Quality control matters when selecting an OEM injection mold supplier because the mold must repeat the same geometry across thousands or millions of cycles, not merely produce an acceptable T1 sample. ISO 20457:2018 addresses tolerances for plastic molded parts, while ASTM D955-21 notes that mold temperature, melt temperature, filling conditions, and packing conditions can materially affect shrinkage. For dimensional characteristics, process capability gives more information than pass/fail inspection: NIST data show that a centered process with Cp 1.33 uses about 75% of its specification width. A supplier should control design, steel, machining, molding parameters, measurement, revisions, and maintenance as one manufacturing system.
The first place to examine is what happens before mold steel is machined. A supplier should review wall thickness, ribs, bosses, draft, undercuts, shut-offs, gate position, venting, ejection, expected shrinkage, cooling layout, parting lines, and dimensions that affect assembly. ISO 20457 was published in 2018 specifically to provide a structured basis for manufacturing tolerances on molded plastic parts. A drawing tolerance still cannot compensate for poor part geometry, so DFM comments should be resolved before CNC or EDM work starts.
A 2.0 mm wall beside a 4.0 mm section, for example, cools and shrinks differently. Increasing holding pressure may reduce one sink mark while increasing residual stress elsewhere. ASTM D955-21 makes the same broader point: shrinkage depends on mold design and processing conditions, and standardized test specimens cannot predict every real component because actual parts have different wall thicknesses, flow paths, pressure gradients, and temperature gradients. That relationship makes the supplier's engineering review more useful than a quotation based only on CAD volume and tool size.
A mold sample that measures within tolerance once does not show whether the process can hold that tolerance after 10,000, 100,000, or 1,000,000 molding cycles.
Once geometry is reviewed, material control becomes the next test of the supplier's quality system. Tool steel should match the purchase specification, with traceable grade, heat-treatment status, hardness, and source documentation where the program requires them. A mold intended for 20,000 prototype shots has different wear requirements from a production tool expected to run 1 million cycles. Glass-filled polymers also place more abrasive wear on gates, runners, cavity surfaces, sliders, and shut-offs than many unfilled grades.
The buyer should therefore request records for components that affect mold life rather than accepting a generic material statement. A practical file may include:
-
steel certificates and insert identification;
-
heat-treatment and hardness records;
-
purchased-component specifications;
-
cavity and core revision numbers;
-
electrode or EDM records when relevant;
-
dimensional inspection results after machining;
-
replacement-part drawings for wear components.
Those records become more useful during machining, because assembled molds can hide individual component errors. CNC milling, EDM, wire EDM, grinding, polishing, fitting, and heat treatment each introduce dimensional variation. If a cavity feature is 0.03 mm oversize before assembly, process adjustment on the molding machine may not provide a stable correction. Measuring inserts and shut-offs before final assembly allows the supplier to correct steel while the affected component remains accessible.
Measurement equipment also needs to match the feature being checked. A caliper may be appropriate for a non-functional external dimension with a wide tolerance, while a CMM, optical system, bore gauge, height gauge, micrometer, gauge pin, or surface instrument may be required elsewhere. ISO 9001:2015, used by more than 1 million organizations in 189 countries according to ISO, requires organizations to manage resources, controlled processes, documented information, monitoring, measurement, and nonconforming outputs. Certification does not guarantee a good mold, but an audit can show whether those controls exist in daily production.
| Area to review | Useful supplier record | What the buyer can verify |
|---|---|---|
| Mold steel | Certificate, hardness result | Grade and treatment match the specification |
| Cavity machining | CMM or dimensional report | Features were checked before assembly |
| Mold trial | Parameter sheet | Settings can be reproduced |
| Molded parts | FAI or cavity report | Dimensions are tied to individual cavities |
| Revisions | Change log | T1, T2, and later changes are identifiable |
| Maintenance | Shot-count service record | Wear parts are inspected at planned intervals |
Inspection becomes more informative when the supplier separates individual cavities. In an 8-cavity mold, measuring one part cannot establish that all 8 cavities produce the same result. Samples should retain cavity identification, especially for dimensions related to sealing, snap fits, connector positions, bearings, alignment, or automated assembly. A cavity that runs 0.08 mm differently from the other seven may still disappear inside an averaged inspection report, so cavity-level records are more useful for troubleshooting.
Pass/fail inspection is only one layer. NIST defines process capability by comparing process performance with specification limits and gives Cp as the specification width divided by six standard deviations. For a centered normal process, NIST's reference example associates Cp 1.00 with about 0.27% outside specification, while Cp 1.33 corresponds to about 64 parts per million and uses roughly 75% of the available tolerance width. Real molding processes are not always perfectly centered or normally distributed, which is why Cpk and the underlying data should be reviewed rather than treating 1.33 as a universal acceptance rule.
