanding remove the peaks of the surface that already exists — they follow the form rather than correcting it, and they round edges and blur detail. Grind or lap for form, then polish for appearance.

### What is grinding burn and how is it detected?

Grinding puts almost all its energy into the workpiece as heat, and excessive removal rates locally retemper the surface, leaving softened material and tensile residual stress that shortens fatigue life. It is controlled with flood coolant, correct wheel dressing and conservative infeed, and detected by nital etch inspection or Barkhausen noise measurement.

## Alternative processes

- [Electropolishing](https://manufacturingprocesses.org/processes/finishing/electropolishing.md): Electropolishing reverses plating: current dissolves the surface fastest at the peaks, leaving a bright, deburred, passive finish.
- [Vibratory Tumbling and Mass Finishing](https://manufacturingprocesses.org/processes/finishing/vibratory-tumbling-and-mass-finishing.md): Vibratory tumbling agitates parts with abrasive media in a bowl so edges are broken and surfaces smoothed in bulk, without hand work.
- [CNC Turning](https://manufacturingprocesses.org/processes/cutting/cnc-turning.md): CNC turning rotates the workpiece against a stationary tool, producing round parts with excellent concentricity and short cycle times.
- [Electrical Discharge Machining](https://manufacturingprocesses.org/processes/cutting/electrical-discharge-machining.md): EDM erodes conductive metal with controlled electrical sparks across a dielectric gap, cutting hardened material and sharp internal corners.
- [Bead Blasting](https://manufacturingprocesses.org/processes/finishing/bead-blasting.md): Bead blasting propels glass or ceramic beads at a surface to produce an even satin matte texture without cutting into it aggressively.

## Related processes

- [Electropolishing](https://manufacturingprocesses.org/processes/finishing/electropolishing.md): Electropolishing reverses plating: current dissolves the surface fastest at the peaks, leaving a bright, deburred, passive finish.
- [Vibratory Tumbling and Mass Finishing](https://manufacturingprocesses.org/processes/finishing/vibratory-tumbling-and-mass-finishing.md): Vibratory tumbling agitates parts with abrasive media in a bowl so edges are broken and surfaces smoothed in bulk, without hand work.
- [Abrasive Blasting](https://manufacturingprocesses.org/processes/finishing/abrasive-blasting.md): Abrasive blasting propels media at a surface with compressed air to clean it, strip scale, or produce a uniform matte texture.
- [Heat Treatment](https://manufacturingprocesses.org/processes/forming/heat-treatment.md): Heat treatment cycles metal through controlled heating and cooling to change its hardness, strength, ductility or internal stress state.
- [CNC Turning](https://manufacturingprocesses.org/processes/cutting/cnc-turning.md): CNC turning rotates the workpiece against a stationary tool, producing round parts with excellent concentricity and short cycle times.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/finishing/grinding-sanding-and-polishing)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Hydro Transfer Printing"
category: "Finishing"
subcategory: "Printing"
materials: ["Plastic", "Metal", "Composite"]
tolerances: "The complete system adds 0.003–0.006 in (75–150 µm) per surface, dominated by the base and clear coats; the printed layer itself is negligible"
volumes: "1 to tens of thousands of parts; manual lines for low volume, automated dipping for automotive trim"
lead_time: "1–3 weeks at a job shop, driven by the spray and cure cycles rather than the dip"
url: https://manufacturingprocesses.org/processes/finishing/hydro-transfer-printing
---

# Hydro Transfer Printing

Hydro transfer printing floats a printed film on water, activates it, and wraps it around a submerged part to cover it in a pattern.

- **Category**: [Finishing](https://manufacturingprocesses.org/processes/finishing.md)
- **Family**: Printing
- **Materials**: Plastic, Metal, Composite
- **Typical tolerances**: The complete system adds 0.003–0.006 in (75–150 µm) per surface, dominated by the base and clear coats; the printed layer itself is negligible
- **Surface finish**: Matte through high gloss, set entirely by the clear coat; the pattern follows the substrate contour
- **Typical volumes**: 1 to tens of thousands of parts; manual lines for low volume, automated dipping for automotive trim
- **Lead time**: 1–3 weeks at a job shop, driven by the spray and cure cycles rather than the dip

## Overview

Hydro transfer printing — also called water transfer printing, hydrographics or immersion printing — wraps a printed pattern around a three-dimensional part. A water-soluble PVA film carrying the printed design floats on a tank of water, a solvent activator dissolves the carrier and reactivates the ink into a floating layer, and the part is pushed down through it so hydrostatic pressure conforms the pattern to every contour.

Its distinguishing capability is coverage of complex geometry without line-of-sight limits — undercuts, compound curves, textured surfaces and internal radii all receive the pattern. What it does not do is provide durability on its own: the finished part is a base coat, a translucent printed layer of negligible thickness, and a clear topcoat that carries all the abrasion, UV and chemical resistance. Total added film is typically 0.003–0.006 in (75–150 µm), essentially all of it from the base and clear coats. Water temperature is held around 75–85 °F (24–29 °C), and any rigid substrate that can be primed can be decorated.

## How it works

1. **Prepare and base coat.** The part is cleaned, abraded and primed, then sprayed with a base color and cured. Because the printed film is translucent, the base color is part of the design — the same wood grain pattern over a tan base and over a gray base produces two visibly different finishes.

2. **Float the film.** The printed PVA film is laid on the water surface in the tank, printed side up, and allowed to hydrate for a controlled time — typically a minute or two — until it softens evenly.

3. **Activate.** A solvent activator is sprayed across the floating film. It dissolves the PVA carrier and returns the printed ink to a wet, mobile state, leaving what is effectively a free-floating layer of ink on the water.

4. **Dip.** The part is pushed through the ink layer at a controlled angle and rate. Hydrostatic pressure forces the ink to wrap around every surface as the part descends. Entry angle and speed determine where the pattern stretches and where it compresses.

5. **Rinse and dry.** Residual PVA and activator are rinsed off with warm water, and the part is dried thoroughly before topcoating.

6. **Clear coat.** A two-component urethane clear is sprayed and cured. This layer provides every durability property the finished part has — abrasion resistance, UV stability, chemical resistance and gloss level.

### Pattern distortion

The film stretches as it conforms. On a flat or gently curved surface the pattern reproduces faithfully; on a deep draw, a sharp corner or a re-entrant feature it stretches, thins and distorts. Repeating organic patterns — carbon fiber weave, wood grain, camouflage, stone — hide this well, which is why they dominate the process. Geometric patterns, logos and anything with a required scale or orientation do not.

## Design guidelines

### Choose the pattern to suit the geometry

Repeating, non-directional, organic patterns tolerate the stretch that comes with wrapping a three-dimensional part. Anything with a fixed scale, a straight line that must stay straight, or a logo that must land in a specific place will distort visibly. If precise registration is required, hydro transfer is the wrong process.

### Design the base color into the finish

The printed layer is translucent, so the base coat shows through and changes the appearance of the same film. Specify the base color as part of the finish specification, not as an incidental primer.

### The clear coat is the finish

Without a topcoat the printed layer has no meaningful abrasion, chemical or UV resistance. Specify the clear coat system, its dry film thickness, gloss level, and the acceptance tests — adhesion by ASTM D3359 and abrasion by Taber where relevant.

### Mask fits, sealing faces and grounding paths

The complete system adds 0.003–0.006 in (75–150 µm) per surface. Threads, bores, gasket faces, bearing seats and electrical bonding surfaces need masking or post-process machining, just as with any sprayed system.

### Avoid deep recesses and sharp internal corners

Although the process has no line-of-sight restriction, the film can bridge across a narrow slot or a deep pocket rather than conforming into it, leaving an unprinted void. Open up narrow features and soften internal corners where the pattern must follow.

### Confirm the substrate can take the process

Any rigid material that can be cleaned, abraded and primed can be decorated — ABS, polycarbonate, glass-filled nylon, aluminum, steel, wood and composites are all routine. Flexible parts, low-energy plastics without pretreatment, and anything that cannot tolerate immersion in warm water are not.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Total added film | 0.003–0.006 in (75–150 µm) | — | Base coat plus clear; the ink layer is negligible |
| Water temperature | 75–85 °F (24–29 °C) | — | Controls film hydration and activation |
| Pattern type | Repeating, organic, non-directional | Avoid fixed-scale geometry and logos | The film stretches as it conforms |
| Base coat | Specified as part of the finish | — | The printed layer is translucent |
| Clear coat | Required, always | Never leave uncoated | It carries all durability properties |
| Narrow slots and deep pockets | Open up or accept voids | — | Film bridges instead of conforming |

## Cost drivers

Hydro transfer printing is a multi-step, largely manual finishing routine. The dipping itself is fast and cheap; the spray and cure operations around it are what cost money.

- **Number of spray operations.** Clean, prime, base coat, cure, dip, rinse, dry, clear coat, cure. Two of those are full spray-booth cycles with their own labor and oven time.
- **Manual dipping.** Entry angle and rate are operator-dependent on manual lines, which limits throughput and drives consistency-related rejects. Automated dipping equipment exists for automotive interior trim volumes.
- **Film cost and waste.** Film is consumed by tank area, not by part area, so small parts dipped one at a time waste a great deal of film. Dipping several parts through one sheet is the standard economy.
- **Masking.** Fitted surfaces, threads and grounding paths are masked by hand.
- **Reject rate.** Voids, bridging, distortion and dust in the clear coat are all visible, so cosmetic parts carry meaningful rework.

Cost-reduction tactics:

1. Dip multiple parts through a single film sheet — film consumed per part is one of the largest variable costs.
2. Choose a base color that matches a standard so a stock paint can be used.
3. Select a forgiving pattern; camouflage and carbon weave hide distortion and dramatically reduce cosmetic rejects.
4. Reduce masked features by relocating fits onto separate components.
5. Where the geometry is simple and the pattern is not required, price spray painting or a printed film wrap instead — those avoid two of the process steps entirely.

## FAQ

### How durable is hydro transfer printing?

Exactly as durable as its clear coat. The printed layer itself has essentially no abrasion, chemical or UV resistance — the two-component urethane topcoat provides all of it. Specify the clear coat system, film thickness and acceptance tests rather than treating the printed pattern as the finish.

### Why does the pattern look different on different parts?

Two reasons. The printed film is translucent, so the base coat color changes the appearance of the same pattern. And the film stretches as it conforms to the geometry, so entry angle, dip rate and part shape all change how the pattern lands. Repeating organic patterns hide this; geometric patterns and logos do not.

### How much thickness does hydro transfer printing add?

Typically 0.003–0.006 in (75–150 µm) per surface for the complete system, essentially all of it from the base coat and clear coat. The transferred ink layer itself is negligible. Threads, bores, sealing faces and grounding surfaces need masking as they would with any sprayed system.

### What materials can be hydro dipped?

Any rigid substrate that can be cleaned, abraded and primed — ABS, polycarbonate, glass-filled nylon, aluminum, steel, wood and composites are all routine. Flexible parts, untreated low-energy plastics and anything that cannot tolerate immersion in warm water are not suitable.

### Can hydro transfer printing place a logo in a specific location?

No, not reliably. The film stretches as it wraps and the pattern lands differently depending on entry angle and geometry, so anything requiring fixed scale, orientation or registration will distort. Pad printing or laser marking is the correct process for a located graphic.

### Why do deep pockets sometimes come out unprinted?

Although the process has no line-of-sight restriction, the floating ink layer can bridge across a narrow slot or a deep pocket instead of conforming into it, leaving a void. Open up narrow features and soften internal corners where the pattern must follow the surface.

## Alternative processes

- [Spray Painting](https://manufacturingprocesses.org/processes/finishing/spray-painting.md): Spray painting atomises liquid coating onto a prepared surface, building a film that cures into a decorative and protective layer.
- [Pad Printing](https://manufacturingprocesses.org/processes/finishing/pad-printing.md): Pad printing lifts ink out of an etched plate with a silicone pad and transfers it onto a curved or recessed surface.
- [Vacuum Metalizing](https://manufacturingprocesses.org/processes/finishing/vacuum-metalizing.md): Vacuum metalizing evaporates aluminum inside a vacuum chamber so it condenses as a mirror-bright film on the part surface.
- [Powder Coating](https://manufacturingprocesses.org/processes/finishing/powder-coating.md): Powder coating sprays electrostatically charged dry polymer powder onto an earthed part, then bakes it into a tough continuous film.
- [UV Inkjet Printing](https://manufacturingprocesses.org/processes/finishing/uv-inkjet-printing.md): UV inkjet printing jets UV-curable ink directly onto a flat substrate and cures it instantly under UV light.

## Related processes

- [Spray Painting](https://manufacturingprocesses.org/processes/finishing/spray-painting.md): Spray painting atomises liquid coating onto a prepared surface, building a film that cures into a decorative and protective layer.
- [Pad Printing](https://manufacturingprocesses.org/processes/finishing/pad-printing.md): Pad printing lifts ink out of an etched plate with a silicone pad and transfers it onto a curved or recessed surface.
- [Vacuum Metalizing](https://manufacturingprocesses.org/processes/finishing/vacuum-metalizing.md): Vacuum metalizing evaporates aluminum inside a vacuum chamber so it condenses as a mirror-bright film on the part surface.
- [Screen Printing](https://manufacturingprocesses.org/processes/finishing/screen-printing.md): Screen printing pushes ink through a stencilled mesh onto the surface below, laying down one opaque color per pass.
- [UV Inkjet Printing](https://manufacturingprocesses.org/processes/finishing/uv-inkjet-printing.md): UV inkjet printing jets UV-curable ink directly onto a flat substrate and cures it instantly under UV light.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/finishing/hydro-transfer-printing)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Laser Marking and Engraving"
category: "Finishing"
subcategory: "Subtractive"
materials: ["Metal", "Plastic", "Wood", "Glass", "Ceramic"]
tolerances: "Annealed marks remove no measurable material; etching removes 0.0001–0.001 in (2.5–25 µm); engraving removes 0.001–0.020 in (0.025–0.5 mm). Beam positioning within the marking field is typically within a few thousandths of an inch"
volumes: "1 to millions of parts; variable data such as serial numbers costs the same as static artwork"
lead_time: "Same day to 5 business days; artwork is a file, so there is no tooling lead time"
url: https://manufacturingprocesses.org/processes/finishing/laser-marking-and-engraving
---

# Laser Marking and Engraving

Laser marking alters the surface with a focused beam — annealing, etching or engraving it — to make permanent, contact-free marks.

- **Category**: [Finishing](https://manufacturingprocesses.org/processes/finishing.md)
- **Family**: Subtractive
- **Materials**: Metal, Plastic, Wood, Glass, Ceramic
- **Typical tolerances**: Annealed marks remove no measurable material; etching removes 0.0001–0.001 in (2.5–25 µm); engraving removes 0.001–0.020 in (0.025–0.5 mm). Beam positioning within the marking field is typically within a few thousandths of an inch
- **Surface finish**: Annealing leaves the surface texture unchanged; etching and engraving leave a matte, micro-roughened cavity
- **Typical volumes**: 1 to millions of parts; variable data such as serial numbers costs the same as static artwork
- **Lead time**: Same day to 5 business days; artwork is a file, so there is no tooling lead time

## Overview

Laser marking uses a focused, galvanometer-steered beam to change a surface permanently — annealing it, oxidizing it, foaming it, bleaching a dye, or ablating material away. It is contactless, needs no consumables or tooling, and the artwork is a file, so serial numbers, date codes and 2D data matrix codes cost the same as a static logo.

The distinction that matters on a drawing is between marking and engraving. Annealed marks on stainless and titanium change color with no measurable material removal and leave the surface passive and crevice-free, which is why they are specified for surgical instruments and implants. Ablative etching removes roughly 0.0001–0.001 in (2.5–25 µm). True engraving removes 0.001–0.020 in (0.025–0.5 mm) and requires many passes. Source selection follows the material: 1064 nm fiber lasers for metals and many engineering plastics, 10.6 µm CO₂ for organics, wood, acrylic and paper, and 355 nm UV for heat-sensitive polymers and medical work.

## How it works

1. **Choose the source.** A fiber laser at 1064 nm couples efficiently into metals and is the default for industrial part marking. CO₂ at 10.6 µm couples into organics — wood, leather, paper, acrylic, coated glass — and does almost nothing on bare metal. UV at 355 nm works by photochemical bond breaking rather than heat, giving clean marks on sensitive polymers with minimal heat-affected zone.

2. **Set the focus.** The beam is focused by an F-theta lens that both focuses and flattens the field. A smaller field gives a smaller spot and higher power density; a larger field marks bigger parts at lower resolution. Depth of field is limited, so the marked surface must sit within roughly ±0.040 in (±1 mm) of focus for consistent results on typical setups.

3. **Mark.** Galvanometer mirrors steer the beam across the field at speeds up to several thousand millimeters per second. Power, frequency, pulse duration, scan speed, line spacing and pass count together determine whether the result is an anneal, an etch or an engraving.

4. **Verify.** Cosmetic marks are inspected visually or by contrast measurement. Data matrix codes on direct part marking are graded against the AIM DPM methodology in ISO/IEC TR 29158, which is the correct thing to specify rather than "must be readable".

### Mark types

**Annealing** heats stainless steel or titanium below the melt point so a controlled oxide layer forms, producing a black or colored mark with essentially no material removal and no disturbance of the surface. **Etching or ablation** melts and vaporizes a thin layer, producing high contrast quickly. **Foaming** on plastics creates gas bubbles that scatter light, giving a light mark on a dark polymer. **Carbon migration** darkens plastics and coated surfaces. On dyed anodized aluminum, a standard fiber laser bleaches the dye to a light mark, while a MOPA fiber laser with adjustable pulse duration can produce a black mark instead — the reason MOPA sources are specified for anodized control panels.

## Design guidelines

### Specify the mark type, not just the artwork

"Laser mark per drawing" leaves the shop to choose between an anneal, an etch and an engraving, which produce very different surfaces. State the method, the required contrast or depth, and the acceptance criterion. For medical and aerospace parts, state whether material removal is permitted at all.

### Keep the mark zone flat and in focus

Depth of field is limited — typically around ±0.040 in (±1 mm). A mark that runs across a curved or stepped surface will change appearance as it goes out of focus. Provide a flat mark pad, or specify rotary-axis marking for cylindrical parts.

### Size text to the process, not to the CAD

Practical minimum character height for a reliably legible fiber-laser mark is around 0.020–0.040 in (0.5–1 mm); go smaller and legibility depends on the specific setup and viewing condition. Data matrix codes need enough cell size and enough contrast to grade — specify cell size and a minimum grade rather than an overall code dimension.

### Decide where marking sits in the routing

Mark after anodizing to cut or bleach through the color for high contrast; mark before anodizing to have the anodic film cover an engraved mark uniformly. On stainless medical devices, marking is typically followed by passivation, because ablative marking can locally disturb the passive layer.

### Know which materials resist marking

Bare aluminum and copper reflect 1064 nm strongly and mark poorly without high power or a marking compound; copper and gold often require a green laser. Clear plastics transmit the beam and need an additive or a coated surface. Glass marks by micro-fracturing and can chip. Confirm on the actual production material and finish, not on a look-alike coupon.

### Do not use engraving depth as a strength feature

Deep engraving removes 0.001–0.020 in (0.025–0.5 mm) and creates a stress raiser. Keep deep marks away from fillets, fatigue-critical surfaces and thin walls, and off any surface subject to cyclic loading.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Annealed mark | No measurable removal | Stainless and titanium only | Oxide color change, surface stays passive |
| Etched mark depth | 0.0001–0.001 in (2.5–25 µm) | — | Single or few passes, high contrast |
| Engraved depth | 0.001–0.020 in (0.025–0.5 mm) | Many passes; slow | Removal rate falls as the cavity deepens |
| Depth of field | ±0.040 in (±1 mm) | Provide a flat mark pad | Appearance changes out of focus |
| Minimum character height | 0.020–0.040 in (0.5–1 mm) | Verify on the actual finish | Legibility depends on contrast and viewing |
| Data matrix codes | Specify cell size and ISO/IEC TR 29158 grade | — | "Readable" is not a specification |

## Cost drivers

Laser marking has essentially no consumable cost and no tooling cost. The economics are machine time plus setup, which makes it cheap at high volume and unusually cheap at volume one.

- **Cycle time.** Marking time scales with marked area and pass count. An annealed logo takes seconds; a deep engraving that removes 0.010 in over a large area takes minutes and dominates the price.
- **Fixturing.** Repeatable part location is the setup cost. A simple nest for a flat part is trivial; multi-axis or rotary fixturing for cylindrical or contoured parts is real tooling.
- **Source type.** Standard fiber marking is the cheapest. MOPA, UV and green sources cost more per hour and are specified only when the material or the required mark demands them.
- **Verification.** Contrast measurement, code grading to ISO/IEC TR 29158 and traceability documentation are separate line items on regulated work.
- **Downstream operations.** Passivation after marking on stainless medical parts, or re-anodizing, adds a process step that often exceeds the marking cost.

Cost-reduction tactics:

1. Mark the smallest area that satisfies the requirement — area drives cycle time directly.
2. Use an anneal or a shallow etch where a legible permanent mark is all that is needed; depth is expensive.
3. Design a flat mark pad into the part so a simple nest fixture works and no rotary axis is required.
4. Consolidate all marking content — logo, part number, serial, code — into one field so the part is loaded once.
5. Where hundreds of identical parts need identical artwork with no variable data, compare against pad or screen printing, which can be faster per piece on large marked areas.

## FAQ

### What is the difference between laser marking, etching and engraving?

Marking by annealing changes the surface color through a controlled oxide with no measurable material removal. Etching or ablation removes roughly 0.0001–0.001 in (2.5–25 µm). Engraving removes 0.001–0.020 in (0.025–0.5 mm) and needs many passes. Annealing is specified where the surface must stay passive and crevice-free, such as on surgical instruments.

### Which laser source do I need for my material?

Fiber at 1064 nm for metals and many engineering plastics — the industrial default. CO₂ at 10.6 µm for organics, wood, acrylic, leather and paper; it does almost nothing on bare metal. UV at 355 nm for heat-sensitive polymers and medical work, because it marks photochemically with a minimal heat-affected zone.

### Can you laser mark anodized aluminum?

Yes, and it is one of the most common applications. A standard fiber laser bleaches the dye to produce a light mark on a dark anodized surface. A MOPA fiber laser with adjustable pulse duration can instead produce a black mark on light anodize, which is why MOPA sources are specified for control panels and nameplates.

### Should I laser mark before or after anodizing?

After, if you want high contrast from cutting or bleaching through the color. Before, if you want the anodic film to cover an engraved mark uniformly for corrosion protection. On stainless medical devices, marking is normally followed by passivation because ablative marking can disturb the passive layer.

### How small can laser-marked text be?

Practical minimum character height for reliably legible fiber-laser marking is around 0.020–0.040 in (0.5–1 mm), though legibility depends on contrast, surface finish and viewing conditions. For 2D data matrix codes, specify cell size and a minimum grade to ISO/IEC TR 29158 rather than an overall code size.

### Which materials are difficult to laser mark?

Bare aluminum and copper reflect 1064 nm strongly and mark poorly without high power or a marking compound; copper and gold often need a green laser. Clear plastics transmit the beam and require an additive or a coated surface. Glass marks by micro-fracturing and can chip, so it is usually marked through a coating.

## Alternative processes

- [CNC Engraving](https://manufacturingprocesses.org/processes/finishing/cnc-engraving.md): CNC engraving cuts text and artwork into a surface with a small pointed or ball-nose tool following a programmed path.
- [Pad Printing](https://manufacturingprocesses.org/processes/finishing/pad-printing.md): Pad printing lifts ink out of an etched plate with a silicone pad and transfers it onto a curved or recessed surface.
- [Screen Printing](https://manufacturingprocesses.org/processes/finishing/screen-printing.md): Screen printing pushes ink through a stencilled mesh onto the surface below, laying down one opaque color per pass.
- [Photo Etching](https://manufacturingprocesses.org/processes/finishing/photo-etching.md): Photo etching masks a surface photographically and etches shallow detail into it for decoration, marking or texture.
- [UV Inkjet Printing](https://manufacturingprocesses.org/processes/finishing/uv-inkjet-printing.md): UV inkjet printing jets UV-curable ink directly onto a flat substrate and cures it instantly under UV light.

## Related processes

- [CNC Engraving](https://manufacturingprocesses.org/processes/finishing/cnc-engraving.md): CNC engraving cuts text and artwork into a surface with a small pointed or ball-nose tool following a programmed path.
- [Photo Etching](https://manufacturingprocesses.org/processes/finishing/photo-etching.md): Photo etching masks a surface photographically and etches shallow detail into it for decoration, marking or texture.
- [Anodizing](https://manufacturingprocesses.org/processes/finishing/anodizing.md): Anodizing grows a hard porous oxide layer on aluminum electrochemically, which can then be dyed and sealed for wear and corrosion resistance.
- [Passivation](https://manufacturingprocesses.org/processes/finishing/passivation.md): Passivation removes free iron from a stainless steel surface with an acid bath so the chromium oxide layer can reform unbroken.
- [Pad Printing](https://manufacturingprocesses.org/processes/finishing/pad-printing.md): Pad printing lifts ink out of an etched plate with a silicone pad and transfers it onto a curved or recessed surface.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/finishing/laser-marking-and-engraving)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "PVD Coating"
category: "Finishing"
subcategory: "Additive"
materials: ["Metal", "Plastic", "Glass", "Ceramic"]
tolerances: "Adds 0.5–5 µm (0.00002–0.0002 in) per surface — dimensionally negligible for most fits; thicker tooling coatings at 4–5 µm should be checked on close-fitting punch and die clearances"
volumes: "1 part to high volume; cost per part is governed by how densely the chamber can be loaded"
lead_time: "3–10 business days at a coating service; cycle time itself is 2–6 hours"
url: https://manufacturingprocesses.org/processes/finishing/pvd-coating
---

# PVD Coating

PVD coating vaporises a solid target inside a vacuum chamber so it condenses on the part as a very thin, extremely hard film.

- **Category**: [Finishing](https://manufacturingprocesses.org/processes/finishing.md)
- **Family**: Additive
- **Materials**: Metal, Plastic, Glass, Ceramic
- **Typical tolerances**: Adds 0.5–5 µm (0.00002–0.0002 in) per surface — dimensionally negligible for most fits; thicker tooling coatings at 4–5 µm should be checked on close-fitting punch and die clearances
- **Surface finish**: Fully conformal — reproduces the incoming finish; arc-deposited films raise Ra slightly, sputtered films less so
- **Typical volumes**: 1 part to high volume; cost per part is governed by how densely the chamber can be loaded
- **Lead time**: 3–10 business days at a coating service; cycle time itself is 2–6 hours

## Overview

Physical vapor deposition converts a solid target material into vapor inside a vacuum chamber and condenses it on the part as an extremely thin, extremely hard ceramic film. Typical coatings are 0.5–5 µm (0.00002–0.0002 in) thick — thin enough that the part comes out of the chamber dimensionally as it went in, and thin enough that it reproduces the incoming surface finish exactly.

The industrial value is hardness: TiN is commonly quoted near 2,300 HV, TiCN near 3,000 HV, and TiAlN near 3,300 HV, against roughly 900 HV for hardened tool steel. Those numbers, plus service temperatures of 600 °C (1,110 °F) for TiN and around 800 °C (1,470 °F) for TiAlN, are why PVD dominates cutting tool and injection mold coating. The same process in thinner decorative films gives the gold, bronze, black and gunmetal finishes on hardware, watches and firearms. The main constraints are line-of-sight deposition, a process temperature that must sit below the substrate's tempering temperature, and batch-based economics.

## How it works

1. **Prepare the surface.** PVD adds no thickness and hides nothing, so the substrate must already be at its final finish and must be metallurgically clean. Grinding, polishing or lapping comes first. EDM white layer, residual plating, oxide, and any organic residue must be removed — adhesion failures nearly always trace to this step.

2. **Clean.** Multi-stage ultrasonic aqueous or solvent cleaning followed by vacuum drying. Fingerprints deposited after cleaning are enough to cause flaking.

3. **Load and pump down.** Parts are fixtured on planetary rotation so every surface presents to the source, then the chamber is evacuated. Pump-down alone takes a significant fraction of the cycle.

4. **Heat and ion-etch.** The load is heated, typically to 400–500 °C (750–930 °F) for arc processes, and bombarded with ions to sputter-clean the surface at the atomic level and improve adhesion. Low-temperature processes below about 250 °C (480 °F) exist for substrates that cannot take the heat.

5. **Deposit.** In cathodic arc evaporation, an arc vaporizes and ionizes the target; in magnetron sputtering, ions knock atoms off the target. Reactive gas — nitrogen for nitrides, acetylene for carbides — is bled in to form the compound on the part. Deposition rates give 0.5–5 µm over roughly 1–4 hours.

6. **Cool and unload.** Total cycle is typically 2–6 hours including pump-down and cooling, and the whole chamber load shares that cycle.

### Arc versus sputtering

Cathodic arc gives high ionization, excellent adhesion and fast deposition, but ejects micro-droplets that leave the surface slightly rougher than it went in. Magnetron sputtering is smoother and better for decorative and optical work, at lower deposition rates. Choose arc for tooling, sputtering where the appearance of the finished surface is critical.

## Design guidelines

### The substrate must be finished before coating

At 0.5–5 µm the film is fully conformal. Every scratch, tool mark, EDM texture and polish line comes through unchanged, and arc-deposited films slightly increase Ra. If the requirement is a mirror mold surface with a hard coating, polish to that finish first — see the [surface finish chart](/charts/surface-finish-chart) for the Ra achievable by each prep method.

### Check the substrate's tempering temperature

Arc PVD at 400–500 °C (750–930 °F) will over-temper any steel tempered below that. High-speed steels tempered around 540 °C (1,000 °F) are safe. Through-hardened tool steels tempered at 200 °C, precipitation-hardened stainless in certain conditions, and most aluminum are not — those need a low-temperature process below roughly 250 °C (480 °F), which typically trades away some hardness and adhesion.

### Design for line of sight

Vapor travels in straight lines. Deep bores, blind holes, undercuts and the inside of tubes coat thin or not at all, even with planetary rotation. As a working rule, expect good coverage down to about one diameter into a hole and progressively less beyond that. Where an internal surface must be coated uniformly, electroless nickel or a CVD process is the better answer.

### Do not use PVD to fix dimensions or fill defects

Unlike hard chrome, PVD cannot be used to build a worn surface back up. It is a surface property modification, not a repair.

### Provide fixturing features

Parts must be held without shadowing the coated surface, and the fixture contact point is uncoated. On tooling, that is usually the shank or a bore. On decorative parts, designate an acceptable contact area.

### Match the coating to the duty

TiN (gold, ~2,300 HV, to about 600 °C) is the general-purpose choice. TiCN (~3,000 HV) is harder and tougher for interrupted cuts and forming tools. TiAlN/AlTiN (~3,300 HV, to about 800 °C) forms a protective alumina layer at high temperature and is the standard for dry and high-speed machining. CrN is more corrosion resistant and less brittle, used on molds and forming dies. DLC gives a very low coefficient of friction for sliding and non-stick applications.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Coating thickness | 2–4 µm on tooling; 0.25–1 µm decorative | 5 µm | Thicker films become brittle and lose adhesion |
| Incoming surface | Final finish, fully cleaned | — | The film is conformal and hides nothing |
| Substrate temper | Above the process temperature | 400–500 °C (750–930 °F) for arc PVD | Coating cycle will over-temper softer-tempered steels |
| Bore coverage | ~1 diameter deep | — | Line-of-sight deposition |
| Fixture contact | 1 designated uncoated area | — | The part must be held somewhere |
| Chamber utilization | Fill the load | — | Cost is per cycle, not per part |

## Cost drivers

PVD is billed per chamber cycle, not per part. A 2–6 hour cycle costs the same whether the fixture holds twelve parts or twelve hundred, so the single largest lever is load density.

- **Chamber utilization.** Small parts that fixture densely coat at a low unit cost. A single large part occupying a whole chamber carries the entire cycle.
- **Substrate preparation.** Polishing, lapping and cleaning to PVD standards frequently cost more than the coating cycle itself, and are non-negotiable.
- **Coating chemistry.** TiN is the least expensive. Multilayer, nanolayer, AlTiN and DLC coatings need longer cycles and more expensive targets.
- **Fixturing.** Custom fixtures for planetary rotation are tooling, amortized over the program.
- **Requalification and stripping.** Recoating worn tools requires stripping the old coating chemically, which is an extra operation and can attack the substrate if repeated too often.

Cost-reduction tactics:

1. Batch parts so the chamber runs full — this dominates every other cost factor.
2. Specify the incoming Ra you actually need; polishing beyond the functional requirement is wasted money.
3. Use TiN unless the temperature or wear duty genuinely requires an aluminum-bearing or DLC coating.
4. Standardize fixturing across a family of parts so one fixture serves many programs.
5. Where a deep internal surface needs coating, price electroless nickel against PVD before committing — line-of-sight limits may make PVD unusable at any price.

## FAQ

### How thick is a PVD coating?

Typically 0.5–5 µm (0.00002–0.0002 in). Cutting tools and molds are usually coated at 2–4 µm; decorative finishes at 0.25–1 µm. At those thicknesses the coating is dimensionally negligible for most fits and reproduces the underlying surface finish exactly.

### How hard is PVD coating?

Common tool coatings are quoted near 2,300 HV for TiN, 3,000 HV for TiCN and 3,300 HV for TiAlN, against roughly 900 HV for hardened tool steel. Maximum service temperature also varies by chemistry — about 600 °C (1,110 °F) for TiN and about 800 °C (1,470 °F) for TiAlN.

### Will PVD coating damage my heat-treated part?

It can. Cathodic arc PVD runs at 400–500 °C (750–930 °F), which over-tempers any steel tempered below that. High-speed steels tempered around 540 °C (1,000 °F) are safe. For substrates that cannot take the heat, low-temperature processes below roughly 250 °C (480 °F) are available, usually with some loss of hardness and adhesion.

### Can PVD coat the inside of a hole or tube?

Poorly. Deposition is line-of-sight, so coverage drops off rapidly beyond about one diameter into a hole and blind pockets and undercuts may not coat at all. Where an internal surface needs uniform coverage, electroless nickel or a CVD process is the appropriate choice.

### Does PVD hide scratches or improve surface finish?

No. At a few microns the film is fully conformal, so every scratch, tool mark and polish line shows through, and arc-deposited coatings raise Ra slightly. The substrate must be brought to its final finish before coating.

### Which PVD coating should I choose for cutting tools?

TiN is the general-purpose default. TiCN is harder and tougher for interrupted cuts and forming. TiAlN or AlTiN forms a protective alumina layer at high temperature and is standard for dry and high-speed machining. CrN suits molds and forming dies where corrosion resistance and toughness matter more than peak hardness.

## Alternative processes

- [Electroplating](https://manufacturingprocesses.org/processes/finishing/electroplating.md): Electroplating deposits a metal coating from solution onto a conductive part by passing current through an electrolyte bath.
- [Cerakote Ceramic Coating](https://manufacturingprocesses.org/processes/finishing/cerakote-ceramic-coating.md): Cerakote sprays a thin ceramic-filled polymer coating onto a prepared part and cures it, giving wear and corrosion resistance in a very thin film.
- [Anodizing](https://manufacturingprocesses.org/processes/finishing/anodizing.md): Anodizing grows a hard porous oxide layer on aluminum electrochemically, which can then be dyed and sealed for wear and corrosion resistance.
- [Vacuum Metalizing](https://manufacturingprocesses.org/processes/finishing/vacuum-metalizing.md): Vacuum metalizing evaporates aluminum inside a vacuum chamber so it condenses as a mirror-bright film on the part surface.
- [Black Oxide](https://manufacturingprocesses.org/processes/finishing/black-oxide.md): Black oxide converts the steel surface to magnetite in a hot alkaline bath, giving a black finish that adds essentially no thickness.

## Related processes

- [Vacuum Metalizing](https://manufacturingprocesses.org/processes/finishing/vacuum-metalizing.md): Vacuum metalizing evaporates aluminum inside a vacuum chamber so it condenses as a mirror-bright film on the part surface.
- [Grinding, Sanding and Polishing](https://manufacturingprocesses.org/processes/finishing/grinding-sanding-and-polishing.md): Grinding, sanding and polishing remove a thin layer of material with bonded or coated abrasive to correct form or improve surface finish.
- [Electroplating](https://manufacturingprocesses.org/processes/finishing/electroplating.md): Electroplating deposits a metal coating from solution onto a conductive part by passing current through an electrolyte bath.
- [Cerakote Ceramic Coating](https://manufacturingprocesses.org/processes/finishing/cerakote-ceramic-coating.md): Cerakote sprays a thin ceramic-filled polymer coating onto a prepared part and cures it, giving wear and corrosion resistance in a very thin film.
- [Passivation](https://manufacturingprocesses.org/processes/finishing/passivation.md): Passivation removes free iron from a stainless steel surface with an acid bath so the chromium oxide layer can reform unbroken.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/finishing/pvd-coating)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Pad Printing"
category: "Finishing"
subcategory: "Printing"
materials: ["Plastic", "Metal", "Glass", "Ceramic"]
tolerances: "Image position typically within ±0.005 in (±0.13 mm); the transferred ink film is a few microns and adds nothing dimensionally"
volumes: "A few hundred to millions of parts; tooling cost is low enough for short runs"
lead_time: "1–2 weeks including cliché and fixture; 2–5 business days for a repeat order"
url: https://manufacturingprocesses.org/processes/finishing/pad-printing
---

# Pad Printing

Pad printing lifts ink out of an etched plate with a silicone pad and transfers it onto a curved or recessed surface.

- **Category**: [Finishing](https://manufacturingprocesses.org/processes/finishing.md)
- **Family**: Printing
- **Materials**: Plastic, Metal, Glass, Ceramic
- **Typical tolerances**: Image position typically within ±0.005 in (±0.13 mm); the transferred ink film is a few microns and adds nothing dimensionally
- **Surface finish**: Adds a very thin ink film that follows the substrate; gloss is set by the ink system
- **Typical volumes**: A few hundred to millions of parts; tooling cost is low enough for short runs
- **Lead time**: 1–2 weeks including cliché and fixture; 2–5 business days for a repeat order

## Overview

Pad printing transfers ink from an etched plate to a part using a soft silicone pad. The pad picks the image out of the etched recess, the ink's surface chemistry changes during the brief transit, and the pad releases the image onto the part as it deforms against the surface. Because a silicone pad conforms to whatever it touches, pad printing is the standard way to put graphics onto curved, spherical, recessed, stepped and textured surfaces that no flat printing process can reach.

The etched plate — the cliché — is typically cut to 20–30 µm (0.0008–0.0012 in) deep, and the ink film that survives onto the part is a fraction of that: a thin, low-build mark that adds essentially nothing dimensionally. One pad and one cliché produce one color, so multi-color work uses multi-station machines with registration typically held within about ±0.005 in (±0.13 mm). Typical applications are keycaps, medical device graduations, cosmetic closures, automotive switchgear, toys and instrument dials.

## How it works

1. **Etch the cliché.** The image is etched into a steel or photopolymer plate to a controlled depth of roughly 20–30 µm (0.0008–0.0012 in). Depth is critical: too shallow and the pad cannot pick up enough ink, too deep and the pad cannot pull all the ink out, leaving a hollow center in solid areas.

2. **Flood and doctor.** Ink floods the plate and a doctor blade or sealed ink cup wipes the surface clean, leaving ink only in the etched image.

3. **Pick up.** The silicone pad presses onto the plate and lifts the ink film. Silicone's very low surface energy is what makes the process work — ink prefers almost any other surface to the pad.

4. **Transit.** As the pad travels, solvent flashes from the exposed outer face of the ink film, making it tackier on the outside than on the pad side. This asymmetry is what drives near-complete release onto the part.

5. **Transfer.** The pad deforms against the part and rolls the image on from the center outward. Pad shape matters: a pointed pad rolls air out ahead of the print, while a flat pad traps it and causes voids.

6. **Cure.** One-component solvent inks air dry or force dry. Two-component catalyzed inks — epoxy and polyurethane — crosslink over hours to days and give the adhesion and chemical resistance needed on hard plastics, glass and metal. UV inks cure in seconds under a lamp.

### Surface energy governs adhesion

Ink will not wet a low-energy surface. Roughly 38 dyne/cm is the working threshold, and polypropylene, polyethylene, acetal and fluoropolymers all fall below it as molded. Those materials require flame, corona or plasma treatment immediately before printing, and the treatment decays over hours to days, so it has to be scheduled as part of the print operation rather than done in advance.

## Design guidelines

### Keep the image inside a printable zone

The pad has to conform to the surface without folding. Compound curves, deep pockets, steps and edges within the image area cause distortion, smearing or voids. Keep artwork on a single continuous surface, stay at least 0.040 in (1 mm) away from an edge or a parting line, and avoid printing across a radius tighter than the pad can follow.

### Size features to the process

Minimum line width of roughly 0.004 in (0.1 mm) is achievable with a well-controlled cliché and a fine pad. Fine serifs, small reversed-out text and tight halftones lose definition. Solid areas larger than about 0.5 in (13 mm) across risk a hollow center — designers break large solids into a screened fill or accept a specialized deep-etch cliché.

### Budget one setup per color

Each color needs its own cliché, pad, ink and station. A four-color graphic on a curved surface is a four-station machine or four passes, with registration typically ±0.005 in (±0.13 mm). Design registration tolerance into the artwork with trapping rather than assuming perfect alignment.

### Specify the ink system and the adhesion test

Ink chemistry, not the printing, determines whether the graphic survives. Specify the ink family (one-component, two-component catalyzed, or UV), any required pretreatment, and an acceptance test — cross-hatch tape adhesion to ASTM D3359 plus whatever solvent, abrasion or sterilization exposure the product will see.

### Account for surface energy at the design stage

If the part is polypropylene, polyethylene, acetal or a fluoropolymer, plan for flame, corona or plasma treatment as an in-line step. Mold release, plasticizers and antistatic additives migrate to the surface and destroy adhesion, so the resin formulation matters as much as the base polymer.

### Provide a locating feature

The part must sit in a nest that positions it repeatably relative to the pad. A flat, a boss, a pair of holes or a molded-in datum makes fixturing simple; a free-form part without one drives fixture cost and registration scatter.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Cliché etch depth | 20–30 µm (0.0008–0.0012 in) | — | Too shallow starves the pad; too deep hollows solids |
| Minimum line width | 0.008 in (0.2 mm) | 0.004 in (0.1 mm) | Fine detail loses definition in transfer |
| Solid area size | Under ~0.5 in (13 mm) across | Screen or deep-etch beyond | Large solids print with a hollow center |
| Distance from edge | 0.040 in (1 mm) minimum | — | Pad folds and smears at an edge |
| Registration (multi-color) | ±0.005 in (±0.13 mm) | Design in trapping | Each color is a separate station |
| Substrate surface energy | ≥ 38 dyne/cm | Pretreat PP, PE, POM, PTFE | Ink cannot wet a low-energy surface |

## Cost drivers

Pad printing carries very low tooling cost and moderate per-part cost, which makes it economical from a few hundred pieces upward.

- **Number of colors.** Each color is a cliché, a pad, an ink station and a setup. Color count is the dominant cost variable.
- **Cliché type.** Photopolymer plates are cheap and fast but wear; steel clichés cost more and last for long production runs.
- **Fixturing.** A simple nest is inexpensive; a multi-cavity or indexing fixture for a contoured part is real tooling amortized over the program.
- **Pretreatment.** Flame, corona or plasma treatment on low-energy plastics is an in-line operation with its own equipment and process control.
- **Ink system and testing.** Two-component catalyzed inks have a pot life and produce waste; qualification testing for adhesion, chemical and abrasion resistance is a fixed program cost.

