Yes. A professional injection molding supplier can produce tight-tolerance parts, but repeatability depends on material, geometry, tool construction, process control, and measurement. A ±0.05 mm requirement on a 50 mm feature equals a total tolerance band of only 0.10 mm, so small changes in shrinkage, cavity temperature, packing pressure, or moisture can move production outside specification. ISO 20457:2026 addresses dimensional and geometrical tolerances for plastic molded parts, while ISO 294-4:2018 defines methods for measuring molding and post-molding shrinkage. Reliable production comes from controlling the full molding system, not from machining the cavity accurately once.
Plastic molding tolerances should be evaluated differently from metal machining tolerances because the part continues changing after the cavity is filled. ISO 20457:2026 specifically notes that molded plastics can show greater variation in size, form, and location because of shrinkage, warpage, part geometry, and non-uniform cooling. A 0.2% dimensional change across a 100 mm housing equals 0.20 mm, already four times larger than a ±0.025 mm drawing tolerance.
That difference makes material behavior the first practical check. Mold shrinkage is not one fixed number for an entire polymer family; resin grade, reinforcement, melt temperature, mold temperature, packing conditions, and flow direction all affect the molded result. ISO 294-4:2018 separately measures shrinkage parallel and normal to melt flow, which matters when a 30% glass-filled resin places fibers mainly along the filling direction.
A drawing may call for ±0.05 mm, but the supplier must first determine whether the material, feature length, flow direction, and measurement condition can hold that range over normal production.
Geometry comes next because dimensions do not move independently. A compact 20 mm boss with uniform surrounding walls is usually easier to control than a 200 mm open housing with thin walls, ribs, clips, and several thickness transitions. A wall changing from 2.0 mm to 4.0 mm contains twice the section thickness, so the thicker area cools more slowly and can pull neighboring surfaces as the polymer contracts.
That cooling difference leads directly to mold design. A precision mold needs more than accurate cavity steel; it also needs balanced water circuits, stable inserts, appropriate gate positions, adequate venting, and enough stiffness to resist pressure. If one cavity surface runs 8°C hotter than another during a long production run, local cooling and shrinkage can change even though the CNC-machined dimensions have not moved.
Gate location has a similar effect because it sets the filling path and packing pattern. In fiber-filled materials, the gate also affects fiber orientation, which can change shrinkage by direction. For an eight-cavity tool, a supplier should therefore compare cavity 1 through cavity 8 separately rather than averaging all measurements together. An average can look acceptable while one cavity repeatedly approaches an upper specification limit.
Multi-cavity comparison also affects how process capability should be interpreted. If a 50.00 ±0.05 mm feature produces cavity averages between 49.97 and 50.04 mm, the total spread already uses 70% of the allowed 0.10 mm tolerance band before normal within-cavity variation is considered. Pooling all cavities can hide the source of that variation.
A professional Custom molded plastic parts supplier should therefore ask which dimensions affect assembly, sealing, bearing location, gear engagement, electrical alignment, or another measurable function. Requiring ±0.03 mm on 25 unrelated dimensions can make a mold far more difficult to qualify than requiring it on three functional features. ISO 20457:2026 was developed partly to help designers and manufacturers agree on technically achievable tolerances instead of applying metal-part assumptions to plastics.
The drawing also needs a defined measurement condition. Some thermoplastics absorb moisture after molding, while nearly all plastics respond to temperature to some degree. A part measured 15 minutes after ejection may not match the same part after 24 hours of conditioning. For a tolerance of ±0.05 mm, the supplier should document when the part is measured, where datums are established, how the part is supported, and what inspection equipment is used.
Common inspection choices differ by feature:
| Feature | Typical measurement approach | Main concern |
|---|---|---|
| Hole or pin diameter | Pin gauge, bore gauge, CMM | Form error and measurement force |
| Datum-to-hole position | CMM or vision system | Datum setup and fixture repeatability |
| Flatness | CMM or optical system | Part support and natural warpage |
| Small molded detail | Vision system | Edge definition and magnification |
| Assembly fit | Functional gauge | Combined dimensional stack |
Measurement quality matters because inspection itself contributes variation. If repeated measurements of the same feature vary by 0.02 mm and the full drawing tolerance is only 0.10 mm, measurement spread alone occupies 20% of that tolerance. Gauge repeatability and reproducibility studies are therefore useful before a supplier uses inspection data to approve a tight process.
