A precision injection molding supplier holds tight tolerances by controlling mold geometry, polymer shrinkage, cavity pressure, temperature, cooling, material condition, and measurement as one production system. ISO 20457:2026 addresses dimensional and geometrical tolerances for molded plastic parts because plastics respond differently from metals to temperature, shrinkage, warpage, and processing conditions. ASTM D955-21 also measures shrinkage at both 24 and 48 hours rather than treating the dimension at ejection as final. For demanding molded features, tolerances around ±0.025–0.05 mm may be achievable when geometry and resin permit, but stable production requires cavity-specific data, calibrated metrology, controlled material lots, and statistically capable processes.
A drawing may specify a narrow dimensional band, but the mold cavity cannot simply be machined to the finished part size. Thermoplastics contract after packing and cooling, and ASTM D955-21 notes that mold and melt temperature, filling conditions, packing, pressure gradients, and geometry all affect shrinkage. The standard uses measurements after 24 and 48 hours because some polymers continue changing after molding. A supplier therefore starts with resin-specific shrinkage information and then adjusts it for the actual wall thickness, flow path, gate position, reinforcement, and expected pressure distribution.
A catalog shrinkage value is useful for estimating cavity size, but it cannot predict the final dimensions of a complex production part by itself.
That limitation matters when dimensional requirements fall below 0.10 mm. A 100 mm feature with 0.5% nominal shrinkage loses about 0.50 mm from cavity size to finished size; a change of only 0.05 percentage points in effective shrinkage changes the result by roughly 0.05 mm. The tolerance can therefore be consumed by material and process behavior before tool-machining error is considered. ASTM D955 specifically warns that data from standard specimens cannot predict absolute dimensions in parts with different wall sections, pressure gradients, flow paths, and process conditions.
Once shrinkage risk is understood, the supplier reviews the geometry rather than treating every drawing dimension equally. Long unsupported walls, thick-to-thin transitions, deep ribs, tall bosses, off-center gates, and large flat surfaces usually require more tolerance allowance than compact features near a stable datum. A 2.0 mm wall next to a 4.0 mm mass of plastic cools at a different rate, so local contraction can pull an otherwise accurate feature away from its nominal position.
For that reason, an early dimensional review usually separates features into functional groups:
-
sealing, bearing, snap-fit, gear, connector, or alignment dimensions that need close control;
-
cosmetic or clearance dimensions that can accept wider variation;
-
dimensions crossing a parting line, slide, lifter, or separate insert;
-
dimensions influenced by post-mold moisture or conditioning;
-
geometric requirements such as flatness, position, perpendicularity, and profile rather than simple length.
ISO 20457:2026 provides a plastics-specific tolerance framework rather than applying metal-part assumptions to molded polymers. The 2026 edition covers general tolerances, directly specified tolerances, acceptance conditions, and technically achievable manufacturing accuracy. That distinction helps keep close tolerance only where the function requires it, reducing unnecessary mold correction and inspection work.
The geometry review then feeds into mold design. Steel dimensions must include expected shrinkage, but mold construction also needs stable datum surfaces, accurate insert location, sufficient backing support, repeatable slide positioning, controlled shutoffs, and predictable cavity-to-cavity alignment. In an 8-cavity mold, eight acceptable parts do not prove the process is balanced if cavity 2 averages 0.04 mm smaller than cavity 7.
Multi-cavity variation is often easier to see when each cavity is treated as a separate data population. A practical qualification might collect 30 parts from every cavity after the process reaches thermal stability, producing 240 measurements for one critical feature in an 8-cavity tool. Means and standard deviations can then reveal cavity offsets that disappear when all 240 values are pooled into one average.
Mold construction accuracy alone still cannot guarantee molded dimensions because the polymer experiences different conditions during filling and packing. Gate size and location determine how easily pressure reaches each region before the gate freezes. If a gate freezes early, additional machine hold pressure may no longer compensate for shrinkage in distant or thick sections.
Machine settings describe what the molding machine is commanded to do; cavity conditions show what the polymer actually experiences.
That difference explains why cavity-pressure monitoring is used in demanding precision programs. RJG notes that machine parameters do not always represent actual conditions inside the cavity because viscosity changes, mold imbalance, and machine behavior can change filling and packing. Pressure sensing inside the mold provides information closer to the point where molded dimensions are formed.
A validated production window commonly controls injection velocity, transfer position, peak pressure, hold pressure, hold time, cushion, melt temperature, mold temperature, cooling time, screw recovery, and cycle time. Changing several settings together may produce one acceptable sample, but it does not show which variable actually controls a dimension.
DOE-style process development can separate those relationships. For example, three mold-temperature settings, three pack-pressure settings, and two hold-time settings create 18 processing combinations before replication. Measuring 10 parts from each condition gives a 180-part data set that can show whether a bore diameter responds mainly to packing pressure, temperature, or an interaction between both.
Temperature deserves separate control because plastic parts do not cool uniformly simply because the machine cycle is repeatable. Water flow restrictions, scale, different cooling-line distances, insert materials, or unequal circuit temperatures can change local heat removal. Even a temperature difference of several degrees Celsius between two mold regions can change local crystallization, residual stress, and contraction in materials sensitive to cooling history.
