Injection molding tolerances, explained

DIN 16742, in plain terms

DIN 16742 is the standard molders use for general plastic part tolerances when a drawing does not call out a tighter figure explicitly. It does not give one number for a whole part. It sets a tolerance group per material, a group that reflects how much and how uniformly that material shrinks in the tool, and then a table of allowed deviation banded by the size of the dimension: a 20 mm feature and a 200 mm feature in the same material do not carry the same absolute tolerance, even inside the same group.

On our own tooling, at the 100 mm reference point used to compare groups:

This is the table published on our injection molding capabilities page; it is the reference to design against before a tighter figure is discussed. We publish DIN 16742 groups for these six resins because they cover the large majority of our production. For a grade not listed here, ask at quote stage rather than assuming a group: the standard's groups are set per material family and a filled or specialty grade can sit outside this table.

Shrinkage percentage is not the same thing as tolerance

It is tempting to read a material's mold shrinkage percentage off its datasheet and assume that number predicts how tight a tolerance it can hold. It does not, on its own. Shrinkage percentage tells you how much a material contracts moving from melt temperature to room temperature. Tolerance tells you how repeatable that contraction is, part after part, once the tool has been cut to compensate for it.

POM is the clearest example on our own material pages. Its mold shrinkage runs 1.8 to 2.5 percent, one of the higher figures we publish, higher than ABS at 0.4 to 0.7 percent or PC at 0.5 to 0.7 percent. Yet POM holds the same tight TG5 group as PC in the table above. The reason is that POM is highly crystalline and picks up very little moisture, so its shrinkage, while larger in absolute terms, is unusually uniform from shot to shot and cavity to cavity. A toolmaker can compensate a large, predictable number with precision. A smaller but more variable number is often the harder one to hold.

The practical takeaway: a wide shrinkage range on a datasheet is a better warning sign than a high shrinkage midpoint. A material quoted as a single fixed percentage is telling you it behaves consistently; a material quoted across a wide band is telling you that band moves with moisture content, wall thickness, or hold pressure, and the tool has to be cut for a compromise somewhere inside it.

Mold shrinkage across the resins we run

Published on each material's own page, reproduced here for comparison. Materials without a published figure are metals run on our CNC capability, not molded, or grades where we confirm the shrinkage window against the specific supplier lot at quote stage rather than publish a fixed range.

PEEK and PEI are run in production but we confirm the shrinkage window by grade and glass loading at quote stage rather than publish one range, because filled and unfilled grades of both differ enough to make a single published number misleading.

What tightening a tolerance actually costs

Nothing about a tight callout is free, and none of it is negotiable once a dimension is flagged. Four things change together:

• Cavity machining time. EDM and hard-milled steel finished to a tight figure takes measurably longer than a standard machining pass, before the tool ever sees plastic.
• Try-out iterations. A "steel safe" cavity, cut slightly undersized on a critical dimension and opened up after first samples confirm the real shrinkage, is the standard way we hold ±0.05 mm on a feature. That is an extra sampling round on the schedule, not a one-shot cut.
• First-article and in-process inspection. A CMM-verified dimension is measured on a fixture, not read off a caliper, and that measurement is repeated through the production run at an agreed frequency, not just once.
• Cavity-to-cavity matching. On a multi-cavity tool, a tight dimension has to be held to the same figure in every cavity, which is a tighter machining and polishing spec across the whole mold base, not just one insert.

None of this is reversible cheaply. It is far easier and cheaper to open up a cavity that was cut steel-safe than to add steel back to one that was cut to a loose tolerance and then needs to tighten. See what drives injection molding part cost for how tolerance sits alongside cavitation, tool steel and finish as a cost driver.

Which dimensions are hardest to hold

How to call out tolerance on your drawing

1. State the general tolerance as a reference to the standard, for example "DIN 16742, medium" in the drawing's general tolerance block, rather than leaving it blank. A blank block gets quoted against our default assumption, which may be looser than you need.
2. Flag only the dimensions that carry a fit or a function with a tight individual callout. A bore that mates with a bearing, a snap feature, a sealing groove. Everything else should sit on the general tolerance.
3. Use basic dimensions with a datum scheme for anything that assembles, not a chain of plus-or-minus dimensions off arbitrary edges. A chain stacks tolerance error at every link; a shared datum does not.
4. Call out flatness or parallelism separately from a linear dimension where a part has to sit against another surface. A linear tolerance alone does not control warp.
5. Tell us the grade, not just the family, on the same drawing. PA66 and PA66-GF30 sit in different DIN 16742 groups, per the first table above, so the material line on the drawing is itself a tolerance input.

A drawing that follows this sequence prices faster because nothing has to be assumed, and it samples faster because the tool was cut against a real target instead of a guess. For the rest of what we ask for at quote stage, see how to get an accurate injection molding quote.