Cost-reduction tactics:

1. Reduce the color count. A single-color mark on a molded-in contrasting surface is dramatically cheaper than a four-color graphic.
2. Design the artwork within one continuous surface so a simple pad and nest will do the job.
3. Choose a printable resin — an ABS or PC part avoids the pretreatment line entirely.
4. Use molded-in text or texture for content that never changes, and print only what varies.
5. Where the surface is flat and the run is long, compare against screen printing, which lays down a heavier, more opaque film per pass at a lower per-piece rate.

## FAQ

### What can pad printing print on that other processes cannot?

Curved, spherical, recessed, stepped and textured surfaces. The silicone pad deforms to whatever it touches and rolls the image on from the center outward, which is why pad printing is used for keycaps, medical device graduations, cosmetic closures and instrument dials that no flat process can reach.

### How fine a detail can pad printing hold?

Roughly 0.004 in (0.1 mm) minimum line width with a well-controlled cliché, though 0.008 in (0.2 mm) is a safer design target. Fine serifs, small reversed-out text and tight halftones lose definition, and solid areas larger than about 0.5 in (13 mm) across risk printing with a hollow center.

### Why does ink not stick to my polypropylene part?

Surface energy. Ink cannot wet a surface below roughly 38 dyne/cm, and polypropylene, polyethylene, acetal and fluoropolymers all fall below that as molded. They require flame, corona or plasma treatment immediately before printing, and the treatment decays over hours to days so it must be an in-line step.

### How many colors can pad printing do?

As many as the machine has stations, but each color requires its own cliché, pad, ink and setup, so color count is the main cost driver. Registration between colors is typically held within about ±0.005 in (±0.13 mm), so artwork should include trapping rather than assume perfect alignment.

### How thick is a pad-printed ink film?

Very thin. The cliché is etched only 20–30 µm (0.0008–0.0012 in) deep and the film that survives onto the part is a fraction of that — a few microns. It adds nothing dimensionally, but it also means pad printing is a poor choice where heavy opacity over a dark substrate is required.

### Pad printing or screen printing?

Pad printing for curved, recessed or textured surfaces and small graphics with low tooling cost. Screen printing for flat or gently curved surfaces where a much heavier, more opaque ink film is needed, or where the image area is large — a screen lays down far more ink per pass at a lower per-piece rate on long runs.

## Alternative processes

- [Screen Printing](https://manufacturingprocesses.org/processes/finishing/screen-printing.md): Screen printing pushes ink through a stencilled mesh onto the surface below, laying down one opaque color per pass.
- [UV Inkjet Printing](https://manufacturingprocesses.org/processes/finishing/uv-inkjet-printing.md): UV inkjet printing jets UV-curable ink directly onto a flat substrate and cures it instantly under UV light.
- [Laser Marking and Engraving](https://manufacturingprocesses.org/processes/finishing/laser-marking-and-engraving.md): Laser marking alters the surface with a focused beam — annealing, etching or engraving it — to make permanent, contact-free marks.
- [Hydro Transfer Printing](https://manufacturingprocesses.org/processes/finishing/hydro-transfer-printing.md): Hydro transfer printing floats a printed film on water, activates it, and wraps it around a submerged part to cover it in a pattern.
- [CNC Engraving](https://manufacturingprocesses.org/processes/finishing/cnc-engraving.md): CNC engraving cuts text and artwork into a surface with a small pointed or ball-nose tool following a programmed path.

## Related processes

- [Screen Printing](https://manufacturingprocesses.org/processes/finishing/screen-printing.md): Screen printing pushes ink through a stencilled mesh onto the surface below, laying down one opaque color per pass.
- [UV Inkjet Printing](https://manufacturingprocesses.org/processes/finishing/uv-inkjet-printing.md): UV inkjet printing jets UV-curable ink directly onto a flat substrate and cures it instantly under UV light.
- [Laser Marking and Engraving](https://manufacturingprocesses.org/processes/finishing/laser-marking-and-engraving.md): Laser marking alters the surface with a focused beam — annealing, etching or engraving it — to make permanent, contact-free marks.
- [Hydro Transfer Printing](https://manufacturingprocesses.org/processes/finishing/hydro-transfer-printing.md): Hydro transfer printing floats a printed film on water, activates it, and wraps it around a submerged part to cover it in a pattern.
- [Foil Blocking and Embossing](https://manufacturingprocesses.org/processes/finishing/foil-blocking-and-embossing.md): Foil blocking and embossing press a heated die into the surface to transfer metallic foil, raise a relief, or do both at once.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/finishing/pad-printing)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Passivation"
category: "Finishing"
subcategory: "Additive"
materials: ["Metal"]
tolerances: "No measurable dimensional change; the passive film is a few nanometers thick"
volumes: "1 part to millions; small parts run in bulk baskets at very low cost per piece"
lead_time: "1–5 business days at a job shop; often same-day when combined with an existing cleaning operation"
url: https://manufacturingprocesses.org/processes/finishing/passivation
---

# Passivation

Passivation removes free iron from a stainless steel surface with an acid bath so the chromium oxide layer can reform unbroken.

- **Category**: [Finishing](https://manufacturingprocesses.org/processes/finishing.md)
- **Family**: Additive
- **Materials**: Metal
- **Typical tolerances**: No measurable dimensional change; the passive film is a few nanometers thick
- **Surface finish**: Unchanged — passivation neither smooths nor brightens the surface
- **Typical volumes**: 1 part to millions; small parts run in bulk baskets at very low cost per piece
- **Lead time**: 1–5 business days at a job shop; often same-day when combined with an existing cleaning operation

## Overview

Passivation is a chemical dip that removes free iron and other exogenous contamination from a stainless steel surface so the chromium oxide film that gives stainless its corrosion resistance can reform continuous and unbroken. Nothing is deposited and essentially nothing is removed — the passive film is only a few nanometers thick, so passivation is dimensionally neutral and invisible.

It is a corrective, not a coating. Machining, grinding, blasting with contaminated media, tumbling with steel media and even handling with carbon-steel tooling smear free iron across a stainless surface. That embedded iron rusts, and the rust spots look like the stainless has failed when the substrate is fine. Two chemistries dominate: nitric acid, typically 20–50% by volume at 70–130 °F (21–54 °C), and citric acid, typically 4–10% by weight at 70–160 °F (21–71 °C). ASTM A967 and AMS 2700 define both, along with the tests used to prove the result.

## How it works

1. **Degrease.** Cutting oil, coolant, drawing compound and fingerprints must be removed first. An acid bath cannot passivate through a film of oil, and this is the most common cause of a failed passivation.

2. **Descale if required.** Passivation does not remove weld heat tint, mill scale or oxide. Those require pickling, mechanical cleaning or blasting with clean non-ferrous media beforehand. A heat-tinted weld zone that has only been passivated will still corrode, because the chromium-depleted layer under the tint is still there.

3. **Immerse in the acid.** Nitric acid dissolves free iron much faster than it dissolves the chromium-rich stainless matrix. Citric acid works by chelating iron instead and is increasingly preferred because it is not an oxidizer, is easier to handle and dispose of, and does not require the sodium dichromate additions that some nitric baths do. Times run roughly 20–30 minutes for nitric and 4–30 minutes for citric depending on grade and temperature.

4. **Rinse thoroughly.** Residual acid in blind holes, threads or crevices causes the exact corrosion the process was meant to prevent. Deionized water for the final rinse on critical work.

5. **Dry and let the film form.** The passive chromium oxide layer reforms on exposure to air within minutes and continues to strengthen over roughly 24–48 hours.

### Choosing the bath

Free-machining grades — 303, 416 and other sulfur- or selenium-bearing steels — and high-carbon martensitic grades can flash-attack in plain nitric acid, etching and darkening instead of passivating. ASTM A967 addresses this with specific nitric types (including dichromate-inhibited baths) and with citric alternatives. Specify the ASTM A967 or AMS 2700 method and the acceptance test rather than just writing "passivate".

## Design guidelines

### Passivation fixes contamination, not design

If a part corrodes because the grade is wrong for the environment, because a crevice traps chlorides, or because a weld was not descaled, passivation will not save it. Choose the alloy first: 304 for general use, 316/316L where chlorides are present, and duplex or higher grades for aggressive service.

### Keep iron off the part in the first place

Specify stainless-only or ceramic blast media, non-ferrous or dedicated stainless tooling, and separate stainless from carbon steel in tumbling, deburring, forming and storage. Passivation removes surface iron but cannot recover iron that has been pressed deep into a soft, worked surface.

### Design out acid traps

Blind holes, threads, crimped joints, spot-welded overlaps and unsealed crevices hold acid through rinsing. Provide drain paths and through-features where possible, and specify a deionized final rinse for parts with internal passages.

### Do not expect an appearance change

Passivation leaves the surface looking exactly as it went in. Machining marks, dullness, discoloration and weld tint all remain. If the requirement is appearance or cleanability, the surface has to be mechanically finished or electropolished; see the [surface finish chart](/charts/surface-finish-chart) for the Ra ranges each prep achieves.

### Specify the test, not just the process

ASTM A967 pairs each treatment with acceptance practices — high humidity, water immersion, salt spray, copper sulfate and free-iron tests. A drawing note that reads "passivate per ASTM A967, citric 3, test per Practice E" is verifiable; "passivate" alone is not.

### Sequence it last

Passivation belongs after all machining, welding, forming, marking and deburring. Any subsequent operation with steel tooling reintroduces free iron, and laser marking can locally disturb the passive layer, which is why medical device routings often place passivation after marking.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Nitric acid bath | 20–50 vol%, 70–130 °F (21–54 °C) | Not for free-machining grades uninhibited | Sulfur-bearing grades flash-attack |
| Citric acid bath | 4–10 wt%, 70–160 °F (21–71 °C) | — | Non-oxidizing, safer handling and disposal |
| Dimensional change | None measurable | — | The passive film is a few nanometers thick |
| Pre-cleaning | Full degrease mandatory | — | Acid cannot passivate through oil |
| Weld heat tint | Remove mechanically or by pickling first | — | Passivation does not restore a chromium-depleted layer |
| Position in routing | After all machining and marking | — | Steel tooling recontaminates the surface |

## Cost drivers

Passivation is among the cheapest metal finishing operations. It is priced per lot or per pound, and the bath cost per part is small, so the price is driven almost entirely by handling and paperwork.

- **Part handling.** Small parts run in baskets at very low cost per piece. Large weldments must be racked, immersed individually or spray-treated.
- **Bath choice.** Citric baths generally carry lower disposal and handling cost than nitric; nitric with dichromate carries the highest.
- **Testing and certification.** Copper sulfate, high humidity, water immersion or salt spray testing per ASTM A967, plus a certificate of conformance, is often the largest single line on a small order.
- **Pre-cleaning exceptions.** Parts arriving with weld tint, mill scale, heavy oil or steel-media embedment need pickling, blasting or descaling first — a separate operation.
- **Rework.** A failed copper sulfate test means re-cleaning and re-running the whole lot.

Cost-reduction tactics:

1. Keep carbon steel away from stainless throughout the shop; contamination avoided is passivation rework avoided.
2. Batch parts so the lot minimum is spread over as many pieces as possible.
3. Specify citric acid unless a customer specification requires nitric.
4. Combine passivation with an existing cleaning step in the routing rather than adding a separate handling cycle.
5. Where the part also needs a bright, low-particle surface, price electropolishing once instead of mechanical polishing plus passivation.

## FAQ

### What does passivation actually do?

It removes free iron and other surface contamination from stainless steel so the chromium oxide film can reform continuous and unbroken. Nothing is deposited and essentially no metal is removed — the passive film is only a few nanometers thick, so the process is dimensionally neutral and visually invisible.

### Citric or nitric acid passivation — which should I specify?

Citric acid, typically 4–10% by weight at 70–160 °F (21–71 °C), is increasingly the default because it is non-oxidizing, easier to handle and cheaper to dispose of. Nitric acid at 20–50% by volume remains common and is required by some customer specifications. Both are defined in ASTM A967 and AMS 2700.

### Does passivation remove weld heat tint or scale?

No. Passivation dissolves free iron, not oxide. Heat tint, mill scale and weld scale must be removed by pickling, blasting with clean non-ferrous media, or mechanical cleaning first. A passivated but still heat-tinted weld will corrode because the chromium-depleted layer beneath the tint remains.

### Does passivation change part dimensions?

No measurable amount. This is the main practical difference from electropolishing, which removes 0.0002–0.001 in (5–25 µm) per surface. Passivation can therefore be applied to finished, toleranced parts without any allowance.

### Why did my 303 stainless part come out etched and dark after passivation?

Free-machining grades such as 303 and 416 contain sulfur or selenium, and they can flash-attack in an uninhibited nitric bath — etching and darkening instead of passivating. ASTM A967 addresses this with dichromate-inhibited nitric types and with citric alternatives; specify the method rather than just writing 'passivate'.

### Where in the routing should passivation go?

Last, after all machining, welding, forming, deburring and marking. Any operation using steel tooling or steel media reintroduces free iron. Medical device routings often place passivation after laser marking specifically because marking can disturb the passive layer.

## Alternative processes

- [Electropolishing](https://manufacturingprocesses.org/processes/finishing/electropolishing.md): Electropolishing reverses plating: current dissolves the surface fastest at the peaks, leaving a bright, deburred, passive finish.
- [Chemical Conversion Coating](https://manufacturingprocesses.org/processes/finishing/chemical-conversion-coating.md): Chemical conversion coating reacts the metal surface with a chemical bath to form a thin protective, conductive or paint-adhering film.
- [Black Oxide](https://manufacturingprocesses.org/processes/finishing/black-oxide.md): Black oxide converts the steel surface to magnetite in a hot alkaline bath, giving a black finish that adds essentially no thickness.
- [Electroplating](https://manufacturingprocesses.org/processes/finishing/electroplating.md): Electroplating deposits a metal coating from solution onto a conductive part by passing current through an electrolyte bath.
- [PVD Coating](https://manufacturingprocesses.org/processes/finishing/pvd-coating.md): PVD coating vaporises a solid target inside a vacuum chamber so it condenses on the part as a very thin, extremely hard film.

## Related processes

- [Electropolishing](https://manufacturingprocesses.org/processes/finishing/electropolishing.md): Electropolishing reverses plating: current dissolves the surface fastest at the peaks, leaving a bright, deburred, passive finish.
- [Abrasive Blasting](https://manufacturingprocesses.org/processes/finishing/abrasive-blasting.md): Abrasive blasting propels media at a surface with compressed air to clean it, strip scale, or produce a uniform matte texture.
- [Vibratory Tumbling and Mass Finishing](https://manufacturingprocesses.org/processes/finishing/vibratory-tumbling-and-mass-finishing.md): Vibratory tumbling agitates parts with abrasive media in a bowl so edges are broken and surfaces smoothed in bulk, without hand work.
- [Laser Marking and Engraving](https://manufacturingprocesses.org/processes/finishing/laser-marking-and-engraving.md): Laser marking alters the surface with a focused beam — annealing, etching or engraving it — to make permanent, contact-free marks.
- [Chemical Conversion Coating](https://manufacturingprocesses.org/processes/finishing/chemical-conversion-coating.md): Chemical conversion coating reacts the metal surface with a chemical bath to form a thin protective, conductive or paint-adhering film.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/finishing/passivation)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Photo Etching"
category: "Finishing"
subcategory: "Subtractive"
materials: ["Metal"]
tolerances: "±10–20% of material thickness is standard; approximately ±0.001 in (±0.025 mm) achievable on thin foil. Half-etch depth is controlled to a residual thickness rather than to a depth dimension"
volumes: "1 to several hundred thousand parts; the same tooling serves prototype and production"
lead_time: "3–10 business days including phototool; prototypes in 1–3 days at some suppliers"
url: https://manufacturingprocesses.org/processes/finishing/photo-etching
---

# Photo Etching

Photo etching masks a surface photographically and etches shallow detail into it for decoration, marking or texture.

- **Category**: [Finishing](https://manufacturingprocesses.org/processes/finishing.md)
- **Family**: Subtractive
- **Materials**: Metal
- **Typical tolerances**: ±10–20% of material thickness is standard; approximately ±0.001 in (±0.025 mm) achievable on thin foil. Half-etch depth is controlled to a residual thickness rather than to a depth dimension
- **Surface finish**: Etched surfaces come out uniformly matte; the incoming sheet finish is preserved on unetched areas
- **Typical volumes**: 1 to several hundred thousand parts; the same tooling serves prototype and production
- **Lead time**: 3–10 business days including phototool; prototypes in 1–3 days at some suppliers

## Overview

Photo etching masks a metal surface photographically and dissolves the unmasked areas with a chemical etchant. Because a photoresist can resolve fine detail and an etchant does not care how complicated the pattern is, artwork complexity costs nothing — a surface covered in thousands of individual features etches in the same time as a single one.

Used as a finishing process, it produces shallow decorative texture, logos, scales, grip patterns and permanent part marking in a controlled depth, typically half-etched to 30–50% of material thickness. Used as a cutting process, the same chemistry produces complete parts from foil 0.001–0.060 in (0.025–1.5 mm) thick. In both roles the defining advantages are that nothing touches the workpiece: no burrs, no work hardening, no heat-affected zone, no cutting forces and no distortion. Tooling is a photographic film rather than a hardened die, so lead time is days and cost is a small fraction of a stamping tool.

## How it works

1. **Make the phototool.** The artwork is imaged onto a film or glass phototool, in a matched pair for double-sided work. This is the only tooling in the process and it is inexpensive and fast to revise.

2. **Clean and laminate.** The sheet is chemically cleaned, then a photosensitive dry-film resist is laminated to one or both faces under heat and pressure. Cleanliness governs resist adhesion and therefore edge quality.

3. **Expose and develop.** UV light through the phototool polymerizes the resist in the areas that must survive. Developing washes away the unexposed resist and leaves bare metal exactly where etching should occur.

4. **Etch.** The sheet passes through a spray etcher. Ferric chloride at roughly 120–130 °F (49–54 °C) is the standard etchant for stainless, carbon steel, copper alloys and nickel; aluminum and titanium use different chemistries. Etch time controls depth. Because chemical attack is essentially isotropic, the etchant cuts sideways as fast as it cuts down, producing a characteristic undercut beneath the resist edge.

5. **Strip and inspect.** The remaining resist is chemically stripped, and the part is rinsed and dried.

### Etch factor and the cusped edge

Isotropic attack means that a through-etched feature is never a straight-sided cut. Etching from one side produces a tapered wall; etching from both sides produces an hourglass profile with a small ridge, or cusp, at mid-thickness. This is the geometric reason minimum feature size scales with material thickness — a rule of thumb is that the smallest reliable hole or slot is about 1.0–1.2 times the sheet thickness.

### Half etching

Stopping the etch part way through gives a controlled recess. Etching to 30–50% of thickness is used for surface decoration, permanent marking, bend lines that let a flat part fold accurately by hand, and countersinks and pockets that would otherwise require machining.

## Design guidelines

### Scale every feature to material thickness

Minimum hole diameter and slot width run about 1.0–1.2 times the material thickness, and minimum web or land width follows the same rule. Achievable tolerance is typically expressed as a fraction of thickness — on the order of ±10–20% of material thickness — with roughly ±0.001 in (±0.025 mm) achievable on thin foils. Thicker material means coarser everything.

### Expect a tapered or cusped edge

A one-sided etch leaves a taper; a two-sided etch leaves an hourglass with a small ridge at mid-thickness. Where a square edge is required — a bearing surface, a press fit, a precision aperture — photo etching is the wrong process or the feature needs a secondary operation.

### Use half etching deliberately

Etching to 30–50% of thickness gives decorative texture, permanent identification that cannot rub off, and fold lines that make an accurate hand-formed part from flat stock. Specify the residual thickness rather than the etch depth, because the process controls how much is removed and the remaining web is what matters structurally.

### Design for the sheet, not the part

Photo etching processes a whole panel at once, so cost scales with panel area and the number of panels, not with the number of features. Nest densely, put every variant on one panel, and add tabs to hold parts in the sheet if they must ship as a array.

### Choose an etchable material

Stainless steels, copper alloys, brass, beryllium copper, nickel alloys, molybdenum and spring steels all etch well and cover most applications. Aluminum and titanium require different etchants and are less commonly offered. Hardened, plated and heavily cold-worked surfaces etch unevenly.

### Take advantage of what it does not do

Because there is no tooling contact and no heat, the material comes out with its temper unchanged and no burr — which is why photo etching is chosen for spring elements, shims, EMI shielding, lead frames, encoder discs and fine mesh where a stamped or laser-cut edge would need deburring or would alter the material.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Material thickness | 0.001–0.060 in (0.025–1.5 mm) | Thicker material coarsens every feature | Etch time and undercut scale with depth |
| Minimum hole / slot | 1.2 × material thickness | 1.0 × thickness | Isotropic undercut limits aspect ratio |
| Tolerance | ±10–20% of material thickness | ±0.001 in (±0.025 mm) on thin foil | Etch rate variation across the panel |
| Half-etch depth | 30–50% of thickness | Specify residual web, not depth | Remaining material carries the load |
| Edge profile | Tapered or cusped | Not square | Chemical attack is isotropic |
| Tooling | Photographic film | Revised in hours | No hard tooling exists |

## Cost drivers

Photo etching is priced by panel area and the number of process passes, which produces a cost structure unlike any mechanical cutting process: complexity is free, and area is everything.

- **Panel area and nesting.** The cost of a panel is fixed, so parts per panel is the dominant variable. A dense nest can cut unit cost several-fold.
- **Material.** Thickness and alloy affect both raw material cost and etch time. Stainless etches slower than copper.
- **Number of etch stages.** A simple through-etch is one pass. A part with half-etched features on one side and through-etched features elsewhere needs additional resist and etch steps.
- **Tooling.** A phototool costs a small fraction of a stamping die and can be revised in hours, which is why photo etching dominates prototyping and design iteration for thin metal parts.
- **Secondary operations.** Forming, plating, tab removal and passivation are all separate operations after etching.

Cost-reduction tactics:

1. Nest as many parts per panel as the geometry allows — this is the single biggest lever.
2. Add features rather than removing them if it improves nesting; complexity genuinely costs nothing.
3. Use a single etch depth across the whole part where possible to avoid extra masking stages.
4. Specify the loosest tolerance the function permits; the tightest tolerances require slower, more controlled etching.
5. For high-volume simple parts, compare against stamping — above roughly a hundred thousand pieces the die cost amortizes and stamping wins on unit price, while photo etching wins everywhere below that and on any burr-free requirement.

## FAQ

### What tolerance can photo etching hold?

Typically ±10–20% of material thickness, with about ±0.001 in (±0.025 mm) achievable on thin foils. Minimum hole and slot sizes run roughly 1.0–1.2 times the material thickness, because chemical attack is isotropic and undercuts sideways as fast as it etches down.

### Why do photo-etched edges have a taper or a ridge?

Etching is isotropic — it removes material sideways at the same rate it removes it downward. A one-sided etch leaves a tapered wall; a two-sided etch leaves an hourglass profile with a small ridge at mid-thickness. Where a square edge is required, the feature needs a secondary operation.

### What is half etching used for?

Etching to 30–50% of material thickness produces decorative surface texture, permanent identification that cannot rub off, bend lines that let a flat part be folded accurately by hand, and shallow pockets or countersinks. Specify the residual web thickness rather than the etch depth, since the remaining material is what carries load.

### Does photo etching leave burrs or change the material?

No on both counts. Nothing touches the workpiece and there is no heat input, so the material comes out with no burr, no work hardening, no heat-affected zone and its temper unchanged. That is why the process is used for spring elements, shims, EMI shielding, encoder discs and fine mesh.

### Which materials can be photo etched?

Stainless steels, carbon and spring steels, copper alloys, brass, beryllium copper, nickel alloys and molybdenum all etch well in ferric chloride at roughly 120–130 °F (49–54 °C). Aluminum and titanium need different etchants and are less commonly offered. Hardened, plated or heavily cold-worked surfaces etch unevenly.

### How does photo etching compare to laser cutting or stamping on cost?

Photo etching prices by panel area rather than by cut length, so pattern complexity is free and dense nesting is the main cost lever. Tooling is a photographic film that costs a small fraction of a stamping die and can be revised in hours. Above roughly a hundred thousand simple parts, a stamping die amortizes and wins on unit price.

## Alternative processes

- [Photochemical Machining](https://manufacturingprocesses.org/processes/cutting/photochemical-machining.md): Photochemical machining masks sheet metal photographically and etches away the unmasked areas, cutting burr-free flat parts with no hard tooling.
- [Laser Cutting](https://manufacturingprocesses.org/processes/cutting/laser-cutting.md): Laser cutting melts, burns or vaporises a narrow kerf through flat stock with a focused beam, producing profiles from a 2D file with no tooling.
- [Punching and Blanking](https://manufacturingprocesses.org/processes/cutting/punching-and-blanking.md): Punching and blanking shear sheet metal between a punch and a die, producing holes and flat outlines at very high speed.
- [Laser Marking and Engraving](https://manufacturingprocesses.org/processes/finishing/laser-marking-and-engraving.md): Laser marking alters the surface with a focused beam — annealing, etching or engraving it — to make permanent, contact-free marks.
- [CNC Engraving](https://manufacturingprocesses.org/processes/finishing/cnc-engraving.md): CNC engraving cuts text and artwork into a surface with a small pointed or ball-nose tool following a programmed path.

## Related processes

- [Photochemical Machining](https://manufacturingprocesses.org/processes/cutting/photochemical-machining.md): Photochemical machining masks sheet metal photographically and etches away the unmasked areas, cutting burr-free flat parts with no hard tooling.
- [Laser Marking and Engraving](https://manufacturingprocesses.org/processes/finishing/laser-marking-and-engraving.md): Laser marking alters the surface with a focused beam — annealing, etching or engraving it — to make permanent, contact-free marks.
- [CNC Engraving](https://manufacturingprocesses.org/processes/finishing/cnc-engraving.md): CNC engraving cuts text and artwork into a surface with a small pointed or ball-nose tool following a programmed path.
- [Electropolishing](https://manufacturingprocesses.org/processes/finishing/electropolishing.md): Electropolishing reverses plating: current dissolves the surface fastest at the peaks, leaving a bright, deburred, passive finish.
- [Passivation](https://manufacturingprocesses.org/processes/finishing/passivation.md): Passivation removes free iron from a stainless steel surface with an acid bath so the chromium oxide layer can reform unbroken.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/finishing/photo-etching)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Powder Coating"
category: "Finishing"
subcategory: "Additive"
materials: ["Metal"]
tolerances: "Adds 2–4 mils (50–100 µm) per surface; budget 2× film thickness on a bore diameter and roughly 4× on thread pitch diameter, or mask"
volumes: "1 to millions of parts; conveyorized lines are most economical above a few hundred parts per run"
lead_time: "3–7 business days at a job shop; 24–48 hours expedited on stock colors; 2–3 weeks for a custom-matched powder"
url: https://manufacturingprocesses.org/processes/finishing/powder-coating
---

# Powder Coating

Powder coating sprays electrostatically charged dry polymer powder onto an earthed part, then bakes it into a tough continuous film.

- **Category**: [Finishing](https://manufacturingprocesses.org/processes/finishing.md)
- **Family**: Additive
- **Materials**: Metal
- **Typical tolerances**: Adds 2–4 mils (50–100 µm) per surface; budget 2× film thickness on a bore diameter and roughly 4× on thread pitch diameter, or mask
- **Surface finish**: Smooth gloss through heavy texture depending on powder; the film smooths but does not fill substrate defects
- **Typical volumes**: 1 to millions of parts; conveyorized lines are most economical above a few hundred parts per run
- **Lead time**: 3–7 business days at a job shop; 24–48 hours expedited on stock colors; 2–3 weeks for a custom-matched powder

## Overview

Powder coating applies a dry thermoset polymer powder to an electrically grounded part with an electrostatically charged spray gun, then fuses and cures it in an oven into a single continuous film. There is no solvent, so film builds that would sag in liquid paint are routine: a typical single coat runs 2–4 mils (50–100 µm) against roughly 1–2 mils for a sprayed liquid topcoat. Cure is typically 10–20 minutes at 350–400 °F (177–204 °C) part metal temperature.

The economics come from reclaim — overspray that misses the part is collected and re-sprayed, so material utilization can exceed 95% against 30–65% for conventional liquid spray, and VOC emissions are essentially zero. The constraints follow from the same physics: the part must conduct to hold a charge and must survive the bake, deep recesses starve from Faraday-cage shielding, and 2–4 mils lands on every reachable surface including threads and bores.

## How it works

1. **Clean and pretreat.** Degrease, then either chemically convert the surface or blast it. Iron phosphate is the common commercial pretreatment; zinc phosphate gives markedly better corrosion performance under the film. On structural steel the alternative is abrasive blasting to an SSPC preparation grade with a 1.5–3 mil (38–75 µm) anchor profile. Pretreatment, not the powder, determines how the finished part performs in salt spray.

2. **Dry off.** A dry-off oven drives water out of seams, weld joints and blind holes. Trapped moisture flashes during cure and blisters the film.

3. **Apply.** A corona gun charges powder to 30–100 kV as it leaves the nozzle; the charged particles follow field lines to the grounded part and cling electrostatically until cured. Tribo guns charge by particle-to-wall friction instead and penetrate recesses better because they produce no free-ion cloud. Fluidized-bed dipping — preheating the part and immersing it in aerated powder — is the alternative route for heavy builds of 10–40 mils (250–1000 µm), used mainly for electrical insulation.

4. **Cure.** The part passes through a convection or infrared oven. The powder melts, flows out, and the thermoset chemistry crosslinks. Schedules are quoted as part metal temperature, not air temperature: 10–20 minutes at 350–400 °F (177–204 °C) is typical, and heavy sections need extra oven time simply to reach temperature. Low-temperature powders cure near 250–300 °F (121–149 °C); UV-cure powders melt thermally and then crosslink under UV in seconds, which is what makes powder coating of MDF and some plastics possible.

5. **Cool and inspect.** Film thickness is verified with a magnetic or eddy-current gauge, adhesion by cross-hatch tape test to ASTM D3359. Reclaimed overspray from the booth cyclone is sieved and blended back into the feed.

### Which powder chemistry?

Epoxy is the toughest and most chemically resistant but chalks quickly in sunlight, so it is an interior and primer material. Polyester — TGIC and TGIC-free — is the standard exterior choice. Polyester-urethane gives a smoother, thinner-looking film for appearance parts. Superdurable polyester and fluoropolymer powders serve architectural work specified to AAMA 2604 and AAMA 2605.

## Design guidelines

### Budget 2–4 mils on every reachable surface

The film goes everywhere the powder can reach. On a 0.250 in bore, 3 mils per wall closes the diameter by about 0.006 in. On a 1/4-20 thread the pitch diameter grows by roughly 4× the radial build — 3 mils becomes about 0.012 in, which no thread class accepts. Mask threads with silicone plugs or caps, or chase them after cure.

### Break every sharp edge

Molten powder pulls away from a sharp edge by surface tension, so an unbroken edge ends up with a fraction of the film thickness of the adjacent flat and becomes the site where corrosion and chipping start. Break outside edges to at least a 0.020 in (0.5 mm) radius; 0.030–0.060 in (0.75–1.5 mm) is better on parts that must pass a long salt-spray requirement.

### Design around the Faraday cage

Electrostatic field lines terminate on the nearest grounded surface, so powder starves in inside corners, deep pockets, channel interiors and slots. As a working rule, a recess deeper than its opening is wide will coat thin. Open up internal corners, provide gun access, or accept a thinner film in the recess and state it on the drawing.

### Provide a hanging point, drainage and venting

The part hangs from a hook that must make bare metal-to-metal contact, so at least one location will carry a mark: add a dedicated hanging hole, typically 0.20–0.38 in (5–10 mm), in a non-cosmetic area. Cups, box sections and blind pockets trap pretreatment solution that boils out during cure and ruins the film, so add a drain hole at the low point in the hanging orientation and vent enclosed volumes.

### Check that everything in the assembly survives the oven

Cure at 350–400 °F (177–204 °C) rules out most thermoplastics, elastomeric seals, bearings, adhesives, electronics and some pressed-in inserts. Coat before assembly, or move to a low-temperature or UV-cure powder.

### Do not expect powder to hide the substrate

A 2–4 mil film smooths but does not fill. Weld spatter, deep tool marks, casting porosity and blast profile telegraph through. Textured and wrinkle powders exist specifically to disguise a poor substrate.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Single-coat film thickness | 2–4 mils (50–100 µm) | 1.5 mils (38 µm) minimum | Below ~1.5 mils coverage becomes discontinuous |
| Outside edge radius | 0.030 in (0.75 mm) | 0.020 in (0.5 mm) | Molten powder pulls back from sharp edges |
| Bore diameter allowance | 2 × film thickness | — | Film builds on both walls |
| Thread allowance | 4 × film thickness on pitch dia. | Mask instead | Thread geometry multiplies radial build |
| Recess depth vs. opening width | ≤ 1 : 1 | — | Faraday cage starves deeper recesses |
| Hanging hole | 0.20–0.38 in (5–10 mm), non-cosmetic face | — | Grounding contact always leaves a mark |
| Component temperature rating | Above cure temperature | 350–400 °F (177–204 °C) | Seals, bearings and plastics fail in the oven |

## Variants

- Electrostatic Spraying
- Fluidized Bed Powder Coating

## Cost drivers

Powder coating is priced per part or per rack over a lot minimum, and the powder itself is rarely the dominant cost.

- **Racking and line density.** Cost tracks how many parts fit on a linear foot of conveyor. A part that hangs flat and nests closely runs at a fraction of the cost of a bulky weldment.
- **Pretreatment level.** A five-stage zinc phosphate washer or an abrasive blast to a specified SSPC grade is a separate operation with its own labor and consumables — and it is the single biggest determinant of corrosion performance.
- **Color changes.** Reclaim booths must be blown down and cleaned between colors, so a short run in a non-stock color pays for that cleanout.
- **Masking.** Silicone plugs, caps and high-temperature tape are applied and removed by hand, twice per part.
- **Rework rate.** Powder cannot be spot-repaired the way liquid paint can. A defect usually means chemical or burn-off stripping and a full recoat.

Cost-reduction tactics:

1. Consolidate to a single stock color — black and white are the cheapest almost everywhere.
2. Design in a hanging hole and orient the part so it drains, which eliminates rework for runs and trapped solution.
3. Replace masked threads with post-cure thread chasing where the thread is accessible.
4. Break edges during machining or deburring rather than paying for hand work at the coater.
5. Release in full-line-load batches — the oven and booth cost the same whether the conveyor is full or half empty.

## FAQ

### How thick is powder coating?

A typical single coat is 2–4 mils (50–100 µm), roughly double a sprayed liquid topcoat. Below about 1.5 mils (38 µm) coverage becomes patchy. Fluidized-bed dipping is used when a much heavier build of 10–40 mils (250–1000 µm) is needed, mainly for electrical insulation.

### What temperature does powder coating cure at?

Most thermoset powders cure at 350–400 °F (177–204 °C) part metal temperature for 10–20 minutes. Low-temperature powders cure near 250–300 °F (121–149 °C), and UV-cure powders melt thermally then crosslink under UV light in seconds, which is what allows MDF and some plastics to be powder coated.

### Will powder coating close up threads and bores?

Yes. At 3 mils per surface a bore loses about 0.006 in on diameter, and a thread's pitch diameter grows by roughly 4× the radial build, or about 0.012 in — enough to bind any standard thread class. Mask threads with silicone plugs and caps, or chase them after cure.

### Can plastic be powder coated?

Only with a low-temperature or UV-cure powder, and only if the part can be made conductive — typically with a conductive primer, because electrostatic application needs a ground path. Standard thermoset powders cure at 350–400 °F (177–204 °C), which destroys most thermoplastics.

### Why does powder coating come out thin on sharp edges?

Molten powder pulls away from a sharp edge by surface tension during flow-out, leaving a fraction of the film thickness found on the adjacent flat. That thin spot is where chipping and corrosion begin. Break outside edges to at least a 0.020 in (0.5 mm) radius, and 0.030 in or more for demanding corrosion requirements.

### Is powder coating more durable than liquid paint?

For impact and abrasion resistance, generally yes — the film is thicker and fully crosslinked. But corrosion performance is set mostly by pretreatment: zinc phosphate under the powder outperforms iron phosphate by a wide margin. Liquid paint still wins where a Class A automotive appearance, field touch-up or a heat-sensitive substrate is required.

### Which powder chemistry should I specify?

Epoxy for interior and primer use — tough and chemically resistant, but it chalks in sunlight. Polyester for general exterior work. Polyester-urethane for appearance parts. Superdurable polyester or fluoropolymer for architectural aluminum specified to AAMA 2604 or 2605.

## Alternative processes

- [Spray Painting](https://manufacturingprocesses.org/processes/finishing/spray-painting.md): Spray painting atomises liquid coating onto a prepared surface, building a film that cures into a decorative and protective layer.
- [E-Coating (Electrophoretic Deposition)](https://manufacturingprocesses.org/processes/finishing/e-coating-electrophoretic-deposition.md): E-coating deposits paint electrically from a bath so it plates evenly into every recess, then cures to a uniform film.
- [Anodizing](https://manufacturingprocesses.org/processes/finishing/anodizing.md): Anodizing grows a hard porous oxide layer on aluminum electrochemically, which can then be dyed and sealed for wear and corrosion resistance.
- [Cerakote Ceramic Coating](https://manufacturingprocesses.org/processes/finishing/cerakote-ceramic-coating.md): Cerakote sprays a thin ceramic-filled polymer coating onto a prepared part and cures it, giving wear and corrosion resistance in a very thin film.
- [Electroplating](https://manufacturingprocesses.org/processes/finishing/electroplating.md): Electroplating deposits a metal coating from solution onto a conductive part by passing current through an electrolyte bath.

## Related processes

- [E-Coating (Electrophoretic Deposition)](https://manufacturingprocesses.org/processes/finishing/e-coating-electrophoretic-deposition.md): E-coating deposits paint electrically from a bath so it plates evenly into every recess, then cures to a uniform film.
- [Spray Painting](https://manufacturingprocesses.org/processes/finishing/spray-painting.md): Spray painting atomises liquid coating onto a prepared surface, building a film that cures into a decorative and protective layer.
- [Abrasive Blasting](https://manufacturingprocesses.org/processes/finishing/abrasive-blasting.md): Abrasive blasting propels media at a surface with compressed air to clean it, strip scale, or produce a uniform matte texture.
- [Chemical Conversion Coating](https://manufacturingprocesses.org/processes/finishing/chemical-conversion-coating.md): Chemical conversion coating reacts the metal surface with a chemical bath to form a thin protective, conductive or paint-adhering film.
- [Galvanizing](https://manufacturingprocesses.org/processes/finishing/galvanizing.md): Galvanizing coats steel with zinc, usually by hot dipping, so the zinc corrodes sacrificially and protects the steel beneath it.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/finishing/powder-coating)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Screen Printing"
category: "Finishing"
subcategory: "Printing"
materials: ["Plastic", "Metal", "Glass", "Wood", "Ceramic"]
tolerances: "Registration typically ±0.005–0.010 in (±0.13–0.25 mm); dried ink film 0.0004–0.002 in (10–50 µm), which is enough to matter on a stacked assembly"
volumes: "A hundred to millions of impressions; setup per color makes very short runs uneconomical"
lead_time: "1–2 weeks including screens and setup; 3–7 business days for repeat orders"
url: https://manufacturingprocesses.org/processes/finishing/screen-printing
---

# Screen Printing

Screen printing pushes ink through a stencilled mesh onto the surface below, laying down one opaque color per pass.

- **Category**: [Finishing](https://manufacturingprocesses.org/processes/finishing.md)
- **Family**: Printing
- **Materials**: Plastic, Metal, Glass, Wood, Ceramic
- **Typical tolerances**: Registration typically ±0.005–0.010 in (±0.13–0.25 mm); dried ink film 0.0004–0.002 in (10–50 µm), which is enough to matter on a stacked assembly
- **Surface finish**: Matte through gloss depending on ink; the heavy film gives a slightly raised, tactile edge on solid areas
- **Typical volumes**: A hundred to millions of impressions; setup per color makes very short runs uneconomical
- **Lead time**: 1–2 weeks including screens and setup; 3–7 business days for repeat orders

## Overview

Screen printing forces ink through the open areas of a stencilled mesh onto the surface below with a squeegee. It lays down the heaviest ink film of any common printing process — typically 0.0004–0.002 in (10–50 µm) dry, against a few microns for pad printing or inkjet — which is why it is the default when a graphic must be fully opaque over a dark or colored substrate, or when the deposited layer has to do a job beyond looking right.

That heavy deposit is also what makes screen printing a functional process, not just a decorative one: conductive silver inks for membrane switches, dielectric layers, resistive elements and solder paste stencils all rely on controlled film thickness. Mesh count governs the trade-off, from around 110 threads per inch for heavy opaque deposits down to 305 and finer for detail work, with minimum line widths of roughly 0.005–0.010 in (0.13–0.25 mm) and registration typically ±0.005–0.010 in (±0.13–0.25 mm). One screen prints one color per pass.

## How it works

1. **Make the screen.** Mesh is tensioned on a frame, coated with photosensitive emulsion, exposed through a film positive and washed out. The open areas of the emulsion are where ink will pass. Mesh count, thread diameter and emulsion thickness together determine deposited film thickness.

2. **Set off-contact.** The screen is held a small distance above the substrate. The squeegee presses it into contact only along the line of the stroke, so the mesh snaps back immediately behind the blade. Without off-contact, the screen sticks to the wet ink and smears the image.

3. **Flood and print.** A flood bar fills the open mesh with ink, then the squeegee sweeps across at a controlled angle, pressure and speed, shearing the ink through the mesh onto the substrate. Squeegee durometer, angle and speed are the primary process variables.

4. **Cure.** Solvent inks air dry or pass through a force-dry tunnel at roughly 150–200 °F (65–95 °C). UV inks cure in seconds under a lamp and are standard for high-speed lines. Two-component epoxies are used on glass and metal for adhesion and chemical resistance. Ceramic frit inks on glass are fired at high temperature and become part of the glass, giving a permanent, dishwasher- and abrasion-proof mark.

5. **Repeat per color.** Each additional color is another screen, another pass and another registration operation.

### Flat, cylindrical and second-surface

Flat-bed printing handles panels, overlays and nameplates. Cylindrical printing rotates bottles and tubes against a fixed squeegee. Second-surface printing puts the ink on the back of a transparent polycarbonate or polyester overlay so the substrate itself protects the graphic — the standard construction for membrane switches, appliance overlays and instrument panels, where the printed layer never wears because it is never touched.

## Design guidelines

### Choose mesh count for the job, not for the detail

Low mesh counts around 110 lay down a heavy, opaque, tactile film and are used for solid colors, white over dark, and functional inks. High mesh counts of 305 and above give fine detail and a thin film. You cannot have maximum opacity and finest detail in the same pass — split them across two screens if both are required.

### Design to a 0.005–0.010 in minimum line

Lines and gaps below roughly 0.005 in (0.13 mm) will not resolve reliably, and 0.010 in (0.25 mm) is a safer production target. Reversed-out text needs to be larger than positive text at the same weight, because ink spread closes small counters.

### Build in registration tolerance

Each color is a separate pass with registration typically ±0.005–0.010 in (±0.13–0.25 mm). Adjacent colors need trapping or a keyline; abutting colors with no overlap will show white gaps or overlaps depending on which way the pass drifts.

### Keep the substrate flat or singly curved

The squeegee must maintain contact along a line. Flat and cylindrical surfaces work; compound curves, deep recesses and stepped surfaces do not. Where the surface is complex, pad printing is the appropriate process.

