CNC Plastic Processing Tolerance Guide

• 1
  What a tolerance really controls on a plastic part
• 2
  Thermal expansion: the largest single error source
• 3
  Moisture, creep and elastic recovery
• 4
  Machining parameters that keep a plastic part in band
• 5
  Realistic tolerance bands by plastic family
• 6
  How to write tolerances that a plastic shop can actually hold
• 7
  Tolerance expectation by plastic family

A tolerance is the band of acceptable variation around a nominal dimension. On a drawing it looks like a pair of numbers, but on a plastic part it is the result of several moving inputs: the moulding or extrusion history of the stock, the heat generated at the cutter, the moisture sitting inside the polymer, and the force the fixture applies while the part is held. Change any one of them and the finished size moves.

Metals are forgiving in one specific way: they are stiff. A 6061 aluminium bracket pushed by a vise springs back by a few micrometres. A POM block of the same size does not. It bends, creeps under the clamp load, and then relaxes after you release it. The dimension you measure on the machine is not always the dimension the customer measures three days later.

That gap is why plastic tolerances have to be written with the material in mind. The same ±0.05 mm callout that is comfortable on a stainless shaft can be a coin flip on a thin-walled PA cover. It is not that the machine is worse. The workpiece itself is moving.

The practical takeaway: decide early whether the critical dimension is a fit, a seal, or appearance. A fit needs a real number and an inspection plan. A seal often needs surface finish more than size. Appearance needs a flatness or waviness limit, not a diameter.

• 1
  Stiffness sets the floorLow-modulus plastics deflect under clamp load and recover slowly.
• 2
  Thermal expansion sets the driftMost polymers expand 5–10× more than steel per degree.
• 3
  Moisture sets the long-term shiftPA and POM absorb water and grow after machining.

Every polymer has a coefficient of linear thermal expansion (CLTE) far above steel. Filled grades sit lower, unfilled grades sit higher. Over a 20 °C workshop swing, a 200 mm unfilled PA part can move more than a 200 mm steel part by a wide margin. That movement happens while the cutter is running, so the first pass and the last pass are not cutting the same geometry.

The countermeasure is not complicated. Rough the part, let it sit, then finish. For tight work we rough to within 0.3–0.5 mm, allow the part to reach room temperature, and take the finishing pass with light depths of cut. Coolant or chilled air helps hold the workpiece steady during the final pass.

Measure at the same temperature you will inspect at. A part pulled hot off the machine and gauged immediately will read small, then grow as it equalises. If your incoming inspection happens on a cold dock in winter and the machine shop sits at 24 °C, add that difference to the tolerance stack before you blame the supplier.

For long parts the error is not uniform. A 500 mm extrusion-machined rail can bow as the top surface warms faster than the core. Symmetric material removal helps. So does flipping the part between passes instead of cutting one face to final depth.

• 1
  Rough, rest, finishLeave 0.3–0.5 mm for the final pass on tight features.
• 2
  Match measurement temperatureInspect at the same temperature the part was cut at.
• 3
  Balance material removalCut both faces evenly to limit bowing on long parts.

PA, POM and to a lesser degree ABS take up water from the air. Machined dry, a PA part can grow measurably over the following days as it reaches equilibrium with the shop or the customer's site. The change is slow but it is real, and it is one reason a drawing that demands a tight bore in PA6 is harder than the same drawing in aluminium.

Creep matters when the part carries load. A plastic boss clamped hard in a fixture will keep deforming while the spindle runs. Reduce clamp pressure, support the part underneath, and use soft jaws or a fitted nest. Where the geometry allows, machine the part in a pocket that matches its outline so the clamp force spreads over a larger area.

Elastic recovery shows up most on thin walls and small bores. The cutter pushes the material away, the material springs back, and the finished hole comes out undersize. A spring pass at the same setting, or a light reaming pass with a sharp tool, usually brings the bore back into band.

Not every plastic behaves the same way here. PEEK and filled grades are stiff and stay put. PP and HDPE are soft and move a lot. PMMA machines cleanly but chips easily at the edges. Match the strategy to the grade, not to the family name on the drawing.

