A professional injection molding supplier controls part tolerances by managing the entire dimensional process, from DFM and mold machining to resin conditioning, cavity pressure, cooling, measurement, and production monitoring. A ±0.05 mm drawing tolerance can be difficult to hold when polymer shrinkage alone may range from about 0.4% to more than 2%, depending on resin, wall thickness, fiber orientation, and processing conditions. Suppliers therefore validate dimensions with molded samples rather than relying only on mold dimensions. CMM inspection, cavity-by-cavity measurements, Gauge R&R, SPC, and Cp/Cpk studies are commonly used to determine whether production can repeatedly remain inside specification.
Tolerance work starts before mold steel is cut because a drawing dimension is not equally difficult in every location. ISO 20457:2018 provides a framework for dimensional tolerances on molded plastic parts, while actual manufacturing capability still depends on resin, geometry, mold construction, and process conditions. A ±0.05 mm requirement across a short bore formed by one insert can be much easier to maintain than the same tolerance across 150 mm of plastic spanning the core and cavity.
A supplier therefore reviews how each important dimension is physically formed. Dimensions crossing a parting line can include alignment and mold-closing variation; dimensions formed by slides add positioning clearance; long dimensions accumulate thermal contraction over a greater distance. A 0.1% dimensional change is only 0.02 mm across 20 mm, but it becomes 0.15 mm across 150 mm. That difference affects how the mold should be designed before machining begins.
Mold accuracy and molded-part accuracy are not the same measurement. A cavity can be machined within ±0.01 mm while the finished plastic part moves much more because of shrinkage, packing, temperature, moisture, or warpage.
Material selection is therefore connected directly to dimensional planning. Typical molding shrinkage may be roughly 0.4–0.7% for some ABS grades, around 0.5–0.7% for many polycarbonate grades, and approximately 1–2% or more for some polypropylene and acetal grades. Grade-specific supplier data must be used because fillers, additives, processing conditions, and geometry can change those ranges.
Glass fiber changes the problem again. A 30% glass-filled polymer can shrink differently along and across the melt-flow direction because fibers become oriented during cavity filling. A nominal 100 mm feature cannot be corrected reliably by applying one universal shrinkage percentage to every axis. Gate position, flow path, wall thickness, fiber orientation, and local packing need to be considered together.
That material behavior explains why professional toolmakers often leave selected dimensions steel-safe during first-tool construction. Rather than machining every critical cavity surface directly to its theoretical final position, removable inserts or adjustable steel conditions provide room for correction after T0 or T1 molded parts have been measured.
For example, assume a functional molded feature is specified at 25.00 ±0.05 mm. Initial samples average 25.08 mm after 24 hours of conditioning. If measurement repeatability and process stability have already been verified, the mold insert can be adjusted toward nominal in a controlled amount. Several small steel corrections are generally preferable to removing too much material in one machining operation.
| Source of dimensional change | Typical engineering check | Production response |
|---|---|---|
| Resin shrinkage | Grade-specific molding data | Adjust cavity allowance |
| Uneven cooling | Mold surface temperature | Balance cooling circuits |
| Packing variation | Pressure and hold-time study | Establish process window |
| Fiber orientation | Flow direction and gate position | Modify gate or geometry |
| Cavity imbalance | Samples separated by cavity | Correct individual cavity |
| Measurement variation | Gauge R&R | Improve gauge or fixture |
Once the mold enters sampling, process development becomes as important as tool dimensions. Injection speed affects filling and molecular orientation; transfer position controls when velocity filling changes to pressure control; packing pressure affects material compensation before gate freeze. Even when machine settings are unchanged, actual cavity conditions can differ if resin viscosity or mold temperature changes.
Holding time should therefore be established around gate sealing rather than selected only from machine cycle targets. A common study increases hold time in small steps and weighs the molded parts after each setting. When part weight stops increasing materially, additional hold time is no longer feeding significant material through the frozen gate. A study might use 6–10 settings at intervals of 0.5–2 seconds, depending on gate size and resin.
Cooling follows because most of the molding cycle is spent removing heat from the polymer. A controller set to 60°C does not prove that every cavity surface is operating at 60°C. Water-flow differences, deposits inside cooling channels, hose restrictions, channel distance, and local steel thickness can create temperature differences across the mold.
If one side of a 120 mm housing runs several degrees warmer than the opposite side, the two regions can contract at different rates. The result may appear as bow, twist, hole-position movement, or flatness variation rather than a simple overall size change. Engineers may therefore measure mold-surface temperatures at multiple locations after thermal equilibrium has been reached.
For a dimensional study, recording only the machine setpoint is incomplete. Actual melt behavior, cavity filling, cooling conditions, and measured part dimensions need to be compared under the same production run.
Cavity pressure sensing can provide another layer of information for tight-tolerance programs. A pressure sensor near the gate or end of fill records what occurs inside the mold rather than only what the molding machine reports. Changes in peak pressure, pressure integral, or the pressure curve can be compared with dimensional results from the same production samples.
This becomes useful when dimensional limits are narrow. If a 40.00 mm feature has a total tolerance band of 0.10 mm, a gradual process shift of 0.02–0.03 mm already consumes 20–30% of that band. Monitoring process behavior allows the supplier to investigate the change before the measurement reaches the drawing limit.
Multi-cavity tooling adds another statistical issue. A 16-cavity mold producing a 10-second cycle can theoretically release 5,760 parts per hour before downtime and rejects, so a small cavity-specific dimensional difference can affect thousands of components within one shift. Mixing all cavities into one inspection sample can make the average look acceptable while one cavity is close to a specification boundary.