### Confirm surface energy and adhesion

As with any ink process, a substrate below roughly 38 dyne/cm — polypropylene, polyethylene, acetal, fluoropolymers — requires flame, corona or plasma treatment before printing. Specify an adhesion test to ASTM D3359 and any chemical, abrasion or UV exposure requirement.

### Use second-surface printing when the graphic must not wear

Printing on the reverse of a clear polycarbonate or polyester overlay puts the substrate between the user and the ink. This is why membrane switch legends survive millions of actuations while a first-surface print on the same product would not.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Dried ink film | 0.0004–0.002 in (10–50 µm) | — | Far heavier than pad printing or inkjet |
| Mesh count | 110–160 for opacity; 230–305 for detail | — | Mesh controls both film build and resolution |
| Minimum line / gap | 0.010 in (0.25 mm) | 0.005 in (0.13 mm) | Finer detail does not resolve reliably |
| Registration | ±0.005–0.010 in (±0.13–0.25 mm) | Design in trapping | Each color is a separate pass |
| Substrate geometry | Flat or singly curved | Compound curves need pad printing | Squeegee contacts along a line |
| Substrate surface energy | ≥ 38 dyne/cm | Pretreat low-energy plastics | Ink cannot wet the surface otherwise |

## Cost drivers

Screen printing has low tooling cost per color and a very low per-piece rate once running, which makes it the economical choice for medium and long runs of flat graphics.

- **Color count.** One screen, one setup and one pass per color. This dominates both tooling and run cost.
- **Screen making and registration setup.** Each color needs a film positive, a coated and exposed screen, and registration. Short runs are dominated by this setup.
- **Ink system.** UV inks cost more per unit but cure instantly and allow much higher line speeds. Two-component and ceramic frit systems carry handling, pot-life and firing costs.
- **Substrate preparation.** Flame or corona treatment on low-energy plastics is an added in-line operation.
- **Part handling.** Flat sheets feed fast; three-dimensional parts need nests and often manual loading.

Cost-reduction tactics:

1. Reduce color count — a two-color design almost always costs less than half of a four-color one.
2. Gang multiple graphics onto one sheet and die cut afterward instead of printing parts individually.
3. Use second-surface printing on a clear overlay rather than a first-surface print plus a protective clear coat.
4. Standardize on one mesh count and one ink family across a product family so screens and setups are reused.
5. For short runs, variable data or photographic imagery, compare against UV inkjet, which has no screens and no per-color setup at all.

## FAQ

### How thick is screen-printed ink?

Typically 0.0004–0.002 in (10–50 µm) dry, far heavier than pad printing or inkjet. That heavy deposit is why screen printing is chosen when a graphic must be fully opaque over a dark substrate, and why it is used for functional layers such as conductive silver inks, dielectrics and solder paste.

### What mesh count should I specify?

Around 110–160 threads per inch for heavy, opaque deposits and functional inks; 230–305 and above for fine detail with a thin film. Mesh count sets both film thickness and resolution, so maximum opacity and finest detail cannot be achieved in the same pass — split them across two screens if both are required.

### What is the minimum line width for screen printing?

Roughly 0.005 in (0.13 mm) is the resolution limit and 0.010 in (0.25 mm) is a safer production target. Reversed-out text must be larger than positive text at the same weight, because ink spread closes small counters in the characters.

### What is second-surface printing?

Printing the graphic on the reverse of a transparent polycarbonate or polyester overlay so the substrate protects the ink. It is the standard construction for membrane switches, appliance overlays and instrument panels, where legends survive millions of actuations because the ink is never touched.

### Can screen printing be used on curved parts?

On singly curved surfaces such as bottles and tubes, yes — the part rotates against a fixed squeegee. Compound curves, deep recesses and stepped surfaces do not work because the squeegee needs to maintain contact along a line. Pad printing is the correct process for those geometries.

### Screen printing or UV inkjet?

Screen printing for heavy opaque film, functional inks and long runs of the same artwork, where the per-piece rate is very low once the screens are made. UV inkjet for short runs, variable data, photographic images and gradients, since it has no screens and no per-color setup at all.

## Alternative processes

- [Pad Printing](https://manufacturingprocesses.org/processes/finishing/pad-printing.md): Pad printing lifts ink out of an etched plate with a silicone pad and transfers it onto a curved or recessed surface.
- [UV Inkjet Printing](https://manufacturingprocesses.org/processes/finishing/uv-inkjet-printing.md): UV inkjet printing jets UV-curable ink directly onto a flat substrate and cures it instantly under UV light.
- [Hydro Transfer Printing](https://manufacturingprocesses.org/processes/finishing/hydro-transfer-printing.md): Hydro transfer printing floats a printed film on water, activates it, and wraps it around a submerged part to cover it in a pattern.
- [Laser Marking and Engraving](https://manufacturingprocesses.org/processes/finishing/laser-marking-and-engraving.md): Laser marking alters the surface with a focused beam — annealing, etching or engraving it — to make permanent, contact-free marks.
- [Photo Etching](https://manufacturingprocesses.org/processes/finishing/photo-etching.md): Photo etching masks a surface photographically and etches shallow detail into it for decoration, marking or texture.

## Related processes

- [Pad Printing](https://manufacturingprocesses.org/processes/finishing/pad-printing.md): Pad printing lifts ink out of an etched plate with a silicone pad and transfers it onto a curved or recessed surface.
- [UV Inkjet Printing](https://manufacturingprocesses.org/processes/finishing/uv-inkjet-printing.md): UV inkjet printing jets UV-curable ink directly onto a flat substrate and cures it instantly under UV light.
- [PCB Assembly (SMT Reflow)](https://manufacturingprocesses.org/processes/joining/pcb-assembly-smt-reflow.md): PCB assembly prints solder paste, places components and reflows the board through a heated oven so every joint forms in one pass.
- [Foil Blocking and Embossing](https://manufacturingprocesses.org/processes/finishing/foil-blocking-and-embossing.md): Foil blocking and embossing press a heated die into the surface to transfer metallic foil, raise a relief, or do both at once.
- [Laser Marking and Engraving](https://manufacturingprocesses.org/processes/finishing/laser-marking-and-engraving.md): Laser marking alters the surface with a focused beam — annealing, etching or engraving it — to make permanent, contact-free marks.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/finishing/screen-printing)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Spray Painting"
category: "Finishing"
subcategory: "Additive"
materials: ["Metal", "Plastic", "Wood", "Composite"]
tolerances: "A complete system adds 4–6 mils (100–150 µm) per coated surface; budget 2× on a bore diameter and roughly 4× on thread pitch diameter, or mask"
volumes: "1 part to continuous automotive lines; the most viable option at volume one"
lead_time: "2–10 business days at a job shop depending on the number of coats and cure schedule; field work is scheduled by weather window"
url: https://manufacturingprocesses.org/processes/finishing/spray-painting
---

# Spray Painting

Spray painting atomises liquid coating onto a prepared surface, building a film that cures into a decorative and protective layer.

- **Category**: [Finishing](https://manufacturingprocesses.org/processes/finishing.md)
- **Family**: Additive
- **Materials**: Metal, Plastic, Wood, Composite
- **Typical tolerances**: A complete system adds 4–6 mils (100–150 µm) per coated surface; budget 2× on a bore diameter and roughly 4× on thread pitch diameter, or mask
- **Surface finish**: Flat through full gloss; the film smooths but does not fill weld spatter, porosity or deep sanding scratches
- **Typical volumes**: 1 part to continuous automotive lines; the most viable option at volume one
- **Lead time**: 2–10 business days at a job shop depending on the number of coats and cure schedule; field work is scheduled by weather window

## Overview

Spray painting atomizes a liquid coating and delivers it to a prepared surface, where it flows out and cures into a decorative and protective film. It remains the most flexible finishing process available: it accepts any substrate that tolerates the cure, produces the full color and gloss range including automotive Class A finishes, can be applied in the field, and can be spot-repaired — none of which powder coating or e-coating can do.

Typical film builds are 1–2 mils (25–50 µm) for a primer, 0.5–1 mil (13–25 µm) for a basecoat, and 1.5–2.5 mils (38–63 µm) for a clearcoat, giving a complete system of roughly 4–6 mils (100–150 µm). Delivery method sets transfer efficiency and therefore both cost and emissions: conventional air spray transfers roughly 30–40% of what leaves the gun, HVLP (defined by an atomizing air pressure of 10 psi or less at the cap) reaches around 65%, and electrostatic bell application goes considerably higher.

## How it works

1. **Clean and abrade.** Solvent wipe to remove oil and release agents, then abrade — typically P320–P400 for a primer key. Any silicone contamination produces fisheyes that cannot be sanded out of a single coat.

2. **Pretreat.** Bare steel needs a phosphate conversion coating or an etch primer; bare aluminum needs a chromate or non-chrome conversion coating or a self-etch primer. As with every coating system, this stage sets corrosion performance far more than the topcoat does.

3. **Prime and block.** A build primer or primer-surfacer fills sanding scratches and minor substrate defects, then is block sanded flat. This step is what separates a Class A finish from an ordinary one.

4. **Spray the topcoat.** Gun setup — fluid tip typically 1.3–1.4 mm for basecoat and 1.4–1.8 mm for primer — plus 6–8 in (150–200 mm) gun distance and 50% pass overlap produce a uniform film. Dry film thickness follows from wet film thickness multiplied by the coating's volume solids, which is the arithmetic used to plan how many coats are needed.

5. **Flash and cure.** Solvent flashes between coats, then the film cures — air dry, force dry at roughly 140–180 °F (60–82 °C), or a full bake. Two-component polyurethanes and epoxies crosslink chemically and reach full hardness over days.

6. **Polish if required.** Class A work is denibbed, compounded and polished after cure to remove dust inclusions and orange peel.

### Delivery methods

**Conventional air spray** atomizes finely but wastes most of the material. **HVLP and LVLP** trade atomization quality for transfer efficiency and are the regulatory default. **Airless** pumps fluid at 1,000–3,000 psi through a small orifice with no atomizing air, delivering high film build fast on large structures. **Air-assisted airless** adds a small amount of shaping air for a better finish. **Electrostatic** charges the atomized droplets so they wrap around the part, giving the highest transfer efficiency on racked metal parts.

### Environmental window

Substrate temperature should be at least 5 °F (3 °C) above the dew point, and most systems specify 65–80 °F (18–27 °C) with relative humidity below roughly 70–80%. Spraying outside that window causes blushing, solvent popping and adhesion failures.

## Design guidelines

### Break sharp edges

Liquid paint, like powder, pulls away from a sharp edge as it flows out and cures. An unbroken edge carries a fraction of the film thickness of the adjacent flat and is where corrosion and chipping start. Break outside edges to at least 0.020 in (0.5 mm), and more on parts that must survive a long salt-spray requirement.

### Design for gun access

Spray is line-of-sight. Deep pockets, channel interiors, blind recesses and the back sides of closely spaced ribs receive little coating. Where the interior of a hollow section must be protected, use e-coating, a cavity wax, or design the part to be coated before assembly.

### Provide drainage, venting and a hanging point

Cups and pockets hold solvent and washer solution, which causes runs and solvent popping. Add drain holes at the low point in the hang orientation, vent enclosed volumes, and provide a hanging feature so the part can be racked without the operator improvising one.

### Mask everything with a fit or a ground path

At 4–6 mils for a full system, threads, bores, bearing seats, sealing faces and grounding pads must be masked or machined afterward. Thread pitch diameter grows by roughly four times the radial film thickness.

### Specify the system, not the color

A drawing note that gives only a color leaves the corrosion performance undefined. State the pretreatment, primer, topcoat chemistry, dry film thickness range and the acceptance tests — adhesion by ASTM D3359 cross-hatch, thickness by ASTM D7091, and a salt spray requirement if one applies.

### Set expectations for substrate telegraphing

Paint at 4–6 mils smooths but does not fill. Weld spatter, porosity, deep sanding scratches and blast profile all read through a gloss finish. Class A appearance requires block-sanded primer and a substrate prepared to match; see the [surface finish chart](/charts/surface-finish-chart) for how prep methods compare.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Primer film thickness | 1–2 mils (25–50 µm) | — | Fills sanding scratches and provides adhesion |
| Topcoat film thickness | 1.5–2.5 mils (38–63 µm) | Excess film runs and sags | Liquid coatings have no self-limiting mechanism |
| Complete system | 4–6 mils (100–150 µm) | — | Typical primer plus base plus clear |
| Outside edge break | 0.020 in (0.5 mm) minimum | — | Coating thins at sharp edges |
| Thread allowance | ~4 × film on pitch diameter | Mask instead | Thread geometry multiplies radial build |
| Substrate vs. dew point | At least 5 °F (3 °C) above | — | Condensation causes blushing and adhesion failure |

## Cost drivers

Spray painting is priced by labor and booth time. Material is a minor share, especially at low transfer efficiency where most of what is sprayed never reaches the part.

- **Surface preparation.** Sanding, filling, priming and block sanding dominate the cost of any appearance finish. The paint itself is the cheapest part of a Class A job.
- **Number of coats and colors.** Every additional coat is another booth cycle plus flash and cure time. Two-tone work adds masking between coats.
- **Masking.** Hand-applied, hand-removed, per part.
- **Transfer efficiency.** Conventional air spray wastes 60–70% of the material and generates the corresponding VOC and waste-disposal burden. Electrostatic and HVLP reduce both.
- **Rework and inspection.** Dust inclusions, runs and orange peel on a gloss finish are visible at a glance, so appearance work carries a high denib-and-polish and rework cost.

Cost-reduction tactics:

1. Specify the gloss and appearance grade the part actually needs — a satin finish hides substrate defects that a high gloss will magnify and hides the cost of correcting them.
2. Use HVLP or electrostatic application to cut material consumption and emissions.
3. Design in a hanging point and drainage so parts rack quickly and do not need rework for runs.
4. Reduce masked features by moving fitted surfaces onto separate components.
5. At volume on metal parts with no heat-sensitive components, price powder coating instead — it is usually cheaper per part, needs no solvent, and is tougher, at the cost of appearance flexibility and repairability.

## FAQ

### How thick is a sprayed paint system?

A typical complete system is 4–6 mils (100–150 µm): roughly 1–2 mils (25–50 µm) of primer, 0.5–1 mil (13–25 µm) of basecoat and 1.5–2.5 mils (38–63 µm) of clearcoat. Dry film thickness equals wet film thickness multiplied by the coating's volume solids, which is how coat counts are planned.

### What is HVLP and why does transfer efficiency matter?

HVLP means high volume, low pressure — defined by an atomizing air pressure of 10 psi or less at the air cap. It transfers roughly 65% of the material to the part, against about 30–40% for conventional air spray. Higher transfer efficiency cuts material cost, VOC emissions and waste disposal, which is why it is the regulatory default.

### Spray painting or powder coating?

Spray painting for heat-sensitive substrates, Class A automotive appearance, field application, spot repair and one-off work. Powder coating for a thicker, tougher film at lower cost per part in volume, with no solvent — but it requires a conductive part that survives a 350–400 °F (177–204 °C) bake and cannot be spot repaired.

### What conditions does spray painting require?

Substrate temperature at least 5 °F (3 °C) above the dew point, with most systems specifying 65–80 °F (18–27 °C) and relative humidity below roughly 70–80%. Spraying outside that window causes blushing, solvent popping and adhesion failures that cannot be corrected without stripping.

### Why does paint fail first at edges and corners?

Liquid coating pulls away from a sharp edge by surface tension as it flows out, so the edge ends up with a fraction of the film thickness on the adjacent flat. Break outside edges to at least 0.020 in (0.5 mm), and more where a long salt-spray life is required.

### How do I specify a paint finish on a drawing?

Name the pretreatment, the primer, the topcoat chemistry, the dry film thickness range and the acceptance tests — adhesion by ASTM D3359, thickness by ASTM D7091, and a salt-spray requirement if one applies. A color alone leaves corrosion performance completely undefined.

## Alternative processes

- [Powder Coating](https://manufacturingprocesses.org/processes/finishing/powder-coating.md): Powder coating sprays electrostatically charged dry polymer powder onto an earthed part, then bakes it into a tough continuous film.
- [E-Coating (Electrophoretic Deposition)](https://manufacturingprocesses.org/processes/finishing/e-coating-electrophoretic-deposition.md): E-coating deposits paint electrically from a bath so it plates evenly into every recess, then cures to a uniform film.
- [Cerakote Ceramic Coating](https://manufacturingprocesses.org/processes/finishing/cerakote-ceramic-coating.md): Cerakote sprays a thin ceramic-filled polymer coating onto a prepared part and cures it, giving wear and corrosion resistance in a very thin film.
- [Hydro Transfer Printing](https://manufacturingprocesses.org/processes/finishing/hydro-transfer-printing.md): Hydro transfer printing floats a printed film on water, activates it, and wraps it around a submerged part to cover it in a pattern.
- [Anodizing](https://manufacturingprocesses.org/processes/finishing/anodizing.md): Anodizing grows a hard porous oxide layer on aluminum electrochemically, which can then be dyed and sealed for wear and corrosion resistance.

## Related processes

- [Powder Coating](https://manufacturingprocesses.org/processes/finishing/powder-coating.md): Powder coating sprays electrostatically charged dry polymer powder onto an earthed part, then bakes it into a tough continuous film.
- [E-Coating (Electrophoretic Deposition)](https://manufacturingprocesses.org/processes/finishing/e-coating-electrophoretic-deposition.md): E-coating deposits paint electrically from a bath so it plates evenly into every recess, then cures to a uniform film.
- [Abrasive Blasting](https://manufacturingprocesses.org/processes/finishing/abrasive-blasting.md): Abrasive blasting propels media at a surface with compressed air to clean it, strip scale, or produce a uniform matte texture.
- [Chemical Conversion Coating](https://manufacturingprocesses.org/processes/finishing/chemical-conversion-coating.md): Chemical conversion coating reacts the metal surface with a chemical bath to form a thin protective, conductive or paint-adhering film.
- [Cerakote Ceramic Coating](https://manufacturingprocesses.org/processes/finishing/cerakote-ceramic-coating.md): Cerakote sprays a thin ceramic-filled polymer coating onto a prepared part and cures it, giving wear and corrosion resistance in a very thin film.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/finishing/spray-painting)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "UV Inkjet Printing"
category: "Finishing"
subcategory: "Printing"
materials: ["Plastic", "Metal", "Wood", "Glass"]
tolerances: "Image placement typically within ±0.010–0.020 in (±0.25–0.5 mm) relative to the substrate datum; printed film 0.0004–0.0008 in (10–20 µm) for a standard CMYK build"
volumes: "1 to a few thousand impressions; variable data and one-offs cost the same as repeats"
lead_time: "Same day to 5 business days; no tooling, so there is no plate or screen lead time"
url: https://manufacturingprocesses.org/processes/finishing/uv-inkjet-printing
---

# UV Inkjet Printing

UV inkjet printing jets UV-curable ink directly onto a flat substrate and cures it instantly under UV light.

- **Category**: [Finishing](https://manufacturingprocesses.org/processes/finishing.md)
- **Family**: Printing
- **Materials**: Plastic, Metal, Wood, Glass
- **Typical tolerances**: Image placement typically within ±0.010–0.020 in (±0.25–0.5 mm) relative to the substrate datum; printed film 0.0004–0.0008 in (10–20 µm) for a standard CMYK build
- **Surface finish**: Matte through gloss depending on the varnish layer; textured and tactile builds up to about 0.02 in (0.5 mm)
- **Typical volumes**: 1 to a few thousand impressions; variable data and one-offs cost the same as repeats
- **Lead time**: Same day to 5 business days; no tooling, so there is no plate or screen lead time

## Overview

UV inkjet printing jets picoliter droplets of UV-curable ink directly onto a substrate and cures them instantly with UV light as the carriage passes. Because the ink cures on contact rather than drying by absorption or evaporation, it prints onto non-porous materials — acrylic, polycarbonate, aluminum, glass, ceramic, wood, foam board and composites — with no drying time and no ink migration.

The process is fully digital: there are no screens, no clichés, no plates and no per-color setup, so a one-off print and a photographic image with a thousand colors cost the same as a single-color logo. Flatbed machines commonly handle sheets up to 4 × 8 ft (1.2 × 2.4 m) with substrate thickness capacity of a few inches. Typical output resolution runs 600–1440 dpi from drop volumes of a few picoliters upward. A standard CMYK pass leaves a film on the order of 0.0004–0.0008 in (10–20 µm), and multi-layer printing can build tactile relief up to roughly 0.02 in (0.5 mm) for braille and textured effects.

## How it works

1. **Prepare and locate the substrate.** The sheet is placed on a vacuum bed and registered against pins or a corner datum. Flatness matters: the printhead runs 1–2 mm above the surface, so a warped or uneven substrate risks a head strike, which is the most expensive failure mode on these machines.

2. **Prime if required.** Glass, some metals and low-energy plastics need an adhesion promoter or primer. UV ink cures into a hard film very quickly, which gives it little time to wet a difficult surface.

3. **Print white if required.** On a non-white or transparent substrate, a white underbase is printed first so the color layer reads correctly. Many machines print white, CMYK and clear varnish in a single pass or in sequential passes.

4. **Jet and cure.** Piezoelectric heads fire drops on demand; LED-UV lamps mounted alongside the carriage cure each swath as it is laid down. Because curing is essentially instantaneous, drops do not spread or absorb, so dot gain is small and the image sits on top of the surface.

5. **Add clear and texture layers.** A clear varnish gives gloss, spot gloss or a matte finish. Repeated passes of clear build physical relief for tactile graphics, braille and simulated textures.

### Where it fits against the other printing processes

UV inkjet replaces screen printing when the run is short, when the artwork changes per piece, or when the image is photographic. It does not replace screen printing when the requirement is a heavy, opaque, functional ink film — a screen lays down 10–50 µm in a single pass against roughly 10–20 µm for a full CMYK inkjet build. And it cannot print onto a curved or three-dimensional part, which is where pad printing and hydro transfer remain the only options.

## Design guidelines

### Design for a flat, rigid substrate

The head-to-substrate gap is only 1–2 mm. Substrates must be flat within roughly 0.06 in (1.5 mm) across the print area and rigid enough not to lift under vacuum. Warped panels, unsupported thin sheet and anything with raised features taller than the gap cannot be printed.

### Specify a white underbase on non-white substrates

UV inks are translucent. On clear acrylic, colored plastic, aluminum or wood, the color layer will shift toward the substrate color unless a white underbase is printed first. State whether white is single-hit or double-hit; opacity over a dark substrate usually needs two.

### Do not expect screen-printing opacity or film build

A full CMYK inkjet build is on the order of 0.0004–0.0008 in (10–20 µm), where a single screen pass lays 0.0004–0.002 in (10–50 µm). For deep saturated solids, functional conductive layers or heavy tactile film, screen printing is the correct process.

### Plan adhesion, not just appearance

Instant cure means the ink has very little time to wet the surface. Glass, bare metals, powder-coated surfaces and low-energy plastics generally need an adhesion promoter, a primer or a surface treatment. Specify an adhesion test to ASTM D3359 and, for outdoor work, the expected service life.

### Set realistic outdoor expectations

Uncoated UV inkjet graphics are commonly rated by manufacturers for roughly 1–2 years outdoors before noticeable fade, extended substantially by a UV-blocking laminate or a clear overprint. Confirm the rating for the specific ink set rather than assuming.

### Use the digital nature of the process

Serial numbers, QR codes, personalization and short-run variants cost nothing extra. Design product families so the only difference between variants is printed content, and the tooling cost of a variant drops to zero.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Substrate flatness | Within 0.06 in (1.5 mm) | Head gap is 1–2 mm | A head strike is the main failure mode |
| CMYK film build | 0.0004–0.0008 in (10–20 µm) | Use screen printing for heavier film | Inkjet drops are small by design |
| Tactile / braille build | Up to ~0.02 in (0.5 mm) | Many passes; slow | Relief is built layer by layer |
| Non-white substrate | Print white underbase first | Double-hit white over dark | UV inks are translucent |
| Resolution | 600–1440 dpi | — | Set by head and drop volume |
| Outdoor life uncoated | Commonly rated 1–2 years | Laminate or clear coat to extend | UV exposure fades unprotected ink |

## Cost drivers

UV inkjet has effectively zero tooling cost and a per-piece cost that is dominated by machine time and ink coverage. That inverts the usual economics: it is the cheapest process at quantity one and among the most expensive per piece at high volume.

- **Print area and coverage.** Cost tracks square footage and how much ink lands on it. A full-bleed photographic image costs several times a line-art logo of the same size.
- **Number of ink layers.** White underbase, CMYK, and clear varnish are separate passes. A print on clear acrylic with white behind and gloss on top is three times the machine time of CMYK alone.
- **Machine time per sheet.** Higher resolutions and multi-pass modes trade speed for quality directly.
- **Substrate handling and priming.** Priming, cleaning and masking are manual operations that often exceed the print time on small parts.
- **Finishing.** Laminating, clear coating, cutting and edge finishing after printing are separate operations.

Cost-reduction tactics:

1. Gang multiple parts onto one full sheet — the machine charges by sheet and by pass, not by part.
2. Print only the layers you need; a white underbase and a clear varnish each add a full pass.
3. Choose a substrate whose color works with the artwork so no white underbase is required.
4. Use a lower-quality, faster print mode where viewing distance permits.
5. Above a few thousand identical impressions of the same artwork, price screen printing — the screens amortize quickly and the per-piece rate is far lower.

## FAQ

### What materials can UV inkjet print on?

Any rigid, reasonably flat substrate — acrylic, polycarbonate, PVC, aluminum, glass, ceramic tile, wood, foam board and composites. Because the ink cures instantly under UV rather than drying by absorption, non-porous materials that would never accept conventional ink print without issue. Glass, bare metal and low-energy plastics generally need a primer or adhesion promoter.

### Why does my print look washed out on colored acrylic?

UV inks are translucent, so the substrate color shows through. A white underbase must be printed first on any non-white or transparent substrate, and a double-hit white is usually needed over dark colors. That underbase is a separate pass and adds directly to cost.

### How thick is UV inkjet ink?

A standard CMYK build is on the order of 0.0004–0.0008 in (10–20 µm), against 0.0004–0.002 in (10–50 µm) for a single screen-printing pass. Repeated clear passes can build tactile relief up to roughly 0.02 in (0.5 mm) for braille and textured effects.

### UV inkjet or screen printing?

UV inkjet for short runs, variable data, photographic images and gradients — there are no screens and no per-color setup, so quantity one costs the same per piece as quantity ten. Screen printing for heavy opaque film, functional inks and long runs of identical artwork, where the screens amortize and the per-piece rate is far lower.

### How long do UV inkjet graphics last outdoors?

Manufacturers commonly rate uncoated UV inkjet output for roughly 1–2 years outdoors before noticeable fade, with a UV-blocking laminate or clear overprint extending it substantially. Confirm the figure for the specific ink set rather than assuming a general number.

### Can UV inkjet print on curved or three-dimensional parts?

Not usefully. The printhead runs only 1–2 mm above the substrate, which must be flat within roughly 0.06 in (1.5 mm) across the print area. For curved, recessed or fully three-dimensional surfaces, pad printing or hydro transfer printing are the appropriate processes.

## Alternative processes

- [Screen Printing](https://manufacturingprocesses.org/processes/finishing/screen-printing.md): Screen printing pushes ink through a stencilled mesh onto the surface below, laying down one opaque color per pass.
- [Pad Printing](https://manufacturingprocesses.org/processes/finishing/pad-printing.md): Pad printing lifts ink out of an etched plate with a silicone pad and transfers it onto a curved or recessed surface.
- [Hydro Transfer Printing](https://manufacturingprocesses.org/processes/finishing/hydro-transfer-printing.md): Hydro transfer printing floats a printed film on water, activates it, and wraps it around a submerged part to cover it in a pattern.
- [Laser Marking and Engraving](https://manufacturingprocesses.org/processes/finishing/laser-marking-and-engraving.md): Laser marking alters the surface with a focused beam — annealing, etching or engraving it — to make permanent, contact-free marks.
- [Spray Painting](https://manufacturingprocesses.org/processes/finishing/spray-painting.md): Spray painting atomises liquid coating onto a prepared surface, building a film that cures into a decorative and protective layer.

## Related processes

- [Screen Printing](https://manufacturingprocesses.org/processes/finishing/screen-printing.md): Screen printing pushes ink through a stencilled mesh onto the surface below, laying down one opaque color per pass.
- [Pad Printing](https://manufacturingprocesses.org/processes/finishing/pad-printing.md): Pad printing lifts ink out of an etched plate with a silicone pad and transfers it onto a curved or recessed surface.
- [Laser Marking and Engraving](https://manufacturingprocesses.org/processes/finishing/laser-marking-and-engraving.md): Laser marking alters the surface with a focused beam — annealing, etching or engraving it — to make permanent, contact-free marks.
- [Hydro Transfer Printing](https://manufacturingprocesses.org/processes/finishing/hydro-transfer-printing.md): Hydro transfer printing floats a printed film on water, activates it, and wraps it around a submerged part to cover it in a pattern.
- [Foil Blocking and Embossing](https://manufacturingprocesses.org/processes/finishing/foil-blocking-and-embossing.md): Foil blocking and embossing press a heated die into the surface to transfer metallic foil, raise a relief, or do both at once.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/finishing/uv-inkjet-printing)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Vacuum Metalizing"
category: "Finishing"
subcategory: "Additive"
materials: ["Plastic", "Metal", "Glass"]
tolerances: "Metal film 0.05–0.15 µm (2–6 µin) is dimensionally negligible; the base and topcoats add roughly 0.5–1.5 mils (13–38 µm) total per coated surface"
volumes: "Hundreds to millions; batch chamber cycles favor high-volume decorative parts"
lead_time: "2–4 weeks for first articles including basecoat qualification; days per lot in production"
url: https://manufacturingprocesses.org/processes/finishing/vacuum-metalizing
---

# Vacuum Metalizing

Vacuum metalizing evaporates aluminum inside a vacuum chamber so it condenses as a mirror-bright film on the part surface.

- **Category**: [Finishing](https://manufacturingprocesses.org/processes/finishing.md)
- **Family**: Additive
- **Materials**: Plastic, Metal, Glass
- **Typical tolerances**: Metal film 0.05–0.15 µm (2–6 µin) is dimensionally negligible; the base and topcoats add roughly 0.5–1.5 mils (13–38 µm) total per coated surface
- **Surface finish**: Mirror bright to satin, entirely determined by the molded surface and basecoat beneath the metal
- **Typical volumes**: Hundreds to millions; batch chamber cycles favor high-volume decorative parts
- **Lead time**: 2–4 weeks for first articles including basecoat qualification; days per lot in production

## Overview

Vacuum metalizing evaporates a metal — almost always aluminum — from resistively heated tungsten filaments inside a vacuum chamber, so the vapor travels in straight lines and condenses on the part as a mirror-bright film. The deposit is extraordinarily thin, on the order of 0.05–0.15 µm (2–6 µin), and has essentially no mechanical strength of its own.

That means the process is really a three-layer system, and the metal is the least of it. A sprayed and cured basecoat, typically 0.4–1 mil (10–25 µm), levels the substrate so the metal film has something optically smooth to sit on. The evaporated aluminum provides the reflectivity. A clear topcoat protects the soft metal from abrasion and handling and carries any tint. Cycle time in the chamber is minutes, the process runs at near room temperature, and it works on ABS, polycarbonate, other thermoplastics, glass and metal. It is the standard way to make automotive lamp reflectors, cosmetic closures and bright decorative trim.

## How it works

1. **Basecoat.** The part is sprayed with a UV-cure or solvent lacquer at roughly 0.4–1 mil (10–25 µm) and cured. Injection-molded surfaces carry flow lines, sink and micro-texture that would destroy a mirror finish; the basecoat fills and levels them. Reflector optics depend almost entirely on this layer.

2. **Load and pump down.** Parts are fixtured on rotating planetary racks inside the chamber, which is then evacuated to a high vacuum. Pump-down is the longest part of the cycle.

3. **Evaporate.** Aluminum wire clips on tungsten filaments are resistively heated until the aluminum melts and evaporates. In the vacuum, atoms travel ballistically to the part and condense. Deposition takes seconds to a minute; the rotation is what gives even coverage on a three-dimensional part.

4. **Vent and unload.** The chamber is vented and the parts come out mirror bright, with a film so thin that it can be rubbed off with a finger.

5. **Topcoat.** A clear lacquer is sprayed over the metal to protect it. Dyes in the topcoat give gold, bronze, smoke and other tints from the same aluminum deposit. Without a topcoat the finish will not survive handling.

### Variants

**Sputtering** in place of evaporation gives denser films and access to target materials that do not evaporate cleanly — chromium and stainless for a true chrome look, or indium for the semi-transparent, electrically non-conductive "chrome" used on radar-transparent vehicle emblems. **Thicker functional deposits** are used for EMI shielding of plastic enclosures, where the requirement is surface conductivity rather than appearance.

## Design guidelines

### The basecoat, not the metal, determines the finish

A 0.1 µm aluminum film is perfectly conformal. Every flow line, weld line, sink mark, gate blush, ejector pin witness and micro-texture in the molded surface reads straight through. Getting a mirror requires a Class A molded surface plus a basecoat that levels what remains. Budget the mold polish accordingly.

### Design for line of sight

Vapor travels in straight lines from point sources. Deep recesses, undercuts, the inside of ribs and any surface shadowed by another feature receive little or no metal, even with planetary rotation. Keep the decorated surface open and convex where possible, and do not expect uniform brightness inside a deep pocket.

### Do not specify it for surfaces that are touched or abraded

The metal film has no wear resistance and depends entirely on the topcoat. It is well suited to reflectors, trim, closures and display parts, and poorly suited to handles, buttons, edges and anything subject to repeated contact or cleaning with abrasives. Where durability is needed, electroplated ABS or PVD is the appropriate choice.

### Provide fixturing that does not shadow the show surface

Every part must be held. The contact area is unmetallized and often shows as a witness. Designate a non-cosmetic gripping location or a feature that later gets hidden by assembly.

### Confirm the substrate is vacuum-compatible

Outgassing is the failure mode. Resins with high moisture uptake, plasticizers, or residual mold release contaminate the chamber and produce hazy, poorly adherent films. ABS and polycarbonate are the standard choices; nylon and other hygroscopic resins need drying and careful process control.

### Do not rely on it for conductivity unless specified as such

A decorative aluminum film is far too thin for meaningful EMI shielding or grounding. Shielding requires a deliberately thicker functional deposit, specified by surface resistivity, not by appearance.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Metal film thickness | 0.05–0.15 µm (2–6 µin) | — | Optical layer only; carries no mechanical load |
| Basecoat thickness | 0.4–1 mil (10–25 µm) | — | Levels molded surface defects for the mirror |
| Topcoat | Always required | Never leave bare | Bare aluminum film rubs off with handling |
| Coverage | Line of sight, convex surfaces | Deep recesses shadow | Ballistic vapor transport in vacuum |
| Substrate | ABS, PC, PC/ABS, glass, metal | Avoid hygroscopic and plasticized resins | Outgassing produces hazy, non-adherent film |
| Fixture contact | 1 designated non-cosmetic area | — | Contact points are unmetallized

## Cost drivers

Vacuum metalizing is a batch process billed per chamber cycle, but the chamber is usually the cheapest part of the routing — the sprayed basecoat and topcoat carry most of the cost.

- **Base and topcoat application.** Two spray-and-cure operations with their own booths, racks and labor. These typically exceed the metalizing cost.
- **Chamber load density.** Cycle time is fixed, so parts that fixture densely and present their show surface outward cost far less per piece.
- **Mold and substrate quality.** A Class A tool costs more up front but eliminates rework and basecoat build; a poor molded surface can make an acceptable mirror unachievable at any coating cost.
- **Reject rate.** Dust inclusions, fisheyes and basecoat defects are visible on a mirror surface at a glance, so this process runs higher inspection and scrap rates than a matte finish.
- **Tinting and multi-color work.** Each tint or masked area is another spray operation.

Cost-reduction tactics:

1. Invest in mold polish and gating so the molded surface needs less basecoat leveling.
2. Design the show surface to be convex and open so parts fixture densely with no shadowing.
3. Use a tinted topcoat rather than a different metal to achieve gold, bronze or smoke.
4. Keep cosmetic requirements to the surfaces that are actually visible in the assembly.
5. Where the part is handled in service, price electroplated ABS or PVD instead — the rework cost of a scuffed metalized part exceeds the difference.

## FAQ

### How thick is vacuum metalizing?

The evaporated aluminum film is only about 0.05–0.15 µm (2–6 µin) — thin enough to rub off with a finger before it is topcoated. The functional thickness in the system comes from the basecoat and topcoat, which together add roughly 0.5–1.5 mils (13–38 µm).

### Why does vacuum metalizing need a basecoat and a topcoat?

The metal film is perfectly conformal and has no mechanical strength. The basecoat, typically 0.4–1 mil (10–25 µm), levels flow lines, sink and texture in the molded surface so there is something optically smooth to mirror. The topcoat protects the soft aluminum from abrasion and carries any tint.

### Is vacuum metalizing as durable as chrome plating?

No. Chrome-plated ABS produces a genuinely hard, wear-resistant surface; vacuum metalizing produces a mirror protected only by a clear lacquer. It suits reflectors, trim, closures and display parts, but not handles, buttons, edges or anything cleaned with abrasives.

### What substrates can be vacuum metalized?

ABS, polycarbonate and PC/ABS are the standard choices, along with glass and metal. Hygroscopic or plasticized resins outgas in the vacuum chamber and produce hazy, poorly adherent films, so they require drying and careful process control if they can be used at all.

### Can vacuum metalizing be used for EMI shielding?

Only with a deliberately thicker functional deposit specified by surface resistivity. A decorative aluminum film at 0.05–0.15 µm is far too thin to provide meaningful shielding or a ground path.

### Why are some areas of my metalized part dull or uncoated?

Deposition is line-of-sight from point sources in a vacuum, so deep recesses, undercuts and surfaces shadowed by other features receive little metal even with planetary rotation. Keep decorated surfaces open and convex, and expect a witness mark wherever the fixture held the part.

## Alternative processes

- [Electroplating](https://manufacturingprocesses.org/processes/finishing/electroplating.md): Electroplating deposits a metal coating from solution onto a conductive part by passing current through an electrolyte bath.
- [PVD Coating](https://manufacturingprocesses.org/processes/finishing/pvd-coating.md): PVD coating vaporises a solid target inside a vacuum chamber so it condenses on the part as a very thin, extremely hard film.
- [Spray Painting](https://manufacturingprocesses.org/processes/finishing/spray-painting.md): Spray painting atomises liquid coating onto a prepared surface, building a film that cures into a decorative and protective layer.
- [Hydro Transfer Printing](https://manufacturingprocesses.org/processes/finishing/hydro-transfer-printing.md): Hydro transfer printing floats a printed film on water, activates it, and wraps it around a submerged part to cover it in a pattern.
- [Powder Coating](https://manufacturingprocesses.org/processes/finishing/powder-coating.md): Powder coating sprays electrostatically charged dry polymer powder onto an earthed part, then bakes it into a tough continuous film.

## Related processes

- [PVD Coating](https://manufacturingprocesses.org/processes/finishing/pvd-coating.md): PVD coating vaporises a solid target inside a vacuum chamber so it condenses on the part as a very thin, extremely hard film.
- [Spray Painting](https://manufacturingprocesses.org/processes/finishing/spray-painting.md): Spray painting atomises liquid coating onto a prepared surface, building a film that cures into a decorative and protective layer.
- [Electroplating](https://manufacturingprocesses.org/processes/finishing/electroplating.md): Electroplating deposits a metal coating from solution onto a conductive part by passing current through an electrolyte bath.
- [Hydro Transfer Printing](https://manufacturingprocesses.org/processes/finishing/hydro-transfer-printing.md): Hydro transfer printing floats a printed film on water, activates it, and wraps it around a submerged part to cover it in a pattern.
- [Screen Printing](https://manufacturingprocesses.org/processes/finishing/screen-printing.md): Screen printing pushes ink through a stencilled mesh onto the surface below, laying down one opaque color per pass.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/finishing/vacuum-metalizing)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: process
name: "Vibratory Tumbling and Mass Finishing"
category: "Finishing"
subcategory: "Subtractive"
materials: ["Metal", "Plastic", "Ceramic"]
tolerances: "Removes 0.0001–0.001 in (2.5–25 µm) per surface in a typical cycle and breaks exposed edges 0.002–0.010 in (0.05–0.25 mm); the edge break is not controllable to a tolerance"
volumes: "Dozens to millions of parts; cost per part is essentially independent of count once the machine is full"
lead_time: "Same day to 5 business days at a job shop; cycle time itself is 30 minutes to 4 hours for vibratory work"
url: https://manufacturingprocesses.org/processes/finishing/vibratory-tumbling-and-mass-finishing
---

# Vibratory Tumbling and Mass Finishing

Vibratory tumbling agitates parts with abrasive media in a bowl so edges are broken and surfaces smoothed in bulk, without hand work.

- **Category**: [Finishing](https://manufacturingprocesses.org/processes/finishing.md)
- **Family**: Subtractive
- **Materials**: Metal, Plastic, Ceramic
- **Typical tolerances**: Removes 0.0001–0.001 in (2.5–25 µm) per surface in a typical cycle and breaks exposed edges 0.002–0.010 in (0.05–0.25 mm); the edge break is not controllable to a tolerance
- **Surface finish**: A cut cycle improves an as-machined 63–125 µin (1.6–3.2 µm) surface substantially; multi-stage or chemically accelerated finishing reaches below 4 µin (0.1 µm) Ra
- **Typical volumes**: Dozens to millions of parts; cost per part is essentially independent of count once the machine is full
- **Lead time**: Same day to 5 business days at a job shop; cycle time itself is 30 minutes to 4 hours for vibratory work

## Overview

Mass finishing agitates parts together with abrasive media in a bowl, tub, barrel or disc so that thousands of small rubbing actions deburr edges, blend surfaces and improve finish without any hand work. A vibratory bowl is the common form: media and parts circulate in a toroidal path, and each part is worked on every surface it exposes.

The characteristic result of a deburr cycle is an edge break of 0.002–0.010 in (0.05–0.25 mm) and stock removal of 0.0001–0.001 in (2.5–25 µm), over 30 minutes to 4 hours. Centrifugal barrel and centrifugal disc machines apply far higher forces and do comparable work in a fraction of that time. At the fine end, chemically accelerated (isotropic) finishing takes gear flanks and bearing races below 4 µin (0.1 µm) Ra. The economics are the point: mass finishing is the only deburring method whose cost per part is essentially independent of part count.

## How it works

1. **Choose media.** Ceramic media cuts fastest and suits steel and hard alloys. Plastic media is gentler and is used on aluminum, zinc and soft metals where ceramic would leave impingement marks. Steel media does not cut at all — it burnishes, work-hardens and brightens. Organic media (corn cob, walnut shell) carries polishing compound or dries parts. Media shape — cone, triangle, wedge, cylinder, angle-cut triangle — is chosen so it reaches into the part's features and does not lodge in them.

2. **Set the media-to-part ratio.** A typical working ratio is 3:1 to 6:1 media to parts by volume, rising toward 10:1 or more for delicate parts that must not touch each other. Too little media and parts impinge on one another; too much and cycle time and machine capacity are wasted.

3. **Add compound and water.** A metered flow of liquid compound controls pH, suspends the removed metal and abrasive fines, prevents rust on ferrous parts, and either cuts or burnishes depending on chemistry. Running dry loads media and stains parts.

