Professional mold solutions improve production accuracy by controlling the sources of dimensional error before full-scale molding begins. Tool machining, resin shrinkage, gate position, cooling balance, cavity pressure, steel temperature, and ejection all affect finished dimensions. One established injection-molding supplier publishes mold-machining capability around ±0.003 in. (±0.08 mm), while resin-related dimensional allowance can range from about 0.2% to 2.5% depending on material and geometry. ISO 20457:2026 separately addresses dimensional and geometrical tolerances for plastic molded parts because metal-part tolerance methods do not adequately represent polymer shrinkage and warpage. Accurate production therefore depends on the mold, material, process settings, measurement method, and repeatability working together.
A mold can match its CAD model and still produce parts outside drawing limits. Plastic changes dimensions after filling because pressure, temperature, molecular orientation, cooling rate, and local wall thickness are not uniform throughout the cavity. ISO 20457:2026 specifically recognizes molding shrinkage, processing conditions, geometry, warpage, and non-uniform cooling when defining achievable tolerances for molded plastics.
That difference becomes easier to see when shrinkage is expressed as a percentage. A published injection-molding design guide reports shrink allowance near 0.002 in./in. for dimensionally stable materials such as ABS and polycarbonate, equivalent to about 0.2%. At the other end, some TPE applications may approach 0.025 in./in., or 2.5%. A nominal 100 mm feature exposed to those two rates can therefore behave very differently after cooling.
Because material movement can exceed machining error, cavity dimensions cannot simply copy finished-part dimensions. Mold engineers normally compensate for expected shrinkage while checking which dimensions are constrained by steel, which cross a parting line, and which depend heavily on packing or cooling. A 0.05 mm machining error and a 1% molding contraction are different problems and require different controls.
Wall thickness then becomes part of the same calculation. A thick rib intersection retains heat longer than a nearby thin wall, so the two regions may contract at different times. When a 2 mm nominal wall grows locally to 4 mm around a boss or rib junction, the cooling history can change enough to affect flatness, sink, or hole location even though the surrounding cavity geometry is accurate.
For that reason, professional mold work normally reviews geometry before cutting steel. Draft, ribs, bosses, deep pockets, shutoffs, slides, lifters, inserts, sealing faces, datum surfaces, and assembly interfaces should be checked together rather than as isolated CAD features. Tolerance should be assigned according to how the part is molded and measured, not only according to nominal geometry.
Tool machining still sets the physical starting point. A commercial precision-machining reference lists general plastic machining deviations around ±0.05 mm for dimensions from 0.5 to 6 mm, ±0.10 mm from over 6 to 30 mm, and ±0.15 mm from over 30 to 120 mm under its published fine-tolerance conditions. Tighter features need specific drawing requirements and appropriate equipment.
Those figures also explain why every dimension should not receive the tightest possible tolerance. Holding ±0.05 mm on a cosmetic cover dimension that could function at ±0.20 mm adds machining, inspection, and adjustment work without improving assembly. ISO 20457:2026 was issued in August 2026 with guidance on technically achievable tolerances, giving designers a plastics-specific framework instead of treating molded polymers like machined metals.
Machining accuracy alone is not enough because the cavity is exposed to injection pressure during every cycle. Large projected areas can place substantial force on mold plates and inserts. If support pillars, plate thickness, insert seating, or interlocks allow small movement under pressure, a dimension may measure correctly on the open mold yet change while the cavity is filled.
Cooling follows the same mechanical logic but works through temperature. Water channels located too far from one cavity area can leave a local hot region, while a nearby channel may cool another surface much faster. A difference of only several degrees Celsius repeated over thousands of cycles can produce a stable but unwanted dimensional bias rather than random inspection noise.
A good dimensional result is not one perfect sample. It is a controlled distribution of parts produced across many cycles, cavities, operators, material lots, and normal machine conditions.
Multi-cavity tools make that requirement harder. An eight-cavity mold does not have one filling path; it has eight cavities that should receive comparable melt conditions. Runner length, gate size, venting, cooling-water distribution, cavity machining, and local pressure can make cavity 1 behave differently from cavity 8 even when every cavity came from the same 3D model.
Published multi-cavity guidance from Protolabs illustrates how geometry is restricted to help manage consistency: allowable surface-area difference between paired parts is listed at 20% for parts below 127 mm, 15% from 127 to below 254 mm, 10% from 254 to below 508 mm, and 5% for larger stated ranges. The figures are supplier-specific, but they show why cavity balance is treated quantitatively rather than visually.
Gate position is tied closely to that balance. A gate near a thick region may support packing differently from a gate feeding through a long thin section. Melt reaching one side of a part earlier can also create unequal orientation and shrinkage. For glass-filled polymers, flow direction matters because fiber orientation can make longitudinal and transverse contraction different.
