A professional injection molding supplier controls manufacturing tolerances by managing the full process from CAD review to production inspection. Mold dimensions alone are not enough. A 100 mm plastic feature with 0.5% molding shrinkage changes by about 0.50 mm, while a shift from 0.5% to 0.7% adds another 0.20 mm. Resin grade, wall thickness, gate position, packing pressure, mold temperature, cooling time, and measurement method all affect the final dimension. Reliable suppliers define CTQ dimensions, machine mold components to measured limits, validate samples with CMM or gauges, establish repeatable processing ranges, and monitor production with statistical methods such as Cp and Cpk.
Tolerance planning starts with the drawing because not every number printed beside a CAD feature can be held economically in molded plastic. ISO 20457:2018 provides a structured reference for dimensional tolerances of molded plastic parts, while ISO 2768 is commonly seen on general engineering drawings but should not be treated as a substitute for plastic-specific tolerance assessment. A supplier reviews nominal dimensions, datum structure, fit conditions, wall thickness, resin, draft, ribs, bosses, holes, and assembly interfaces before machining the tool.
A ±0.05 mm requirement on a 6 mm locating pin is very different from ±0.05 mm across a 180 mm molded housing. The second feature is more exposed to shrinkage and warpage because dimensional change accumulates over a longer flow and cooling distance. A supplier should therefore separate functional dimensions from general geometry instead of assigning the same inspection effort to 100% of drawing characteristics.
A tolerance that passes on three first-shot samples is not automatically suitable for a 100,000-part production run. Repeatability matters more than one favorable measurement.
Material behavior is checked next because molded polymers contract after filling the cavity. Published shrinkage ranges vary by resin grade and processing condition. Many amorphous engineering plastics have lower and more uniform shrinkage than semi-crystalline materials, while reinforced grades can show directional behavior because fibers align with melt flow.
Representative engineering ranges can look like this, although the resin manufacturer's current data sheet should always be used for the selected commercial grade:
| Material family | Typical molding shrinkage range | Main dimensional concern |
|---|---|---|
| ABS | about 0.4–0.8% | wall-thickness and packing differences |
| Polycarbonate | about 0.5–0.7% | residual stress and mold temperature |
| Polypropylene | about 1.0–2.5% | higher shrinkage and warpage |
| POM/acetal | about 1.8–2.5% | crystallization and post-mold change |
| Glass-filled nylon | often about 0.2–1.0% depending on direction | anisotropic shrinkage from fiber orientation |
A 150 mm polypropylene part at 1.5% nominal shrinkage requires roughly 2.25 mm of dimensional compensation at the tooling stage. If real production shrinkage reaches 1.7%, the difference is about 0.30 mm over the same length, which can exceed an assembly allowance even though the mold itself was machined accurately.
For that reason, a Custom mold engineering company should not apply one shrinkage percentage to an entire part without reviewing flow direction, wall sections, gate location, and resin reinforcement. Mold-flow simulation can help estimate filling pressure, weld-line position, temperature distribution, fiber orientation, and possible deformation before steel cutting, but simulation data still needs confirmation during actual mold trials.
Tool design then determines how evenly the material fills, packs, cools, and leaves the cavity. Gate position affects pressure distribution. A gate placed too far from a tolerance-sensitive area may leave that area under-packed, while an oversized or poorly positioned gate can increase local stress or create directional shrinkage.
Cooling layout is equally important because the mold removes most of the thermal energy before ejection. A part released at 90°C does not behave like the same part released at 65°C. If one wall cools faster than the opposite wall, shrinkage can become uneven and produce bow or twist even when every cavity surface is within machining tolerance.
For a production cycle of 30 seconds, increasing cooling by only 3 seconds adds 10% to cycle time. Suppliers therefore cannot solve every dimensional issue by extending cooling indefinitely. Cooling channel position, diameter, coolant flow, mold temperature, and thermal balance need to be designed so that dimensional stability and production rate can coexist.
Machining accuracy establishes the physical starting point. Modern precision tooling may combine high-speed CNC milling, grinding, wire EDM, sinker EDM, jig grinding, and polishing, depending on the feature. Mold inserts that control a tight bore or shutoff surface are normally inspected before final assembly rather than waiting for the first molded parts.
A tool component measured at 20.010 mm when the drawing target is 20.000 ±0.005 mm is already outside its machining requirement before resin enters the cavity. Finding the error during insert inspection is much cheaper than discovering it after T0 samples, disassembly, correction, reassembly, and another molding trial.
CMM inspection is often used for three-dimensional features, datum relationships, hole positions, flatness, and profile requirements. Pin gauges can be faster for high-volume hole checks, while optical systems may suit small or easily distorted features where contact pressure could change the reading.
Measurement conditions also need definition. Plastics continue changing temperature after molding, and some materials absorb moisture from the environment. A dimension checked 5 minutes after molding can differ from the same feature after 24 hours of conditioning. For hygroscopic polymers such as nylon, moisture conditioning may cause additional dimensional change, so drawing requirements should state the expected condition when it affects fit.