4. **Run the cycle.** A vibratory bowl runs a deburr cycle in roughly 30 minutes to 4 hours. Centrifugal barrel finishing generates much higher effective forces and completes comparable work roughly an order of magnitude faster, at higher equipment cost. Drag and stream finishing hold parts in fixtures and drag them through a bed of media, which suits high-value parts that must not touch each other at all.

5. **Separate, rinse and dry.** Parts are separated from media by screen, magnet or manually, rinsed, and dried in a hot air or corn cob dryer. Any media lodged in a hole or slot has to be removed here.

### Multi-stage processes

Aggressive cut media first, then a finer cut stage, then a burnish or polish stage with steel or organic media produces a bright, deburred, low-Ra surface. Chemically accelerated finishing adds a conversion chemistry that softens the surface so the media removes only the peaks, and it is the standard route to sub-4 µin (0.1 µm) Ra on gear teeth and bearing components.

## Design guidelines

### Match media size to every hole and slot

Media that is close in size to a hole, slot or pocket will lodge in it, and removing lodged media is manual work on every part. Choose media that is either clearly smaller than the smallest feature so it passes through, or clearly larger than the largest so it cannot enter. Where a part has both a small hole and a large pocket, expect a compromise or a dedicated fixture process.

### Design out nesting and interlocking

Parts that nest — cupped stampings, rings, thin discs, hooks, springs — trap each other and finish unevenly on the mating faces. Break the nesting geometry, run a higher media ratio, or move to drag finishing where parts are held individually.

### Accept an uncontrolled edge break

Mass finishing rounds every exposed edge, and it rounds outside corners faster than inside ones. A typical deburr cycle gives 0.002–0.010 in (0.05–0.25 mm), and long cycles go further. Where a sharp edge is functional — a sealing land, a shear edge, a mating datum — mass finishing is the wrong process or the feature must be produced afterward.

### Budget the stock loss on toleranced features

Removal is 0.0001–0.001 in (2.5–25 µm) in a typical cycle. That is inside most machining tolerances but not inside a press-fit or a gage tolerance. Check close fits, thread pitch diameters and thin walls before committing.

### Protect fragile features

Thin fins, small pins, delicate tabs and threads on small parts can be bent or broken by media impact and by part-on-part contact. Raise the media ratio, drop the amplitude, or fixture the parts.

### Specify the finish achievable, not the finish hoped for

A cut cycle on an as-machined 63–125 µin (1.6–3.2 µm) surface typically improves it substantially; a burnish stage brightens it further; but reaching below about 8 µin (0.2 µm) requires multiple stages or chemically accelerated finishing. See the [surface finish chart](/charts/surface-finish-chart) to set a realistic callout.

| Feature | Recommended | Limit | Why |
| --- | --- | --- | --- |
| Media-to-part ratio | 3:1 to 6:1 by volume | 10:1+ for delicate parts | Low ratios cause part-on-part impingement |
| Media size vs. features | Clearly smaller or clearly larger | Never close to hole size | Similar-sized media lodges in holes |
| Edge break from a deburr cycle | 0.002–0.010 in (0.05–0.25 mm) | Not controllable to a tolerance | Every exposed edge is worked equally |
| Stock removal | 0.0001–0.001 in (2.5–25 µm) | Check press fits and threads | Cumulative over the cycle |
| Cycle time | 30 min–4 h vibratory | Centrifugal barrel is far faster | Force level sets removal rate |
| Sub-4 µin (0.1 µm) Ra | Chemically accelerated finishing | Not achievable with plain media | Requires conversion chemistry |

## Cost drivers

Mass finishing is the cheapest deburring method available per part, because one machine cycle processes an entire batch. Cost per part falls almost linearly with batch size until the machine is full.

- **Batch density.** The bowl or barrel costs the same to run whether it holds fifty parts or five hundred. Under-filling is the most common source of a high unit price.
- **Cycle time.** Longer cycles cost machine hours and consume media and compound. Centrifugal equipment cuts cycle time dramatically but costs more per hour.
- **Media and compound consumption.** Media wears down and is topped up continuously; compound flows through the machine. These are real per-hour costs, not negligible.
- **Separation and de-media labor.** Parts with holes that trap media require hand picking, which can exceed the cost of the cycle itself.
- **Multi-stage processes.** Each additional stage — cut, then fine cut, then burnish — is a separate machine setup and cycle.

Cost-reduction tactics:

1. Fill the machine. Batch parts across jobs where the media and compound are compatible.
2. Choose feature sizes at design time so media cannot lodge; this eliminates the largest hidden labor cost.
3. Use the shortest cycle that achieves the required edge break rather than defaulting to a long polish cycle.
4. Consider centrifugal barrel finishing when a vibratory cycle exceeds a few hours — the higher hourly rate is usually offset by the shorter cycle.
5. Where only a light edge break is needed, a single cut stage is enough; skip the burnish stage unless appearance is specified.

## FAQ

### How much edge break does vibratory tumbling produce?

A typical deburr cycle gives 0.002–0.010 in (0.05–0.25 mm) on exposed edges, and longer cycles go further. The break is not controllable to a tolerance — every exposed edge is worked, and outside corners round faster than inside ones — so functional sharp edges must be produced after tumbling or protected.

### How much material does mass finishing remove?

Typically 0.0001–0.001 in (2.5–25 µm) per surface in a normal cycle. That is inside most machining tolerances but not inside a press fit, a thread pitch diameter tolerance or a gage tolerance, so close fits should be checked before committing to the process.

### How do I stop media from getting stuck in holes?

Choose media that is either clearly smaller than the smallest hole, so it passes through freely, or clearly larger than the largest, so it cannot enter. Media close in size to a feature will lodge, and hand-picking it out on every part frequently costs more than the finishing cycle itself.

### What media should I use?

Ceramic cuts fastest and suits steel and hard alloys. Plastic is gentler for aluminum, zinc and soft metals. Steel media does not cut at all — it burnishes and work-hardens. Organic media such as corn cob and walnut shell carries polishing compound or dries parts. Media shape is chosen so it reaches into the part's features without lodging in them.

### Is centrifugal barrel finishing worth the extra cost?

Usually, when a vibratory cycle would run more than a few hours. Centrifugal barrel and disc machines generate much higher effective forces and complete comparable work roughly an order of magnitude faster, so the higher hourly rate is typically offset by the shorter cycle.

### Can mass finishing produce a mirror finish?

Multi-stage processes — cut, fine cut, then burnish with steel or organic media — produce a bright, low-Ra surface. Reaching below about 4 µin (0.1 µm) Ra requires chemically accelerated (isotropic) finishing, which adds a conversion chemistry so the media removes only the surface peaks. That route is standard for gear flanks and bearing races.

## Alternative processes

- [Electropolishing](https://manufacturingprocesses.org/processes/finishing/electropolishing.md): Electropolishing reverses plating: current dissolves the surface fastest at the peaks, leaving a bright, deburred, passive finish.
- [Abrasive Blasting](https://manufacturingprocesses.org/processes/finishing/abrasive-blasting.md): Abrasive blasting propels media at a surface with compressed air to clean it, strip scale, or produce a uniform matte texture.
- [Grinding, Sanding and Polishing](https://manufacturingprocesses.org/processes/finishing/grinding-sanding-and-polishing.md): Grinding, sanding and polishing remove a thin layer of material with bonded or coated abrasive to correct form or improve surface finish.
- [Bead Blasting](https://manufacturingprocesses.org/processes/finishing/bead-blasting.md): Bead blasting propels glass or ceramic beads at a surface to produce an even satin matte texture without cutting into it aggressively.
- [Photo Etching](https://manufacturingprocesses.org/processes/finishing/photo-etching.md): Photo etching masks a surface photographically and etches shallow detail into it for decoration, marking or texture.

## Related processes

- [Abrasive Blasting](https://manufacturingprocesses.org/processes/finishing/abrasive-blasting.md): Abrasive blasting propels media at a surface with compressed air to clean it, strip scale, or produce a uniform matte texture.
- [Bead Blasting](https://manufacturingprocesses.org/processes/finishing/bead-blasting.md): Bead blasting propels glass or ceramic beads at a surface to produce an even satin matte texture without cutting into it aggressively.
- [Grinding, Sanding and Polishing](https://manufacturingprocesses.org/processes/finishing/grinding-sanding-and-polishing.md): Grinding, sanding and polishing remove a thin layer of material with bonded or coated abrasive to correct form or improve surface finish.
- [Electropolishing](https://manufacturingprocesses.org/processes/finishing/electropolishing.md): Electropolishing reverses plating: current dissolves the surface fastest at the peaks, leaving a bright, deburred, passive finish.
- [Passivation](https://manufacturingprocesses.org/processes/finishing/passivation.md): Passivation removes free iron from a stainless steel surface with an acid bath so the chromium oxide layer can reform unbroken.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/processes/finishing/vibratory-tumbling-and-mass-finishing)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: index
title: "Reference Charts"
charts: 12
url: https://manufacturingprocesses.org/charts
---

# Reference Charts

12 engineering reference charts — gauge, tolerance, drill and finish data, in full, with sources.

- [Bolt Torque Chart](https://manufacturingprocesses.org/charts/bolt-torque-chart.md): Torque and clamp load for SAE Grade 2, 5 and 8 inch bolts and metric class 8.8, 10.9 and 12.9 bolts, dry and lubricated, at 75% of proof load.
- [Common Engineering Material Properties](https://manufacturingprocesses.org/charts/material-properties.md): Density, tensile and yield strength, hardness and Young's modulus for 16 common engineering metals and plastics, in both SI and imperial units, with per-row verification status.
- [Drill Size Chart](https://manufacturingprocesses.org/charts/drill-size-chart.md): Every number, letter, fractional and common metric drill with its decimal inch and millimeter diameter, per ANSI/ASME B94.11M.
- [GD&T Symbols Chart (ASME Y14.5)](https://manufacturingprocesses.org/charts/gdt-symbols.md): All geometric characteristic symbols with category, datum requirement and a plain-English meaning, flagged for the 2018 withdrawal of concentricity and symmetry, plus the modifier symbols.
- [ISO 286 Fits & Tolerance Chart (Hole Basis)](https://manufacturingprocesses.org/charts/iso-286-fits-tolerances.md): Preferred hole-basis fits, IT6–IT11 standard tolerance grades to 120 mm, and full deviation tables for H7/g6, H7/h6, H8/f7, H7/k6, H7/p6 and H7/s6.
- [Injection Molding Design Guidelines](https://manufacturingprocesses.org/charts/injection-molding-design-guidelines.md): Recommended wall thickness by polymer, draft angle by surface finish and texture, and the rib, boss and corner ratios that keep a molded part free of sink and warp.
- [Melting Points of Metals Chart](https://manufacturingprocesses.org/charts/metal-melting-points.md): Melting points and melting ranges in °C and °F for 40 pure metals and engineering alloys, with solidus and liquidus given separately where they differ.
- [Sheet Metal Bend Radius & K-Factor Chart](https://manufacturingprocesses.org/charts/sheet-metal-bend-radius-k-factor.md): Minimum inside bend radius by alloy and temper as multiples of thickness, K-factor against R/t, and K by bending method — with the two conflicting published K sets shown side by side.
- [Sheet Metal Gauge Chart](https://manufacturingprocesses.org/charts/sheet-metal-gauge-chart.md): Gauge-to-thickness conversions for steel, galvanized steel, stainless and aluminum, gauges 3 to 30, in inches and millimeters — plus the mill tolerance band.
- [Surface Finish Chart](https://manufacturingprocesses.org/charts/surface-finish-chart.md): ISO 1302 N-grades with Ra, RMS, Rz and Rt equivalents, the conversion factors between roughness parameters, and the Ra range each manufacturing process holds.
- [Tap Drill Chart](https://manufacturingprocesses.org/charts/tap-drill-chart.md): Tap drill sizes for UNC and UNF threads from #0 to 1/2 in at ~75% thread engagement, and for metric coarse threads M1.6 to M12 by the D−P rule.
- [Thread Size Chart](https://manufacturingprocesses.org/charts/thread-size-chart.md): Unified inch threads #0 to 1-1/2 with UNC and UNF pitches, and ISO metric threads M1.6 to M24 with coarse and fine pitches and basic minor diameters.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/charts)*


<!-- ---------------- -->

---
type: chart
title: "Bolt Torque Chart"
tables: 3
url: https://manufacturingprocesses.org/charts/bolt-torque-chart
---

# Bolt Torque Chart

Torque and clamp load for SAE Grade 2, 5 and 8 inch bolts and metric class 8.8, 10.9 and 12.9 bolts, dry and lubricated, at 75% of proof load.

Every value on this page is computed from the standard short-form
relationship **T = K · D · F**, where T is torque, K is the nut factor
(a dimensionless friction coefficient), D is nominal bolt diameter and F is
the target clamp load.

The assumptions behind the numbers, which you must read before using them:
target clamp load is **75% of proof load**, the standard industry basis for
reusable joints. **K = 0.20 dry** (plain or zinc-plated) and **K = 0.15
lubricated** (oil, wax or tapping fluid); cadmium and PTFE coatings run
nearer 0.12 and are not tabulated. The joint is assumed non-gasketed,
metal-to-metal and rigid, with a hardened washer under the turned element,
and the nut or tapped hole is assumed at least as strong as the bolt. Proof
loads used are Grade 2 = 55,000 psi, Grade 5 = 85,000 psi and Grade 8 =
120,000 psi.

**Torque is an indirect proxy for tension, and K scatters by roughly ±25%**
even under controlled conditions. Treat these as estimates, not
specifications.

## SAE inch bolts, coarse thread (UNC) — ft-lb

| Size | TPI | Stress area (in²) | Gr 2 clamp (lb) | Gr 2 dry (ft-lb) | Gr 2 lubed (ft-lb) | Gr 5 clamp (lb) | Gr 5 dry (ft-lb) | Gr 5 lubed (ft-lb) | Gr 8 clamp (lb) | Gr 8 dry (ft-lb) | Gr 8 lubed (ft-lb) |
| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
| 1/4-20 | 20 | 0.0318 | 1312 | 5.5 | 4.1 | 2029 | 8.5 | 6.3 | 2864 | 11.9 | 8.9 |
| 5/16-18 | 18 | 0.0524 | 2162 | 11.3 | 8.4 | 3342 | 17.4 | 13.1 | 4719 | 24.6 | 18.4 |
| 3/8-16 | 16 | 0.0775 | 3197 | 20 | 15 | 4940 | 30.9 | 23.2 | 6974 | 43.6 | 32.7 |
| 7/16-14 | 14 | 0.1063 | 4385 | 32 | 24 | 6777 | 49.4 | 37.1 | 9568 | 69.8 | 52.3 |
| 1/2-13 | 13 | 0.1419 | 5853 | 48.8 | 36.6 | 9046 | 75.4 | 56.5 | 12771 | 106.4 | 79.8 |
| 9/16-12 | 12 | 0.182 | 7508 | 70.4 | 52.8 | 11599 | 108.7 | 81.6 | 16375 | 153.5 | 115.1 |
| 5/8-11 | 11 | 0.226 | 9322 | 97.1 | 72.8 | 14408 | 150.1 | 112.6 | 20340 | 211.9 | 158.9 |
| 3/4-10 | 10 | 0.3345 | 13798 | 172.5 | 129.4 | 21322 | 266.5 | 199.9 | 30101 | 376.3 | 282.2 |

## Metric bolts, coarse thread — dry, K = 0.20 (N·m)

| Thread | Pitch (mm) | 8.8 clamp (kN) | 8.8 dry (N·m) | 10.9 clamp (kN) | 10.9 dry (N·m) | 12.9 clamp (kN) | 12.9 dry (N·m) |
| --- | --- | --- | --- | --- | --- | --- | --- |
| M6x1 | 1 | 8.8 | 10.5 | 12.5 | 15 | 14.6 | 17.6 |
| M8x1.25 | 1.25 | 15.9 | 25.5 | 22.8 | 36.5 | 26.6 | 42.6 |
| M10x1.5 | 1.5 | 25.2 | 50.5 | 36.1 | 72.2 | 42.2 | 84.4 |
| M12x1.75 | 1.75 | 36.7 | 88 | 52.5 | 125.9 | 61.3 | 147.1 |
| M14x2 | 2 | 50.2 | 140.6 | 71.9 | 201.2 | 84 | 235.1 |
| M16x2 | 2 | 68.1 | 218.1 | 97.5 | 312.1 | 114 | 364.7 |
| M18x2.5 | 2.5 | 83.7 | 301.4 | 119.8 | 431.3 | 140 | 504.1 |
| M20x2.5 | 2.5 | 106.5 | 425.9 | 152.4 | 609.5 | 178.1 | 712.3 |

## Metric bolts, coarse thread — lubricated, K = 0.15 (N·m)

| Thread | 8.8 lubed (N·m) | 10.9 lubed (N·m) | 12.9 lubed (N·m) |
| --- | --- | --- | --- |
| M6x1 | 7.9 | 11.3 | 13.2 |
| M8x1.25 | 19.1 | 27.3 | 32 |
| M10x1.5 | 37.8 | 54.1 | 63.3 |
| M12x1.75 | 66 | 94.4 | 110.3 |
| M14x2 | 105.5 | 150.9 | 176.4 |
| M16x2 | 163.6 | 234.1 | 273.5 |
| M18x2.5 | 226.1 | 323.5 | 378 |
| M20x2.5 | 319.4 | 457.1 | 534.2 |

## Notes

**Do not use this chart for structural steel connections (use RCSC/AISC),
flanged pressure joints (use ASME PCC-1) or wheel fasteners (use the OEM
spec).** For anything safety-critical, control tension directly:
turn-of-nut, bolt stretch, ultrasonic measurement or load-indicating
washers.

**Many shop charts list Grade 2 1/2-13 dry at about 55 ft-lb; this chart
gives 48.8.** The difference is the clamp-load basis — those charts
typically assume 85% of proof load, or a 57,000 psi proof value. Neither is
wrong, but the two bases must never be mixed inside one joint analysis. The
conservative 75%-of-proof basis is used here for consistency across all
three grades.

**Grade 2 is excluded above 3/4 in on purpose.** SAE J429 Grade 2 proof load
drops from 55,000 psi to 33,000 psi above 3/4 in, so extrapolating the table
would produce dangerously high values.

**A third-source metric spot-check gives M10 dry as 48 / 68 / 80 N·m for
8.8 / 10.9 / 12.9 against this chart's 50 / 72 / 84** — a roughly 5% gap
caused by a different clamp-load basis (some charts use 70% of proof, or
65% of yield). Both are defensible; the basis here is stated explicitly.

**Small sizes are usually specified in inch-pounds in the field.** Multiply
the ft-lb value by 12: 1/4-20 Grade 5 dry is 101 in-lb, Grade 8 dry is
143 in-lb.

Clamp loads were independently re-derived as 0.75 × proof × tensile stress
area and agreed with the published Fastenal values to within 0.1%.

## FAQ

### How much torque for a 1/2-13 Grade 8 bolt?

106.4 ft-lb dry and 79.8 ft-lb lubricated, at a clamp load of 12,771 lb (75% of proof). Grade 5 at the same size is 75.4 ft-lb dry, and Grade 2 is 48.8 ft-lb dry.

### How much torque for an M10 class 8.8 bolt?

50 N·m dry and 38 N·m lubricated, at a clamp load of 25.2 kN. Class 10.9 is 72 N·m dry and class 12.9 is 84 N·m dry.

### Why is lubricated torque lower than dry torque?

Because lubrication lowers the nut factor K from about 0.20 to about 0.15, so less of the applied torque is lost to friction and more of it becomes bolt tension. Applying the dry figure to a lubricated bolt over-tensions it by roughly a third and can yield the fastener.

### What does T = K · D · F mean?

Torque equals the nut factor times the nominal bolt diameter times the target clamp load. K bundles all the friction in the joint — thread friction and under-head friction — into one empirical number, which is why torque is only ever an indirect proxy for the tension you actually want.

### How accurate are torque charts?

K scatters by roughly ±25% even under controlled laboratory conditions, so the clamp load you actually achieve varies by about the same amount. Where preload matters, measure it directly with turn-of-nut, bolt stretch, ultrasonic measurement or load-indicating washers.

### Why are there no Grade 2 values above 3/4 inch?

SAE J429 Grade 2 proof strength drops from 55,000 psi to 33,000 psi above 3/4 in. Extending the table with the 55,000 psi basis would produce torque values well above what the fastener can carry.

### Can I use these numbers for structural steel or wheel bolts?

No. Structural steel connections follow RCSC/AISC, flanged pressure joints follow ASME PCC-1, and wheel fasteners follow the vehicle manufacturer's specification. This chart is for general non-gasketed metal-to-metal joints.

## Sources

- [Fastenal — Torque-Tension Reference Guide (clamp loads, stress areas, T = KDF basis)](https://www.fastenal.com/content/merch_rules/images/fcom/content-library/Torque-Tension%20Reference%20Guide.pdf)
- [Portland Bolt — Bolt Torque Chart](https://www.portlandbolt.com/technical/tools/bolt-torque-chart/)

## Related processes

- [Mechanical Fastening](https://manufacturingprocesses.org/processes/joining/mechanical-fastening.md)
- [Sheet Metal Fabrication](https://manufacturingprocesses.org/processes/forming/sheet-metal-fabrication.md)

## Related charts

- [Thread Size Chart](https://manufacturingprocesses.org/charts/thread-size-chart.md)
- [Tap Drill Chart](https://manufacturingprocesses.org/charts/tap-drill-chart.md)

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/charts/bolt-torque-chart)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: chart
title: "Common Engineering Material Properties"
tables: 2
url: https://manufacturingprocesses.org/charts/material-properties
---

# Common Engineering Material Properties

Density, tensile and yield strength, hardness and Young's modulus for 16 common engineering metals and plastics, in both SI and imperial units, with per-row verification status.

Sixteen materials an engineer meets constantly: the structural aluminiums,
the workhorse steels, the common stainless grades, titanium, brass, copper
and four engineering thermoplastics.

Values are **typical for the stated temper or condition**, and that
qualifier is doing a great deal of work — heat treatment moves these numbers
far more than alloy choice does. 4140 goes from 655 MPa annealed to 1020 MPa
quenched and tempered without changing composition at all.

Metal hardness is quoted in Brinell (HB) or Rockwell C (HRC). Plastic
hardness is Rockwell M or R, **a different scale entirely** — an R110 on ABS
and a 110 HB on steel have nothing to do with each other.

The ksi and lb/in³ columns are computed from the SI values (6.894757 MPa per
ksi, 0.0361273 lb/in³ per g/cm³), so no source rounding propagates into
them.

## Mechanical properties

| Material | Density (g/cm³) | Density (lb/in³) | UTS (MPa) | UTS (ksi) | Yield (MPa) | Yield (ksi) | Hardness | Young's modulus (GPa) | Verified |
| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
| Aluminum 6061-T6 | 2.7 | 0.0975 | 310 | 45 | 270 | 39.2 | 95 HB (93 HB MakeItFrom) | 69 | Yes |
| Aluminum 7075-T6 | 2.81 | 0.1015 | 572 | 83 | 503 | 73 | 150 HB | 71.7 | Yes |
| Steel 1018 (cold drawn) | 7.87 | 0.2843 | 440 | 63.8 | 370 | 53.7 | 126 HB | 205 | Indicative |
| Steel 4140 (annealed) | 7.85 | 0.2836 | 655 | 95 | 415 | 60.2 | 197 HB | 205 | Indicative |
| Steel 4140 (Q&T, 28-32 HRC) | 7.85 | 0.2836 | 1020 | 147.9 | 900 | 130.5 | 28-32 HRC (~285-302 HB) | 205 | Indicative |
| Steel A36 (structural) | 7.85 | 0.2836 | 400 | 58 | 250 | 36.3 | ~119-159 HB | 200 | Indicative |
| Stainless 304 (annealed) | 8 | 0.289 | 515 | 74.7 | 205 | 29.7 | 201 HB max / 92 HRB max | 193 | Yes |
| Stainless 316 (annealed) | 8 | 0.289 | 515 | 74.7 | 205 | 29.7 | 217 HB max / 95 HRB max | 193 | Indicative |
| Stainless 17-4PH (H900) | 7.75 | 0.28 | 1310 | 190 | 1170 | 169.7 | 40-47 HRC | 197 | Indicative |
| Titanium Ti-6Al-4V (Gr 5, annealed) | 4.43 | 0.16 | 950 | 137.8 | 880 | 127.6 | 334 HB / ~36 HRC | 113.8 | Yes |
| Brass C360 (free-cutting, H02) | 8.49 | 0.3067 | 400 | 58 | 310 | 45 | 78 HRB | 97 | Yes |
| Copper C110 (ETP, H04 hard) | 8.94 | 0.323 | 345 | 50 | 310 | 45 | ~50 HRF (annealed) / 60 HRB (hard) | 117 | Indicative |
| ABS (unfilled) | 1.04 | 0.0376 | 40 | 5.8 | 40 | 5.8 | R105-R110 Rockwell | 2.3 | Yes |
| Polycarbonate (PC, unfilled) | 1.2 | 0.0434 | 65 | 9.4 | 62 | 9 | M70 / R118 Rockwell | 2.4 | Yes |
| Nylon PA66 (dry, unfilled) | 1.14 | 0.0412 | 82 | 11.9 | 82 | 11.9 | M90 / R120 Rockwell | 3 | Yes |
| PEEK (unfilled) | 1.32 | 0.0477 | 100 | 14.5 | 100 | 14.5 | M99 / R126 Rockwell | 3.9 | Yes |

## Selection notes

| Material | Character |
| --- | --- |
| Aluminum 6061-T6 | Workhorse structural aluminum. Weldable, good machinability, anodizes well. |
| Aluminum 7075-T6 | Aerospace. High strength, poor weldability, poor corrosion resistance vs 6061. |
| Steel 1018 (cold drawn) | Low-carbon mild steel. Good weldability and formability; case-hardenable. |
| Steel 4140 (annealed) | Chrome-moly. Annealed condition for machining before heat treat. |
| Steel 4140 (Q&T, 28-32 HRC) | Common delivered condition for shafts and tooling. |
| Steel A36 (structural) | Structural plate/shape. 400-550 MPa UTS range; 250 MPa min yield. |
| Stainless 304 (annealed) | Min values per ASTM A240. Typical bar: 580 UTS / 230 YS, 170 HB. |
| Stainless 316 (annealed) | Adds Mo for chloride resistance. Same min strengths as 304. |
| Stainless 17-4PH (H900) | Precipitation-hardening. High strength plus decent corrosion resistance. |
| Titanium Ti-6Al-4V (Gr 5, annealed) | Best strength-to-weight of the common metals. Poor thermal conductivity - machines slowly. |
| Brass C360 (free-cutting, H02) | Machinability rating 100 - the benchmark all other alloys are rated against. |
| Copper C110 (ETP, H04 hard) | Highest practical electrical/thermal conductivity. Gummy - machinability rating ~20. |
| ABS (unfilled) | Tough, cheap, easy to mold and machine. Poor UV and solvent resistance. |
| Polycarbonate (PC, unfilled) | High impact strength, optically clear. Notch-sensitive; stress-cracks in some solvents. |
| Nylon PA66 (dry, unfilled) | Good wear and fatigue. Absorbs moisture - properties and dimensions shift with humidity. |
| PEEK (unfilled) | High-temperature engineering thermoplastic. Continuous service to ~250 C. Expensive. |

## Notes

**This table is spot-verified, not fully verified.** Sixteen materials by
six properties is 96 values. Nine rows were checked against at least one
independent datasheet and are marked verified; the other seven — 1018 cold
drawn, 4140 annealed, 4140 Q&T, A36, 316 annealed, 17-4PH H900 and copper
C110 — are long-established handbook values (ASTM A108, A29, A36, A240,
AMS 5643 and B152 respectively) that were not cross-checked cell by cell.
**Verify them before using them in a stress calculation.**

**304 is listed at its ASTM A240 minimums, not its typicals.** The minimums
are 515 MPa UTS and 205 MPa yield; typical bar runs nearer 580 and 230 MPa.
Designing to typical rather than minimum values is a classic and dangerous
error — the minimum is what the material certificate guarantees.

**7075-T6 density is 2.81 g/cm³, not the 3.0 some databases publish.** The
3.0 figure is a two-significant-figure rounding artefact and would introduce
a 7% mass error into any weight calculation. Its UTS and yield are listed at
the standard ASM values of 572/503 MPa against a rounded 560/480 elsewhere.

**6061-T6 hardness is quoted two ways:** 95 HB (ASM / Aluminum Association)
and 93 HB (MakeItFrom). The difference is within normal lot scatter and both
are shown.

**A36 is a specification, not a fixed alloy.** UTS is a range (400–550 MPa)
and yield is a minimum (250 MPa). Real plate routinely exceeds both.

**Copper and brass must always be quoted with a temper.** C110 spans roughly
220 MPa UTS / 70 MPa yield annealed through 380 / 340 MPa hard-drawn — yield
strength changes by more than 4×. The row here is the H04 hard condition.

**All polymer values are for unfilled, dry-as-molded resin at room
temperature.** Glass filling roughly doubles tensile strength and modulus.
Nylon in particular loses 30–40% of its strength and grows dimensionally
once it reaches equilibrium moisture content, so the dry values here are
optimistic for most real service environments.

## FAQ

### What is the yield strength of 6061-T6 aluminum?

270 MPa (39.2 ksi), with an ultimate tensile strength of 310 MPa (45 ksi), density 2.70 g/cm³ and Young's modulus 69 GPa. Hardness is about 95 HB.

### How much stronger is 7075-T6 than 6061-T6?

About 85% stronger in tension: 572 MPa UTS against 310 MPa, and 503 MPa yield against 270 MPa. Density is only 4% higher at 2.81 g/cm³. The trade-off is that 7075 is not practically weldable and corrodes more readily.

### What is the density of steel in lb/in³?

0.2836 lb/in³ (7.85 g/cm³) for 4140, A36 and most alloy steels; 1018 is very slightly higher at 0.2843 lb/in³ (7.87 g/cm³). Austenitic stainless 304 and 316 are denser at 0.2890 lb/in³ (8.00 g/cm³).

### Why are the 304 stainless values lower than what I see elsewhere?

Because these are the ASTM A240 minimums — 515 MPa UTS and 205 MPa yield — rather than typical values. Typical bar runs nearer 580 and 230 MPa. The minimum is what a material certificate guarantees, and it is what you should design to.

### How does heat treatment change 4140?

Annealed 4140 is 655 MPa UTS, 415 MPa yield and 197 HB. Quenched and tempered to 28–32 HRC the same alloy reaches 1020 MPa UTS and 900 MPa yield — a 56% increase in tensile strength with no change in composition.

### Can I compare plastic hardness to metal hardness?

No. Metals here are quoted in Brinell or Rockwell C; plastics in Rockwell M or R. They are separate scales measured with different indenters and loads, and there is no meaningful conversion between them.

### Which of these materials has the best strength-to-weight ratio?

Ti-6Al-4V: 950 MPa UTS at a density of 4.43 g/cm³. That beats 7075-T6 (572 MPa at 2.81 g/cm³) on absolute strength per unit volume and is comparable per unit mass, while holding strength at far higher temperatures.

### Do these plastic values apply to glass-filled grades?

No. All polymer rows are unfilled, dry-as-molded resin at room temperature. Glass filling roughly doubles tensile strength and modulus. Nylon also loses 30–40% of its strength once it absorbs equilibrium moisture, so the dry values are optimistic in service.

## Sources

- [MakeItFrom — per-material property pages (6061-T6, 7075-T6, annealed 304)](https://www.makeitfrom.com/material-properties/6061-T6-Aluminum)
- [The World Material — AISI 304 Stainless Steel Properties (ASTM A240 minimums)](https://www.theworldmaterial.com/type-304-grade-stainless-steel/)
- [ATI — ATI Ti-6Al-4V, Grade 5 datasheet](https://www.atimaterials.com/Products/Documents/datasheets/titanium/alloyed/ati_6-4_tds_en_v1.pdf)
- [National Bronze — C36000 free-cutting brass datasheet](https://www.nationalbronze.com/pdfs/C36000.pdf)
- [Direct Plastics — PEEK data sheet](https://www.directplastics.co.uk/pdf/datasheets/PEEK%20Data%20Sheet.pdf)
- [Smith Metal — Nylon 6/66 data sheet](https://www.smithmetal.com/pdf/plastics/nylon-6-66.pdf)

## Related processes

- [CNC Machining](https://manufacturingprocesses.org/processes/cutting/cnc-machining.md)
- [Sheet Metal Fabrication](https://manufacturingprocesses.org/processes/forming/sheet-metal-fabrication.md)
- [Heat Treatment](https://manufacturingprocesses.org/processes/forming/heat-treatment.md)
- [Injection Molding](https://manufacturingprocesses.org/processes/forming/injection-molding.md)

## Related charts

- [Melting Points of Metals Chart](https://manufacturingprocesses.org/charts/metal-melting-points.md)
- [Sheet Metal Gauge Chart](https://manufacturingprocesses.org/charts/sheet-metal-gauge-chart.md)

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/charts/material-properties)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: chart
title: "Drill Size Chart"
tables: 4
url: https://manufacturingprocesses.org/charts/drill-size-chart
---

# Drill Size Chart

Every number, letter, fractional and common metric drill with its decimal inch and millimeter diameter, per ANSI/ASME B94.11M.

Four drill size systems coexist on a US shop floor and none of them
supersedes the others: number drills (#1 to #80, larger number = smaller
drill), letter drills (A to Z, larger letter = larger drill), fractional
drills in 64ths of an inch, and metric drills. A single tap chart will
routinely send you to all four.

Diameters below are the nominal values from ANSI/ASME B94.11M. The
millimeter columns on the inch systems, and the inch columns on the metric
system, are computed at exactly 25.4 mm/in rather than transcribed.
Fractional decimals are computed from the fraction itself, so they are
exact — a 64th of an inch is exactly representable in decimal.

Note that letter E and 1/4 in are the same nominal diameter (0.2500 in).
That duplication is in the standard, not an error in the table.

## Number drills (#1 – #80)

| Drill | Diameter (in) | Diameter (mm) |
| --- | --- | --- |
| #80 | 0.0135 | 0.343 |
| #79 | 0.0145 | 0.368 |
| #78 | 0.016 | 0.406 |
| #77 | 0.018 | 0.457 |
| #76 | 0.02 | 0.508 |
| #75 | 0.021 | 0.533 |
| #74 | 0.0225 | 0.571 |
| #73 | 0.024 | 0.61 |
| #72 | 0.025 | 0.635 |
| #71 | 0.026 | 0.66 |
| #70 | 0.028 | 0.711 |
| #69 | 0.0292 | 0.742 |
| #68 | 0.031 | 0.787 |
| #67 | 0.032 | 0.813 |
| #66 | 0.033 | 0.838 |
| #65 | 0.035 | 0.889 |
| #64 | 0.036 | 0.914 |
| #63 | 0.037 | 0.94 |
| #62 | 0.038 | 0.965 |
| #61 | 0.039 | 0.991 |
| #60 | 0.04 | 1.016 |
| #59 | 0.041 | 1.041 |
| #58 | 0.042 | 1.067 |
| #57 | 0.043 | 1.092 |
| #56 | 0.0465 | 1.181 |
| #55 | 0.052 | 1.321 |
| #54 | 0.055 | 1.397 |
| #53 | 0.0595 | 1.511 |
| #52 | 0.0635 | 1.613 |
| #51 | 0.067 | 1.702 |
| #50 | 0.07 | 1.778 |
| #49 | 0.073 | 1.854 |
| #48 | 0.076 | 1.93 |
| #47 | 0.0785 | 1.994 |
| #46 | 0.081 | 2.057 |
| #45 | 0.082 | 2.083 |
| #44 | 0.086 | 2.184 |
| #43 | 0.089 | 2.261 |
| #42 | 0.0935 | 2.375 |
| #41 | 0.096 | 2.438 |
| #40 | 0.098 | 2.489 |
| #39 | 0.0995 | 2.527 |
| #38 | 0.1015 | 2.578 |
| #37 | 0.104 | 2.642 |
| #36 | 0.1065 | 2.705 |
| #35 | 0.11 | 2.794 |
| #34 | 0.111 | 2.819 |
| #33 | 0.113 | 2.87 |
| #32 | 0.116 | 2.946 |
| #31 | 0.12 | 3.048 |
| #30 | 0.1285 | 3.264 |
| #29 | 0.136 | 3.454 |
| #28 | 0.1405 | 3.569 |
| #27 | 0.144 | 3.658 |
| #26 | 0.147 | 3.734 |
| #25 | 0.1495 | 3.797 |
| #24 | 0.152 | 3.861 |
| #23 | 0.154 | 3.912 |
| #22 | 0.157 | 3.988 |
| #21 | 0.159 | 4.039 |
| #20 | 0.161 | 4.089 |
| #19 | 0.166 | 4.216 |
| #18 | 0.1695 | 4.305 |
| #17 | 0.173 | 4.394 |
| #16 | 0.177 | 4.496 |
| #15 | 0.18 | 4.572 |
| #14 | 0.182 | 4.623 |
| #13 | 0.185 | 4.699 |
| #12 | 0.189 | 4.801 |
| #11 | 0.191 | 4.851 |
| #10 | 0.1935 | 4.915 |
| #9 | 0.196 | 4.978 |
| #8 | 0.199 | 5.055 |
| #7 | 0.201 | 5.105 |
| #6 | 0.204 | 5.182 |
| #5 | 0.2055 | 5.22 |
| #4 | 0.209 | 5.309 |
| #3 | 0.213 | 5.41 |
| #2 | 0.221 | 5.613 |
| #1 | 0.228 | 5.791 |

## Letter drills (A – Z)

| Drill | Diameter (in) | Diameter (mm) |
| --- | --- | --- |
| A | 0.234 | 5.944 |
| B | 0.238 | 6.045 |
| C | 0.242 | 6.147 |
| D | 0.246 | 6.248 |
| E | 0.25 | 6.35 |
| F | 0.257 | 6.528 |
| G | 0.261 | 6.629 |
| H | 0.266 | 6.756 |
| I | 0.272 | 6.909 |
| J | 0.277 | 7.036 |
| K | 0.281 | 7.137 |
| L | 0.29 | 7.366 |
| M | 0.295 | 7.493 |
| N | 0.302 | 7.671 |
| O | 0.316 | 8.026 |
| P | 0.323 | 8.204 |
| Q | 0.332 | 8.433 |
| R | 0.339 | 8.611 |
| S | 0.348 | 8.839 |
| T | 0.358 | 9.093 |
| U | 0.368 | 9.347 |
| V | 0.377 | 9.576 |
| W | 0.386 | 9.804 |
| X | 0.397 | 10.084 |
| Y | 0.404 | 10.262 |
| Z | 0.413 | 10.49 |

## Fractional drills (1/64 in – 1 in, by 64ths)

| Fraction | Diameter, exact (in) | Diameter (mm) |
| --- | --- | --- |
| 1/64 | 0.015625 | 0.3969 |
| 1/32 | 0.03125 | 0.7937 |
| 3/64 | 0.046875 | 1.1906 |
| 1/16 | 0.0625 | 1.5875 |
| 5/64 | 0.078125 | 1.9844 |
| 3/32 | 0.09375 | 2.3812 |
| 7/64 | 0.109375 | 2.7781 |
| 1/8 | 0.125 | 3.175 |
| 9/64 | 0.140625 | 3.5719 |
| 5/32 | 0.15625 | 3.9688 |
| 11/64 | 0.171875 | 4.3656 |
| 3/16 | 0.1875 | 4.7625 |
| 13/64 | 0.203125 | 5.1594 |
| 7/32 | 0.21875 | 5.5562 |
| 15/64 | 0.234375 | 5.9531 |
| 1/4 | 0.25 | 6.35 |
| 17/64 | 0.265625 | 6.7469 |
| 9/32 | 0.28125 | 7.1437 |
| 19/64 | 0.296875 | 7.5406 |
| 5/16 | 0.3125 | 7.9375 |
| 21/64 | 0.328125 | 8.3344 |
| 11/32 | 0.34375 | 8.7312 |
| 23/64 | 0.359375 | 9.1281 |
| 3/8 | 0.375 | 9.525 |
| 25/64 | 0.390625 | 9.9219 |
| 13/32 | 0.40625 | 10.3187 |
| 27/64 | 0.421875 | 10.7156 |
| 7/16 | 0.4375 | 11.1125 |
| 29/64 | 0.453125 | 11.5094 |
| 15/32 | 0.46875 | 11.9062 |
| 31/64 | 0.484375 | 12.3031 |
| 1/2 | 0.5 | 12.7 |
| 33/64 | 0.515625 | 13.0969 |
| 17/32 | 0.53125 | 13.4937 |
| 35/64 | 0.546875 | 13.8906 |
| 9/16 | 0.5625 | 14.2875 |
| 37/64 | 0.578125 | 14.6844 |
| 19/32 | 0.59375 | 15.0812 |
| 39/64 | 0.609375 | 15.4781 |
| 5/8 | 0.625 | 15.875 |
| 41/64 | 0.640625 | 16.2719 |
| 21/32 | 0.65625 | 16.6687 |
| 43/64 | 0.671875 | 17.0656 |
| 11/16 | 0.6875 | 17.4625 |
| 45/64 | 0.703125 | 17.8594 |
| 23/32 | 0.71875 | 18.2562 |
| 47/64 | 0.734375 | 18.6531 |
| 3/4 | 0.75 | 19.05 |
| 49/64 | 0.765625 | 19.4469 |
| 25/32 | 0.78125 | 19.8438 |
| 51/64 | 0.796875 | 20.2406 |
| 13/16 | 0.8125 | 20.6375 |
| 53/64 | 0.828125 | 21.0344 |
| 27/32 | 0.84375 | 21.4312 |
| 55/64 | 0.859375 | 21.8281 |
| 7/8 | 0.875 | 22.225 |
| 57/64 | 0.890625 | 22.6219 |
| 29/32 | 0.90625 | 23.0187 |
| 59/64 | 0.921875 | 23.4156 |
| 15/16 | 0.9375 | 23.8125 |
| 61/64 | 0.953125 | 24.2094 |
| 31/32 | 0.96875 | 24.6062 |
| 63/64 | 0.984375 | 25.0031 |
| 1 | 1 | 25.4 |

## Common metric drills (0.5 – 13.0 mm)

| Drill | Diameter (mm) | Diameter (in) |
| --- | --- | --- |
| 0.5 mm | 0.5 | 0.0197 |
| 0.6 mm | 0.6 | 0.0236 |
| 0.7 mm | 0.7 | 0.0276 |
| 0.8 mm | 0.8 | 0.0315 |
| 0.9 mm | 0.9 | 0.0354 |
| 1 mm | 1 | 0.0394 |
| 1.1 mm | 1.1 | 0.0433 |
| 1.2 mm | 1.2 | 0.0472 |
| 1.25 mm | 1.25 | 0.0492 |
| 1.3 mm | 1.3 | 0.0512 |
| 1.4 mm | 1.4 | 0.0551 |
| 1.5 mm | 1.5 | 0.0591 |
| 1.6 mm | 1.6 | 0.063 |
| 1.7 mm | 1.7 | 0.0669 |
| 1.75 mm | 1.75 | 0.0689 |
| 1.8 mm | 1.8 | 0.0709 |
| 1.9 mm | 1.9 | 0.0748 |
| 2 mm | 2 | 0.0787 |
| 2.1 mm | 2.1 | 0.0827 |
| 2.2 mm | 2.2 | 0.0866 |
| 2.25 mm | 2.25 | 0.0886 |
| 2.3 mm | 2.3 | 0.0906 |
| 2.4 mm | 2.4 | 0.0945 |
| 2.5 mm | 2.5 | 0.0984 |
| 2.6 mm | 2.6 | 0.1024 |
| 2.7 mm | 2.7 | 0.1063 |
| 2.75 mm | 2.75 | 0.1083 |
| 2.8 mm | 2.8 | 0.1102 |
| 2.9 mm | 2.9 | 0.1142 |
| 3 mm | 3 | 0.1181 |
| 3.1 mm | 3.1 | 0.122 |
| 3.2 mm | 3.2 | 0.126 |
| 3.25 mm | 3.25 | 0.128 |
| 3.3 mm | 3.3 | 0.1299 |
| 3.4 mm | 3.4 | 0.1339 |
| 3.5 mm | 3.5 | 0.1378 |
| 3.6 mm | 3.6 | 0.1417 |
| 3.7 mm | 3.7 | 0.1457 |
| 3.75 mm | 3.75 | 0.1476 |
| 3.8 mm | 3.8 | 0.1496 |
| 3.9 mm | 3.9 | 0.1535 |
| 4 mm | 4 | 0.1575 |
| 4.1 mm | 4.1 | 0.1614 |
| 4.2 mm | 4.2 | 0.1654 |
| 4.25 mm | 4.25 | 0.1673 |
| 4.3 mm | 4.3 | 0.1693 |
| 4.4 mm | 4.4 | 0.1732 |
| 4.5 mm | 4.5 | 0.1772 |
| 4.6 mm | 4.6 | 0.1811 |
| 4.7 mm | 4.7 | 0.185 |
| 4.75 mm | 4.75 | 0.187 |
| 4.8 mm | 4.8 | 0.189 |
| 4.9 mm | 4.9 | 0.1929 |
| 5 mm | 5 | 0.1969 |
| 5.1 mm | 5.1 | 0.2008 |
| 5.2 mm | 5.2 | 0.2047 |
| 5.25 mm | 5.25 | 0.2067 |
| 5.3 mm | 5.3 | 0.2087 |
| 5.4 mm | 5.4 | 0.2126 |
| 5.5 mm | 5.5 | 0.2165 |
| 5.6 mm | 5.6 | 0.2205 |
| 5.7 mm | 5.7 | 0.2244 |
| 5.75 mm | 5.75 | 0.2264 |
| 5.8 mm | 5.8 | 0.2283 |
| 5.9 mm | 5.9 | 0.2323 |
| 6 mm | 6 | 0.2362 |
| 6.1 mm | 6.1 | 0.2402 |
| 6.2 mm | 6.2 | 0.2441 |
| 6.25 mm | 6.25 | 0.2461 |
| 6.3 mm | 6.3 | 0.248 |
| 6.4 mm | 6.4 | 0.252 |
| 6.5 mm | 6.5 | 0.2559 |
| 6.6 mm | 6.6 | 0.2598 |
| 6.7 mm | 6.7 | 0.2638 |
| 6.75 mm | 6.75 | 0.2657 |
| 6.8 mm | 6.8 | 0.2677 |
| 6.9 mm | 6.9 | 0.2717 |
| 7 mm | 7 | 0.2756 |
| 7.1 mm | 7.1 | 0.2795 |
| 7.2 mm | 7.2 | 0.2835 |
| 7.25 mm | 7.25 | 0.2854 |
| 7.3 mm | 7.3 | 0.2874 |
| 7.4 mm | 7.4 | 0.2913 |
| 7.5 mm | 7.5 | 0.2953 |
| 7.6 mm | 7.6 | 0.2992 |
| 7.7 mm | 7.7 | 0.3031 |
| 7.75 mm | 7.75 | 0.3051 |
| 7.8 mm | 7.8 | 0.3071 |
| 7.9 mm | 7.9 | 0.311 |
| 8 mm | 8 | 0.315 |
| 8.25 mm | 8.25 | 0.3248 |
| 8.5 mm | 8.5 | 0.3346 |
| 8.75 mm | 8.75 | 0.3445 |
| 9 mm | 9 | 0.3543 |
| 9.25 mm | 9.25 | 0.3642 |
| 9.5 mm | 9.5 | 0.374 |
| 9.75 mm | 9.75 | 0.3839 |
| 10 mm | 10 | 0.3937 |
| 10.5 mm | 10.5 | 0.4134 |
| 11 mm | 11 | 0.4331 |
| 11.5 mm | 11.5 | 0.4528 |
| 12 mm | 12 | 0.4724 |
| 12.5 mm | 12.5 | 0.4921 |
| 13 mm | 13 | 0.5118 |

## Notes

**Letter K is 0.2810 in, not 0.2811 in.** Engineers Edge publishes 0.2811;
Wikipedia and *Machinery's Handbook* give 0.2810, which is the value in
ASME B94.11M and in every drill manufacturer's catalog. The Engineers Edge
figure looks like a transcription artefact from the adjacent 9/32 = 0.28125
row. The 0.0001 in difference sits below drill manufacturing tolerance, so
it will never matter at the machine — but this chart publishes four-place
decimals, so it matters here.