Before steel modification, simulation can compare gate locations, filling sequence, pressure distribution, weld-line position, temperature, and likely warpage. Simulation does not replace molding trials, but it can narrow the number of layouts that need to be tested physically. For a mold planned for 500,000 or 1,000,000 parts, avoiding one avoidable steel revision has much more production relevance than saving a few hours during early design review.
Venting affects dimensions for a similar reason. Air that cannot leave the cavity raises resistance near the end of fill. Operators may respond by increasing pressure or speed, which can alter packing and flash behavior. A venting problem therefore may appear first as a filling defect but later show up as inconsistent weight, local dimensions, or parting-line condition.
Ejection can introduce another error after the cavity has already formed the part. A thin housing pushed by four small ejector pins may flex more than the same housing supported by twelve well-positioned pins or a stripper system. If the polymer is still hot when released, deformation can remain after the component reaches room temperature.
Measurement therefore needs a defined timing and method. A plastic dimension taken 30 seconds after ejection may not match the same feature after 24 hours of conditioning. Moisture-sensitive resins can change further after environmental exposure, so the drawing, inspection plan, conditioning requirement, and measuring equipment should describe the same state of the part.
A practical inspection plan separates dimensions by function. Datum locations, sealing surfaces, bearing seats, connector interfaces, snap features, hole spacing, and flatness may justify CMM or optical measurement, while less demanding external dimensions can be checked with gauges or conventional instruments. A 30-cavity dimensional report is more useful when measurements are identified by cavity number rather than pooled into one average.
| Production source | Example quantity to control | What engineers compare |
|---|---|---|
| Tool machining | ±0.05 to ±0.15 mm in published fine machining ranges | Steel measurement vs. CAD |
| Material contraction | About 0.2% to 2.5% in one published molding guide | Molded dimension vs. conditioned dimension |
| Multi-cavity balance | 2, 4, 8, 16+ cavities | Cavity-to-cavity weight and dimensions |
| Inspection | 30-piece or larger study sets where appropriate | Mean, range, standard deviation |
| Long production | 100,000 to 1,000,000+ cycles | Wear, dimensional drift, maintenance history |
The machining figures and molding allowances in the table are reference examples rather than universal promises; achievable tolerance changes with resin, geometry, mold construction, machine condition, and inspection method. Protolabs, for example, publishes about ±0.003 in. (±0.08 mm) mold-machining accuracy while separately stating that resin allowance may be greater than the machining component.
Statistical production checks add information that a first-article report cannot provide. If 30 consecutive parts all meet a ±0.10 mm drawing limit but cluster close to the upper boundary, the process has less room for normal production variation than another 30-part sample centered near nominal. Average dimension, standard deviation, cavity identity, machine settings, and sampling time give engineers a better view of repeatability.
Process settings should then be recorded with the dimensional results. Melt temperature, mold temperature, injection speed, transfer position, holding pressure, holding time, cooling time, and cushion can change how much polymer enters and remains packed inside the cavity. Changing holding pressure by 10% may have little effect on one dimension and a measurable effect on another near the gate.
A production-ready mold should therefore be checked across more than one machine cycle at a single setting. Engineers commonly compare samples after the process has reached thermal stability, then review whether small permitted setting changes push dimensions toward specification limits. A mold that only produces acceptable parts at one narrow setting is harder to reproduce across long production runs.
The same principle applies when tooling is transferred between facilities. Machine screw diameter, clamp behavior, temperature control, dryer performance, water supply, and measurement practice may differ. Documented process conditions reduce the risk of treating the original machine’s numerical settings as if they were universally transferable.
Suppliers such as Qlution Manufacturing can support accuracy by connecting mold design, machining, molding trials, dimensional inspection, and later tooling adjustments within the same manufacturing workflow. The practical benefit comes from comparing steel measurements, molded-part data, cavity identification, and processing records before changing the tool.
Tool wear becomes more important as production moves from 10,000 cycles toward several hundred thousand cycles. Gates can wear, vents can become contaminated, slides and lifters can develop clearance, ejector components can loosen, and glass-filled resin can abrade exposed steel surfaces. A dimension approved in the first qualification run should not be assumed to remain unchanged throughout the mold’s service life.
Maintenance records help connect dimensional movement with physical mold condition. If cavity 4 gradually produces a larger flash line while cavities 1–3 remain stable, inspecting the local shutoff and alignment surfaces is more informative than changing global machine settings. Similar cavity-level records can reveal whether a recurring dimensional difference follows one insert, one cooling circuit, or one wear location.
Accuracy therefore comes from controlling several measurable sources at the same time: cavity geometry, polymer contraction, filling balance, temperature distribution, pressure response, ejection, inspection timing, and tool condition. ISO 20457:2026 formalizes why molded plastics require their own tolerance approach, while current manufacturing references show that machining capability around ±0.08 mm may still sit beside resin movement measured in tenths of a percent or more.