Cooling design therefore affects tolerance as much as cycle time. Flow should remain repeatable, circuits should be monitored and maintained, and production should begin only after the mold reaches a stable thermal condition. Parts collected during the first 10 cycles after a cold startup may not represent the dimensions reached after 50 or 100 cycles of steady production.
Material preparation comes next because resin is another measurable source of dimensional variation. Hygroscopic polymers such as many polyamide, polycarbonate, PBT, PET, and TPU grades must be processed within supplier-recommended moisture limits. Moisture can affect melt viscosity and molecular integrity during processing, while some finished parts also absorb moisture after molding and change size over time.
For reinforced materials, fiber orientation adds another variable. A 30% glass-fiber-filled grade may shrink differently parallel and perpendicular to polymer flow, so one universal shrinkage allowance across a complex housing is rarely adequate. Gate location, weld regions, local thickness, and flow direction should therefore be considered before final steel dimensions are approved.
The molding machine also needs to suit the shot rather than simply provide enough clamp force. Very small shots in an oversized barrel can produce longer residence time and less stable material delivery, while worn screws, check rings, or barrels can change shot repeatability. When a molded part weighs only 2 g, a 0.02 g change represents 1% of part mass; that scale of variation may matter when tight packed dimensions depend on consistent material delivery.
Machine consistency connects directly to Consumer product injection molding because housings, closures, wearable-device parts, personal-care components, and small mechanical assemblies often need both dimensional fit and visible surface quality. A part may pass length and width requirements yet fail assembly because snap features, insert locations, shutoff geometry, or flatness moved outside the functional range.
Measurement therefore has to match the feature being controlled. A ±0.03 mm specification should not be treated like a ±0.30 mm clearance dimension. CMMs can measure datum-based position and profile; optical systems can measure small non-contact features; pin gauges can verify holes; dedicated fixtures can check assembled function quickly during production.
Measurement repeatability should also be quantified. A Gauge R&R study commonly uses multiple operators, repeated measurements, and a sample covering the production range. A study with 3 operators measuring 10 parts twice creates 60 readings, enough to show whether apparent manufacturing variation is partly coming from the inspection method rather than the molding process.
Plastic conditioning must be fixed before those measurements are compared. ASTM D955-21 includes 24-hour and 48-hour shrinkage measurements because dimensions can continue changing after molding. Inspection instructions should therefore define elapsed time after molding, temperature, humidity where relevant, and any required conditioning rather than allowing one batch to be measured after 30 minutes and another after 24 hours.
Once the measurement method is stable, capability data becomes more useful than pass/fail inspection alone. Cp compares process spread with the tolerance width, while Cpk also accounts for how far the process mean sits from the specification center. A dimension can have low scatter and still carry rejection risk if its average sits close to one limit.
Consider a feature specified at 20.00 ±0.10 mm. The full tolerance width is 0.20 mm. If the process standard deviation is 0.020 mm, six standard deviations occupy 0.120 mm, leaving some tolerance space for normal variation. Moving the average from 20.00 to 20.06 mm does not change the spread, but it moves production much closer to the upper 20.10 mm limit.
Sampling must also reflect cavity and time. Measuring 5 parts from one cavity during mold approval says little about a 16-cavity tool scheduled for hundreds of thousands of cycles. A stronger plan samples all cavities, multiple production intervals, restart conditions, and more than one resin lot when the program risk justifies the additional data.
Traceability then connects an out-of-specification measurement to its manufacturing history. Useful records include material lot, mold ID, cavity number, machine, date, process revision, dimensional results, maintenance work, and approved parameter limits. If cavity 6 alone begins increasing by 0.03 mm after 200,000 cycles, cavity identification reduces the amount of production that has to be investigated.
Tool maintenance closes the loop because qualified steel does not remain unchanged indefinitely. Gates wear, shutoffs move, slides develop clearance, ejector components wear, vents collect residue, and cooling channels accumulate deposits. A dimensional trend detected at 100,000-cycle intervals can therefore reveal gradual tooling change before parts exceed the drawing limit.
A capable supplier ties those records back to controlled tool correction instead of cutting steel after the first failed inspection. When a dimension moves 0.05 mm, engineers first check measurement repeatability, material condition, thermal stability, cavity-specific behavior, packing response, and tool condition. Steel adjustment is appropriate only after the source of the dimensional offset has been separated from normal processing variation.
For a buyer evaluating a supplier, useful evidence is measurable rather than promotional: recent CMM capability, cavity-specific inspection records, process-window documentation, Gauge R&R results, mold-maintenance history, resin handling records, and Cp/Cpk data from comparable parts. Asking for 30 or more consecutive measurements from a representative production run provides more information than reviewing several hand-selected samples.
ISO 20457:2026 also helps the supplier and customer use a common framework for dimensional and geometrical acceptance. The standard was published in August 2026 as the second edition and replaced the 2018 edition, reflecting the need to define molded-plastic tolerances around actual manufacturing behavior rather than machining assumptions.
A supplier able to maintain tight tolerances will usually be able to explain the full path from drawing tolerance to cavity dimension, resin shrinkage allowance, process window, inspection method, cavity data, and long-term maintenance history. Repeatability across cavities, material lots, restarts, and long production runs matters more than producing one dimensionally perfect first article.