**Several published fractional decimals are truncated rather than rounded.**
Engineers Edge shows 11/32 as .3437 (correctly .3438), 15/32 as .4687
(.4688), 19/32 as .5937 (.5938), 23/32 as .7187 (.7188), 27/32 as .8437
(.8438) and 31/32 as .9687 (.9688). Because this chart computes fractional
decimals from the fraction, none of that propagates.

**Shop-floor names are not a second size.** Machinists call fractional
drills by an abbreviated decimal — 27/64 is "a four-twenty-one drill", 1/16
is "a sixty-two". That shorthand is inconsistent (it rounds in some places
and follows tradition in others), so it is deliberately not tabulated here.
The exact decimal is the one to work from.

All 80 number drills and all 26 letter drills were cross-checked cell by
cell against two independent sources.

## FAQ

### What size is a #7 drill?

A #7 drill is 0.2010 in (5.105 mm). It is the standard tap drill for a 1/4-20 UNC thread at roughly 75% engagement.

### Do number drills get bigger or smaller as the number goes up?

Smaller. #1 is the largest at 0.2280 in and #80 is the smallest at 0.0135 in. Letter drills run the other way: A is the smallest at 0.2340 in and Z the largest at 0.4130 in.

### What is the metric equivalent of a 1/4 in drill?

1/4 in is exactly 6.350 mm. The nearest stocked metric drills are 6.3 mm (0.2480 in) and 6.4 mm (0.2520 in), so neither is an exact substitute.

### Why do letter E and 1/4 in have the same diameter?

Because they genuinely do: both are 0.2500 in. The letter and fractional series overlap in places and ASME B94.11M lists both designations. It is a duplicate name, not a duplicate size.

### What is the largest number drill and the smallest letter drill?

The largest number drill is #1 at 0.2280 in; the smallest letter drill is A at 0.2340 in. The series are contiguous, so the letter drills pick up where the number drills leave off.

### How precise are these decimals?

Number and letter drill diameters are the nominal values from ANSI/ASME B94.11M, cross-checked cell by cell against two sources. Fractional decimals are computed exactly from the fraction, and the metric-to-inch conversions use exactly 25.4 mm/in.

## Sources

- [Wikipedia — Drill bit sizes (sourced to ASME B94.11M)](https://en.wikipedia.org/wiki/Drill_bit_sizes)
- [Engineers Edge — Machinist Drill Bit Size Table (ANSI/ASME B94.11M-1993)](https://www.engineersedge.com/drill_sizes.htm)

## Related processes

- [CNC Milling](https://manufacturingprocesses.org/processes/cutting/cnc-milling.md)
- [CNC Machining](https://manufacturingprocesses.org/processes/cutting/cnc-machining.md)
- [CNC Turning](https://manufacturingprocesses.org/processes/cutting/cnc-turning.md)

## Related charts

- [Tap Drill Chart](https://manufacturingprocesses.org/charts/tap-drill-chart.md)
- [Thread Size Chart](https://manufacturingprocesses.org/charts/thread-size-chart.md)

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/charts/drill-size-chart)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: chart
title: "GD&T Symbols Chart (ASME Y14.5)"
tables: 3
url: https://manufacturingprocesses.org/charts/gdt-symbols
---

# GD&T Symbols Chart (ASME Y14.5)

All geometric characteristic symbols with category, datum requirement and a plain-English meaning, flagged for the 2018 withdrawal of concentricity and symmetry, plus the modifier symbols.

**The widely repeated "14 geometric characteristic symbols" figure is out of
date.** It comes from ASME Y14.5-2009. **ASME Y14.5-2018 removed
concentricity and symmetry, reducing the set to 12.** Both are still listed
below and clearly flagged, because they remain on a large amount of legacy
drawing stock and inspectors still have to read them — but neither should
appear on a new drawing. The replacement in both cases is **position**, or
runout for coaxial features.

The symbols split into five categories. Form controls (4) never take a
datum: they describe a feature against itself. Profile controls (2) may take
a datum. Orientation (3) and runout (2) controls always take one. Location
now has a single member, position, where it had three in 2009.

The plain-English meanings in the table below were written for this site.
The category assignments, datum requirements and 2018 statuses are factual
and were verified across five independent sources.

## Geometric characteristic symbols

| Symbol | Characteristic | Category | Datum required | ASME Y14.5-2018 | Meaning |
| --- | --- | --- | --- | --- | --- |
| ⏤ | Straightness | Form | No | Current | Keeps a surface line, or the feature's own axis, from bowing — it must stay inside a straight zone of the stated width. Says nothing about where the feature sits. |
| ⏥ | Flatness | Form | No | Current | Every point on the surface has to fall between two parallel planes set the tolerance apart. Self-contained: no datum, and no relationship to any other face. |
| ○ | Circularity (roundness) | Form | No | Current | Slice the feature perpendicular to its axis and the outline of that one slice must sit between two concentric circles whose radii differ by the tolerance. Each slice is judged on its own. |
| ⌭ | Cylindricity | Form | No | Current | The whole cylindrical surface must fit between two coaxial cylinders separated by the tolerance. One control that simultaneously limits roundness, straightness of the axis, and taper. |
| ⌒ | Profile of a line | Profile | Optional | Current | A single cross-sectional outline must stay inside a two-dimensional band that hugs the theoretically exact shape. Checked one slice at a time, so it does not police the surface as a whole. |
| ⌓ | Profile of a surface | Profile | Optional | Current | The entire surface must lie inside a three-dimensional shell wrapped around the ideal shape. The most powerful symbol in the set — with datums it can control size, form, orientation and location at once. |
| ∠ | Angularity | Orientation | Yes | Current | Holds a surface or axis at a stated basic angle to a datum. It fixes the tilt only; where the feature sits along the part is left to other controls. |
| ⊥ | Perpendicularity | Orientation | Yes | Current | Holds a surface or axis square to a datum. Mathematically just angularity at 90°, so no basic angle is written. |
| ∥ | Parallelism | Orientation | Yes | Current | Holds a surface or axis the same distance from a datum along its whole length. Angularity at 0° — again, no basic angle is written. |
| ⌖ | Position | Location | Yes (usually) | Current | Limits how far a feature's center, axis, or median plane may wander from the exact spot defined by basic dimensions. The workhorse control for hole patterns and the correct replacement for concentricity and symmetry. |
| ◎ | Concentricity | Location | Yes | Withdrawn | Required the median points of opposing surface elements to hug a datum axis. Withdrawn in ASME Y14.5-2018 — the median-point derivation is slow and poorly repeatable to inspect. Use position or runout. |
| ⌯ | Symmetry | Location | Yes | Withdrawn | Required the median points of a feature to straddle a datum plane evenly. Withdrawn in ASME Y14.5-2018 for the same measurement difficulty as concentricity. Use position. |
| ↗ | Circular runout | Runout | Yes | Current | Spin the part about its datum axis and, at any one fixed cross-section, the indicator sweep must stay inside the tolerance. Catches roundness plus off-center position together, in one cheap measurement. |
| ⌰ | Total runout | Runout | Yes | Current | Same rotation, but the indicator travels the full length or across the whole face while the part turns. The entire surface must stay within a single tolerance band, so it also catches taper and profile error. |

## Category summary — ASME Y14.5-2018 against Y14.5-2009

| Category | Count (2018) | Count (2009) | Members | Datum |
| --- | --- | --- | --- | --- |
| Form | 4 | 4 | Straightness, Flatness, Circularity, Cylindricity | Never |
| Profile | 2 | 2 | Profile of a line, Profile of a surface | Optional |
| Orientation | 3 | 3 | Angularity, Perpendicularity, Parallelism | Always |
| Location | 1 | 3 | Position — concentricity and symmetry withdrawn in 2018 | Usually |
| Runout | 2 | 2 | Circular runout, Total runout | Always |

## Modifiers and supplementary symbols

| Symbol | Name | Meaning |
| --- | --- | --- |
| Ⓜ | Maximum Material Condition (MMC) | Apply the tolerance at the feature's most-material size; bonus tolerance is earned as the feature departs from MMC. Use for assembly clearance. |
| Ⓛ | Least Material Condition (LMC) | Same bonus mechanism anchored at the least-material size. Use to protect wall thickness or edge distance. |
| (none) | Regardless of Feature Size (RFS) | ASME default with no modifier: tolerance applies at every size, no bonus. |
| Ⓟ | Projected tolerance zone | Moves the tolerance zone into the mating part, typically above a tapped hole. |
| Ⓕ | Free state | Inspect unrestrained; for non-rigid parts. |
| Ⓣ | Tangent plane | Control the plane resting on the high points instead of every surface point. |
| Ⓤ | Unequally disposed profile | Profile zone split unevenly about true shape; outward amount stated after the symbol. |
| Ⓘ | Independency | Releases the envelope requirement so form is not limited by size. |
| ⟨ST⟩ | Statistical tolerance | Valid only under statistical process control. |
| ⟨CF⟩ | Continuous feature | Treat interrupted surfaces as one feature. |
| △ | Dynamic profile | New in Y14.5-2018; profile zone controls form only and may expand/contract uniformly. |
| ⌀ / S⌀ | Diameter / spherical diameter | Declares a cylindrical or spherical tolerance zone or dimension. |
| R / CR / SR | Radius / controlled radius / spherical radius | CR requires a fair curve with no flats or reversals. |
| ▷ | Translation | Datum feature simulator is free to translate rather than being locked. |
| ↔ | Between | Applies the tolerance only across the stated span. |
| Ⓔ | Envelope requirement (ISO) | ISO-only; imposes the envelope condition that ASME applies by default. |

## Notes

**Concentricity and symmetry were withdrawn in ASME Y14.5-2018.** Four
independent sources confirm it. Both required deriving median points from
opposing surface elements, which is slow and poorly repeatable to inspect.
Use position instead, or runout for coaxial features. Symbol reference pages
that still list all 14 without flagging the withdrawal are how the outdated
count keeps propagating.

**Counts reconcile:** 4 + 2 + 3 + 3 + 2 = 14 under Y14.5-2009, and
4 + 2 + 3 + 1 + 2 = 12 under Y14.5-2018.

**Dynamic profile (△) is a Y14.5-2018 addition**, confirmed by two sources.
The profile zone controls form only and is free to expand or contract
uniformly, so size and location are handled elsewhere.

**Several GD&T symbols have no exact Unicode codepoint** — all-around,
all-over, the datum feature symbol, and the runout arrows in particular. The
characters shown here are the closest available and are adequate as database
keys, but they are approximations of the drawn glyphs, not the glyphs
themselves.

**Not verified against the standard document itself.** ASME Y14.5-2018 is
paywalled; all verification here is against secondary sources, four of which
are independent and in agreement.

## FAQ

### How many GD&T symbols are there?

Twelve geometric characteristic symbols under ASME Y14.5-2018. The commonly quoted figure of 14 is from Y14.5-2009; the 2018 revision withdrew concentricity and symmetry.

### Why were concentricity and symmetry removed from ASME Y14.5?

Both required deriving median points from opposing surface elements, which is slow to measure and poorly repeatable. Y14.5-2018 withdrew them. Use position instead, or runout for coaxial features. They still appear on legacy drawings, so they are listed here flagged as withdrawn.

### Which GD&T symbols need a datum?

Orientation (angularity, perpendicularity, parallelism) and runout (circular, total) always need one. Form controls (straightness, flatness, circularity, cylindricity) never take one. Profile controls may take one. Position usually does.

### What is the difference between circular runout and total runout?

Circular runout judges the indicator sweep at one fixed cross-section as the part rotates. Total runout moves the indicator along the full length or across the whole face while the part turns, so the entire surface must fall inside a single tolerance band — which also catches taper and profile error.

### What does the MMC modifier do?

Maximum Material Condition applies the tolerance at the feature's most-material size — the smallest hole or largest shaft — and grants bonus tolerance in proportion as the feature departs from that size. Use it when the concern is assembly clearance.

### What is the default if no modifier is shown?

Regardless of Feature Size (RFS). The stated tolerance applies at every size with no bonus. It is the ASME default and needs no symbol.

### What is profile of a surface used for?

It is the most powerful control in the set: the whole surface must lie inside a three-dimensional shell wrapped around the ideal shape. Combined with datums it can control size, form, orientation and location simultaneously.

### What is dynamic profile?

A Y14.5-2018 addition. The profile zone controls form only and is free to expand or contract uniformly, so size and location are left to other controls.

## Sources

- [GD&T Basics — GD&T Symbols Explained: Full Reference Guide](https://www.gdandtbasics.com/gdt-symbols/)
- [GD&T Basics — ASME Y14.5 2009 vs. 2018: Standard Changes](https://www.gdandtbasics.com/asme-2009-vs-2018/)
- [Mitutoyo — GD&T and the new ASME Y14.5-2018 (Salsbury)](https://www.mitutoyo.com/webfoo/wp-content/uploads/ASME_Y14.5-2018_Salsbury.pdf)
- [GeoTol — ASME Y14.5-2018 vs. 2009: Changes & Latest GD&T Standards](https://geotol.com/symbol/2018-standards/)
- [Measure Metrology — ASME has just released Y14.5-2018: what's new?](https://measuremetrology.com/asme-has-just-released-y14-5-2018-whats-new/)

## Related processes

- [CNC Machining](https://manufacturingprocesses.org/processes/cutting/cnc-machining.md)
- [CNC Milling](https://manufacturingprocesses.org/processes/cutting/cnc-milling.md)
- [CNC Turning](https://manufacturingprocesses.org/processes/cutting/cnc-turning.md)

## Related charts

- [ISO 286 Fits & Tolerance Chart (Hole Basis)](https://manufacturingprocesses.org/charts/iso-286-fits-tolerances.md)
- [Surface Finish Chart](https://manufacturingprocesses.org/charts/surface-finish-chart.md)

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/charts/gdt-symbols)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: chart
title: "ISO 286 Fits & Tolerance Chart (Hole Basis)"
tables: 8
url: https://manufacturingprocesses.org/charts/iso-286-fits-tolerances
---

# ISO 286 Fits & Tolerance Chart (Hole Basis)

Preferred hole-basis fits, IT6–IT11 standard tolerance grades to 120 mm, and full deviation tables for H7/g6, H7/h6, H8/f7, H7/k6, H7/p6 and H7/s6.

In a **hole-basis** system the hole is the fixed member — always an H class,
lower deviation zero — and the shaft is varied to create the fit. This is
the default for general machining because reamers, broaches and plug gauges
come in fixed sizes while a shaft can be turned to any diameter. Shaft-basis
equivalents are listed alongside for the case where one ground shaft passes
through several different components.

The sign convention used throughout: **maximum clearance = hole upper
deviation − shaft lower deviation**, and **minimum clearance = hole lower
deviation − shaft upper deviation**. A negative result is interference.
All deviations are in micrometers.

As a rough process-capability guide: IT6 is precision grinding or honing,
IT7 is good reaming, fine turning or grinding, IT8 is normal turning, boring
and reaming, IT9 is general turning and milling, and IT10–IT11 covers
drilling, rough machining and cold drawing. H7 is the practical default for
a machined and reamed hole.

## Preferred hole-basis fits

| Fit (hole basis) | Shaft-basis equivalent | Category | Name | Typical use |
| --- | --- | --- | --- | --- |
| H11/c11 | C11/h11 | Clearance | Loose running | Wide clearance, accuracy irrelevant. Pivots, latches, parts exposed to heat, corrosion or contamination. |
| H9/d9 | D9/h9 | Clearance | Free running | Large clearance. High running speeds, big temperature swings, heavy journal pressure. |
| H8/f7 | F8/h7 | Clearance | Close running | Moderate accuracy. Plain bearings, spindles, sliding rods. |
| H7/g6 | G7/h6 | Clearance | Sliding | Minimal clearance with high accuracy; assembles easily and turns or slides freely. Guided shafts, sliding gears. |
| H7/h6 | H7/h6 | Clearance (zero-line) | Locational clearance | Very close clearance, no force to assemble, turns and slides when lubricated. Precise shaft location. |
| H7/k6 | K7/h6 | Transition | Locational transition (similar fit) | Negligible clearance or interference; mallet assembly. Hubs, gears, pulleys, bushes, bearings. |
| H7/n6 | N7/h6 | Transition | Locational transition (fixed fit) | Small interference, light press. Plugs, driven bushes, armatures on shafts. |
| H7/p6 | P7/h6 | Interference | Press fit | Light interference, cold press. Hubs, bearings, bushings, retainers. |
| H7/s6 | S7/h6 | Interference | Driving fit | Medium interference, hot press or heavy cold press. Permanent gear and shaft mounting. |
| H7/u6 | U7/h6 | Interference | Forced / shrink fit | High interference needing a large temperature differential. Permanent couplings that cannot be separated without damage. |

## IT grade standard tolerance values, IT6 – IT11, 0 – 120 mm

| Nominal size (mm) | IT6 (µm) | IT7 (µm) | IT8 (µm) | IT9 (µm) | IT10 (µm) | IT11 (µm) |
| --- | --- | --- | --- | --- | --- | --- |
| 0–3 | 6 | 10 | 14 | 25 | 40 | 60 |
| 3–6 | 8 | 12 | 18 | 30 | 48 | 75 |
| 6–10 | 9 | 15 | 22 | 36 | 58 | 90 |
| 10–18 | 11 | 18 | 27 | 43 | 70 | 110 |
| 18–30 | 13 | 21 | 33 | 52 | 84 | 130 |
| 30–50 | 16 | 25 | 39 | 62 | 100 | 160 |
| 50–80 | 19 | 30 | 46 | 74 | 120 | 190 |
| 80–120 | 22 | 35 | 54 | 87 | 140 | 220 |

## H7/g6 — Sliding / close clearance

| Nominal size (mm) | H7 hole (µm) | g6 shaft (µm) | Max clearance (µm) | Min clearance (µm) | Result |
| --- | --- | --- | --- | --- | --- |
| 0–3 | +10/+0 | -2/-8 | 18 | 2 | clearance |
| 3–6 | +12/+0 | -4/-12 | 24 | 4 | clearance |
| 6–10 | +15/+0 | -5/-14 | 29 | 5 | clearance |
| 10–18 | +18/+0 | -6/-17 | 35 | 6 | clearance |
| 18–30 | +21/+0 | -7/-20 | 41 | 7 | clearance |
| 30–50 | +25/+0 | -9/-25 | 50 | 9 | clearance |
| 50–80 | +30/+0 | -10/-29 | 59 | 10 | clearance |
| 80–120 | +35/+0 | -12/-34 | 69 | 12 | clearance |

## H7/h6 — Locational clearance

| Nominal size (mm) | H7 hole (µm) | h6 shaft (µm) | Max clearance (µm) | Min clearance (µm) | Result |
| --- | --- | --- | --- | --- | --- |
| 0–3 | +10/+0 | +0/-6 | 16 | 0 | clearance |
| 3–6 | +12/+0 | +0/-8 | 20 | 0 | clearance |
| 6–10 | +15/+0 | +0/-9 | 24 | 0 | clearance |
| 10–18 | +18/+0 | +0/-11 | 29 | 0 | clearance |
| 18–30 | +21/+0 | +0/-13 | 34 | 0 | clearance |
| 30–50 | +25/+0 | +0/-16 | 41 | 0 | clearance |
| 50–80 | +30/+0 | +0/-19 | 49 | 0 | clearance |
| 80–120 | +35/+0 | +0/-22 | 57 | 0 | clearance |

## H8/f7 — Close running

| Nominal size (mm) | H8 hole (µm) | f7 shaft (µm) | Max clearance (µm) | Min clearance (µm) | Result |
| --- | --- | --- | --- | --- | --- |
| 0–3 | +14/+0 | -6/-16 | 30 | 6 | clearance |
| 3–6 | +18/+0 | -10/-22 | 40 | 10 | clearance |
| 6–10 | +22/+0 | -13/-28 | 50 | 13 | clearance |
| 10–18 | +27/+0 | -16/-34 | 61 | 16 | clearance |
| 18–30 | +33/+0 | -20/-41 | 74 | 20 | clearance |
| 30–50 | +39/+0 | -25/-50 | 89 | 25 | clearance |
| 50–80 | +46/+0 | -30/-60 | 106 | 30 | clearance |
| 80–120 | +54/+0 | -36/-71 | 125 | 36 | clearance |

## H7/k6 — Locational transition

| Nominal size (mm) | H7 hole (µm) | k6 shaft (µm) | Max clearance (µm) | Min clearance (µm) | Result |
| --- | --- | --- | --- | --- | --- |
| 0–3 | +10/+0 | +6/+0 | 10 | -6 | transition |
| 3–6 | +12/+0 | +9/+1 | 11 | -9 | transition |
| 6–10 | +15/+0 | +10/+1 | 14 | -10 | transition |
| 10–18 | +18/+0 | +12/+1 | 17 | -12 | transition |
| 18–30 | +21/+0 | +15/+2 | 19 | -15 | transition |
| 30–50 | +25/+0 | +18/+2 | 23 | -18 | transition |
| 50–80 | +30/+0 | +21/+2 | 28 | -21 | transition |
| 80–120 | +35/+0 | +25/+3 | 32 | -25 | transition |

## H7/p6 — Press / locational interference

| Nominal size (mm) | H7 hole (µm) | p6 shaft (µm) | Max clearance (µm) | Min clearance (µm) | Result |
| --- | --- | --- | --- | --- | --- |
| 0–3 | +10/+0 | +12/+6 | 4 | -12 | transition |
| 3–6 | +12/+0 | +20/+12 | 0 | -20 | interference |
| 6–10 | +15/+0 | +24/+15 | 0 | -24 | interference |
| 10–18 | +18/+0 | +29/+18 | 0 | -29 | interference |
| 18–30 | +21/+0 | +35/+22 | -1 | -35 | interference |
| 30–50 | +25/+0 | +42/+26 | -1 | -42 | interference |
| 50–80 | +30/+0 | +51/+32 | -2 | -51 | interference |
| 80–120 | +35/+0 | +59/+37 | -2 | -59 | interference |

## H7/s6 — Driving / medium interference

| Nominal size (mm) | H7 hole (µm) | s6 shaft (µm) | Max clearance (µm) | Min clearance (µm) | Result |
| --- | --- | --- | --- | --- | --- |
| 0–3 | +10/+0 | +20/+14 | -4 | -20 | interference |
| 3–6 | +12/+0 | +27/+19 | -7 | -27 | interference |
| 6–10 | +15/+0 | +32/+23 | -8 | -32 | interference |
| 10–18 | +18/+0 | +39/+28 | -10 | -39 | interference |
| 18–30 | +21/+0 | +48/+35 | -14 | -48 | interference |
| 30–50 | +25/+0 | +59/+43 | -18 | -59 | interference |
| 50–65 | +30/+0 | +72/+53 | -23 | -72 | interference |
| 65–80 | +30/+0 | +78/+59 | -29 | -78 | interference |
| 80–100 | +35/+0 | +93/+71 | -36 | -93 | interference |
| 100–120 | +35/+0 | +101/+79 | -44 | -101 | interference |

## Notes

**A web-search summary returned IT6 = 0.8, IT7 = 1.2 and IT8 = 2 µm for the
0–3 mm range. Those are wrong** — they are the IT1/IT2/IT3 values, off by
five grade columns. The correct 0–3 mm values are IT6 = 6, IT7 = 10,
IT8 = 14 µm. This is worth flagging because it is exactly the kind of error
that propagates into AI-generated reference tables.

**The s6 table subdivides where the others do not.** ISO 286 splits 50–80 mm
into 50–65 and 65–80, and 80–120 mm into 80–100 and 100–120, for the s and u
shaft classes only. Those four bands are preserved here; every other table
collapsing them is correct behaviour, not an omission.

**Every deviation on this page was independently re-derived from ISO 286
first principles** (fundamental deviation plus IT grade width) and
cross-checked against a second source. Examples: g6 at 50–80 → deviation
−10 µm, IT6 = 19 → −10/−29. k6 at 80–120 → +3, IT6 = 22 → +25/+3. p6 at
50–80 → +32, IT6 = 19 → +51/+32. f7 at 80–120 → −36, IT7 = 35 → −36/−71.
The deviation table and the IT grade table came from different sources and
agree to the last micrometer.

**IT10 is single-source.** IT6, IT7, IT8, IT9 and IT11 were confirmed by two
independent sources; IT10 appears in only one, but sits correctly in the
geometric progression between the two-source-verified IT9 and IT11 rows.

## FAQ

### What is an H7 tolerance?

H7 is a hole with a lower deviation of zero and an upper deviation equal to the IT7 grade width for that size. At 10–18 mm that is +18/+0 µm; at 30–50 mm it is +25/+0 µm. It is the practical default for a machined and reamed hole.

### What is the difference between H7/g6 and H7/h6?

H7/h6 is a zero-line clearance fit: minimum clearance is exactly zero, so the parts assemble without force but are not intended to run. H7/g6 pulls the shaft down slightly to guarantee a small clearance — +2 to +18 µm at 0–3 mm — so the joint slides and turns freely when lubricated.

### Which fit should I use for a press fit?

H7/p6 for a light cold press — at 18–30 mm it gives −1 to −35 µm — and H7/s6 for medium interference needing heat or heavy press force, which at the same size gives −14 to −48 µm. H7/u6 is the forced or shrink fit for permanent couplings.

### What does hole basis mean?

The hole is held at a fixed tolerance class (always H, with zero lower deviation) and the fit is changed by varying the shaft. It is the default in general machining because reamers, broaches and plug gauges come in fixed sizes while shafts can be turned to any diameter.

### How do I calculate clearance from a fit?

Maximum clearance = hole upper deviation − shaft lower deviation; minimum clearance = hole lower deviation − shaft upper deviation. A negative answer is interference. For H8/f7 at Ø50, that gives +89 and +25 µm.

### Is H7/p6 always an interference fit?

No. In the 0–3 mm range H7/p6 is technically a transition fit: it runs from +4 µm clearance to −12 µm interference. From 3 mm upward it is interference throughout.

### What machining process holds an IT7 tolerance?

Good reaming, fine turning or grinding. IT6 needs precision grinding or honing, IT8 is achievable with normal turning, boring and reaming, and IT10–IT11 covers drilling, rough machining and cold-drawn stock.

## Sources

- [Engineers Edge — International Tolerance (IT) Grades Table Chart, ISO 286-1:2010](https://www.engineersedge.com/international_tol.htm)
- [Wikipedia — Engineering fit (cited to Engineers Black Book 2nd ed., p. 70)](https://en.wikipedia.org/wiki/Engineering_fit)

## Related processes

- [CNC Turning](https://manufacturingprocesses.org/processes/cutting/cnc-turning.md)
- [CNC Milling](https://manufacturingprocesses.org/processes/cutting/cnc-milling.md)
- [Electrical Discharge Machining](https://manufacturingprocesses.org/processes/cutting/electrical-discharge-machining.md)
- [Grinding, Sanding and Polishing](https://manufacturingprocesses.org/processes/finishing/grinding-sanding-and-polishing.md)

## Related charts

- [Surface Finish Chart](https://manufacturingprocesses.org/charts/surface-finish-chart.md)
- [Drill Size Chart](https://manufacturingprocesses.org/charts/drill-size-chart.md)

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/charts/iso-286-fits-tolerances)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: chart
title: "Injection Molding Design Guidelines"
tables: 3
url: https://manufacturingprocesses.org/charts/injection-molding-design-guidelines
---

# Injection Molding Design Guidelines

Recommended wall thickness by polymer, draft angle by surface finish and texture, and the rib, boss and corner ratios that keep a molded part free of sink and warp.

**Uniform wall thickness is the single highest-leverage decision in a
molded part.** Non-uniform walls cause sink, voids, warp and differential
shrink. Where a thickness change is unavoidable, transition gradually — a
3:1 taper minimum — and keep the step under 25% of nominal for amorphous
resins (ABS, PC, PS) and under 15% for semi-crystalline resins (PA, PP, PBT,
POM), which shrink more.

Most consumer and industrial parts land at 2.0–3.0 mm (0.080–0.120 in).
High-flow resins such as PP, POM and LCP hold 1.0–1.5 mm; low-flow resins
such as PC and glass-filled nylon generally want 2.0–3.0 mm.

Draft is measured **per side**, from the direction of mold opening. Zero
draft is possible only on shallow, highly polished faces and always risks
drag marks and ejector damage. The texture rule that matters: add **1–1.5°
of draft for every 0.001 in (0.025 mm) of texture depth**, on top of the base
draft — and ask the texture house for the actual depth rather than guessing
from the SPI or MT code. For deep cores over 50 mm of draw, add roughly 0.5°
per additional 25 mm of depth.

All rib and boss ratios below are relative to the **nominal wall thickness
T**.

## Recommended wall thickness by polymer

| Polymer | Min (in) | Max (in) | Min (mm) | Max (mm) | Flow behaviour |
| --- | --- | --- | --- | --- | --- |
| ABS | 0.045 | 0.14 | 1.14 | 3.56 | Medium flow, forgiving |
| Acetal (POM) | 0.03 | 0.12 | 0.76 | 3.05 | High flow, thin walls OK, high shrink |
| Acrylic (PMMA) | 0.025 | 0.5 | 0.64 | 12.7 | Tolerates thick sections — better than PC for thick optics |
| Liquid crystal polymer (LCP) | 0.03 | 0.12 | 0.76 | 3.05 | Very high flow, can go very thin |
| Long-fiber reinforced | 0.075 | 1 | 1.9 | 25.4 | Thick sections; strong but warp-prone |
| Nylon (PA) | 0.03 | 0.115 | 0.76 | 2.92 | High flow unfilled; glass-filled needs thicker walls |
| Polycarbonate (PC) | 0.04 | 0.15 | 1.02 | 3.81 | Low flow, needs thicker walls and short flow paths |
| Polyester (PBT / PET) | 0.025 | 0.125 | 0.64 | 3.17 | Medium-high flow, high shrink |
| Polyethylene (PE) | 0.03 | 0.2 | 0.76 | 5.08 | High flow, high shrink |
| Polyphenylene sulfide (PPS) | 0.02 | 0.18 | 0.51 | 4.57 | High flow, very thin capable |
| Polypropylene (PP) | 0.025 | 0.15 | 0.64 | 3.81 | High flow, thin walls OK, high shrink |
| Polystyrene (PS) | 0.035 | 0.15 | 0.89 | 3.81 | Easy flow, low shrink |
| Polyurethane (PU) | 0.08 | 0.75 | 2.03 | 19.05 | Thick sections, elastomeric |

## Draft angle by surface condition (degrees per side)

| Surface condition | Min draft (°) | Max draft (°) |
| --- | --- | --- |
| Shutoff / shallow feature (<1mm) | 0.25 | 0.5 |
| SPI A-1/A-2/A-3 diamond polish | 1.5 | 2 |
| SPI B-1/B-2/B-3 paper finish | 1 | 2 |
| SPI C-1/C-2/C-3 stone finish | 1.5 | 2.5 |
| General default, untextured | 1 | 2 |
| SPI D-1/D-2/D-3 dry blast | 2 | 3 |
| Light texture (MT-11010/11020) | 3 | 3 |
| Medium texture (MT-11030) | 3 | 5 |
| Heavy texture / leather grain | 5 | 7 |
| Ribs and internal bosses | 0.5 | 1.5 |

## Rib, boss and corner ratios (relative to nominal wall T)

| Feature | Guideline | Why |
| --- | --- | --- |
| Rib thickness at base | 0.5 – 0.6 × T | Above 0.6 T a sink mark shows on the opposite face |
| Rib thickness (textured or cosmetic face) | 0.4 – 0.5 × T | Extra margin against sink |
| Rib height | ≤ 3 × T | Taller ribs are hard to fill and to eject |
| Rib spacing, center to center | ≥ 2 × T | Closer spacing traps heat and creates a thick section |
| Rib base fillet | 0.25 – 0.5 × T | Smaller cracks; larger creates a thick spot |
| Rib draft | 0.5 – 1.5° per side | Needed to eject the rib |
| Boss outside diameter | 2 – 2.5 × screw major diameter | Resists hoop stress from the screw |
| Boss wall thickness | 0.5 – 0.6 × T | Same sink rule as ribs |
| Boss standoff from a wall | Connect with a rib or gusset, not a thick blend | Avoids a heavy junction that sinks |
| Internal corner radius | ≥ 0.5 × T | Sharp internal corners are the number one stress riser |
| External corner radius | Internal radius + T | Keeps the wall uniform through the corner |
| Hole to edge, hole to hole | ≥ 2 × hole diameter | Avoids weak, hard-to-fill webs |

## Notes

**Polishing does not reliably reduce the draft requirement.** Some sources
say an SPI A polished surface can go as low as 0.5°; others state a 1.5–2°
minimum for SPI A-1/A-2. The higher figure is used here, because a high
polish increases vacuum adhesion between part and core. The sub-0.5° case is
listed separately and restricted to shallow features. Under-drafting a deep
polished core is a common and expensive mistake.

**The wall thickness table comes from a single primary source** (Protolabs),
with the millimeter columns computed at 25.4 mm/in. Five rows — ABS, PC, PP,
nylon and acetal — were independently spot-checked and matched. The
remaining eight rows (acrylic, LCP, long-fiber, polyester, PE, PPS, PS, PU)
are single-source.

**The SPI finish grade to draft mapping is industry practice, not a
standard.** SPI defines the mold finish, not a draft requirement. Confirm
with the moldmaker for the specific texture and draw depth.

**Sink-mark rule of thumb:** any local section thicker than about 60% of the
nominal wall will show on the opposite surface of a cosmetic face. Core it
out or rib it instead of leaving solid mass.

The rib thickness ratio of 0.5–0.6 × T was confirmed by two sources. The
tightened 0.4–0.5 × T figure for cosmetic and textured faces is common shop
practice rather than a published standard, as are the rib height, spacing,
fillet, boss and corner-radius ratios.

## FAQ

### What wall thickness should I use for an injection molded part?

Most parts land at 2.0–3.0 mm. ABS runs 1.14–3.56 mm, polycarbonate 1.02–3.81 mm and polypropylene 0.64–3.81 mm. The stronger rule is uniformity: hold variation within 25% of nominal for amorphous resins and 15% for semi-crystalline ones.

### How much draft angle does an injection molded part need?

1–2° per side is the general default on an untextured wall. A dry-blast SPI D finish wants 2–3°, light texture wants at least 3°, and heavy leather grain wants 5–7° or more. Ribs and internal bosses can run 0.5–1.5°.

### How much extra draft does texture need?

Add 1–1.5° for every 0.001 in (0.025 mm) of texture depth, on top of the base draft. Ask the texture house for the actual depth rather than inferring it from the SPI or MT code.

### How thick should a rib be?

0.5–0.6 × the nominal wall at the base, dropping to 0.4–0.5 × on a textured or cosmetic face. Anything above 0.6 × will show as a sink mark on the opposite surface. Keep rib height at or below 3 × wall and spacing at 2 × wall or more.

### What causes sink marks?

A local section thicker than roughly 60% of the nominal wall. It cools last, shrinks after the surface has frozen, and pulls the opposite face inward. The fix is to core the thick section out and replace it with ribs.

### How big should a boss be for a self-tapping screw?

Outside diameter 2–2.5 × the screw's major diameter, with a boss wall of 0.5–0.6 × the nominal part wall. Connect the boss to a nearby wall with a rib or gusset rather than a thick blend, which would create a heavy junction that sinks.

### What internal corner radius should I use?

At least 0.5 × the nominal wall thickness. Sharp internal corners are the leading stress riser in a molded part and also restrict melt flow. Make the external radius the internal radius plus one wall thickness so the wall stays uniform through the corner.

## Sources

- [Protolabs — Improving Part Design with Uniform Wall Thickness (Plastic Wall Thickness Chart)](https://www.protolabs.com/resources/design-tips/improving-part-design-with-uniform-wall-thickness/)
- [3D DFM — Injection Molding Design Guidelines: Wall Thickness, Ribs & Draft](https://3ddfm.com/)
- [Evok Polymer — Injection Molding Draft Angle: Complete Design Guide](https://evokpoly.com/)

## Related processes

- [Injection Molding](https://manufacturingprocesses.org/processes/forming/injection-molding.md)
- [Overmolding](https://manufacturingprocesses.org/processes/forming/overmolding.md)
- [Insert Molding](https://manufacturingprocesses.org/processes/forming/insert-molding.md)
- [Compression Molding](https://manufacturingprocesses.org/processes/forming/compression-molding.md)
- [Liquid Silicone Rubber (LSR) Molding](https://manufacturingprocesses.org/processes/forming/liquid-silicone-rubber-lsr-molding.md)

## Related charts

- [Surface Finish Chart](https://manufacturingprocesses.org/charts/surface-finish-chart.md)
- [Common Engineering Material Properties](https://manufacturingprocesses.org/charts/material-properties.md)

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/charts/injection-molding-design-guidelines)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: chart
title: "Melting Points of Metals Chart"
tables: 1
url: https://manufacturingprocesses.org/charts/metal-melting-points
---

# Melting Points of Metals Chart

Melting points and melting ranges in °C and °F for 40 pure metals and engineering alloys, with solidus and liquidus given separately where they differ.

Pure metals melt at a point. **Alloys melt over a range**, from the solidus
(where melting starts) to the liquidus (where it finishes). Where the low
and high figures below differ, that is the melting range; where they are
equal, the source gives a single value.

That distinction matters in practice. Aluminum 7075 has a solidus of 480 °C
against 6061's 580 °C, which is why 7075 is far more prone to incipient
melting during solution heat treatment and is not practically weldable.

Fahrenheit columns are computed as °C × 9/5 + 32, not transcribed. Elemental
rows are cross-checked against CRC values; alloy rows come primarily from
engineering property tables.