That distinction matters during mold trials. T0, T1, and T2 should generate more than molded samples and photographs. The supplier should record resin grade, drying condition where required, barrel temperatures, mold temperature, injection speed, transfer position, peak pressure, holding pressure, holding time, cooling time, screw speed, back pressure, cushion, shot weight, and total cycle time. Without those records, two sample sets produced in 2026 may have the same dimensions for very different reasons.
A stable trial also needs enough consecutive cycles to show whether the mold reaches thermal balance. Five hand-selected parts taken immediately after startup provide less information than samples collected after the mold temperature has stabilized. Buyers can specify a production run and sampling plan that fits program risk: for example, measurements from every cavity at multiple time points rather than 5 pieces from one cavity. The required sample size should be agreed before mold acceptance instead of chosen after a dimensional problem appears.
Cooling deserves particular attention because published injection-molding research has reported that cooling can account for up to 75% of the molding cycle. Poor cooling-channel placement can create local hot areas, longer cycles, warpage, and cavity-to-cavity differences. Consider a tool reduced from a 30-second cycle to 25 seconds without changing part quality. At 1 million cycles, machine time falls from about 8,333 hours to 6,944 hours, a difference of roughly 1,389 machine hours.
That calculation explains why mold quality should not be evaluated only by the tooling invoice. Suppose Supplier A quotes $45,000 and Supplier B quotes $50,000. A 10% lower tool price looks attractive, but a longer cycle, 2% scrap rate, repeated local repairs, or an extra qualification round can use that difference quickly. Buyers working with an Engineering plastic injection molding supplier should compare the assumptions behind cycle time, mold life, steel selection, cavity count, inspection, spare components, and process qualification rather than comparing the final quotation line alone.
Process records also matter because polymer dimensions continue to reflect material and molding conditions after the part leaves the cavity. ASTM D955-21 specifies shrinkage measurements at 24 and 48 hours and states that melt temperature, mold temperature, fill time, and packing conditions can significantly affect molding shrinkage. Measuring a dimension immediately after ejection and approving the mold from that result may therefore be unsuitable when the material or application requires dimensional conditioning before inspection.
The same discipline applies to engineering changes. A customer may change a boss by 0.20 mm after T1, adjust a sealing surface after T2, or replace one resin grade with another before production approval. The part drawing, mold drawing, insert revision, CNC data, inspection report, sample label, and molding record should all refer to the same revision. In a 6-month development program with several modifications, an uncontrolled old insert can cause dimensional disagreement even when the molding machine is set correctly.
Quality records should also follow the mold after approval. Ejector pins, sliders, lifters, gates, shut-offs, leader components, seals, cooling channels, springs, and textured surfaces wear at different rates. A supplier expecting a tool to exceed 1 million cycles should identify serviceable areas, recommended spare parts, lubrication points, cleaning requirements, and inspection intervals. Shot-count-based maintenance provides a clearer reference than waiting until flash, sticking, dimensional change, or water-flow reduction appears during scheduled production.
For overseas mold transfers, documentation becomes even more practical. A receiving molder may use a different press brand, controller, water system, dryer, or plant environment. Providing the approved process sheet, cooling connections, resin information, mold dimensions, required clamp force range, shot size, sample reports, spare-part drawings, and 2026 revision status gives the receiving team a known starting condition. Missing records force technicians to recreate settings while production equipment is already consuming machine time.
Before issuing a purchase order, an OEM buyer can ask the supplier for one completed project file with commercial information removed. The file should allow the buyer to follow one part from drawing revision through DFM, steel preparation, machining inspection, mold trial, dimensional approval, correction, and release. A functioning quality system leaves measurable records at every manufacturing stage, not only a final inspection sheet created before shipment.
During an audit, several checks can be completed without complicated scoring:
-
Select 3 measuring instruments and verify calibration status.
-
Select 3 mold inserts and compare identification with drawings.
-
Review one T1 report and confirm recorded process settings.
-
Compare one dimensional report with the customer drawing revision.
-
Check samples from every cavity of one multi-cavity mold.
-
Review one nonconformance and the documented correction.
-
Compare the latest mold revision with the physical insert installed.
A supplier that can provide those records consistently offers the buyer more information about repeatability than showroom appearance, machine quantity, or a low tooling quote. When the mold enters serial production, dimensional capability, cycle time, scrap percentage, maintenance frequency, cavity balance, material handling, and revision control continue affecting the cost and reliability of every part produced.