Once measurement is reliable, the molding process has to remain stable. Filling speed, transfer position, melt temperature, mold temperature, holding pressure, holding time, cooling time, and material preparation all affect final dimensions. A supplier producing 100,000 parts cannot rely on one good setup from the first trial; parameters need controlled limits that operators can reproduce after a machine stop, mold service, or material-lot change.
Part weight is often a useful process indicator. If a molded component normally weighs 42.0 g and begins running at 41.4 g, the difference is about 1.4%. That does not automatically prove a dimensional defect, but it can point to changes in packing, material feed, or cavity filling before dimensional failures become widespread.
Holding pressure deserves particular attention because it compensates for material contraction while the gate remains open. After the gate freezes, extra holding time no longer adds material to the cavity. A 12-second hold may therefore give the same dimensional result as 18 seconds if gate freeze occurs at 10 seconds, while the extra 6 seconds only increases cycle time.
Mold temperature should be treated with the same discipline. A process specified at 80°C should not be assumed stable merely because the controller displays 80°C. Cooling channels can foul, flow rates can differ, and inserts can create local temperature differences. Checking actual mold temperatures and coolant flow is more useful than relying on one controller value during a 20-hour production run.
Tool correction should happen only after the process is reasonably stable. If a hole is 0.08 mm undersize because of excessive local shrinkage, changing steel immediately can compensate for the first trial while leaving the underlying process unstable. A better sequence is to establish filling, packing, cooling, and conditioning first, then modify the insert if the dimension remains consistently offset.
Making one sample at 50.00 mm is relatively easy. Keeping a production population near 50.00 mm across multiple cavities, shifts, material lots, and maintenance cycles is the harder engineering task.
Process capability data can show whether that population is comfortably inside specification. For example, if a 40.00 ±0.10 mm dimension has a measured standard deviation of 0.02 mm, six standard deviations span 0.12 mm, or 60% of the total 0.20 mm tolerance band. If the process average then moves by only 0.04 mm, available margin becomes much smaller.
A supplier should also separate dimensional variation caused by the process from variation caused by tool wear. A mold producing 1 million cycles may experience measurable wear at gates, shutoffs, slides, lifters, or frequently contacted inserts. Preventive maintenance records help determine whether a dimension is slowly changing because of process settings or because tooling surfaces are physically changing.
Material control supports the same goal. Hygroscopic resins need controlled drying, and reinforced grades should be run within the material supplier’s processing recommendations. A 30% glass-filled grade should not be treated as interchangeable with an unfilled version of the same polymer because stiffness, shrinkage, flow, and anisotropy differ substantially.
For demanding programs, the supplier should normally provide several forms of production information rather than only a first-article report:
-
cavity-specific dimensional results from an agreed sample size, such as 5 parts from each of 8 cavities;
-
documented molding parameters and acceptable operating ranges;
-
material lot and drying records where applicable;
-
calibrated inspection records;
-
capability data for selected CTQ dimensions;
-
mold maintenance history after defined cycle intervals.
Sample quantity matters when evaluating repeatability. Measuring 3 parts can show whether a dimension is obviously wrong, but it says little about a run of 50,000 units. A study using 30 or more observations gives a much better view of process spread, especially when samples are collected across different periods instead of consecutively from one short production window.
Not every very tight dimension should be molded directly. A plastic housing may be economical to mold to ±0.10 mm overall while one bearing bore requires ±0.015 mm. In that case, post-mold reaming or CNC machining may cost less than forcing the entire molding process, tool, cooling system, inspection plan, and scrap allowance to support the tighter requirement.
The commercial effect can be large. If tightening a specification increases rejection from 1% to 6% on an annual volume of 500,000 parts, scrap rises from 5,000 to 30,000 parts before rework, inspection labor, or machine time is counted. Tight tolerances should therefore be placed on dimensions that affect measured product function rather than applied broadly across the drawing.
Supplier qualification should focus on evidence from similar work. Ask for examples involving comparable polymers, feature sizes, cavity counts, and tolerance levels; then ask how the supplier controlled temperature, cavity balance, conditioning, and measurement. A company that has held ±0.05 mm on a 15 mm PBT connector feature has useful experience, but that result does not automatically prove it can hold ±0.05 mm across a 250 mm glass-filled nylon housing.
ISO 20457:2026 reinforces that plastic molded parts need a plastics-specific approach to dimensional and geometrical tolerancing, while ISO 294-4:2018 provides a standardized way to determine molding and post-molding shrinkage. Using both ideas in production leads to a practical question: not whether a supplier can make one dimension once, but whether its material controls, mold design, processing limits, and inspection system can keep the required dimension inside tolerance over the planned production volume.