## Melting points and melting ranges of metals and alloys

| Metal / alloy | Solidus (°C) | Liquidus (°C) | Solidus (°F) | Liquidus (°F) | Type | Note |
| --- | --- | --- | --- | --- | --- | --- |
| Aluminum, pure (1100) | 660 | 660 | 1220 | 1220 | Element | Pure Al melts at 660.3 °C (CRC) |
| Aluminum alloys (general) | 463 | 671 | 865 | 1240 | Alloy | Wide range; check the specific alloy |
| Aluminum 6061 | 580 | 650 | 1076 | 1202 | Alloy | Solidus 580 / liquidus 650 |
| Aluminum 7075 | 480 | 640 | 896 | 1184 | Alloy | Solidus 480 / liquidus 640 — note the low solidus |
| Aluminum bronze | 1027 | 1038 | 1881 | 1900 | Alloy | — |
| Brass, yellow (C270 / C268) | 905 | 932 | 1661 | 1710 | Alloy | — |
| Brass, red (C230) | 990 | 1025 | 1814 | 1877 | Alloy | — |
| Brass, free-cutting (C360) | 885 | 900 | 1625 | 1652 | Alloy | Cu-Zn-Pb; lower than plain brass |
| Bronze, manganese | 865 | 890 | 1589 | 1634 | Alloy | — |
| Beryllium copper | 865 | 955 | 1589 | 1751 | Alloy | — |
| Copper, pure (C110) | 1084 | 1084 | 1983 | 1983 | Element | 1084.6 °C (CRC) |
| Cupronickel | 1170 | 1240 | 2138 | 2264 | Alloy | — |
| Steel, carbon (general) | 1425 | 1540 | 2597 | 2804 | Alloy | Falls with carbon content |
| Steel, stainless 304 | 1400 | 1450 | 2552 | 2642 | Alloy | Solidus 1400 / liquidus 1450 |
| Steel, stainless (general) | 1510 | 1510 | 2750 | 2750 | Alloy | Engineering ToolBox single figure |
| Iron, wrought | 1482 | 1593 | 2700 | 2899 | Alloy | — |
| Iron, gray cast | 1127 | 1204 | 2061 | 2199 | Alloy | — |
| Iron, ductile | 1149 | 1149 | 2100 | 2100 | Alloy | — |
| Iron, pure | 1538 | 1538 | 2800 | 2800 | Element | CRC value |
| Titanium, pure (Gr 2) | 1668 | 1670 | 3034 | 3038 | Element | CRC 1668; Engineering ToolBox 1670 |
| Titanium Ti-6Al-4V | 1604 | 1660 | 2919 | 3020 | Alloy | Solidus ~1604 / liquidus ~1660 |
| Nickel, pure | 1453 | 1455 | 2647 | 2651 | Element | CRC 1455; Engineering ToolBox 1453 |
| Inconel (600/625 family) | 1390 | 1425 | 2534 | 2597 | Alloy | — |
| Monel 400 | 1300 | 1350 | 2372 | 2462 | Alloy | — |
| Hastelloy C | 1320 | 1350 | 2408 | 2462 | Alloy | — |
| Zinc | 419.5 | 419.5 | 787 | 787 | Element | 419.53 °C (CRC) |
| Zinc die-cast alloy (Zamak 3) | 381 | 387 | 718 | 729 | Alloy | — |
| Magnesium, pure | 650 | 650 | 1202 | 1202 | Element | CRC 650 |
| Magnesium alloys (AZ91 etc.) | 349 | 649 | 660 | 1200 | Alloy | Wide range |
| Lead | 327.5 | 327.5 | 622 | 622 | Element | 327.46 °C (CRC) |
| Tin | 232 | 232 | 450 | 450 | Element | 231.9 °C (CRC) |
| Solder 50/50 Sn-Pb | 183 | 215 | 361 | 419 | Alloy | Eutectic 63/37 melts at 183 °C |
| Silver, pure | 961 | 961 | 1762 | 1762 | Element | 961.8 °C (CRC) |
| Gold, 24K | 1063 | 1064 | 1945 | 1947 | Element | 1064.2 °C (CRC) |
| Chromium | 1860 | 1907 | 3380 | 3465 | Element | ⚠ Engineering ToolBox 1860; CRC/modern 1907 — large gap |
| Molybdenum | 2620 | 2623 | 4748 | 4753 | Element | CRC 2623 |
| Tungsten | 3400 | 3422 | 6152 | 6192 | Element | CRC 3422 |
| Cobalt | 1495 | 1495 | 2723 | 2723 | Element | CRC 1495 |
| Manganese | 1244 | 1246 | 2271 | 2275 | Element | CRC 1246 |
| Platinum | 1768 | 1770 | 3214 | 3218 | Element | CRC 1768 |

## Notes

**Two sources disagree on stainless steel by 60–110 °C.** One engineering
table gives a single figure of 1510 °C for "stainless steel"; per-alloy data
gives 1400–1450 °C for annealed 304. Both rows are shown and labelled.
1400–1450 °C is the correct range for austenitic 304; 1510 °C is closer to a
ferritic or martensitic grade, or is a generic figure. **Use the
grade-specific row, not the generic one.**

**Chromium carries a genuine 47 °C conflict.** One widely used engineering
table publishes 1860 °C; the modern accepted value (CRC, IUPAC) is 1907 °C.
The 1860 figure is an older measurement that persists in many engineering
tables. Both are shown in the row.

**Smaller element-level differences are resolved in favour of CRC** and noted
inline: titanium (1670 versus CRC 1668), nickel (1453 versus 1455), tungsten
(3400 versus 3422), molybdenum (2620 versus 2623), manganese (1244 versus
1246) and platinum (1770 versus 1768). All are within about 1% and none
changes a process decision, but the differences are recorded because this
chart publishes exact numbers.

**Alloy rows verified against a second source:** 6061 (580/650 °C), 7075
(480/640 °C) and 304 stainless (1400/1450 °C) all matched.

**Single-source rows, treat as indicative:** aluminum bronze, the brasses,
manganese bronze, beryllium copper, cupronickel, wrought/gray/ductile iron,
Inconel, Monel, Hastelloy C, magnesium alloys, Zamak, solder, C360 brass and
the Ti-6Al-4V solidus/liquidus.

## FAQ

### What is the melting point of aluminum?

Pure aluminum (1100) melts at 660 °C (1220 °F). Alloys melt lower and over a range: 6061 runs 580–650 °C and 7075 runs 480–640 °C. The alloy family as a whole spans roughly 463–671 °C.

### What is the melting point of steel?

Carbon steel melts over roughly 1425–1540 °C (2597–2804 °F), falling as carbon content rises. Annealed 304 stainless melts at 1400–1450 °C. Pure iron melts at 1538 °C.

### Why do alloys have a melting range instead of a melting point?

Because the constituents do not all change state at the same temperature. Melting begins at the solidus and finishes at the liquidus, and between the two the alloy is a mush of solid and liquid. Only pure elements and eutectic compositions melt at a single temperature.

### Which metal in this chart has the highest melting point?

Tungsten, at 3400–3422 °C (6152–6192 °F). Molybdenum is next at 2620–2623 °C, then platinum at 1768–1770 °C.

### What is the melting point of copper and brass?

Pure copper (C110) melts at 1084 °C (1983 °F). Brasses melt considerably lower: free-cutting C360 at 885–900 °C, yellow brass at 905–932 °C and red brass at 990–1025 °C.

### Why does 7075 aluminum have such a low solidus?

Its zinc, magnesium and copper content creates low-melting phases, giving a solidus of 480 °C against 580 °C for 6061. That narrow margin below the solution treatment temperature is one reason 7075 is prone to incipient melting and is not practically weldable.

### What temperature does solder melt at?

50/50 tin-lead solder melts over 183–215 °C. The 63/37 eutectic composition melts at a single temperature of 183 °C, which is why it is preferred for electronics.

## Sources

- [Engineering ToolBox — Metals and Alloys: Melting Temperatures](https://www.engineeringtoolbox.com/melting-temperature-metals-d_860.html)
- [MakeItFrom — 6061-T6 Aluminum material properties](https://www.makeitfrom.com/material-properties/6061-T6-Aluminum)
- [MakeItFrom — 7075-T6 Aluminum material properties](https://www.makeitfrom.com/material-properties/7075-T6-Aluminum)
- [MakeItFrom — Annealed 304 Stainless Steel material properties](https://www.makeitfrom.com/material-properties/Annealed-304-Stainless-Steel)
- CRC / IUPAC standard pure-element melting points

## Related processes

- [Sand Casting](https://manufacturingprocesses.org/processes/forming/sand-casting.md)
- [Die Casting](https://manufacturingprocesses.org/processes/forming/die-casting.md)
- [Investment Casting](https://manufacturingprocesses.org/processes/forming/investment-casting.md)
- [Soldering and Brazing](https://manufacturingprocesses.org/processes/joining/soldering-and-brazing.md)
- [Arc Welding](https://manufacturingprocesses.org/processes/joining/arc-welding.md)
- [Heat Treatment](https://manufacturingprocesses.org/processes/forming/heat-treatment.md)

## Related charts

- [Common Engineering Material Properties](https://manufacturingprocesses.org/charts/material-properties.md)

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/charts/metal-melting-points)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: chart
title: "Sheet Metal Bend Radius & K-Factor Chart"
tables: 4
url: https://manufacturingprocesses.org/charts/sheet-metal-bend-radius-k-factor
---

# Sheet Metal Bend Radius & K-Factor Chart

Minimum inside bend radius by alloy and temper as multiples of thickness, K-factor against R/t, and K by bending method — with the two conflicting published K sets shown side by side.

All radii on this page are **inside radius, on a 90° bend, at room
temperature, bending across the grain**, expressed as a multiple of material
thickness t.

Two radius columns are given because the sources genuinely disagree, and the
disagreement is meaningful. **Absolute min** is the crack threshold reported
by fabricators: achievable, with no margin. **Design default** is what to put
on a drawing — it adds margin for sheet thickness tolerance, lot-to-lot
ductility variation and the punch tips a shop actually owns.

The **K-factor** is the position of the neutral axis: distance from the
inside surface to the neutral axis, divided by thickness. It feeds the bend
allowance, BA = π · (R + K·t) · A / 180, with A in degrees. K can never
exceed 0.5 — that would put the neutral axis past mid-thickness, which is
physically impossible — and in practice it runs 0.30 to 0.50.

## Minimum inside bend radius by material (90°, across grain, multiples of thickness t)

| Material | Absolute min | Design default | Representative K-factor |
| --- | --- | --- | --- |
| Aluminum 1100-O / 3003-O | 0 × t | 0.5 × t | 0.38 |
| Aluminum 3003-H14 | 0.5 × t | 1 × t | 0.4 |
| Aluminum 5052-O | 0 × t | 0.5 × t | 0.4 |
| Aluminum 5052-H32 | 0.5 × t | 1 × t | 0.42 |
| Aluminum 6061-T4 | 1 × t | 1.5 × t | 0.41 |
| Aluminum 6061-T6 | 2 × t | 2.5 × t | 0.4 |
| Aluminum 6063-T5 | 1.5 × t | 2 × t | 0.41 |
| Aluminum 7075-T6 | 4 × t | 5 × t | 0.33 |
| Mild steel / CRS 1008-1018 | 0.5 × t | 1 × t | 0.41 |
| Hot-rolled steel | 1 × t | 1.5 × t | 0.41 |
| Galvanized steel G90 | 1 × t | 1.5 × t | 0.41 |
| HSLA 50 ksi | 1 × t | 1.5 × t | 0.42 |
| DP590 | 1.5 × t | 2 × t | 0.43 |
| DP780 | 3 × t | 4 × t | 0.45 |
| DP1180 | 5 × t | 6 × t | 0.45 |
| Stainless 304 / 304L | 0.5 × t | 1 × t | 0.38 |
| Stainless 316 / 316L | 0.5 × t | 1 × t | 0.38 |
| Stainless 430 | 1 × t | 1.5 × t | 0.4 |
| Copper C110 soft | 0 × t | 0.5 × t | 0.35 |
| Copper half-hard | 0.5 × t | 1 × t | 0.38 |
| Brass C260 | 0.5 × t | 1 × t | 0.38 |
| Titanium Grade 2 | 2.5 × t | 3 × t | 0.4 |

## K-factor against R/t — the dominant relationship

| R/t | K-factor |
| --- | --- |
| 0.1 | 0.21 |
| 0.2 | 0.22 |
| 0.3 | 0.23 |
| 0.4 | 0.24 |
| 0.5 | 0.25 |
| 0.6 | 0.26 |
| 0.7 | 0.27 |
| 0.8 | 0.3 |
| 1 | 0.31 |
| 1.2 | 0.33 |
| 1.5 | 0.36 |
| 2 | 0.37 |
| 2.5 | 0.4 |
| 3 | 0.42 |
| 5 | 0.46 |
| 75 | 0.5 |

## Typical K-factor by material — two published sets that disagree

| Material | Machinery's Handbook, 90° tables | Fabricator practical values |
| --- | --- | --- |
| Soft brass, soft copper | 0.35 | 0.35–0.38 |
| Hard brass/copper, mild steel, aluminum | 0.41 | — |
| Aluminum 5052 (soft) | — | 0.42 |
| Aluminum 6061-T6 | — | 0.4 |
| Cold-rolled steel | 0.45 | 0.38 |
| Stainless steel | — | 0.38 |
| Hard brass, bronze, spring steel | 0.45 | — |
| Hardened steel, 7075 | — | 0.33 |

## K-factor by bending method

| Method | K min | K max |
| --- | --- | --- |
| Air bending | 0.33 | 0.42 |
| Bottoming | 0.4 | 0.45 |
| Coining | 0.45 | 0.5 |

## Notes

**The dominant variable for K is R/t, not the material.** The neutral axis
shifts inward as the bend gets tighter, from K ≈ 0.21 at R/t = 0.1 to the
0.50 asymptote at very large R/t. Pick K from R/t first, then calibrate
against a test bend on the actual stock and tooling. A 0.05 error in K
produces roughly 0.3 mm of deviation per bend on 3 mm stock — four bends in
series and the part is 1.2 mm off.

**K-factor by material is genuinely contested.** *Machinery's Handbook* gives
cold-rolled steel K = 0.45; production fabricators use 0.38 for the same
material. Both are published in good faith: the difference is bending method
(bottoming and coining versus air bending) and R/t. Both columns are shown
rather than averaged.

**Minimum bend radius is the other place the sources conflict.** For mild
steel one fabricator source says 0.5 t while two design guides say 1.0 t;
for 304 stainless the spread is 0.5 t to 1.5 t; for copper it runs from
0 t (soft) to 1.0 t. Rather than average — which would produce a number no
source supports and no fabricator would honour — both positions are shown.
The one place a design guide was rejected is 6061-T6, where it claims 1.5 t
against 2.0 t from two other sources; 8–10% elongation makes 2.0 t the
defensible figure.

**Terminology trap.** Some guides label the bend-radius multiplier itself
"k factor" (min radius = k × t). That is not the K-factor. The K-factor is
the neutral-axis ratio and is always ≤ 0.5. The two are unrelated quantities
that happen to share a letter.

**Grain direction:** bending parallel to the rolling direction stretches the
weak axis. Add 50–100% to the minimum radius for with-grain bends, or rotate
the flat pattern. **Feature clearance:** keep holes and slots at least
2.5 × t + bend radius from the bend line, measured to the near edge, or they
distort. **Springback:** expect 1–3°, so a 90° bend is typically overbent to
about 87° in cold-rolled steel or 88° in 5052.

**Thickness matters too.** Minimum radius grows with thickness even in the
same alloy. A working rule for steel: up to about 6 mm use 0.8–1.0 t, 6–12 mm
use about 1.2 t, above 12 mm expect 1.5 t or more. And the bend takes the
punch tip radius plus springback correction — if the theoretical minimum is
1.0 mm but the smallest tip on the floor is 1.5 mm, you get 1.5 mm.

## FAQ

### What is the minimum bend radius for 6061-T6 aluminum?

2.0 × t as an absolute minimum and 2.5 × t as a design default. 6061-T6 has only 8–10% elongation and cracks easily. If you need a tighter radius, bend in the T4 or O temper and age afterwards, or switch to 5052-H32, which bends at 0.5 × t.

### What is a good default K-factor?

Pick it from R/t rather than from the material: about 0.31 at R/t = 1.0, 0.36 at 1.5 and 0.42 at 3.0. If you have to pick blind, 0.40 to 0.42 is a reasonable air-bending default — then calibrate against a test bend.

### Can the K-factor be greater than 0.5?

No. K is the distance from the inside surface to the neutral axis divided by thickness, so K > 0.5 would put the neutral axis past mid-thickness. In practice it runs 0.30 to 0.50, approaching 0.50 only at very large bend radii.

### What is the bend allowance formula?

BA = π · (R + K·t) · A / 180, where R is the inside radius, t the material thickness, K the K-factor and A the bend angle in degrees.

### How far should a hole be from a bend line?

At least 2.5 × t plus the bend radius, measured from the bend line to the near edge of the hole. Closer than that and the hole distorts into an oval as the material stretches through the bend.

### Does grain direction change the minimum bend radius?

Yes, substantially. Bending parallel to the rolling direction stretches the weak axis; add 50–100% to the minimum radius for a with-grain bend, or rotate the flat pattern so the bend runs across the grain.

### How much springback should I allow?

One to three degrees for common sheet metals, so a 90° bend is typically overbent to roughly 87° in cold-rolled steel or 88° in 5052 aluminum. Higher-strength alloys spring back more.

## Sources

- [PrecisionSmith — Sheet Metal Bend Radius Chart](https://precisionsmith.com/guides/sheet-metal-bend-radius-chart/)
- [MachineMFG — K Factor Calculator for Sheet Metal Bending](https://www.machinemfg.com/k-factor-calculator/)
- [DurmaPress — Minimum Bend Radius for Sheet Metal: Chart, Formula & Design Guide](https://www.durmapress.com/minimum-bend-radius-for-sheet-metal-chart-formula-design-guide/)

## Related processes

- [Press Braking](https://manufacturingprocesses.org/processes/forming/press-braking.md)
- [Sheet Metal Fabrication](https://manufacturingprocesses.org/processes/forming/sheet-metal-fabrication.md)
- [Laser Cutting](https://manufacturingprocesses.org/processes/cutting/laser-cutting.md)
- [Roll Forming](https://manufacturingprocesses.org/processes/forming/roll-forming.md)

## Related charts

- [Sheet Metal Gauge Chart](https://manufacturingprocesses.org/charts/sheet-metal-gauge-chart.md)
- [Common Engineering Material Properties](https://manufacturingprocesses.org/charts/material-properties.md)

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/charts/sheet-metal-bend-radius-k-factor)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: chart
title: "Sheet Metal Gauge Chart"
tables: 5
url: https://manufacturingprocesses.org/charts/sheet-metal-gauge-chart
---

# Sheet Metal Gauge Chart

Gauge-to-thickness conversions for steel, galvanized steel, stainless and aluminum, gauges 3 to 30, in inches and millimeters — plus the mill tolerance band.

Sheet metal gauge is not a unit. It is an index into a table, and there are
four different tables in daily use: carbon steel follows the Manufacturers'
Standard Gauge (MSG), which is weight-derived — 16 ga is 2.5 lb/ft².
Galvanized sheet uses the steel MSG value plus 0.0037 in for the zinc
coating. Stainless follows the fraction-based US Standard Gauge, in exact
1/64 in steps from gauge 3 to 14 and 1/128 in steps from 15 to 30.
Aluminum, brass and copper follow the Brown & Sharpe (AWG) geometric
series, ratio 92^(1/39) per step.

The result is that one gauge number means four different thicknesses.
18 ga steel is 0.0478 in, 18 ga galvanized is 0.0516 in, 18 ga stainless is
0.0500 in and 18 ga aluminum is 0.0403 in — a 28% spread. In every
standard, a larger gauge number means thinner material.

Millimeter columns here are computed at exactly 25.4 mm/in and rounded to
three decimals, not transcribed, so they carry none of the rounding errors
that circulate in published charts.

## Carbon steel — Manufacturers' Standard Gauge (MSG)

| Gauge | Thickness (in) | Thickness (mm) |
| --- | --- | --- |
| 3 | 0.2391 | 6.073 |
| 4 | 0.2242 | 5.695 |
| 5 | 0.2092 | 5.314 |
| 6 | 0.1943 | 4.935 |
| 7 | 0.1793 | 4.554 |
| 8 | 0.1644 | 4.176 |
| 9 | 0.1495 | 3.797 |
| 10 | 0.1345 | 3.416 |
| 11 | 0.1196 | 3.038 |
| 12 | 0.1046 | 2.657 |
| 13 | 0.0897 | 2.278 |
| 14 | 0.0747 | 1.897 |
| 15 | 0.0673 | 1.709 |
| 16 | 0.0598 | 1.519 |
| 17 | 0.0538 | 1.367 |
| 18 | 0.0478 | 1.214 |
| 19 | 0.0418 | 1.062 |
| 20 | 0.0359 | 0.912 |
| 21 | 0.0329 | 0.836 |
| 22 | 0.0299 | 0.759 |
| 23 | 0.0269 | 0.683 |
| 24 | 0.0239 | 0.607 |
| 25 | 0.0209 | 0.531 |
| 26 | 0.0179 | 0.455 |
| 27 | 0.0164 | 0.417 |
| 28 | 0.0149 | 0.378 |
| 29 | 0.0135 | 0.343 |
| 30 | 0.012 | 0.305 |

## Galvanized steel — Galvanized Sheet Gauge (steel MSG + 0.0037 in)

| Gauge | Thickness (in) | Thickness (mm) |
| --- | --- | --- |
| 8 | 0.1681 | 4.27 |
| 9 | 0.1532 | 3.891 |
| 10 | 0.1382 | 3.51 |
| 11 | 0.1233 | 3.132 |
| 12 | 0.1084 | 2.753 |
| 13 | 0.0934 | 2.372 |
| 14 | 0.0785 | 1.994 |
| 15 | 0.071 | 1.803 |
| 16 | 0.0635 | 1.613 |
| 17 | 0.0575 | 1.46 |
| 18 | 0.0516 | 1.311 |
| 19 | 0.0456 | 1.158 |
| 20 | 0.0396 | 1.006 |
| 21 | 0.0366 | 0.93 |
| 22 | 0.0336 | 0.853 |
| 23 | 0.0306 | 0.777 |
| 24 | 0.0276 | 0.701 |
| 25 | 0.0247 | 0.627 |
| 26 | 0.0217 | 0.551 |
| 27 | 0.0202 | 0.513 |
| 28 | 0.0187 | 0.475 |
| 29 | 0.0172 | 0.437 |
| 30 | 0.0157 | 0.399 |

## Stainless steel — US Standard Gauge (fraction-based)

| Gauge | Thickness (in) | Thickness (mm) |
| --- | --- | --- |
| 7 | 0.1875 | 4.762 |
| 8 | 0.1719 | 4.366 |
| 9 | 0.1563 | 3.97 |
| 10 | 0.1406 | 3.571 |
| 11 | 0.125 | 3.175 |
| 12 | 0.1094 | 2.779 |
| 13 | 0.0938 | 2.383 |
| 14 | 0.0781 | 1.984 |
| 15 | 0.0703 | 1.786 |
| 16 | 0.0625 | 1.587 |
| 17 | 0.0563 | 1.43 |
| 18 | 0.05 | 1.27 |
| 19 | 0.0438 | 1.113 |
| 20 | 0.0375 | 0.952 |
| 21 | 0.0344 | 0.874 |
| 22 | 0.0313 | 0.795 |
| 23 | 0.0281 | 0.714 |
| 24 | 0.025 | 0.635 |
| 25 | 0.0219 | 0.556 |
| 26 | 0.0188 | 0.478 |
| 27 | 0.0172 | 0.437 |
| 28 | 0.0156 | 0.396 |
| 29 | 0.0141 | 0.358 |
| 30 | 0.0125 | 0.318 |

## Aluminum, brass and copper — Brown & Sharpe (AWG)

| Gauge | Thickness (in) | Thickness (mm) |
| --- | --- | --- |
| 3 | 0.2294 | 5.827 |
| 4 | 0.2043 | 5.189 |
| 5 | 0.1819 | 4.62 |
| 6 | 0.162 | 4.115 |
| 7 | 0.1443 | 3.665 |
| 8 | 0.1285 | 3.264 |
| 9 | 0.1144 | 2.906 |
| 10 | 0.1019 | 2.588 |
| 11 | 0.0907 | 2.304 |
| 12 | 0.0808 | 2.052 |
| 13 | 0.072 | 1.829 |
| 14 | 0.0641 | 1.628 |
| 15 | 0.0571 | 1.45 |
| 16 | 0.0508 | 1.29 |
| 17 | 0.0453 | 1.151 |
| 18 | 0.0403 | 1.024 |
| 19 | 0.0359 | 0.912 |
| 20 | 0.032 | 0.813 |
| 21 | 0.0285 | 0.724 |
| 22 | 0.0253 | 0.643 |
| 23 | 0.0226 | 0.574 |
| 24 | 0.0201 | 0.511 |
| 25 | 0.0179 | 0.455 |
| 26 | 0.0159 | 0.404 |
| 27 | 0.0142 | 0.361 |
| 28 | 0.0126 | 0.32 |
| 29 | 0.0113 | 0.287 |
| 30 | 0.01 | 0.254 |

## Mill thickness tolerance — cold-rolled steel sheet

| Gauge | Nominal (in) | Max (in) | Min (in) | Total band (in) |
| --- | --- | --- | --- | --- |
| 10 | 0.1345 | 0.1405 | 0.1285 | 0.012 |
| 11 | 0.1196 | 0.1256 | 0.1136 | 0.012 |
| 12 | 0.1046 | 0.1106 | 0.0986 | 0.012 |
| 14 | 0.0747 | 0.0797 | 0.0697 | 0.01 |
| 16 | 0.0598 | 0.0648 | 0.0548 | 0.01 |
| 18 | 0.0478 | 0.0518 | 0.0438 | 0.008 |
| 20 | 0.0359 | 0.0389 | 0.0329 | 0.006 |
| 22 | 0.0299 | 0.0329 | 0.0269 | 0.006 |
| 24 | 0.0239 | 0.0269 | 0.0209 | 0.006 |
| 26 | 0.0179 | 0.0199 | 0.0159 | 0.004 |
| 28 | 0.0149 | 0.0169 | 0.0129 | 0.004 |

## Notes

**Dimension in decimals, not in gauge.** The mill tolerance band on
cold-rolled steel sheet is frequently wider than the step between adjacent
gauges: 16 ga is nominally 0.0598 in but ships anywhere from 0.0548 to
0.0648 in, a 0.010 in band, while the gap from 16 ga to 15 ga is only
0.0075 in. A drawing that reads "16 GA" without naming the alloy can
legitimately be built to three different thicknesses.

**Aluminum values published by Engineers Edge contain errors.** Its second
aluminum table lists gauge 26 as 0.017 in; the Brown & Sharpe series and
Metal Supermarkets both give 0.01594 in, which rounds to 0.016. Gauges 17,
21 and 23 are also rounded to 2–3 decimals in that table (0.045, 0.028 and
0.023 against the exact 0.0453, 0.0285 and 0.0226). This chart uses the
exact Brown & Sharpe series throughout.

**Stainless gauges 3–6 and galvanized gauges 3–7 are genuinely absent, not
missing.** Neither source lists them: stainless in that thickness is sold as
plate in fractional sizes, and the galvanized gauge series starts at 8.

**Internal check on the galvanized column:** every galvanized value equals
the steel MSG value plus 0.0037 in to within 0.0001 in across all 23 gauges,
which independently corroborates both source charts.

## FAQ

### How thick is 16 gauge steel?

16 gauge carbon steel is 0.0598 in (1.519 mm) under the Manufacturers' Standard Gauge. The same gauge number is 0.0635 in in galvanized steel, 0.0625 in in stainless and 0.0508 in in aluminum, so the material has to be stated alongside the gauge.

### Is 16 gauge thicker than 18 gauge?

Yes. Gauge numbers run backwards: the larger the number, the thinner the sheet. 16 ga steel is 0.0598 in and 18 ga steel is 0.0478 in, so 16 ga is about 25% thicker.

### Why is aluminum gauge different from steel gauge?

They come from unrelated standards. Steel sheet uses the Manufacturers' Standard Gauge, which was derived from weight per square foot, while aluminum follows the Brown & Sharpe wire gauge, a geometric series with a ratio of 92^(1/39) per step. The numbering coincides but the thicknesses do not.

### Why is galvanized sheet thicker than plain steel at the same gauge?

The galvanized gauge series adds a fixed 0.0037 in allowance for the zinc coating on top of the steel MSG value. A 16 ga galvanized sheet is 0.0635 in against 0.0598 in for 16 ga plain steel.

### Should I specify gauge or thickness on a drawing?

Specify thickness with units. Gauge is ambiguous across material families, and on cold-rolled steel the mill tolerance band — 0.010 in at 16 gauge — is wider than the step to the next gauge, so the gauge number alone does not pin the thickness down.

### What is the thinnest gauge in these tables?

Gauge 30, which is 0.0120 in (0.305 mm) in steel, 0.0157 in galvanized, 0.0125 in stainless and 0.0100 in aluminum.

## Sources

- [Engineers Edge — Sheet Metal Wire Gauge Sizes Data Chart](https://www.engineersedge.com/gauge.htm)
- [Metal Supermarkets — Sheet Metal Gauge Chart](https://www.metalsupermarkets.com/sheet-metal-gauge-chart/)

## Related processes

- [Press Braking](https://manufacturingprocesses.org/processes/forming/press-braking.md)
- [Sheet Metal Fabrication](https://manufacturingprocesses.org/processes/forming/sheet-metal-fabrication.md)
- [Laser Cutting](https://manufacturingprocesses.org/processes/cutting/laser-cutting.md)
- [Punching and Blanking](https://manufacturingprocesses.org/processes/cutting/punching-and-blanking.md)

## Related charts

- [Sheet Metal Bend Radius & K-Factor Chart](https://manufacturingprocesses.org/charts/sheet-metal-bend-radius-k-factor.md)
- [Common Engineering Material Properties](https://manufacturingprocesses.org/charts/material-properties.md)

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/charts/sheet-metal-gauge-chart)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: chart
title: "Surface Finish Chart"
tables: 3
url: https://manufacturingprocesses.org/charts/surface-finish-chart
---

# Surface Finish Chart

ISO 1302 N-grades with Ra, RMS, Rz and Rt equivalents, the conversion factors between roughness parameters, and the Ra range each manufacturing process holds.

Surface roughness is specified against ISO 1302 (N-grades), ISO 4287 (the
Ra and Rz parameters), ASME B46.1 and ASME Y14.36 (drawing symbols). Each
N-grade step doubles Ra: N1 is 0.025 µm, N6 is 0.8 µm, N12 is 50 µm.

The RMS and Rz columns here are **computed, not transcribed** — RMS as
Ra × 1.11 and Rz as the shop-rule Ra × 4. That is deliberate, and the reason
is the most important item on this page (see the notes).

**Ra and Rz measure genuinely different things and cannot be converted
exactly.** Ra averages the whole profile; Rz is a peak-to-valley measure and
is far more sensitive to isolated deep scratches. A surface can pass an Ra
spec and fail an Rz spec badly. For sealing and fatigue-critical surfaces,
specify Rz — or both — rather than Ra alone. ASME Y14.36 prefers an explicit
Ra value over an N-grade; N-grades appear mostly on older or ISO/JIS
drawings.

## ISO 1302 N-grades with Ra, RMS, Rz and Rt equivalents

| N-grade | Ra (µm) | Ra (µin) | RMS (µin) | Rz approx (µin) | Rt (µm) | Cut-off (in) | Cut-off (mm) |
| --- | --- | --- | --- | --- | --- | --- | --- |
| N1 | 0.025 | 1 | 1.1 | 4 | 0.3 | 0.003 | 0.08 |
| N2 | 0.05 | 2 | 2.2 | 8 | 0.5 | 0.01 | 0.25 |
| N3 | 0.1 | 4 | 4.4 | 16 | 0.8 | 0.01 | 0.25 |
| N4 | 0.2 | 8 | 8.9 | 32 | 1.2 | 0.01 | 0.25 |
| N5 | 0.4 | 16 | 17.8 | 64 | 2 | 0.01 | 0.25 |
| N6 | 0.8 | 32 | 35.5 | 128 | 4 | 0.03 | 0.8 |
| N7 | 1.6 | 63 | 69.9 | 252 | 8 | 0.03 | 0.8 |
| N8 | 3.2 | 125 | 138.8 | 500 | 13 | 0.1 | 2.5 |
| N9 | 6.3 | 250 | 277.5 | 1000 | 25 | 0.1 | 2.5 |
| N10 | 12.5 | 500 | 555 | 2000 | 50 | 0.1 | 2.5 |
| N11 | 25 | 1000 | 1110 | 4000 | 100 | 0.3 | 8 |
| N12 | 50 | 2000 | 2220 | 8000 | 200 | 0.3 | 8 |

## Conversions between Ra, RMS, Rz and Rt

| Relationship | Formula | Reliability |
| --- | --- | --- |
| Ra ↔ CLA | Identical — CLA (Center Line Average) is the older term for Ra | Exact |
| µm ↔ µin | Ra (µin) = Ra (µm) × 39.37 (commonly rounded to × 40) | Exact |
| Ra → RMS (Rq) | RMS ≈ Ra × 1.11 | Good for typical machined surfaces; assumes a roughly sinusoidal profile |
| Ra → Rz | Rz ≈ Ra × 4 (shop rule), up to Ra × 7 depending on process | Approximate only — not a conversion |
| Ra → Rz (ISO) | Rz (ISO) ≈ Ra × 7.6 | Approximate |
| Ra → Rt | Rt ≈ Ra × 8.7 | Approximate |
| Ra → Rmax | Rmax ≈ Ra × 8.0 | Approximate |
| Ra → Rp | Rp ≈ Ra × 3.6 | Approximate |

## Typical Ra achievable by manufacturing process

| Process | Ra typical (µin) | Ra typical (µm) | N-grade band | Verified |
| --- | --- | --- | --- | --- |
| Flame / plasma cutting | 500–2000 | 12.5–50 | N10-N12 | Indicative |
| Sand casting | 250–1000 | 6.3–25 | N9-N11 | Indicative |
| Hot rolling | 250–1000 | 6.3–25 | N9-N11 | Indicative |
| Sawing | 250–1000 | 6.3–25 | N9-N11 | Indicative |
| Forging | 125–500 | 3.2–12.5 | N8-N10 | Indicative |
| Rough milling / hogging | 250–500 | 6.3–12.5 | N9-N10 | Two sources |
| Rough turning | 125–250 | 3.2–6.3 | N8-N9 | Two sources |
| Investment casting | 63–250 | 1.6–6.3 | N7-N9 | Indicative |
| Drilling | 63–250 | 1.6–6.3 | N7-N9 | Indicative |
| Laser cutting (cut edge) | 32–250 | 0.8–6.3 | N6-N9 | Two sources |
| Die casting | 32–125 | 0.8–3.2 | N6-N8 | Two sources |
| Cold rolling / drawing | 32–125 | 0.8–3.2 | N6-N8 | Indicative |
| Standard CNC milling (as-machined) | 63–125 | 1.6–3.2 | N7-N8 | Two sources |
| Standard CNC turning | 63–125 | 1.6–3.2 | N7-N8 | Two sources |
| Boring | 32–125 | 0.8–3.2 | N6-N8 | Indicative |
| Reaming | 32–125 | 0.8–3.2 | N6-N8 | Indicative |
| Broaching | 32–125 | 0.8–3.2 | N6-N8 | Indicative |
| EDM (sinker / wire) | 32–250 | 0.8–6.3 | N6-N9 | Indicative |
| Fine milling / fine turning | 16–32 | 0.4–0.8 | N5-N6 | Two sources |
| Surface & cylindrical grinding | 8–63 | 0.2–1.6 | N4-N7 | Two sources |
| Honing | 4–32 | 0.1–0.8 | N3-N6 | Indicative |
| Lapping | 1–16 | 0.025–0.4 | N1-N5 | Two sources |
| Superfinishing / polishing | 1–8 | 0.025–0.2 | N1-N4 | Two sources |

## Notes

**Several well-known chart pages list N6 RMS as 32.5 µin and N7 as 64.3 µin,
contradicting the ×1.11 rule those same pages state.** Applying the rule
gives 32 × 1.11 = **35.5** and 63 × 1.11 = **69.9**. Every other row on those
charts does follow ×1.11 exactly (1→1.1, 2→2.2, 4→4.4, 8→8.8, 125→137.5,
250→275, 500→550, 1000→1100, 2000→2200) — only N6 and N7 break it. The three
charts that print 32.5 and 64.3 are not independent of each other; they
share an original with a transcription error. This chart computes RMS =
Ra × 1.11 for every row, which agrees with the two sources that do not carry
the error.

**The Rz column is labelled approximate on purpose.** It is computed as
Ra × 4, the convention on machining-shop charts, but published factor tables
also give Rz = Ra × 7.2 and Rz(ISO) = Ra × 7.6. These conflict and none of
them is wrong — the true ratio depends on the process. This is the least
deterministic number on the page.

**The process-versus-Ra table is the least tightly verified data here.** The
rows marked verified were independently confirmed by two sources: standard
CNC milling and turning, fine milling and turning, rough milling, grinding,
lapping and superfinishing, laser cut edge and die casting. The remaining
rows — sand casting, hot rolling, forging, sawing, flame cutting,
investment casting, drilling, boring, reaming, broaching, EDM and honing —
come from general references only and should be treated as indicative.

**Cost note:** finishes tighter than Ra 1.6 µm (63 µin) typically add 40–60%
or more to cycle time, and below Ra 0.8 µm (32 µin) usually require a
secondary grinding or polishing operation. Specify the loosest Ra that meets
the function.

## FAQ

### What Ra does standard CNC machining produce?

Ra 63–125 µin (1.6–3.2 µm), which is the N7–N8 band, for both milling and turning. That is the default as-machined finish and costs nothing extra. Fine milling or turning with sharp tools and a light stepover reaches 16–32 µin (0.4–0.8 µm).

### What is N6 in Ra?

N6 is Ra 0.8 µm, which is 32 µin, and an RMS of 35.5 µin. It is achievable by boring, reaming, broaching, cold rolling, die casting or a fine machining pass.

### How do I convert Ra to RMS?

RMS ≈ Ra × 1.11. It is a good approximation for typical machined surfaces because it assumes a roughly sinusoidal profile. Ra 32 µin gives RMS 35.5 µin, not the 32.5 µin printed on several widely copied charts.

### Can I convert Ra to Rz?

Not exactly. Ra averages the whole profile while Rz is a peak-to-valley measure, so the ratio depends on the process. The shop rule is Rz ≈ Ra × 4, but published factor tables also give ×7.2 and ×7.6. If Rz is what matters, specify Rz and measure it.

### What is the difference between Ra and CLA?

Nothing. CLA (Center Line Average) is the older British and ISO term for the same parameter now called Ra.

### When does surface finish start costing real money?

Below Ra 1.6 µm (63 µin) you typically add 40–60% or more to cycle time, and below Ra 0.8 µm (32 µin) you generally need a secondary grinding or polishing operation. Specify the loosest Ra the function tolerates.

### Should I put an N-grade or an Ra value on a drawing?

An explicit Ra value. ASME Y14.36 prefers it, and it is unambiguous. N-grades still appear on older drawings and on ISO/JIS drawings, which is why the mapping is on this page.

## Sources

- [Engineers Edge — Surface Roughness Conversion Chart Tables](https://www.engineersedge.com/manufacturing/surface-roughness-conversion.htm)
- [Engineers Edge — Surface Roughness (Finish) Review and Equations (ASME B46.1 / ISO 4287)](https://www.engineersedge.com/surface_finish.htm)
- [Mactech Inc. — Surface Roughness Conversion Chart (PDF)](https://www.mactechonsite.com/wp-content/uploads/Surface-Roughness-Conversion-Chart.pdf)
- [Rembar — Surface Conversion Chart (PDF)](https://www.rembar.com/wp-content/uploads/2020/01/Rembar-Surface-Conversion-Chart.pdf)
- [PREMSA Industries — Surface Roughness Chart](https://premsaindustries.com/en/resources/surface-finish-chart)
- [RapidDirect — Surface Roughness Chart](https://www.rapiddirect.com/blog/surface-roughness-chart/)

## Related processes

- [CNC Milling](https://manufacturingprocesses.org/processes/cutting/cnc-milling.md)
- [CNC Turning](https://manufacturingprocesses.org/processes/cutting/cnc-turning.md)
- [Grinding, Sanding and Polishing](https://manufacturingprocesses.org/processes/finishing/grinding-sanding-and-polishing.md)
- [Bead Blasting](https://manufacturingprocesses.org/processes/finishing/bead-blasting.md)
- [Electropolishing](https://manufacturingprocesses.org/processes/finishing/electropolishing.md)

## Related charts

- [ISO 286 Fits & Tolerance Chart (Hole Basis)](https://manufacturingprocesses.org/charts/iso-286-fits-tolerances.md)
- [Common Engineering Material Properties](https://manufacturingprocesses.org/charts/material-properties.md)

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/charts/surface-finish-chart)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: chart
title: "Tap Drill Chart"
tables: 2
url: https://manufacturingprocesses.org/charts/tap-drill-chart
---

# Tap Drill Chart

Tap drill sizes for UNC and UNF threads from #0 to 1/2 in at ~75% thread engagement, and for metric coarse threads M1.6 to M12 by the D−P rule.

A tap drill is sized to leave a chosen percentage of full thread depth in
the hole. The inch column below uses the *Machinery's Handbook* / ASME B1.1
standard of approximately **75% thread engagement**; the metric column uses
the ISO 2306 recommended drills, which follow the **major diameter minus
pitch** rule and land at roughly 76–77% engagement.

**Why 75% and not 100%.** Going from 75% to 100% engagement buys only about
5% more thread strength but roughly triples tapping torque, and is the
leading cause of broken taps. In tough or gummy materials — stainless,
titanium, Inconel — drop to 60–65% engagement by moving up one drill size.

Drill decimals here are the true drill diameters from the ASME B94.11M
drill chart, not the three-place approximations most tap charts print.

## UNC / UNF tap drills, #0 to 1/2 in (~75% thread engagement)

| Size | Major dia (in) | UNC TPI | UNC tap drill | UNC drill (in) | UNC drill (mm) | UNF TPI | UNF tap drill | UNF drill (in) | UNF drill (mm) |
| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
| #0 | 0.06 | — | — | — | — | 80 | 3/64 | 0.046875 | 1.191 |
| #1 | 0.073 | 64 | #53 | 0.0595 | 1.511 | 72 | #53 | 0.0595 | 1.511 |
| #2 | 0.086 | 56 | #50 | 0.07 | 1.778 | 64 | #50 | 0.07 | 1.778 |
| #3 | 0.099 | 48 | #47 | 0.0785 | 1.994 | 56 | #45 | 0.082 | 2.083 |
| #4 | 0.112 | 40 | #43 | 0.089 | 2.261 | 48 | #42 | 0.0935 | 2.375 |
| #5 | 0.125 | 40 | #38 | 0.1015 | 2.578 | 44 | #37 | 0.104 | 2.642 |
| #6 | 0.138 | 32 | #36 | 0.1065 | 2.705 | 40 | #33 | 0.113 | 2.87 |
| #8 | 0.164 | 32 | #29 | 0.136 | 3.454 | 36 | #29 | 0.136 | 3.454 |
| #10 | 0.19 | 24 | #25 | 0.1495 | 3.797 | 32 | #21 | 0.159 | 4.039 |
| #12 | 0.216 | 24 | #16 | 0.177 | 4.496 | 28 | #14 | 0.182 | 4.623 |
| 1/4 | 0.25 | 20 | #7 | 0.201 | 5.105 | 28 | #3 | 0.213 | 5.41 |
| 5/16 | 0.3125 | 18 | F | 0.257 | 6.528 | 24 | I | 0.272 | 6.909 |
| 3/8 | 0.375 | 16 | 5/16 | 0.3125 | 7.938 | 24 | Q | 0.332 | 8.433 |
| 7/16 | 0.4375 | 14 | U | 0.368 | 9.347 | 20 | 25/64 | 0.390625 | 9.922 |
| 1/2 | 0.5 | 13 | 27/64 | 0.421875 | 10.716 | 20 | 29/64 | 0.453125 | 11.509 |

## Metric coarse tap drills, M1.6 to M12 (D − P rule)

| Thread | Major dia (mm) | Pitch (mm) | Tap drill (mm) | D − P (mm) | Tap drill (in) | Nearest inch drill |
| --- | --- | --- | --- | --- | --- | --- |
| M1.6x0.35 | 1.6 | 0.35 | 1.25 | 1.25 | 0.0492 | #55 (0.0520) |
| M2x0.4 | 2 | 0.4 | 1.6 | 1.6 | 0.063 | 1/16 (0.0625) |
| M2.5x0.45 | 2.5 | 0.45 | 2.05 | 2.05 | 0.0807 | #46 (0.0810) |
| M3x0.5 | 3 | 0.5 | 2.5 | 2.5 | 0.0984 | #40 (0.0980) |
| M3.5x0.6 | 3.5 | 0.6 | 2.9 | 2.9 | 0.1142 | #33 (0.1130) |
| M4x0.7 | 4 | 0.7 | 3.3 | 3.3 | 0.1299 | #30 (0.1285) |
| M5x0.8 | 5 | 0.8 | 4.2 | 4.2 | 0.1654 | #19 (0.1660) |
| M6x1 | 6 | 1 | 5 | 5 | 0.1969 | #9 (0.1960) |
| M7x1 | 7 | 1 | 6 | 6 | 0.2362 | 15/64 (0.2344) |
| M8x1.25 | 8 | 1.25 | 6.8 | 6.75 | 0.2677 | H (0.2660) |
| M10x1.5 | 10 | 1.5 | 8.5 | 8.5 | 0.3346 | Q (0.3320) |
| M12x1.75 | 12 | 1.75 | 10.2 | 10.25 | 0.4016 | Y (0.4040) |

## Notes

**The metric rule is exact enough to memorise.** Tap drill = major diameter
− pitch. Every listed metric drill is within 0.05 mm of D − P, and only M8
(6.80 against a computed 6.75) and M12 (10.20 against 10.25) are rounded to
the nearest stocked drill. Both of those are the standard catalog sizes.