• 1
  Dry or condition firstAsk for the moisture state before tight boring on PA or POM.
• 2
  Lower the clamp loadSoft jaws and fitted nests reduce creep during the cycle.
• 3
  Allow a spring passA second light pass corrects undersize bores from recovery.

Sharp tooling is the first rule. A dull carbide end mill rubs instead of shearing, and rubbing generates heat right at the cut. Use polished flutes and high rake angles where the material allows it. Two-flute cutters clear chips well in soft plastics; three-flute tools work better in filled and stiff grades.

Spindle speed should be high and feed per tooth moderate. For most unfilled plastics a surface speed in the range of 200–500 m/min with a chip load of 0.05–0.15 mm per tooth keeps the cut cool. Flood coolant or strong air blast carries heat away and stops chips from welding back onto the finished face.

Hold the part rigidly but gently. Vacuum chucks work well on thin plates. For prismatic parts, machine soft jaws to the part outline so the contact is broad rather than a few points. Never clamp a plastic part across an unsupported span.

Inspect with the right tool. A micrometer squeezed onto a soft plastic wall reads whatever the anvil forces it to read. Use non-contact measurement, a light-touch gauge, or a fixture that supports the wall from behind. Document the method so the number means the same thing on both sides of the shipment.

• 1
  Sharp, polished toolsReplace or touch up cutters before the finish pass.
• 2
  High speed, moderate chip load200–500 m/min surface speed keeps heat out of the cut.
• 3
  Support, do not crushBroad contact area beats high clamp pressure every time.

General machining tolerances on plastic parts are looser than on metal, and that is a material fact rather than a shop limitation. For most unfilled thermoplastics, ±0.05 mm on a small feature is a reasonable working target and ±0.10 mm is comfortable. Filled and stiff grades such as PEEK, POM-C and glass-filled PA can hold ±0.02–0.05 mm on stable geometry.

Very tight callouts are still possible on the right part. A small, thick, well-supported PEEK component can be held to ±0.005 mm when the geometry cooperates, because the material is stiff and dimensionally calm. The same callout on a thin PA housing is not a machining problem, it is a physics problem.

Geometry limits tolerance more than the material grade does. Thin walls under 1 mm, long unsupported spans, sharp internal corners and deep pockets all move under cutting load. When a drawing has several of these together, expect the achievable band to widen even on a good grade.

The honest answer for a tight callout is a conversation, not a number. Send the drawing, the material, the wall thicknesses and the function of the critical dimensions. That is enough to say whether ±0.02 mm is routine, possible with extra passes, or something to redesign.

• 1
  Unfilled grades±0.05 mm workable, ±0.10 mm comfortable on most features.
• 2
  Filled and stiff grades±0.02–0.05 mm on stable, well-supported geometry.
• 3
  Thin walls and long spansExpect the achievable band to widen regardless of grade.
• 4
  Verify before quotingFunction of the critical dimension decides the real target.

Put the tight tolerance where the function is, and only there. A mounting hole that locates a bearing needs a real number. A clearance hole for an M4 screw does not. Every extra tight callout adds inspection time and scrap risk, and it does not make the part better.

Call out datum features that exist on the finished part, not on the stock. Plastics move during machining, so a datum taken from a rough face may not relate to the final geometry. Pick stable, machined faces for your datum scheme and say so on the drawing.

State the material grade and the moisture condition. 'PA6' is not enough. The difference between dry-as-moulded and conditioned nylon shows up directly in the finished bore size. If the part will live in a humid or wet environment, tell the shop, because that changes the target dimension.

Add a measurement note. Say what instrument, what temperature and whether the part is supported during gauging. This single line removes most of the arguments that happen when a shipment arrives and the numbers do not match the certificate.

• 1
  Tight only where it mattersReserve close callouts for fits and seals.
• 2
  Datums on machined facesRough stock faces move and make poor references.
• 3
  Name the grade and stateDry vs conditioned nylon changes the final bore size.
• 4
  Write the measurement methodInstrument, temperature and support should be on the drawing.