A measurement system should have enough resolution and repeatability to separate process variation from gauge variation. Using a 0.01 mm-resolution tool to judge a ±0.01 mm feature provides very little practical measurement margin.
Trial molding connects tooling dimensions with real material behavior. T0 parts are useful for detecting major mold, filling, venting, ejection, and dimensional issues, but a stable process normally needs further adjustment. Suppliers record melt temperature, mold temperature, injection speed, V/P transfer position, peak pressure, holding pressure, holding time, cooling time, screw recovery, and cycle time instead of relying on operator memory.
Packing deserves close attention because material continues entering the cavity after initial filling. If holding pressure changes from 40 MPa to 60 MPa, part weight and local shrinkage may also change until the gate freezes. Increasing hold time after gate freeze generally provides little additional dimensional benefit because material can no longer flow through the gate.
Part weight is often useful during process development because it gives a fast indication of filling and packing consistency. If a 32.00 g component moves repeatedly between 31.60 g and 32.40 g, the 2.5% spread deserves investigation before dimensional approval, especially when the changing mass is associated with critical packed sections.
Mold temperature also needs a controlled range rather than a single display setting. A controller set to 80°C does not prove that every cavity surface is at 80°C. Cooling-channel restriction, scale, flow imbalance, ambient conditions, and distance from the heater or coolant line can create local temperature differences.
Multi-cavity molds add another layer of control. An 8-cavity mold produces eight sets of dimensions each cycle, not one. If seven cavities average 10.00 mm and cavity 6 repeatedly averages 10.08 mm, changing the molding machine settings may shift all eight cavities and create new problems elsewhere.
Cavity identification allows parts to be measured separately. During qualification, a supplier may collect 5 parts from each of 8 cavities, producing a 40-part dimensional dataset rather than pooling the pieces together. The results can show whether variation is mainly cavity-to-cavity, cycle-to-cycle, or related to one mold position.
Statistical capability becomes useful after the process is stable. Cp compares process spread with the specification width, while Cpk also considers how well the process is centered. A process can have a reasonable Cp but a weaker Cpk if the average dimension sits close to one specification limit.
For example, a feature specified at 20.00 ±0.10 mm has a total tolerance width of 0.20 mm. If process standard deviation is 0.02 mm, six standard deviations occupy 0.12 mm. That provides more room than a process with a 0.04 mm standard deviation, where six standard deviations occupy 0.24 mm and already exceed the drawing tolerance.
Many industrial quality systems use capability targets such as Cpk 1.33 for established production, although customer requirements can be higher or lower depending on product risk and contract terms. A Cpk of 1.33 corresponds to the nearest specification limit being about four standard deviations from the process mean when the process assumptions are appropriate.
The number itself is not enough. Capability should be calculated from a process that is stable and from a sample plan large enough to represent normal production. Thirty measurements provide more information than three, but high-volume qualification may use 50, 100, or more observations across cavities, shifts, machine restarts, or resin lots when the application warrants it.
Long production runs also introduce tool wear. Gates can erode, vents can become contaminated, slides can wear, ejector systems can loosen, and cooling passages can lose efficiency through deposits. A 0.02 mm change at a shutoff or insert may be enough to affect flash, fit, or cavity-specific dimensions on a precision component.
Preventive maintenance therefore belongs in tolerance control. A mold may be inspected after a defined number of cycles such as 50,000, 100,000, or another interval based on resin abrasiveness, mold construction, and historical wear. Glass-filled materials can increase wear on gates and cavity surfaces compared with many unfilled grades, so one universal service interval is not appropriate.
Raw-material control continues during production. Switching from one resin grade to another because both carry the same polymer family name can change viscosity, filler content, shrinkage, and moisture behavior. Even a 0.2% change in effective shrinkage over a 200 mm feature represents 0.40 mm of dimensional movement.
Drying also matters for hygroscopic materials. Resin suppliers publish specific drying temperatures, times, and allowable moisture levels for each grade. Production records should therefore identify the approved material, lot, drying condition, permitted regrind level, and any color or additive package that can affect processing behavior.
Dimensional inspection during production should focus more frequently on CTQ features than on dimensions with wide functional allowance. A supplier may inspect selected dimensions at startup, after machine interruption, after material changes, at scheduled intervals, and at the end of the run, while simpler go/no-go checks can be performed more often when suitable.
Process records make those measurements useful months later. When a 2026 production batch differs from an approved 2025 batch, engineers should be able to compare resin lot, machine, cavity, tool-maintenance history, mold temperature, packing settings, cycle time, and dimensional results instead of attempting to reconstruct the conditions from memory.
A buyer assessing molding capability can ask for more than a statement that ±0.05 mm is possible. Ask how the tolerance was established, over what feature size, in which resin, using how many cavities, under what conditioning method, with what inspection equipment, and from how many measured production parts.
A supplier that can show a 50-part dimensional study, cavity-separated measurements, an approved processing window, calibrated CMM records, material traceability, and Cp/Cpk data provides much more useful information than a supplier quoting one tolerance number without the measurement conditions behind it.