**Published tap-drill decimals are usually rounded to three places.** Common
charts print #42 as .094, #38 as .102, #36 as .107 and #25 as .150. The true
diameters are 0.0935, 0.1015, 0.1065 and 0.1495. There is no conflict here,
only precision — but if you are computing thread engagement, use the true
value.

**#0-80 has two tap drills in circulation.** *Machinery's Handbook* gives
3/64 in (0.0469), while several shop charts give #56 (0.0465). The two
differ by 0.0004 in, about 3% of thread engagement, and either works. This
chart uses the *Machinery's Handbook* value.

Numbered-size major diameters were independently confirmed by formula:
major diameter = (screw number × 0.013) + 0.060 in.

## FAQ

### What is the tap drill for 1/4-20?

A #7 drill, 0.2010 in (5.105 mm), for approximately 75% thread engagement. For 1/4-28 UNF the tap drill is a #3, 0.2130 in.

### What is the tap drill for M6?

5.00 mm for M6 × 1 coarse. That follows directly from the metric rule: tap drill = major diameter − pitch, so 6 − 1 = 5 mm. The nearest inch drill is a #9 at 0.1960 in, which is 0.4% undersize.

### How do I work out a metric tap drill without a chart?

Subtract the pitch from the major diameter. M8 × 1.25 gives 6.75 mm, which rounds to the stocked 6.8 mm drill; M10 × 1.5 gives 8.5 mm exactly. The rule holds across the whole coarse series and produces about 76–77% thread engagement.

### Why is 75% thread engagement the standard target?

Because the last 25% is not worth paying for. Going from 75% to 100% engagement adds only about 5% to thread strength while roughly tripling tapping torque, which is the main cause of broken taps.

### What tap drill should I use in stainless steel?

Go one drill size larger than the 75% value to drop to roughly 60–65% engagement. The same applies to titanium and Inconel. The loss in thread strength is small; the reduction in tapping torque is not.

### What is the tap drill for 10-32?

A #21 drill, 0.1590 in (4.039 mm). For 10-24 UNC it is a #25, 0.1495 in.

## Sources

- [Wikipedia — Unified Thread Standard (cited to Machinery's Handbook 29th ed., p. 1816)](https://en.wikipedia.org/wiki/Unified_Thread_Standard)
- [Wikipedia — ISO metric screw thread (ISO 261/262 pitch table and the D − P rule)](https://en.wikipedia.org/wiki/ISO_metric_screw_thread)
- [Fastenal — Torque-Tension Reference Guide (diameter/TPI pairings, tensile stress areas)](https://www.fastenal.com/content/merch_rules/images/fcom/content-library/Torque-Tension%20Reference%20Guide.pdf)
- [Machining Doctor — Unified Threads (UNC/UNF/UNEF): Dimensions & Formulas](https://www.machiningdoctor.com/charts/unified-inch-threads-charts/)

## Related processes

- [CNC Milling](https://manufacturingprocesses.org/processes/cutting/cnc-milling.md)
- [Mechanical Fastening](https://manufacturingprocesses.org/processes/joining/mechanical-fastening.md)
- [CNC Machining](https://manufacturingprocesses.org/processes/cutting/cnc-machining.md)

## Related charts

- [Drill Size Chart](https://manufacturingprocesses.org/charts/drill-size-chart.md)
- [Thread Size Chart](https://manufacturingprocesses.org/charts/thread-size-chart.md)
- [Bolt Torque Chart](https://manufacturingprocesses.org/charts/bolt-torque-chart.md)

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/charts/tap-drill-chart)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: chart
title: "Thread Size Chart"
tables: 2
url: https://manufacturingprocesses.org/charts/thread-size-chart
---

# Thread Size Chart

Unified inch threads #0 to 1-1/2 with UNC and UNF pitches, and ISO metric threads M1.6 to M24 with coarse and fine pitches and basic minor diameters.

Both thread systems on this page use the same 60° symmetric V profile; only
the diameter and pitch pairings differ. Unified inch threads (ASME B1.1)
are specified by threads per inch, so pitch is the reciprocal of TPI. ISO
metric threads (ISO 261 general plan, ISO 262 selected sizes) are specified
by pitch in millimeters directly.

The **basic minor diameter** column, D1 = D − 1.0825 × P, is the number that
governs tap drill selection and thread shear area. It is computed here, not
transcribed. Pitch columns on the inch table are likewise computed as 1/TPI
and then converted at exactly 25.4 mm/in.

Numbered inch sizes follow their own rule: major diameter = (screw number ×
0.013) + 0.060 in. That gives #0 = 0.060, #6 = 0.138, #10 = 0.190 in.

## Unified inch threads (UNC / UNF), ASME B1.1

| Size | Major dia (in) | Major dia (mm) | UNC TPI | UNC pitch (in) | UNC pitch (mm) | UNF TPI | UNF pitch (in) | UNF pitch (mm) |
| --- | --- | --- | --- | --- | --- | --- | --- | --- |
| #0 | 0.06 | 1.524 | — | — | — | 80 | 0.0125 | 0.3175 |
| #1 | 0.073 | 1.8542 | 64 | 0.015625 | 0.3969 | 72 | 0.013889 | 0.3528 |
| #2 | 0.086 | 2.1844 | 56 | 0.017857 | 0.4536 | 64 | 0.015625 | 0.3969 |
| #3 | 0.099 | 2.5146 | 48 | 0.020833 | 0.5292 | 56 | 0.017857 | 0.4536 |
| #4 | 0.112 | 2.8448 | 40 | 0.025 | 0.635 | 48 | 0.020833 | 0.5292 |
| #5 | 0.125 | 3.175 | 40 | 0.025 | 0.635 | 44 | 0.022727 | 0.5773 |
| #6 | 0.138 | 3.5052 | 32 | 0.03125 | 0.7937 | 40 | 0.025 | 0.635 |
| #8 | 0.164 | 4.1656 | 32 | 0.03125 | 0.7937 | 36 | 0.027778 | 0.7056 |
| #10 | 0.19 | 4.826 | 24 | 0.041667 | 1.0583 | 32 | 0.03125 | 0.7937 |
| #12 | 0.216 | 5.4864 | 24 | 0.041667 | 1.0583 | 28 | 0.035714 | 0.9071 |
| 1/4 | 0.25 | 6.35 | 20 | 0.05 | 1.27 | 28 | 0.035714 | 0.9071 |
| 5/16 | 0.3125 | 7.9375 | 18 | 0.055556 | 1.4111 | 24 | 0.041667 | 1.0583 |
| 3/8 | 0.375 | 9.525 | 16 | 0.0625 | 1.5875 | 24 | 0.041667 | 1.0583 |
| 7/16 | 0.4375 | 11.1125 | 14 | 0.071429 | 1.8143 | 20 | 0.05 | 1.27 |
| 1/2 | 0.5 | 12.7 | 13 | 0.076923 | 1.9538 | 20 | 0.05 | 1.27 |
| 9/16 | 0.5625 | 14.2875 | 12 | 0.083333 | 2.1167 | 18 | 0.055556 | 1.4111 |
| 5/8 | 0.625 | 15.875 | 11 | 0.090909 | 2.3091 | 18 | 0.055556 | 1.4111 |
| 3/4 | 0.75 | 19.05 | 10 | 0.1 | 2.54 | 16 | 0.0625 | 1.5875 |
| 7/8 | 0.875 | 22.225 | 9 | 0.111111 | 2.8222 | 14 | 0.071429 | 1.8143 |
| 1 | 1 | 25.4 | 8 | 0.125 | 3.175 | 12 | 0.083333 | 2.1167 |
| 1-1/8 | 1.125 | 28.575 | 7 | 0.142857 | 3.6286 | 12 | 0.083333 | 2.1167 |
| 1-1/4 | 1.25 | 31.75 | 7 | 0.142857 | 3.6286 | 12 | 0.083333 | 2.1167 |
| 1-3/8 | 1.375 | 34.925 | 6 | 0.166667 | 4.2333 | 12 | 0.083333 | 2.1167 |
| 1-1/2 | 1.5 | 38.1 | 6 | 0.166667 | 4.2333 | 12 | 0.083333 | 2.1167 |

## ISO metric threads M1.6 – M24 (ISO 261 / ISO 262)

| Thread | Major dia (mm) | Coarse pitch (mm) | Fine pitch(es) (mm) | Coarse designation | Minor dia D1, coarse (mm) |
| --- | --- | --- | --- | --- | --- |
| M1.6 | 1.6 | 0.35 | 0.2 | M1.6 × 0.35 | 1.221 |
| M2 | 2 | 0.4 | 0.25 | M2 × 0.4 | 1.567 |
| M2.5 | 2.5 | 0.45 | 0.35 | M2.5 × 0.45 | 2.013 |
| M3 | 3 | 0.5 | 0.35 | M3 × 0.5 | 2.459 |
| M3.5 | 3.5 | 0.6 | 0.35 | M3.5 × 0.6 | 2.851 |
| M4 | 4 | 0.7 | 0.5 | M4 × 0.7 | 3.242 |
| M5 | 5 | 0.8 | 0.5 | M5 × 0.8 | 4.134 |
| M6 | 6 | 1 | 0.75 | M6 × 1 | 4.918 |
| M7 | 7 | 1 | 0.75 | M7 × 1 | 5.918 |
| M8 | 8 | 1.25 | 1 / 0.75 | M8 × 1.25 | 6.647 |
| M10 | 10 | 1.5 | 1.25 / 1 | M10 × 1.5 | 8.376 |
| M12 | 12 | 1.75 | 1.5 / 1.25 | M12 × 1.75 | 10.106 |
| M14 | 14 | 2 | 1.5 | M14 × 2 | 11.835 |
| M16 | 16 | 2 | 1.5 | M16 × 2 | 13.835 |
| M18 | 18 | 2.5 | 2 / 1.5 | M18 × 2.5 | 15.294 |
| M20 | 20 | 2.5 | 2 / 1.5 | M20 × 2.5 | 17.294 |
| M22 | 22 | 2.5 | 2 / 1.5 | M22 × 2.5 | 19.294 |
| M24 | 24 | 3 | 2 | M24 × 3 | 20.753 |

## Notes

**1 in UNF is 1-12, but 1-14 is what is actually stocked.** The Unified
standard fine pitch for 1 in is 12 TPI. 1-14 is far more widely available
and is commonly — and incorrectly — called "1 inch UNF". If you specify a
1 in fine thread, state the TPI explicitly.

**Several metric diameters have more than one fine pitch.** M8, M10, M12,
M18, M20 and M22 each list two; the first shown is the more common. M24 has
only one fine pitch, 2.0 mm.

**A widely republished 9/16 row contains typos.** One source lists
"9/16-12 2A 0.5609" against a basic major diameter of 0.5625, and
"9/16-12 3A … 0.511". The basic major diameter for 9/16 is 0.5625 in by
definition (9 ÷ 16); the TPI data is unaffected.

All 24 UNC/UNF diameter–TPI pairings were cross-verified between two
independent sources, and the computed pitches were spot-checked against a
published pitch column at #6-32, 1/4-20 and 1/2-13 with exact agreement.

## FAQ

### What is the coarse pitch of an M6 thread?

1.0 mm. The full designation is M6 × 1, its basic minor diameter is 4.918 mm, and its fine pitch alternative is 0.75 mm.

### What is the difference between UNC and UNF?

Only the number of threads per inch at the same major diameter. 1/4-20 is UNC (coarse) and 1/4-28 is UNF (fine). Fine threads have a larger minor diameter and so a larger tensile stress area, and they resist loosening slightly better; coarse threads tolerate damage and dirt better and assemble faster.

### How do I convert threads per inch to pitch?

Pitch is the reciprocal of TPI. 1/4-20 has a pitch of 1/20 = 0.050000 in, which is 1.2700 mm. 1/2-13 gives 0.076923 in, or 1.9538 mm.

### What is the minor diameter of a thread and why does it matter?

The basic internal minor diameter is D1 = D − 1.0825 × P — the diameter at the root of an internal thread. It sets the thread shear area and it is the number tap drill selection works back from. For M10 × 1.5 it is 8.376 mm.

### How big is a #10 screw?

0.1900 in (4.826 mm) major diameter. Numbered inch sizes follow major diameter = (number × 0.013) + 0.060 in, so #10 = (10 × 0.013) + 0.060 = 0.190 in.

### Is 1 inch UNF 12 or 14 threads per inch?

The standard says 12. 1-14 is much more widely stocked and is often loosely called "1 inch UNF", so a drawing that says 1 in UNF without a TPI is ambiguous. State the TPI.

## Sources

- [Wikipedia — Unified Thread Standard (cited to Machinery's Handbook 29th ed., p. 1816)](https://en.wikipedia.org/wiki/Unified_Thread_Standard)
- [Engineers Edge — ANSI External Screw Threads Size & Tolerances Chart](https://www.engineersedge.com/screw_threads_chart.htm)
- [Wikipedia — ISO metric screw thread (ISO 262 selected sizes)](https://en.wikipedia.org/wiki/ISO_metric_screw_thread)
- [Machining Doctor — Unified Threads (UNC/UNF/UNEF): Dimensions & Formulas](https://www.machiningdoctor.com/charts/unified-inch-threads-charts/)

## Related processes

- [Mechanical Fastening](https://manufacturingprocesses.org/processes/joining/mechanical-fastening.md)
- [Swiss Screw Machining](https://manufacturingprocesses.org/processes/cutting/swiss-screw-machining.md)
- [CNC Turning](https://manufacturingprocesses.org/processes/cutting/cnc-turning.md)

## Related charts

- [Tap Drill Chart](https://manufacturingprocesses.org/charts/tap-drill-chart.md)
- [Drill Size Chart](https://manufacturingprocesses.org/charts/drill-size-chart.md)
- [Bolt Torque Chart](https://manufacturingprocesses.org/charts/bolt-torque-chart.md)

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/charts/thread-size-chart)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: index
title: "Process Comparisons"
comparisons: 8
url: https://manufacturingprocesses.org/compare
---

# Process Comparisons

8 head-to-head comparisons between manufacturing processes, each ending in a straight recommendation.

- [6061 vs 7075 Aluminum](https://manufacturingprocesses.org/compare/6061-vs-7075.md): 6061-T6 is the weldable, corrosion-resistant, easily anodized workhorse; 7075-T6 is roughly 85% stronger in tension but is not practically weldable and corrodes more readily.
- [CNC Machining vs 3D Printing](https://manufacturingprocesses.org/compare/cnc-machining-vs-3d-printing.md): CNC machining cuts a part out of certified wrought stock and holds tight tolerances; 3D printing builds it up layer by layer and charges nothing for geometric complexity.
- [Casting vs Forging](https://manufacturingprocesses.org/compare/casting-vs-forging.md): Casting melts metal and lets the mold define the shape, so almost any geometry is possible; forging deforms solid metal, so the grain follows the part and fatigue strength is higher.
- [Injection Molding vs 3D Printing](https://manufacturingprocesses.org/compare/injection-molding-vs-3d-printing.md): Injection molding pays for a steel tool up front and then makes parts for pennies; 3D printing has no tooling cost at all but a unit price that never falls with volume.
- [Investment Casting vs Die Casting](https://manufacturingprocesses.org/compare/investment-casting-vs-die-casting.md): Investment casting works in almost any alloy, including steel and superalloys, with modest tooling; die casting is limited to low-melting alloys but is far faster per part once the die exists.
- [MIG vs TIG Welding](https://manufacturingprocesses.org/compare/mig-vs-tig-welding.md): MIG feeds a consumable wire continuously and lays metal down fast; TIG uses a non-consumable tungsten electrode with separate filler, trading speed for control and weld appearance.
- [Powder Coating vs Paint](https://manufacturingprocesses.org/compare/powder-coating-vs-paint.md): Powder coating bakes a thick, tough, solvent-free film onto conductive parts; wet paint gives thinner films, unlimited color matching, and works on substrates that cannot go in an oven.
- [Sand Casting vs Die Casting](https://manufacturingprocesses.org/compare/sand-casting-vs-die-casting.md): Sand casting has almost no tooling cost and no size limit but a rough surface and loose tolerances; die casting inverts every one of those trade-offs.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/compare)*


<!-- ---------------- -->

---
type: comparison
title: "6061 vs 7075 Aluminum"
url: https://manufacturingprocesses.org/compare/6061-vs-7075
---

# 6061 vs 7075 Aluminum

6061-T6 is the weldable, corrosion-resistant, easily anodized workhorse; 7075-T6 is roughly 85% stronger in tension but is not practically weldable and corrodes more readily.

## Comparison

Use 6061-T6 unless you have a specific reason not to — it welds, it anodizes cleanly, it resists corrosion, it comes in every mill form, and it costs less. Move to 7075-T6 when a strength-limited part has to get lighter or smaller: its 572 MPa (83 ksi) tensile and 503 MPa (73 ksi) yield are roughly 85% above 6061-T6's. If the part is stiffness-limited or must be welded, 7075 buys you nothing.

Both are heat-treatable wrought aluminum alloys, but from different families. 6061 is Al-Mg-Si (6xxx), strengthened by magnesium silicide. 7075 is Al-Zn-Mg-Cu (7xxx), the aerospace family, strengthened by zinc-magnesium precipitates with copper added.

## Head-to-head

| Property | 6061-T6 | 7075-T6 |
|---|---|---|
| Alloy system | Al-Mg-Si (6xxx) | Al-Zn-Mg-Cu (7xxx) |
| Density | 2.70 g/cm³ (0.0975 lb/in³) | 2.81 g/cm³ (0.1015 lb/in³) |
| Ultimate tensile strength | 310 MPa (45 ksi) | 572 MPa (83 ksi) |
| Yield strength | 270 MPa (39.2 ksi) | 503 MPa (73 ksi) |
| Hardness | 95 HB | 150 HB |
| Young's modulus | 69 GPa (10.0 Msi) | 71.7 GPa (10.4 Msi) |
| Melting range | Solidus 580 °C, liquidus 650 °C (1,076–1,202 °F) | Solidus 480 °C, liquidus 640 °C (896–1,184 °F) |
| Weldability | Readily fusion welded with 4043 or 5356 filler; the heat-affected zone loses temper unless the part is re-solution treated and aged | Effectively non-weldable by fusion — the weld metal hot cracks. Join mechanically, adhesively, or by friction stir welding |
| Corrosion resistance | Good general atmospheric resistance | Lower. T6 is susceptible to stress corrosion cracking; the overaged T73 and T7351 tempers trade strength for SCC resistance, and aerospace sheet is often supplied Alclad |
| Anodizing response | Anodizes cleanly and consistently — clear or dyed Type II, predictable Type III hardcoat | Anodizes, but the zinc and copper give a darker and sometimes hazier film; hardcoat comes out noticeably darker |
| Machinability | Machines well but runs gummy; needs sharp tooling and good chip evacuation | Chips more cleanly and is generally the easier of the two to machine |
| Formability | Bends in T6 at moderate radii | Essentially unformable in T6; sheet is formed in O or W temper and heat treated afterward |
| Availability | Every mill form — extrusion, structural shape, bar, plate, sheet, tube. The commodity structural aluminum | Mainly plate, bar, and sheet; far fewer stock extruded shapes |
| Relative cost | Baseline | Commonly quoted at roughly twice 6061 per pound, and higher for aerospace-certified plate. Ratio varies with form and market |

Property values above come from the [material properties chart](/charts/material-properties); melting ranges from the [metal melting points chart](/charts/metal-melting-points).

## When to choose 6061-T6

Choose it whenever the joint is welded. Fabricated frames, tube structures, pressure housings, marine hardware, and heat exchangers rely on [arc welding](/processes/joining/arc-welding), and 6061 takes it. Design for the heat-affected zone: as-welded strength near the bead drops well below T6 unless the assembly is re-solution treated and artificially aged, so size the joint for the weakened condition or move the weld out of the high-stress region.

Choose it whenever the part is stiffness-driven. The two alloys differ by roughly 4% in Young's modulus — 69 GPa against 71.7 GPa — so a deflection-limited bracket, plate, or beam deflects essentially the same in either alloy. Paying twice the price for 7075 in a stiffness-limited part buys nothing at all. This is the single most common material selection error in aluminum design.

Choose it for anything extruded. 6061 is the extrusion workhorse and is stocked as angle, channel, tube, and thousands of custom [extrusion](/processes/forming/metal-extrusion) profiles. Choose it for anything formed on a [press brake](/processes/forming/press-braking) — see the [bend radius and K-factor chart](/charts/sheet-metal-bend-radius-k-factor) — and for anything anodized to a consistent cosmetic finish, since 6061 gives [anodizing](/processes/finishing/anodizing) a clean, repeatable color.

## When to choose 7075-T6

Choose it when the part is strength-limited and weight or envelope is fixed. Aircraft fittings, wing spars, landing gear components, rifle receivers, high-load structural brackets, competition bicycle and motorsport parts, and mold plates for tooling all use 7075 because roughly 85% more tensile and yield strength at only 4% more density means a smaller, lighter part that carries the same load.

Choose it for hardness and wear on unlubricated surfaces. At 150 HB against 95 HB, 7075 galls and dents less — useful for fixture plates, jaws, and tooling that gets handled hard.

Then design around what it costs you. There is no fusion welding, so the assembly runs on [mechanical fastening](/processes/joining/mechanical-fastening), [adhesive bonding](/processes/joining/adhesive-bonding), or [friction welding](/processes/joining/friction-welding). There is no forming in T6, so sheet parts are formed soft and heat treated after. Corrosion protection is not optional in a marine or humid environment, and stress corrosion cracking in the short-transverse direction is a real failure mode in T6 — if sustained tensile stress and moisture are both present, specify a T73-type overaged temper and accept the strength penalty.

## Cost comparison

| Situation | Pick | Why |
|---|---|---|
| Welded frame or weldment | 6061-T6 | 7075 cannot be fusion welded at all |
| Deflection-limited plate or bracket | 6061-T6 | Modulus differs by about 4%; the extra strength is unused |
| Strength-limited part, weight critical | 7075-T6 | About 85% more yield strength at 4% more density |
| Extruded profile or bent sheet | 6061-T6 | Stock shapes and formability in temper |
| Cosmetic anodized finish | 6061-T6 | Cleaner, more repeatable anodic film |
| Marine or high-humidity service | 6061-T6, or 7075 in a T73 temper | 7075-T6 is vulnerable to stress corrosion cracking |
| Fixture plates, tooling, hard-use surfaces | 7075-T6 | 150 HB against 95 HB |

Stock price is only part of the delta. 7075 is stocked in fewer forms, so a part that would have been an off-the-shelf 6061 extrusion may have to be [machined](/processes/cutting/cnc-machining) from plate — which adds material removal, cycle time, and scrap on top of the higher per-pound cost. Where the strength really is needed and volume supports it, a near-net 7075 [forging](/processes/forming/forging) cuts the buy-to-fly ratio dramatically compared with hogging the shape out of plate.

## Verdict

Default to 6061-T6 and justify any move away from it. Switch to 7075-T6 only when the part is strength-limited and the envelope or weight budget is fixed — its 503 MPa (73 ksi) yield against 6061-T6's 270 MPa (39.2 ksi) is about 85% more load capacity for 4% more density. Two conditions rule 7075 out outright: any fusion-welded joint, because 7075 hot cracks, and any stiffness-limited part, because the two alloys are within about 4% on Young's modulus (69 vs 71.7 GPa) and the extra strength goes unused. If you do specify 7075 in a humid or marine environment under sustained tensile stress, move to a T73-type overaged temper and accept the strength penalty rather than risk stress corrosion cracking.

## FAQ

### How much stronger is 7075 than 6061?

About 85% stronger in both measures. 7075-T6 runs 572 MPa (83 ksi) ultimate tensile and 503 MPa (73 ksi) yield, against 310 MPa (45 ksi) and 270 MPa (39.2 ksi) for 6061-T6. Density rises only from 2.70 to 2.81 g/cm³ (0.0975 to 0.1015 lb/in³), so the strength-to-weight gain is nearly the full strength gain.

### Can you weld 7075 aluminum?

Not by fusion welding, in practice. 7075 hot cracks in the weld metal and heat-affected zone, so it is not treated as a weldable alloy. Assemblies in 7075 are joined with mechanical fasteners, structural adhesive, or friction stir welding, which stays below the melting point. If the design requires a welded joint, use 6061 instead.

### Is 7075 stiffer than 6061?

Barely, and not usefully. Young's modulus is 69 GPa (10.0 Msi) for 6061-T6 and 71.7 GPa (10.4 Msi) for 7075-T6 — about 4% apart. A deflection-limited part made from 7075 will deflect essentially the same as one made from 6061 with the same geometry, so the alloy upgrade buys nothing on a stiffness-driven design.

### Which anodizes better, 6061 or 7075?

6061. It takes a clean, consistent Type II anodic film that dyes predictably and hardcoats to a repeatable appearance. The zinc and copper in 7075 produce a darker and sometimes hazier finish, and Type III hardcoat on 7075 comes out noticeably darker. If color consistency across a batch matters cosmetically, specify 6061.

### Does 7075 corrode more than 6061?

Yes. 6061 has good general atmospheric corrosion resistance, while 7075 is less resistant and, in the T6 temper, is susceptible to stress corrosion cracking under sustained tensile stress in the short-transverse direction. The usual answers are an overaged T73 or T7351 temper, which trades strength for SCC resistance, Alclad sheet, or a protective anodic coating.

### Can you bend 7075 sheet?

Not in the T6 temper — it cracks at radii 6061-T6 tolerates. Sheet parts in 7075 are formed in the annealed (O) or freshly quenched (W) temper and then heat treated to T6 afterward, which adds process steps and dimensional risk. For a bent part with no exceptional strength requirement, 6061-T6 is the straightforward choice.

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/compare/6061-vs-7075)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: comparison
title: "CNC Machining vs 3D Printing"
a: "cnc-machining"
b: "rapid-prototyping"
url: https://manufacturingprocesses.org/compare/cnc-machining-vs-3d-printing
---

# CNC Machining vs 3D Printing

CNC machining cuts a part out of certified wrought stock and holds tight tolerances; 3D printing builds it up layer by layer and charges nothing for geometric complexity.

## Comparison

Print it when you need the part this week, the geometry is complex, and quantity is under about ten. Machine it when the part has to hold a real tolerance, carry load in every direction, or be made from a certified alloy such as 6061, 7075, 304, or Ti-6Al-4V. Past roughly fifty parts, machining usually wins on price too.

[CNC machining](/processes/cutting/cnc-machining) removes material from certified wrought stock with a rotating cutter or a turning tool, so the part inherits the properties of the bar it came from. [3D printing](/processes/forming/rapid-prototyping) builds the part up in layers from filament, resin, or powder, so it inherits the properties of whatever the layer bond gives you.

## Head-to-head

| Dimension | CNC machining | 3D printing |
|---|---|---|
| Typical tolerance | ±0.005 in (±0.13 mm) standard; ±0.001 in (±0.025 mm) precision, tighter on selected features | Service-bureau specs run about ±0.5% with a ±0.020 in (±0.5 mm) floor for FDM, ±0.3% with a ±0.012 in (±0.3 mm) floor for SLS and MJF, ±0.006 in (±0.15 mm) for SLA, and ±0.2% with roughly a ±0.004 in (±0.1 mm) floor for DMLS |
| Material structure | Isotropic wrought stock, with mill certs and lot traceability available | Layer-bonded and anisotropic; Z-direction strength sits below in-plane strength, worst in FDM |
| Materials | Any machinable metal or plastic sold in bar, plate, or tube | The vendor's catalog of resins, powders, and filaments |
| Surface finish | 125 µin Ra (3.2 µm) as-machined is the default; 32 µin (0.8 µm) with finishing passes | Layer stepping is visible: FDM 0.004–0.012 in (0.1–0.3 mm) layers, SLA 0.001–0.004 in (25–100 µm), SLS about 0.004 in (0.1 mm) |
| Tooling cost | None beyond cutters and fixturing | None |
| Setup cost | Real — CAM programming, workholding, first-article inspection. This is the fixed cost that amortizes | Near zero; slicing takes minutes |
| Per-part cost vs volume | Falls steadily as setup spreads across the lot | Essentially flat — part 1 and part 500 cost about the same |
| Lead time | 3–10 business days from a service bureau | 1–3 days |
| Geometry limits | The cutter must physically reach every surface; internal corners carry the tool radius; no enclosed cavities | Internal channels, lattices, conformal cooling, and undercuts cost nothing extra |
| Size | Common 3-axis envelopes reach roughly 48 × 24 × 24 in (1,200 × 600 × 600 mm) | Typical SLA and SLS build volumes run around 10 × 10 × 12 in (250 × 250 × 300 mm); large-format FDM goes bigger |

## When to choose CNC machining

Choose machining the moment a drawing carries a fit. Bearing seats, dowel-pin holes, shaft journals, and sealing faces are called out with ISO 286 fits — see the [ISO 286 fits and tolerances chart](/charts/iso-286-fits-tolerances) — and no common printing process holds an H7 bore without a machining operation afterward anyway. The same goes for threads: printed threads strip, so either machine and tap them using the [tap drill chart](/charts/tap-drill-chart) or design in heat-set inserts.

Choose it for load-bearing metal parts. A machined 7075-T6 bracket has the tensile and yield strength listed in the [material properties chart](/charts/material-properties) in every direction. A printed equivalent does not, and if the load has a component normal to the build plane the layer bond becomes the design allowable.

Choose it when documentation matters. Aerospace, medical, defense, and pressure work generally want a material certificate traceable to a heat lot, which is routine with bar stock and awkward with powder. Typical machined parts: motor mounts, manifold bodies, optical mounts, fixture plates, shafts on a [CNC lathe](/processes/cutting/cnc-turning), and contoured impellers on [5-axis](/processes/cutting/5-axis-cnc-machining) equipment.

## When to choose 3D printing

Choose printing when complexity is the point. Cooling channels that follow a contour, topology-optimized brackets, lattice-filled energy absorbers, and manifolds with curved runners are cheaper printed than machined — and often impossible machined, because a cutter cannot get inside a closed volume.

Choose it for iteration speed. Form-and-fit checks, ergonomics mockups, jigs, and soft-jaw prototypes are printed overnight, evaluated, and reprinted. Three design revisions in a week is normal in printing and unrealistic when a programming-and-setup cycle sits in front of every part.

Choose it for one-offs and short runs of low-stress parts: enclosures, cable guides, sensor mounts, tooling aids. And choose it for consolidated assemblies — printing five brackets as one piece removes fasteners, joints, and the machining setups that went with them. For metal, [DMLS](/processes/forming/direct-metal-laser-sintering-dmls) covers aluminum, stainless, and titanium; for tough nylon parts, [SLS](/processes/forming/selective-laser-sintering-sls) and [MJF](/processes/forming/multi-jet-fusion-mjf); for smooth cosmetic parts, [SLA](/processes/forming/stereolithography-sla).

## Cost comparison

| Quantity | Usually cheaper | Why |
|---|---|---|
| 1–5 | 3D printing | Machining setup and programming spread over too few parts |
| 5–50 | Crossover zone | Simple prismatic parts tip to machining early; intricate ones stay printed |
| 50–500 | CNC machining | Setup is amortized and cycle time, not fixed cost, sets the price |
| 500+ | Machining, or change process | For plastics evaluate [injection molding](/processes/forming/injection-molding); for metal, casting |

The mechanism behind the crossover is simple: machining has a real fixed cost and a low marginal cost, printing has almost no fixed cost and a marginal cost that never falls. Two variables move the crossover point. Geometric complexity moves it up, because complexity adds machining setups and fixtures while adding nothing to print time. Part size moves it down, because printing bills by material volume and build-plate hours, so a big solid block is expensive to print and cheap to mill.

## Verdict

Start with the tolerance and the load path. If any feature is tighter than about ±0.010 in (±0.25 mm), or the part is loaded normal to where the layers would run, machine it — printed accuracy and layer anisotropy are the two things no amount of post-processing fixes cheaply. If the geometry contains internal channels, lattices, or consolidated features a cutter cannot reach, print it, because machining that shape costs several setups or is outright impossible. Between those poles let quantity decide: under about ten parts printing is faster and cheaper, over about fifty machining is cheaper, and in between the more complex part should be printed. Prototype in print and produce in machining is a good default, not a rule — verify that the printed prototype's fits are representative before you trust it.

## FAQ

### Is CNC machining more accurate than 3D printing?

Yes, by roughly an order of magnitude. Standard machining holds ±0.005 in (±0.13 mm) and precision work ±0.001 in (±0.025 mm), while typical printing specs are ±0.5% with a ±0.020 in (±0.5 mm) floor for FDM and ±0.3% with a ±0.012 in (±0.3 mm) floor for SLS and MJF. Printed parts that need a real fit get the critical feature machined after printing.

### At what quantity does CNC machining become cheaper than 3D printing?

Commonly somewhere between 10 and 50 parts. Machining carries a fixed setup and programming cost with a low per-part cost after that, while printing has almost no setup and a per-part cost that does not fall with volume. Complex geometry pushes the crossover higher; large simple parts pull it lower.

### Are 3D printed parts strong enough for functional use?

Often, but not in every direction. Layer-bonded parts are weaker in the build (Z) direction than in plane, most severely in FDM, so a printed part has to be oriented with the principal load in plane and derated for the layer bond. When the design allowables need to come straight off a material datasheet in all directions, machine from wrought stock.

### Can you 3D print in aluminum or steel?

Yes — DMLS, SLM, and binder jetting produce metal parts in alloys such as AlSi10Mg, 316L, 17-4PH, and Ti-6Al-4V. Expect to machine critical features afterward, to heat treat, and to pay considerably more per part than machining the same shape from bar stock unless the geometry justifies it.

### Which process is faster for a prototype?

Printing, typically 1–3 days against 3–10 business days for machining, because there is no CAM programming, workholding, or first-article step in front of it. The gap widens for intricate geometry and narrows for simple prismatic parts a shop can set up in one operation.

### Can you machine a 3D printed part?

Yes, and it is a common hybrid. Print the complex body, leave stock on the features that need accuracy, then mill or turn the bores, faces, and threads. You get internal channels and consolidated geometry from printing while holding machined tolerances where the drawing requires them.

## Process pages

- [CNC Machining](https://manufacturingprocesses.org/processes/cutting/cnc-machining.md)
- [Rapid Prototyping](https://manufacturingprocesses.org/processes/forming/rapid-prototyping.md)

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/compare/cnc-machining-vs-3d-printing)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: comparison
title: "Casting vs Forging"
a: "sand-casting"
b: "forging"
url: https://manufacturingprocesses.org/compare/casting-vs-forging
---

# Casting vs Forging

Casting melts metal and lets the mold define the shape, so almost any geometry is possible; forging deforms solid metal, so the grain follows the part and fatigue strength is higher.

## Comparison

Forge the part when it carries cyclic or impact loads and the shape is simple enough to squeeze between two dies — crankshafts, lifting hooks, landing gear. Cast it when the geometry has internal passages, thin webs, or undercuts no die can form, when the alloy is cast iron, or when annual volume is too low to pay for a forging die.

[Sand casting](/processes/forming/sand-casting) pours molten metal into a cavity packed in bonded sand; the mold is destroyed to get the part out. [Forging](/processes/forming/forging) hammers or presses solid metal — hot, warm, or cold — between shaped dies until it fills them. That single difference in how the metal arrives at final shape drives everything below.

## Head-to-head

| Dimension | Casting (sand) | Forging (closed-die) |
|---|---|---|
| Grain structure | Dendritic solidification structure with no directionality; shrinkage and gas porosity are inherent risks to design around | Wrought grain flow bends to follow the part contour; forging closes porosity carried in from the billet |
| Strength and fatigue | Static strength is fully specifiable, but fatigue life and impact toughness are limited by porosity and inclusions | Higher fatigue strength and impact toughness in the same alloy — no porosity, and flow lines run along the load path |
| Tolerances | ISO 8062 grade DCTG 11–14; roughly ±0.03 in (±0.8 mm) on small dimensions, widening with part size | Conventional closed-die roughly ±0.03 in (±0.8 mm); precision near-net forging to about ±0.010 in (±0.25 mm) |
| Tooling | Wood, urethane, or aluminum pattern plus core boxes — the cheapest tooling of any metal-forming route | Sunk die blocks in hot-work tool steel, plus trim and coining dies |
| Tooling lead time | 2–4 weeks for a pattern and core boxes | 8–16 weeks for a die set is typical |
| Per-part cost at volume | Falls slowly — every part consumes a fresh sand mold, so mold-making labor never goes away | Falls sharply once the die is paid for; press cycle is seconds per part |
| Materials | Anything a foundry can pour: gray and ductile iron, carbon and stainless steel, aluminum, bronze, magnesium | Wrought alloys only: carbon and alloy steel, stainless, aluminum 6061 and 7075, titanium, nickel superalloys. Cast iron cannot be forged |
| Size limit | Effectively none; multi-ton castings are routine | Bounded by press capacity — the largest closed-die presses in service are in the 50,000-ton class |
| Minimum section | About 0.125–0.25 in (3–6 mm) in aluminum, 0.25 in (6 mm) in steel | Thin webs are hard to fill; ribs need generous fillets and 3–7° draft |
| Surface finish | 250–900 µin Ra (6.3–23 µm) as-cast | 125–500 µin Ra (3.2–12.5 µm) as-forged, under scale |
| Geometry freedom | Internal passages via sand cores, undercuts, organic shapes | No enclosed cavities; the shape has to release from a two-part die |

See the [surface finish chart](/charts/surface-finish-chart) for how those Ra bands compare with machined and ground surfaces.

## When to choose casting

Casting wins whenever the shape does the work. An engine block with cast-in water jackets, a pump volute, a valve body with a branched flow path, a gearbox housing with integral bosses and ribs — none of these can be forged, because a die cannot form an enclosed cavity. Sand cores can.

It also wins on economics at low volume. A wood pattern is cheap and quick, so one-off replacement parts, short-run machine bases, and prototype housings all go to a foundry. Gray iron machine tool bases are cast partly for cost and partly because gray iron's graphite flakes damp vibration in a way no forging does.

If the casting route is right but the tolerances and finish are not, move up the casting family rather than across to forging: [investment casting](/processes/forming/investment-casting) holds roughly ±0.005 in (±0.13 mm) on the first inch at 63–125 µin Ra (1.6–3.2 µm), and [die casting](/processes/forming/die-casting) is finer still in aluminum and zinc.

## When to choose forging

Forge when a fatigue or impact failure is unacceptable. Connecting rods, crankshafts, wheel hubs, spindles, wrenches, lifting hooks, and shackles are forged because the load is cyclic or shock and the consequence of a crack is a person. Pressure-piping flanges and fittings are commonly specified as forged for the same reason, and some codes and customer specifications simply require it — check the spec before you value-engineer a forging into a casting.

Forging is also the right call for high-strength aluminum and titanium airframe parts. 7075 and Ti-6Al-4V bulkheads, fittings, and landing-gear components are forged near-net and then machined, which preserves grain flow and cuts the buy-to-fly ratio compared with hogging the part out of plate.

## Cost comparison

| Annual volume | Usually cheaper | Why |
|---|---|---|
| 1–10 | Sand casting, or machining from solid | Pattern cost is trivial; a forging die never pays back |
| 10–500 | Sand casting | Mold-making labor per part still beats amortized die cost |
| 500–5,000 | Depends on the part | Crossover zone. Simple axisymmetric shapes tip toward forging early; cored geometry stays cast |
| 5,000+ | Forging, where geometry allows | Die cost spreads thin and press cycles run in seconds |

Two costs engineers routinely miss. First, forging stock is wrought bar or billet, which costs more per pound than a foundry's ingot-and-return charge, and closed-die flash is scrap — plan on losing a meaningful fraction of the input weight at the trim press. Second, castings need more machining stock left on functional faces to clear the wider as-cast tolerance band, so the machining bill partly offsets the cheaper tooling. Compare finished-part cost, not process cost.

## Verdict

Decide on the load case first: if the part sees cyclic or impact loading and a crack is a safety event, forge it — grain flow and the absence of porosity are worth the die cost. If the geometry contains an enclosed passage, a cored cavity, or an undercut, cast it, because no two-part die can form those. Below roughly 500 parts a year sand casting almost always wins on total cost, whichever way the strength argument points. When casting is right on geometry but wrong on tolerance or finish, move up the casting family to investment or die casting rather than across to forging.

## FAQ

### Is forging stronger than casting?

In the same alloy, a forging has better fatigue strength and impact toughness because forging closes porosity and bends the grain flow to follow the part contour. Static tensile strength is less of a differentiator — a heat-treated steel casting can be specified to the same UTS. The gap shows up under cyclic and shock loads, which is why crankshafts, connecting rods, and lifting hooks are forged.

### Can you forge cast iron?

No. Gray and ductile iron have too little ductility to deform without cracking, so anything specified in cast iron is a casting by definition. If you need a forged part in an iron-based alloy, the material changes to a carbon or alloy steel.

### What tolerance can a sand casting hold?

Plan on ISO 8062 grade DCTG 11–14, which is roughly ±0.03 in (±0.8 mm) on small dimensions and looser as the part grows. Any surface that has to locate, seal, or mate gets machining stock added and is cut after casting.

### Which is cheaper, casting or forging?

Sand casting is cheaper below a few hundred parts a year because the pattern is inexpensive and quick to make. Forging overtakes it somewhere in the 500–5,000 range for shapes a die can form, since the die amortizes and press cycles run in seconds. Compare finished-part cost including machining, not tooling cost alone.

### Do forged parts still need machining?

Almost always on functional surfaces. Conventional closed-die forgings hold roughly ±0.03 in (±0.8 mm), so bearing bores, sealing faces, and threads are machined afterward. Precision or near-net forging tightens that to about ±0.010 in (±0.25 mm) and reduces, but rarely eliminates, the machining.

### When does investment casting beat forging?

When the geometry is too intricate for a die but the tolerance and finish requirements are too tight for sand casting, and volume runs from a few hundred to a few thousand. Investment casting holds roughly ±0.005 in (±0.13 mm) on the first inch at 63–125 µin Ra (1.6–3.2 µm) in steels, stainless, and superalloys, with far cheaper tooling than a forging die.

