Why the tolerance block decides your mold budget
Send the same 3D file to three molders and the quotes will rarely land within 20% of each other. The gap is almost never the resin. It is the tolerance block on the drawing. A ±0.1 mm callout on a non-critical wall costs almost nothing to hit. Tighten that same wall to ±0.02 mm and the conversation changes: more cavity steel for safe corrections, trial shots after every steel-safe adjustment, SPC charts, and a quote that grows 30–60% before anyone cuts metal. On plastic parts, a tolerance is not just a spec — it is a budget line.
What injection molding can actually hold
Most molders work from published standards similar to SPI or DIN 16901, and the numbers are size-dependent. On a 50 mm feature, the following ranges are realistic for a competent shop:
- Commercial: roughly ±0.1 mm on small dimensions, drifting toward ±0.2–0.3 mm once parts pass 150–200 mm, because shrinkage accumulates with scale.
- Tight: around ±0.05 mm, achievable with solid mold construction and a stable process window.
- Precision: ±0.01–0.02 mm, possible but only with controlled material lots, disciplined drying, and often 100% inspection. Every dimension held at this level needs a reason.
Note what is usually missing from incoming CAD files: a datum scheme. Calling every face critical guarantees that none of them will be. Molders build cavities around defined datums; if your drawing has none, they will pick their own, and their choice will not match your assembly fixture.
Shrinkage is the physics you are fighting
Every tolerance problem in molding starts with shrinkage — the part comes out smaller than the cavity and keeps moving for hours after ejection. Typical values: ABS and PC sit near 0.4–0.7%; PA12 runs higher, often 1–2%, and the nylon family shifts again as it absorbs moisture over several days. Glass filling cuts linear shrinkage roughly in half but introduces anisotropy — flow-direction shrink differs from cross-flow shrink — which is why glass-filled flat panels warp, and why the molder asks about gate location before promising any flatness number.
This is also why a dimension that "measured fine on the samples" can drift at production scale. A different press, a different cooling loop, even a new resin lot moves the mean. The tolerance you pick has to absorb that ongoing variation, not just the first-shot spread.
Five DFM moves that keep tolerances affordable
1. Tolerance only what functions
Walk the drawing and ask one question per dimension: what fails if this is off by 0.1 mm? If the honest answer is "nothing," leave it on the default tolerance. On a typical enclosure, maybe a dozen features truly need better than ±0.1 mm — connector cutouts, bearing seats, sealing surfaces. Everything else is cost without benefit.
2. Kill thick sections before tightening tolerances
Uneven walls are the number-one cause of sink marks, voids, and the warpage that drags your ±0.05 mm features out of spec. Aim for uniform wall thickness — 1.5 to 3 mm covers most consumer and industrial parts — and design ribs at 50–60% of the nominal wall. A molder running a stable process on a uniform part will hold tight numbers far more cheaply than a heroic process fighting a lumpy one.
3. Design for steel-safe iteration
Removing steel from a mold (which makes the part bigger) is easy; adding it back is expensive. If a critical bore lands near the tolerance edge after T1 samples, the correction is to re-cut the cavity larger and re-try. Experienced designers deliberately leave room on non-visible, non-functional surfaces so corrections move steel out, not in. Raise this with your molder before tool kick-off — it changes how they build the cavity and how much margin they leave on it.
4. Keep critical dimensions off the parting line
Features straddling the parting line inherit flash variation from every shot; features formed by sliding cores inherit slide wear over the life of the mold. A sealing groove toleranced across a slide is a promise the mold can only keep for a while. Where the geometry allows, move critical geometry onto surfaces formed entirely within one mold half.
5. Respect post-mold movement
Parts keep shrinking — and sometimes bending — for 24 to 48 hours after molding, and hygroscopic resins change dimension again as they take on moisture. If you gauge parts an hour off the press, you are measuring a transient condition. Agree on a measurement window (24 hours at room temperature is a common compromise) and on conditioning rules for nylons, or your incoming inspection and the molder's outgoing inspection will be arguing forever over parts that are both "correct."
Material quick notes
- ABS: forgiving, dimensionally stable, paints and glues well. The default for housings, and the easiest material in which to hold tight callouts.
- PC: tougher and more stable than ABS, but it demands high melt temperatures and careful drying; watch for molded-in stress around tight-tolerance holes.
- PA12: low friction and tough, but moisture-driven dimensional change makes it a poor partner for ±0.02 mm callouts unless conditioning is controlled end to end.
When molding precision is not enough
Some features genuinely cannot be molded to spec: small bearing bores, optical mounting pads, deep threads in hard resins. The standard answer is to mold the geometry close and finish the critical features in a secondary operation. Our plastic injection molding team quotes parts this way routinely. For pre-tool validation, adding machined features to printed blanks via CNC machining tests the drawing long before steel is cut, and vacuum casting fills the gap for appearance prototypes that need production-like surfaces without production tooling cost.
Getting the tolerance block right is the cheapest engineering hour you will spend on a molded part. Spend it before the mold steel is ordered, not after.