Professional suppliers therefore keep cavity identification where dimensional requirements justify it. If 30 parts from cavity 7 average 19.96 mm while 30 parts from cavity 12 average 20.03 mm, the 0.07 mm difference points toward a cavity-specific condition rather than random process variation. Gate size, venting, cooling, insert position, or local wear can then be checked separately.
Measurement timing also matters because plastic continues changing after ejection. Parts cool toward room temperature, internal stress can relax, and hygroscopic polymers can absorb moisture. A measurement taken 10 minutes after molding cannot automatically be compared with a customer measurement taken after 24 or 48 hours unless the inspection method defines the conditioning state.
For that reason, inspection documentation should state the conditioning period, ambient temperature, measurement equipment, datum system, fixture, and sampling method. ISO 291 specifies standard atmospheres for conditioning and testing plastics, and 23°C is widely used as a reference laboratory temperature. A supplier should not mix hot, freshly molded measurements with stabilized inspection results.
Equipment must also match the tolerance being evaluated. A handheld caliper with 0.01 mm display resolution is not automatically suitable for proving a ±0.02 mm feature. Bore gauges, micrometers, pin gauges, optical systems, dedicated fixtures, or a CMM may provide better repeatability depending on feature geometry and GD&T requirements.
Gauge R&R is used when measurement variation needs to be separated from manufacturing variation. One common study structure uses 10 parts, 3 operators, and 2 or 3 repeated measurements, producing 60 or 90 observations. The study shows whether differences between readings are coming mainly from the molded parts or from the measurement method and operators.
After the measurement system has been checked, capability data can describe production performance. Cp compares process spread with the specification width, while Cpk also accounts for the location of the process mean. Many manufacturing programs use 1.33 as a common Cpk acceptance reference, although the required level depends on customer, application, feature risk, and qualification rules.
A supplier should not claim capability from five first-article parts. A larger sample taken under stable production conditions gives a more useful picture of the distribution. For example, 100 sequential measurements can reveal whether a nominal 50.00 ±0.10 mm dimension is centered near 50.00 mm or operating persistently near 50.08 mm despite every measured part still passing inspection.
Statistical process control extends that work into production. Measurements collected by shift, lot, machine, or cavity can show movement in the process before rejected parts appear. A 0.03 mm shift over 20 consecutive samples deserves investigation even when every value remains within a ±0.10 mm drawing tolerance.
The cause may not be the mold. A different resin lot, moisture level, dryer performance, machine condition, coolant flow, or startup procedure can change molded dimensions. Hygroscopic resins require particular attention because moisture affects processing behavior, while post-molding moisture uptake can also change the final dimensions of materials such as polyamide.
Material records therefore belong beside dimensional records. A controlled production lot can document resin manufacturer, grade, batch number, drying temperature, drying duration, allowable regrind percentage, molding machine, mold number, cavity number, and inspection time. When a dimensional shift appears after 50,000 cycles, engineers can compare production history instead of relying on memory.
Tool wear becomes more relevant as cycle counts increase. Gates, shutoffs, slides, lifters, leader components, ejector systems, and replaceable inserts do not remain in their original condition indefinitely. A mold that passed qualification in 2025 may require dimensional verification after substantial production in 2026 even if the original process settings have not changed.
Preventive maintenance records should therefore connect cycle count with tooling condition. If a wear-sensitive insert controls a ±0.03 mm dimension, checking it at planned intervals can be less expensive than sorting tens of thousands of molded parts after wear has already moved the process. Cooling circuits also need maintenance because deposits can reduce heat transfer and alter previously stable dimensions.
Supplier evaluation should cover the entire control system rather than a statement that a factory can “hold ±0.05 mm.” Buyers can ask for the tolerance review method, mold correction strategy, material-conditioning requirements, cavity identification plan, measurement equipment, Gauge R&R approach, capability target, sampling frequency, maintenance records, and dimensional reports from representative production runs.
For international sourcing, a Plastic injection molding supplier China or a supplier in the United States, Germany, Mexico, or another manufacturing region should be evaluated against the same measurable requirements. Location does not establish dimensional capability; documented mold construction, validated processing conditions, calibrated metrology, traceable material control, and production data do.
A useful approval package may include 30–100 measured samples for selected characteristics, cavity-separated results for multi-cavity molds, material certificates, process settings, dimensional reports, and capability data where required. Automotive or medical programs may add PPAP-related documentation, measurement-system studies, control plans, or customer-specific validation requirements.
When a dimension falls outside tolerance, mold modification should not be the automatic first response. The supplier should first confirm the measurement method and conditioning state, then compare process data, resin lot, cavity number, machine condition, and recent maintenance history. If 4 of 64 cavities move while the remaining 60 remain stable, individual tooling or cooling conditions deserve more attention than a global process adjustment.
If every cavity shifts by approximately the same percentage after a material change, resin or processing conditions deserve more attention. If one operator obtains readings 0.04 mm different from two other operators across a 30-part sample, the inspection method needs review before steel is altered. Separating those sources prevents a correct mold from being modified to compensate for a measurement or process problem.
For production quantities of 100,000 or 1 million parts, tolerance control is therefore demonstrated through repeatability over time rather than one successful first-article inspection. Stable dimensions require the mold, resin, molding process, cooling system, measurement method, cavity condition, and maintenance history to remain within defined operating ranges, with recorded data showing when any part of that system begins to change.