## Process pages

- [Sand Casting](https://manufacturingprocesses.org/processes/forming/sand-casting.md)
- [Forging](https://manufacturingprocesses.org/processes/forming/forging.md)

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/compare/casting-vs-forging)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: comparison
title: "Injection Molding vs 3D Printing"
a: "injection-molding"
b: "rapid-prototyping"
url: https://manufacturingprocesses.org/compare/injection-molding-vs-3d-printing
---

# Injection Molding vs 3D Printing

Injection molding pays for a steel tool up front and then makes parts for pennies; 3D printing has no tooling cost at all but a unit price that never falls with volume.

## Comparison

If your annual volume is in the hundreds or more and the design is frozen, injection mold it — the tool is the only large cost and it amortizes fast. If you need parts this week, expect the design to change, or need fewer than about a hundred pieces, print them. The decision is almost entirely about quantity and design maturity, not about part quality.

[Injection molding](/processes/forming/injection-molding) pays for a machined steel or aluminum tool once and then produces near-identical parts on a 15–60 second cycle. [3D printing](/processes/forming/rapid-prototyping) has no tooling at all, but the per-part cost is the same on part 10 and part 10,000.

## Head-to-head

| Dimension | Injection molding | 3D printing |
|---|---|---|
| Part structure | Fully dense, isotropic within the limits of flow-induced orientation and weld lines | Layer-bonded and anisotropic; Z-direction strength trails in-plane strength, most severely in FDM |
| Typical tolerance | ±0.005 in (±0.13 mm) commercial; ±0.002 in (±0.05 mm) on small dimensions with a well-tuned tool | About ±0.5% with a ±0.020 in (±0.5 mm) floor for FDM, ±0.3% with a ±0.012 in (±0.3 mm) floor for SLS and MJF, ±0.006 in (±0.15 mm) for SLA |
| Tooling cost | The dominant cost. An aluminum bridge tool is a fraction of a hardened multi-cavity production tool, and single-cavity family tools cut it further | None |
| Tooling lead time | 2–8 weeks before the first shot | None; printing starts the day the file is ready |
| Cycle time | 15–60 s for a typical part, all cavities at once | Hours per build, and the build plate is shared across the batch |
| Per-part cost at volume | Falls steeply until the tool is amortized, then flattens near material plus machine time | Flat. Volume buys almost nothing |
| Materials | Any moldable thermoplastic or thermoset: ABS, PC, PP, PA66, POM, PEEK, TPE, glass- and mineral-filled grades | The vendor's qualified catalog — far narrower, and mechanical data is usually process-specific |
| Size limits | Bounded by clamp tonnage and shot size; machines commonly run 30–1,000+ tons | Typical SLA and SLS build volumes near 10 × 10 × 12 in (250 × 250 × 300 mm); large-format FDM goes bigger |
| Surface finish | Whatever the tool has — SPI A-1 polish to a molded-in texture, reproduced exactly on every shot | Layer stepping is visible: FDM 0.004–0.012 in (0.1–0.3 mm), SLA 0.001–0.004 in (25–100 µm). Smoothing is a secondary operation |
| Repeatability | Shot-to-shot variation is small and statistically controllable | Varies between builds, machines, and powder lots |

## When to choose injection molding

Mold it when quantity justifies a tool and the geometry is settled. Consumer product housings, connector bodies, closures, medical disposables, automotive interior clips, and anything sold by the thousand are molded because the per-part cost drops to material plus a few seconds of machine time.

Mold it when the material spec is not negotiable. Flame-rated V-0 grades, 30% glass-filled nylon, food-contact polypropylene, USP Class VI silicone — these exist as molding pellets with published datasheets and certifications. The printing catalog for the equivalent property set is thin and the data is process-specific.

Mold it when cosmetics matter. A molded part carries the tool's finish exactly, including SPI A-2 polish or a MoldTech texture, on every shot with no post-processing. Design to the standard rules before you cut steel — uniform walls of about 0.060–0.140 in (1.5–3.5 mm), a minimum 1–2° of draft, ribs at roughly 60% of the nominal wall to avoid sink — and account for shrinkage, which runs roughly 0.4–0.7% for ABS, 0.5–0.7% for PC, 1–2% for PA66, and 1.5–2.5% for PP. The [injection molding design guidelines chart](/charts/injection-molding-design-guidelines) collects the numbers.

## When to choose 3D printing

Print when the design is still moving. Every geometry change after the tool is cut is a steel change order with a lead time attached; in printing it is a new file. Anything in the validation loop — form-and-fit checks, ergonomic mockups, functional test articles — belongs in printing.

Print for low quantities and for spares. A run of 25 enclosures, a discontinued bracket, or a fixture for the assembly line will never repay a tool. So will a part that ships in ones and twos as a configurable option.

Print when the geometry cannot be molded. Enclosed internal channels, lattice structures, and undercuts that would need side actions or collapsible cores are free in printing and expensive in steel. If the part is nearly moldable, remember that side actions and unscrewing cores are available — they add tool cost and cycle time, but they may still beat printing at volume.

## Cost comparison

| Annual volume | Usually cheaper | Notes |
|---|---|---|
| 1–100 | 3D printing | No tool will amortize this fast |
| 100–1,000 | Crossover zone | An aluminum bridge tool often wins here, especially for parts with many features |
| 1,000–10,000 | Injection molding | Tool cost per part is small and falling |
| 10,000+ | Injection molding | Tooling is noise; material, cycle time, and cavitation set the price |

Two levers move the crossover. Tool class is the first: an aluminum bridge tool good for a few thousand shots costs far less and arrives far sooner than a hardened steel multi-cavity production tool, and it moves the breakeven down into the hundreds. Part size is the second: printing charges by material volume and build hours, so large parts get expensive to print quickly, while a molder's cost per part rises much more slowly with size. For metal parts, the same logic maps onto [metal injection molding](/processes/forming/metal-injection-molding) versus [DMLS](/processes/forming/direct-metal-laser-sintering-dmls).

## Verdict

Count the parts and check whether the design is frozen. Below about 100 pieces, or with any open design questions, print — a tool cut against a moving design is money spent twice. Above about 1,000 pieces of a settled design, mold, because per-part cost drops to material plus a 15–60 second cycle and no printing process follows it down. In the 100–1,000 band, price an aluminum bridge tool: it usually beats printing on unit cost and beats a hardened steel tool on both cost and lead time. Material requirements can override all of this — if the part needs a V-0 rating, a filled engineering grade, or a specific medical certification, molding is often the only route with the data to support it.

## FAQ

### At what quantity is injection molding cheaper than 3D printing?

Usually somewhere between 100 and 1,000 parts. Molding front-loads nearly all its cost into the tool and then makes parts for material plus a 15–60 second cycle, while printing has no tooling cost and a per-part cost that never falls. An aluminum bridge tool moves the breakeven down toward the low hundreds; a hardened multi-cavity steel tool pushes it up.

### Can you 3D print an injection mold?

Yes, for short runs. Printed tool inserts in high-temperature resin typically survive tens to a few hundred shots in low-temperature resins, which is enough to prove a design or bridge to steel. They will not hold up to glass-filled materials or long runs, and cycle times are longer because printed resin conducts heat poorly.

### Are injection molded parts stronger than 3D printed parts?

Generally yes, because a molded part is fully dense and does not have layer bonds as its weak axis, and because filled engineering grades are widely available as molding pellets. Molded parts still have direction-dependent behavior from flow-induced orientation and have weld lines where flow fronts meet, so place gates to keep weld lines out of the load path.

### What tolerance can injection molding hold?

Plan on ±0.005 in (±0.13 mm) for commercial work, with ±0.002 in (±0.05 mm) achievable on small dimensions when the tool is well built and the process is capped. Tolerances are tied to the resin's shrinkage — roughly 0.4–0.7% for ABS, 1–2% for PA66, and 1.5–2.5% for PP — so a high-shrink material widens what the tool can hold.

### How long does injection molding tooling take?

Commonly 2–8 weeks from approved design to first shot, depending on tool class, cavity count, and whether the part needs side actions. That lead time, not the tooling price, is often the real reason a program prints its first hundreds of parts while the tool is being cut.

### Which process should I use for a prototype that becomes a product?

Print the early revisions, then design for molding before you commit. Add uniform walls of about 0.060–0.140 in (1.5–3.5 mm), 1–2° of draft, and ribs at roughly 60% of nominal wall thickness while the part is still printed, so the transition to a tool does not require a redesign. Consider an aluminum bridge tool for the first production run.

## Process pages

- [Injection Molding](https://manufacturingprocesses.org/processes/forming/injection-molding.md)
- [Rapid Prototyping](https://manufacturingprocesses.org/processes/forming/rapid-prototyping.md)

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/compare/injection-molding-vs-3d-printing)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: comparison
title: "Investment Casting vs Die Casting"
a: "investment-casting"
b: "die-casting"
url: https://manufacturingprocesses.org/compare/investment-casting-vs-die-casting
---

# Investment Casting vs Die Casting

Investment casting works in almost any alloy, including steel and superalloys, with modest tooling; die casting is limited to low-melting alloys but is far faster per part once the die exists.

## Comparison

Alloy usually decides this before volume does. If the part has to be steel, stainless, or a nickel or cobalt superalloy, use investment casting — a die casting machine cannot handle those melting points. If the part is aluminum, zinc, or magnesium and you need tens of thousands a year, die cast it: the cycle is under two minutes and the as-cast finish and tolerances are better.

[Investment casting](/processes/forming/investment-casting) builds a ceramic shell around an injected wax pattern, melts the wax out, and pours metal into the shell. [Die casting](/processes/forming/die-casting) injects molten metal under high pressure into a permanent hardened steel die. One tool is destroyed each cycle; the other lasts for a hundred thousand.

## Head-to-head

| Dimension | Investment casting | Die casting |
|---|---|---|
| Alloys | Carbon and alloy steel, stainless including 17-4PH, aluminum, bronze, cobalt and nickel superalloys, titanium under vacuum | Practically limited to aluminum, zinc, and magnesium; brass is possible but destroys dies quickly |
| Structure and heat treatment | Can be fully heat treated and hot isostatically pressed; weldable | Conventional high-pressure die castings trap gas, so they blister on solution heat treatment and are hard to fusion weld. Vacuum and squeeze casting variants relieve this |
| Typical tolerance | About ±0.005 in (±0.13 mm) on the first inch, plus roughly ±0.003 in per additional inch | NADCA standard linear tolerance for aluminum is about ±0.010 in (±0.25 mm) on the first inch plus ±0.002 in per additional inch; zinc holds tighter, and precision tolerances are available |
| Surface finish | 63–125 µin Ra (1.6–3.2 µm) as-cast | 32–63 µin Ra (0.8–1.6 µm) as-cast — the best of any casting process |
| Tooling | An aluminum wax-injection die. Cheaper and faster than a die casting die | A hardened hot-work tool steel die with cooling lines, ejectors, and often slides — the single largest cost |
| Tooling lead time | 4–8 weeks typical | 10–16 weeks typical |
| Cycle time | A batch process: shell build alone takes days of dip-and-dry coats | 60–200 shots per hour on small parts |
| Tool life | Wax dies are lightly loaded and last a long time | Commonly quoted around 100,000 shots for aluminum; zinc dies run far longer because the metal is cooler |
| Per-part cost at volume | Falls modestly — wax injection, shell building, and cutoff are labor per part | Falls sharply once the die is amortized |
| Size limits | Most parts are under about 20 lb (9 kg); specialized foundries pour much larger | Bounded by machine locking force, commonly 200–4,000 tons |
| Minimum wall | About 0.030–0.060 in (0.75–1.5 mm) | About 0.040–0.100 in (1–2.5 mm) in aluminum; zinc goes thinner |
| Draft | Near zero on many features, because the shell is broken away | 1–3° typical, more on interior walls |

The alloy constraint traces straight back to melting point — see the [metal melting points chart](/charts/metal-melting-points). Aluminum 7075 has a solidus of 480 °C (896 °F) and Zamak 3 melts at 381–387 °C (718–729 °F), both survivable for a steel die; carbon steel poured well above 1,400 °C (2,550 °F) is not.

## When to choose investment casting

Choose it whenever the metallurgy rules out die casting. Turbine and compressor blades, valve and pump internals in 316 and CF8M, firearm receivers, surgical instruments, and 17-4PH structural fittings are investment cast because they must be steel or a superalloy and because the part can then be solution treated and aged to full properties.

Choose it for intricate geometry at moderate volume. Because the ceramic shell is destroyed, features do not have to release from a die: near-zero draft, re-entrant details, and thin cored passages are all practical. Runs of a few hundred to a few thousand parts a year are the sweet spot, where tooling cost stays low and per-part labor is tolerable.

Choose it when the part will be welded into an assembly or must pass radiographic inspection, since a die casting's entrapped gas porosity typically disqualifies it from both.

## When to choose die casting

Choose it for high-volume aluminum, zinc, and magnesium parts where the as-cast surface is close to final. Transmission and engine housings, electronics enclosures with cast-in EMI walls, power tool bodies, lighting heat sinks, and zinc hardware are die cast because 60–200 shots per hour makes anything else uncompetitive.

Choose it when you want a good finish straight out of the machine. At 32–63 µin Ra (0.8–1.6 µm) many die cast surfaces go directly to [powder coating](/processes/finishing/powder-coating) or chromate conversion with no machining — see the [surface finish chart](/charts/surface-finish-chart) for context.

Choose it when the design can absorb the constraints: uniform thin walls, 1–3° of draft, and no requirement for heat treatment or structural welding on a conventionally cast part.

## Cost comparison

| Annual volume | Usually cheaper | Why |
|---|---|---|
| Under 500 | Investment casting | Wax die cost is modest and pays back quickly |
| 500–5,000 | Investment casting | Die casting tooling still dominates the unit price |
| 5,000–20,000 | Crossover zone, alloy permitting | Depends on part complexity, secondary machining, and finish requirements |
| 20,000+ | Die casting | The die is amortized and cycle time sets the price |

Do the crossover math on finished cost. Investment castings often go straight to final dimensions on non-critical features because the as-cast tolerance band is tighter; die castings often need trim, deflash, and machining of sealing faces. If the part is stainless or steel, none of this arithmetic applies — the alloy has already made the decision. For lower-volume aluminum work that needs die-casting-like detail, compare [sand casting](/processes/forming/sand-casting) as well.

## Verdict

Check the alloy first — it settles most of these decisions on its own. Anything in steel, stainless, or a nickel or cobalt superalloy has to be investment cast, because die casting dies cannot survive those pour temperatures. For aluminum, zinc, and magnesium, let volume decide: below roughly 5,000 parts a year investment casting's cheap wax tooling wins, above roughly 20,000 the die casting cycle wins decisively. Two requirements override volume entirely: if the part must be solution heat treated to full properties, or fusion welded into an assembly, choose investment casting, because conventional high-pressure die castings blister on heat treatment and their entrapped gas porosity makes welding unreliable.

## FAQ

### Can you die cast steel or stainless steel?

Not in practice. High-pressure die casting is limited to alloys that melt well below the tool steel dies that contain them — aluminum, zinc, and magnesium. Carbon steel pours above 1,400 °C (2,550 °F) and would erode an H13 die almost immediately, so steel and stainless parts go to investment casting or sand casting instead.

### Which has tighter tolerances, investment casting or die casting?

Investment casting, on the first inch: roughly ±0.005 in (±0.13 mm), against the NADCA standard aluminum linear tolerance of about ±0.010 in (±0.25 mm) on the first inch. Die casting closes the gap on longer dimensions because it adds only about ±0.002 in per additional inch, and precision tolerance classes are available on both processes.

### Why can't die castings be heat treated?

Conventional high-pressure die casting entrains air and die lubricant vapor during the fast fill, leaving fine gas porosity distributed through the part. On a solution heat treatment cycle that trapped gas expands and blisters the surface. Vacuum die casting, squeeze casting, and semi-solid processes reduce the porosity enough that heat treatment becomes possible.

### Which process has the better surface finish?

Die casting, at 32–63 µin Ra (0.8–1.6 µm) as-cast, against 63–125 µin Ra (1.6–3.2 µm) for investment casting. Die cast surfaces are frequently good enough to paint or powder coat with no machining at all.

### How many parts do I need to justify die casting tooling?

Usually tens of thousands. The die is a hardened tool steel assembly with cooling, ejection, and often slides, and it typically takes 10–16 weeks to build, so the crossover against investment casting generally lands somewhere between 5,000 and 20,000 parts a year depending on part complexity and secondary operations.

### Can investment castings be welded?

Yes. Investment castings are gravity poured, so they do not carry the entrained gas that makes conventional die castings blister and gas out in the weld pool. They can also be hot isostatically pressed to close residual shrinkage porosity before welding or before radiographic acceptance.

## Process pages

- [Investment Casting](https://manufacturingprocesses.org/processes/forming/investment-casting.md)
- [Die Casting](https://manufacturingprocesses.org/processes/forming/die-casting.md)

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/compare/investment-casting-vs-die-casting)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: comparison
title: "MIG vs TIG Welding"
a: "arc-welding"
url: https://manufacturingprocesses.org/compare/mig-vs-tig-welding
---

# MIG vs TIG Welding

MIG feeds a consumable wire continuously and lays metal down fast; TIG uses a non-consumable tungsten electrode with separate filler, trading speed for control and weld appearance.

## Comparison

Use MIG when you have length of weld to lay down and the material is 1/8 in (3 mm) or thicker — structural steel, frames, brackets, general fabrication. Use TIG when the material is thin, the alloy is fussy (aluminum, titanium, thin stainless), the weld will be seen, or a root pass has to be radiographically clean. MIG buys throughput; TIG buys control.

Both are [arc welding](/processes/joining/arc-welding) processes with gas shielding. MIG — formally GMAW — feeds a consumable wire electrode through the torch continuously, so the electrode is the filler. TIG — GTAW — strikes the arc from a non-consumable tungsten electrode and the welder adds filler rod separately, or runs autogenous with no filler at all.

## Head-to-head

| Dimension | MIG (GMAW) | TIG (GTAW) |
|---|---|---|
| Deposition rate | Typically 2–12 lb/hr (0.9–5.4 kg/hr) depending on wire size and transfer mode | Typically 0.5–2 lb/hr (0.2–0.9 kg/hr) |
| Travel speed | Fast; a continuous wire feed means no stopping to re-grip filler | Slow and deliberate — the limiting factor on cost |
| Thin-material floor | About 0.030–0.040 in (0.8–1.0 mm) in short-circuit transfer; pulsed MIG goes thinner | About 0.020 in (0.5 mm) and below with pulsed current and a foot pedal |
| Thick material | Multi-pass with no practical upper limit given a qualified procedure | Capable but uneconomical; commonly used only for the root pass |
| Heat control | Set at the machine and largely fixed during the pass | Amperage varied live with a foot pedal or torch control, independent of filler addition |
| Distortion | Fast travel puts less total heat into a long seam, so long welds often distort less | Excellent for thin sheet where the risk is burn-through, but slow travel can soak more total heat into a long joint |
| Materials | Carbon steel, stainless, aluminum with a spool gun or push-pull feeder | Aluminum, titanium, magnesium, thin stainless, copper alloys, plus everything MIG does |
| Polarity and gas | DCEP; 75/25 Ar/CO₂ or straight CO₂ on carbon steel, argon-rich mixes on stainless, 100% argon on aluminum | DCEN on steel and stainless, AC on aluminum and magnesium; 100% argon, or argon/helium on thick sections |
| Base-metal cleanliness | Tolerates mill scale, light rust, and oil better | Demands clean, degreased, oxide-free metal; contamination shows up immediately |
| Appearance and cleanup | Spatter is normal, especially in short-circuit transfer and with CO₂; expect grinding | No spatter and a stacked, uniform bead; often no cleanup at all |
| Operator skill | Learnable in days for acceptable fillets | Two hands plus a foot pedal; months to become productive |
| Equipment cost | Lower for equivalent capacity | Higher — AC/DC inverters with pulse and balance control cost more |
| Labor cost per unit length | The low-cost option, by a wide margin | Commonly several times MIG's, driven almost entirely by travel speed |

The AC requirement on aluminum is a materials fact, not a preference: aluminum oxide melts around 2,072 °C (3,762 °F) while the aluminum underneath melts at 660 °C (1,220 °F), per the [metal melting points chart](/charts/metal-melting-points). The electrode-positive half of the AC cycle blasts that refractory oxide skin off so the arc can reach clean metal.

## When to choose MIG

Choose MIG for production fabrication where weld length drives the cost: trailer frames, equipment skids, handrail, weldments assembled in a fixture, and any repetitive joint that a semi-automatic or robotic torch can run. The deposition rate difference is the whole argument — a job that takes an hour in MIG can take several in TIG for a joint that performs identically once ground and painted.

Choose it for thicker steel. Multi-pass MIG on 1/4 in (6 mm) and up is standard structural practice, and spray transfer gives deep, well-fused beads in the flat and horizontal positions. Out of position, drop to short-circuit or pulsed spray.

Choose it when the base metal is not pristine. Shop steel arrives with mill scale, storage rust, and cutting oil, and MIG tolerates all three far better than TIG, which will reject a contaminated joint with porosity and a dirty arc.

## When to choose TIG

Choose TIG for thin material. Sheet metal enclosures, exhaust tubing, instrument housings, and 0.030 in (0.8 mm) stainless are TIG work because the foot pedal lets a welder taper current into and out of the puddle without blowing through.

Choose it for reactive and oxide-forming alloys. Aluminum on AC, titanium with a trailing shield and back purge, and magnesium are TIG's territory. So is thin stainless, where a purged root prevents sugaring on the back side.

Choose it when the weld is the product or has to pass inspection. Architectural stainless, food and pharmaceutical tube, motorcycle frames, and show welds get TIG for appearance. Pressure piping frequently uses a TIG root pass followed by MIG or flux-cored fill and cap — you buy the clean, fully fused root where it matters and the fast fill everywhere else.

## Cost comparison

| Situation | Usually cheaper | Why |
|---|---|---|
| Long fillet welds on 1/8 in (3 mm)+ steel | MIG | Deposition rate and travel speed dominate |
| Sheet metal under 1/16 in (1.6 mm) | TIG | MIG burns through and rework costs more than the slow pass |
| Aluminum, occasional | TIG | Avoids buying a spool gun or push-pull feeder |
| Aluminum, production volume | MIG with a spool gun or push-pull | Deposition rate wins once the setup is paid for |
| Code-critical pipe root | TIG root, MIG or flux-cored fill | Buys weld integrity only where the inspection looks |
| Cosmetic and architectural | TIG | No spatter means no grinding, polishing, or rework labor |

Weld cost is labor hours, not machine price. A TIG machine costs more than a MIG machine, but that difference disappears on the first large job — the recurring cost is the welder's time per foot of weld, and that is where MIG's several-fold speed advantage shows up. Where neither process fits the volume, look at [resistance welding](/processes/joining/resistance-welding) for sheet assemblies, [friction welding](/processes/joining/friction-welding) for round parts, and [soldering and brazing](/processes/joining/soldering-and-brazing) for joints that should not melt the base metal.

## Verdict

Look at thickness, alloy, and who will see the weld. Under about 1/16 in (1.6 mm), or in aluminum, titanium, or thin stainless, use TIG — the foot pedal and separate filler are what keep you from burning through or contaminating the puddle. At 1/8 in (3 mm) and above in carbon steel or stainless, use MIG, because deposition rate of 2–12 lb/hr against 0.5–2 lb/hr decides the job cost and both processes produce a code-qualifiable weld. When the weld is visible or has to pass radiography, TIG's cleanliness pays for its slow travel. On code pipe, do both: TIG the root, then fill and cap with MIG or flux-cored.

## FAQ

### Is TIG welding stronger than MIG welding?

Not inherently. Both are qualifiable to the same codes, and joint strength comes from the filler alloy, joint prep, penetration, and procedure rather than from the process. TIG produces cleaner welds with less porosity and inclusion risk on reactive alloys and thin sections, which is why it is specified for critical root passes — but a properly executed MIG weld in structural steel is fully as strong.

### Can you MIG weld aluminum?

Yes, but not with a standard torch and liner. Aluminum wire is soft and birdnests in a long conventional feed path, so you need a spool gun or a push-pull system, 100% argon shielding, and ER4043 or ER5356 filler. For occasional aluminum work TIG is usually the cheaper answer; for production volume the MIG setup pays for itself.

### Why does TIG use AC for aluminum?

Aluminum carries a tenacious oxide skin that melts near 2,072 °C (3,762 °F), while the aluminum beneath it melts at 660 °C (1,220 °F). The electrode-positive half of an AC cycle strips that oxide off the surface so the arc can wet clean metal; the electrode-negative half puts heat into the workpiece. Balance control on the machine sets the ratio between cleaning and penetration.

### Which welding process is faster?

MIG, by a wide margin. Typical deposition rates run 2–12 lb/hr (0.9–5.4 kg/hr) for MIG against 0.5–2 lb/hr (0.2–0.9 kg/hr) for TIG, and MIG's continuous wire feed removes the stop-and-reposition rhythm of feeding filler rod by hand. On long production welds that difference is the entire cost argument.

### Which process distorts the part less?

It depends on the joint. On thin sheet TIG distorts less because the pedal lets you meter current precisely and avoid burn-through. On a long seam MIG often distorts less, because its fast travel speed puts less total heat into the part even at higher arc energy. Fixture the assembly and stagger or back-step your passes either way.

### Can you TIG weld a MIG-welded joint, or vice versa?

Yes, and combining them is standard practice on pressure piping: a TIG root pass gives a clean, fully fused inside surface, then MIG or flux-cored fills and caps the joint quickly. Just make sure the filler metals are compatible and the procedure is qualified for the combination.

## Process pages

- [Arc Welding](https://manufacturingprocesses.org/processes/joining/arc-welding.md)

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/compare/mig-vs-tig-welding)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: comparison
title: "Powder Coating vs Paint"
a: "powder-coating"
b: "spray-painting"
url: https://manufacturingprocesses.org/compare/powder-coating-vs-paint
---

# Powder Coating vs Paint

Powder coating bakes a thick, tough, solvent-free film onto conductive parts; wet paint gives thinner films, unlimited color matching, and works on substrates that cannot go in an oven.

## Comparison

Powder coat it if the part is bare metal, fits in an oven, and you want a tough, thick, single-coat finish — brackets, frames, enclosures, wheels. Paint it if the substrate is plastic or wood, the assembly contains heat-sensitive components, the part is too big to bake, you need an exact custom color match, or you are finishing one or two pieces.

[Powder coating](/processes/finishing/powder-coating) sprays a dry, electrostatically charged thermoset powder onto a grounded part and fuses it in an oven. [Spray painting](/processes/finishing/spray-painting) atomizes a liquid coating that flows out and cures by solvent evaporation, oxidation, or a two-part chemical reaction.

## Head-to-head

| Dimension | Powder coating | Liquid paint |
|---|---|---|
| Film thickness | 2–4 mils (50–100 µm) is typical in one pass; 1.5–10 mils (38–250 µm) is achievable | 1–2 mils (25–50 µm) per coat; multi-coat systems build 3–6 mils (75–150 µm) |
| Cure | 10–20 minutes at 350–400 °F (177–204 °C) part metal temperature; low-temperature powders cure from about 250 °F (121 °C) | Ambient air dry, force dry at 150–200 °F (65–93 °C), or 2K chemical cure at room temperature |
| Substrates | Electrically conductive and oven-stable — steel, aluminum, most castings. Specialized systems handle MDF and a few high-temperature plastics | Anything: metal, plastic, wood, composites, assembled products |
| Assemblies | Bearings, seals, elastomers, electronics, and plastics will not survive the oven | No thermal restriction |
| Material utilization | Very high with a reclaim booth — oversprayed powder that never cured is recovered and re-sprayed | Overspray is lost. Conventional air spray transfers roughly 30–50%; HVLP and electrostatic liquid do better |
| VOC and permitting | Essentially zero VOC | Solvent-borne systems carry VOC and need permitted booths and often abatement |
| Durability | Better impact, abrasion, and edge coverage at the same film thickness, because the melted powder builds a thicker, more continuous film on corners | Thinner films chip and wear sooner, though high-solids 2K polyurethanes are very durable |
| Color and appearance | Excellent solids and textures; bonded metallics exist but limit reclaim. Color changes require a booth purge | Unlimited custom matching, basecoat/clearcoat metallics, Class A automotive finishes, and easy small-batch color changes |
| Repair and touch-up | Poor — a spot repair usually means recoating and re-baking the whole part | Easy. Spot repair, blend, and refinish in place |
| Masking | Mandatory on threads, bearing bores, and ground points; a 2–4 mil build will bind a Class 2B thread | Still needed, but a 1–2 mil film is more forgiving of tight features |
| Part size | Bounded by oven and conveyor size | Effectively unbounded — bridges, tanks, and structural steel are painted in place |

## When to choose powder coating

Choose it for volume runs of bare metal parts in one color. Fabricated steel brackets, machine guards, electrical enclosures, exercise equipment, wheels, shelving, and appliance panels are powder coated because reclaim keeps material utilization high and a single automated pass replaces a primer-plus-topcoat wet system.

Choose it when the finish takes abuse. At the same nominal thickness a powder film resists impact and abrasion better than most wet coatings and covers sharp edges more completely — the failure point where thin liquid films break down first.

Pick the chemistry to match the exposure. Epoxy powders are interior-only because they chalk under UV; TGIC and TGIC-free polyesters are the standard exterior choice; super-durable polyester and fluoropolymer systems cover architectural work under the AAMA 2603, 2604, and 2605 performance tiers, in ascending order of weathering resistance.

Pretreatment decides corrosion life far more than the coating does. Degrease, then apply an iron or zinc phosphate on steel or a [chemical conversion coating](/processes/finishing/chemical-conversion-coating) on aluminum, or [abrasive blast](/processes/finishing/abrasive-blasting) to a specified profile — see the [surface finish chart](/charts/surface-finish-chart) for how blast profiles compare. Salt spray performance is verified to ASTM B117, and the number a system achieves tracks the pretreatment much more closely than the powder.

## When to choose spray painting

Choose paint when the part cannot be baked. Any assembly containing seals, bearings, wiring, batteries, or molded plastic goes to liquid, as does anything already fitted with heat-sensitive hardware.

Choose it for non-conductive substrates. Plastics, wood, composites, and fiberglass have no path to ground for electrostatic powder deposition; liquid coatings, with an adhesion promoter where needed, cover all of them.

Choose it for color-critical and cosmetic work. Automotive refinish, custom color matching to a customer's Pantone or existing fleet color, metallic basecoat with clearcoat, and multi-color graphics are all liquid work. Changing color between two parts is a gun flush rather than a booth purge.

Choose it for very large structures and for field work. Structural steel, tanks, and equipment already installed on site get liquid coatings because no oven exists that fits them, and because a damaged area can be spot repaired without stripping the whole assembly.

## Cost comparison

| Situation | Usually cheaper | Why |
|---|---|---|
| One-off or a handful of parts | Liquid paint | No oven to heat, no color changeover, no booth purge |
| Mixed colors, small batches | Liquid paint | Powder color changes cost more setup than the parts are worth |
| Batches of like parts, same color | Powder coating | Reclaim keeps material utilization high and a single pass replaces two |
| Continuous production line | Powder coating | Lowest cost per square foot, no solvent, no VOC abatement |
| Field application or repair | Liquid paint | Powder cannot be applied or cured outside a shop |

Weigh the full system, not the coating. Liquid work often needs a primer coat, a topcoat, flash time between coats, VOC permitting, and solvent disposal; powder needs an oven and a color changeover procedure. For corrosion protection on aluminum, compare [anodizing](/processes/finishing/anodizing) as an integral rather than applied finish; for complete coverage inside complex weldments, compare [e-coating](/processes/finishing/e-coating-electrophoretic-deposition), which is frequently used as a primer underneath a powder topcoat.

## Verdict

Ask two questions before anything else: is the substrate electrically conductive, and can the whole assembly survive 20 minutes at 400 °F (204 °C)? If either answer is no, it is a liquid paint job — plastics, wood, and assemblies with seals, bearings, or electronics have no powder option. If both answers are yes and you are running batches of like parts in one color, powder coat: reclaim keeps material utilization high, one 2–4 mil (50–100 µm) pass replaces a primer-plus-topcoat system, and edge coverage and impact resistance are better. Reach for liquid anyway when the color has to match something exactly, when the finish must be Class A, or when the part will need field touch-up, since powder cannot be spot repaired.

## FAQ

### Is powder coating more durable than paint?

At the same film thickness, generally yes. Powder builds a 2–4 mil (50–100 µm) film in one pass and flows to cover sharp edges more completely, giving better impact and abrasion resistance than a 1–2 mil (25–50 µm) liquid coat. High-solids two-part polyurethanes close much of that gap. In both cases, corrosion life is set mostly by pretreatment, not by the coating chemistry.

### Can you powder coat plastic?

Not in the general case. Electrostatic powder deposition needs a conductive, grounded part, and the standard cure of 10–20 minutes at 350–400 °F (177–204 °C) will destroy most thermoplastics. Specialized low-temperature and UV-cure powders on conductive-primed MDF and high-temperature plastics exist, but liquid paint is the normal route for polymer substrates.

### What temperature does powder coating cure at?

Most thermoset powders cure at 350–400 °F (177–204 °C) part metal temperature for 10–20 minutes. Low-temperature chemistries start around 250 °F (121 °C) and UV-cure powders lower still, but the mainstream process schedule rules out any assembly with seals, bearings, wiring, or molded plastic in it.

### Can powder coating be touched up?

Not well. Because the film is a fused thermoset, a spot repair generally means stripping and recoating the part, then re-baking it. Field touch-up is usually done with a matched liquid paint, which will not match the powder's texture or gloss exactly. If the part will be damaged and repaired in service, that is an argument for liquid from the start.

### Do I need to mask threads before powder coating?

Yes, on any thread, bearing bore, press fit, or electrical ground point. A 2–4 mil (50–100 µm) build on both flanks is enough to bind a Class 2B thread and to shift a press fit out of tolerance. Silicone plugs, caps, and high-temperature tape are the standard masking; alternatively, chase the threads after coating.

### Which is cheaper, powder coating or painting?

Powder is cheaper per square foot on batch or line production of like parts in one color, because oversprayed powder is reclaimed and a single pass replaces primer plus topcoat. Liquid is cheaper for one-offs, small mixed-color batches, and field work, where powder's color changeover and oven time have nothing to amortize against.

## Process pages

- [Powder Coating](https://manufacturingprocesses.org/processes/finishing/powder-coating.md)
- [Spray Painting](https://manufacturingprocesses.org/processes/finishing/spray-painting.md)

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/compare/powder-coating-vs-paint)*

*Last updated: August 11, 2026*


<!-- ---------------- -->

---
type: comparison
title: "Sand Casting vs Die Casting"
a: "sand-casting"
b: "die-casting"
url: https://manufacturingprocesses.org/compare/sand-casting-vs-die-casting
---

# Sand Casting vs Die Casting

Sand casting has almost no tooling cost and no size limit but a rough surface and loose tolerances; die casting inverts every one of those trade-offs.

## Comparison

Sand cast it if the alloy is iron or steel, the part is large, or annual volume is in the hundreds or low thousands. Die cast it if the part is aluminum, zinc, or magnesium, has thin uniform walls, and you need tens of thousands a year with a good as-cast surface. Nothing else in casting has such a clean split between the two.

[Sand casting](/processes/forming/sand-casting) packs bonded sand around a reusable pattern and destroys the mold to release each part. [Die casting](/processes/forming/die-casting) injects molten metal at high pressure into a permanent hardened steel die that runs for a hundred thousand cycles. Every trade-off below follows from consumable mold versus permanent die.

## Head-to-head

| Dimension | Sand casting | Die casting |
|---|---|---|
| Structure | Slow solidification gives a coarser grain; shrinkage porosity is managed with risers and chills. Fully heat treatable and weldable | Fast solidification gives a fine, chill-cast skin with good as-cast strength, but conventional high-pressure castings trap gas — they blister on solution heat treatment and are hard to fusion weld |
| Tolerances | ISO 8062 grade DCTG 11–14, roughly ±0.03 in (±0.8 mm) on small dimensions and looser as the part grows | NADCA standard linear tolerance for aluminum is about ±0.010 in (±0.25 mm) on the first inch plus ±0.002 in per additional inch; zinc holds tighter |
| Surface finish | 250–900 µin Ra (6.3–23 µm) — you can see and feel the sand | 32–63 µin Ra (0.8–1.6 µm), often good enough to coat with no machining |
| Tooling | A wood, urethane, or aluminum pattern plus core boxes. The cheapest tooling in metal casting | A hardened hot-work tool steel die with water lines, ejector pins, and slides. The dominant cost |
| Tooling lead time | 2–4 weeks | 10–16 weeks is typical |
| Tool life | Patterns last for thousands of molds, but a new sand mold is built for every single part | Commonly quoted around 100,000 shots for aluminum; zinc dies run far longer because the melt is cooler |
| Rate | Bounded by mold-making, which is per part — hand-rammed in a jobbing shop, automated on a green sand line | 60–200 shots per hour on small parts |
| Per-part cost at volume | Falls slowly; mold-making labor and sand handling never go away | Falls sharply once the die is amortized |
| Materials | Anything pourable: gray and ductile iron, carbon and stainless steel, aluminum, bronze, magnesium | Aluminum, zinc, and magnesium in practice. Brass is possible but consumes dies |
| Size limits | Effectively none; multi-ton castings are routine | Bounded by machine locking force, commonly 200–4,000 tons |
| Minimum wall | About 0.125–0.25 in (3–6 mm) in aluminum, 0.25 in (6 mm) in steel | About 0.040–0.100 in (1–2.5 mm) in aluminum; zinc goes thinner |
| Internal passages | Sand cores make enclosed cavities and branched passages routine | Formed only by retracting steel slides — enclosed passages are effectively out, and undercuts cost slides |
| Draft | 1–3° typical | 1–3°, with more on interior walls than exterior |

The alloy split comes down to melting point. Aluminum 6061 has a solidus of 580 °C (1,076 °F) and Zamak 3 melts at 381–387 °C (718–729 °F) — both comfortable for a steel die. Iron and steel are poured hundreds of degrees higher and would wreck one, as the [metal melting points chart](/charts/metal-melting-points) makes clear. See the [surface finish chart](/charts/surface-finish-chart) for how those two Ra bands compare with machined surfaces.

## When to choose sand casting

Choose it when the alloy rules die casting out. Gray iron machine bases and brake components, ductile iron housings, steel valve bodies, and bronze marine hardware all have to be sand cast or investment cast — the melt would destroy a die.

Choose it when the part is large or heavy. Engine blocks, pump casings, gearbox housings, and machine tool bases go to a foundry because no die casting machine has the shot size or locking force for them, and sand imposes no upper limit.

Choose it for cored geometry. Water jackets, oil galleries, and branched flow passages are made with sand cores that shake out afterward. A die casting die cannot form an enclosed cavity, so that geometry never survives the conversion.

Choose it at low volume. A wood pattern is cheap and arrives in weeks, so one-off replacements, short production runs, and prototype housings are economical in quantities of one.

## When to choose die casting

Choose it for high-volume aluminum, zinc, and magnesium parts. Transmission housings, electronics enclosures with cast-in EMI walls and heat sink fins, power tool bodies, and zinc latches and hinges are die cast because 60–200 shots per hour is unmatchable any other way.

Choose it when walls are thin and uniform. Die casting fills a 0.040–0.100 in (1–2.5 mm) aluminum wall that sand casting cannot even attempt, which cuts part weight substantially on the same envelope.

Choose it when the as-cast surface is the finished surface. At 32–63 µin Ra (0.8–1.6 µm), die cast faces often go straight to [powder coating](/processes/finishing/powder-coating) or a chromate conversion coating with no machining, eliminating operations that sand castings always need.

## Cost comparison

| Annual volume | Usually cheaper | Why |
|---|---|---|
| 1–100 | Sand casting | A pattern costs a fraction of a die and arrives in weeks |
| 100–1,000 | Sand casting | Die tooling has nothing like enough parts to amortize against |
| 1,000–10,000 | Sand casting, usually | Unless thin walls or as-cast finish force the die |
| 10,000+ | Die casting | The die is amortized and cycle time sets the price |
| 100,000+ | Die casting, decisively | Per-part cost approaches metal plus seconds of machine time |

Compare finished cost, not casting cost. Sand castings need extra stock left on functional faces to clear the wide as-cast band, so the machining bill is higher; die castings need trim and deflash but often skip machining entirely on non-critical surfaces. If your volume lands in the awkward middle and the part is intricate, price [investment casting](/processes/forming/investment-casting) as well — it holds roughly ±0.005 in (±0.13 mm) on the first inch at 63–125 µin Ra (1.6–3.2 µm) on modest tooling, and it works in steel.

## Verdict

Alloy and volume settle this in two steps. If the part is iron, steel, or bronze, sand cast it — die casting cannot handle those pour temperatures at all. If it is aluminum, zinc, or magnesium, the crossover is volume: below about 1,000 parts a year sand casting wins because a pattern costs a fraction of a hardened die and arrives in 2–4 weeks instead of 10–16, and above about 10,000 die casting wins because 60–200 shots per hour and a 32–63 µin Ra (0.8–1.6 µm) as-cast finish erase the tooling premium. Two design details override volume: enclosed internal passages need sand cores and cannot be die cast, and any part that must be solution heat treated or fusion welded should not be a conventional high-pressure die casting.

## FAQ

### Which is more accurate, sand casting or die casting?

Die casting, by a wide margin. Sand castings fall in ISO 8062 grade DCTG 11–14, roughly ±0.03 in (±0.8 mm) on small dimensions, while the NADCA standard linear tolerance for aluminum die casting is about ±0.010 in (±0.25 mm) on the first inch plus ±0.002 in per additional inch. Sand castings therefore carry more machining stock on every functional face.

### Can you die cast iron or steel?

No. High-pressure die casting is limited to alloys that melt far below the tool steel die containing them — aluminum, zinc, and magnesium in practice. Iron and steel are poured hundreds of degrees hotter and would erode a die immediately, so those alloys go to sand casting or investment casting.

### How thin can each process cast?

Sand casting bottoms out around 0.125–0.25 in (3–6 mm) in aluminum and about 0.25 in (6 mm) in steel, because the melt freezes before it fills a thinner section. Die casting fills 0.040–0.100 in (1–2.5 mm) walls in aluminum and thinner still in zinc, which is why die cast parts are so much lighter on the same envelope.

### How many parts justify die casting tooling?

Usually tens of thousands. A die is a hardened tool steel assembly with cooling, ejection, and often slides, and it takes 10–16 weeks to build, so the crossover against sand casting typically falls somewhere between 1,000 and 10,000 parts a year — earlier if thin walls or the as-cast finish are requirements rather than preferences.

### Can die castings be heat treated or welded?

Conventional high-pressure die castings usually cannot. The fast fill entrains air and die lubricant vapor, leaving fine gas porosity that blisters on a solution heat treatment cycle and gases out in a weld pool. Sand castings are gravity poured, so they can be fully heat treated and welded. Vacuum die casting and squeeze casting reduce porosity enough to change that answer.

### What surface finish does each process produce?

Sand casting produces 250–900 µin Ra (6.3–23 µm) — visibly and audibly rough. Die casting produces 32–63 µin Ra (0.8–1.6 µm), which is frequently good enough to powder coat or conversion coat with no machining at all.

## Process pages

- [Sand Casting](https://manufacturingprocesses.org/processes/forming/sand-casting.md)
- [Die Casting](https://manufacturingprocesses.org/processes/forming/die-casting.md)

---

*Source: [Manufacturing Processes](https://manufacturingprocesses.org/compare/sand-casting-vs-die-casting)*

*Last updated: August 11, 2026*