A precision injection molding supplier should be evaluated by measurable production capability, not by machine count or the lowest quotation. Review whether the company can repeatedly hold specified dimensions, document resin lots, control drying and mold temperatures, measure parts with suitable equipment, and reproduce an approved process across several production runs. For a ±0.03 mm feature, ask for dimensional data from at least 30–50 parts rather than five selected samples. Review Cpk by cavity, tool-maintenance records, calibration history, reject rates, machine utilization, and production traceability. A supplier that produces one acceptable sample has not yet demonstrated repeatable production.
The drawing should be reviewed before the factory is evaluated. Mark dimensions that affect fit, sealing, alignment, electrical contact, mechanical movement, or assembly. A housing with ±0.20 mm general tolerances presents a different manufacturing requirement from a connector containing four dimensions at ±0.03 mm and a flatness requirement of 0.05 mm.
Material behavior needs to be included in that review. A 30% glass-filled polyamide, unfilled polycarbonate, PBT, PPS, and polypropylene do not shrink, absorb moisture, or respond to mold temperature in the same way. The supplier should explain how fiber orientation, gate position, packing pressure, cooling time, and wall-thickness changes can affect the dimensions shown on the drawing.
A useful supplier review starts with one question: which dimensions are difficult to reproduce after the mold has reached normal production temperature, not which dimensions can be measured on the first sample.
That discussion should continue into mold design because dimensional consistency is partly established before the first molding cycle. Ask how cavity steel dimensions are developed from the nominal part size, expected resin shrinkage, mold temperature, gate layout, and correction allowance.
For an eight-cavity mold, dimensional information should identify each cavity separately. Measuring 40 parts without cavity identification can hide a situation where seven cavities are centered while one cavity runs close to the upper specification limit. A more useful study might contain five parts from each of eight cavities, followed by another set after several hours of production.
Cooling deserves equal attention because uneven mold temperature changes local shrinkage. Review cooling-line placement around deep cores, thick bosses, slides, and areas where one side of the part contains substantially more steel than the other. A cycle reduced from 30 seconds to 24 seconds raises theoretical output by 25%, but it is not an improvement if the shorter cooling period increases warpage or dimensional rejection.
The molding machine then needs to match the mold rather than merely provide enough clamping force. Compare shot size, screw diameter, injection-pressure capacity, plasticizing rate, tie-bar spacing, platen size, injection speed, and control resolution.
A very small shot in a large barrel can increase residence time and make material exposure less consistent. A process operating close to the machine's maximum injection-pressure capability leaves little operating margin. A supplier should be able to explain why a particular 80-ton, 150-ton, or 300-ton press was assigned to the tool rather than simply saying that the mold fits.
Machine condition should also appear in the assessment. Review preventive-maintenance records, screw and barrel inspection practices, heater checks, hydraulic or servo-system maintenance, and the history of alarms or unplanned stops. A machine purchased in 2024 is not automatically more repeatable than a well-maintained machine installed in 2018.
Once machine selection is understood, look at how the process is established. The production record should include more than barrel-temperature settings and cycle time. Useful data can include fill time, transfer position, peak injection pressure, holding pressure, holding time, cushion, screw-recovery time, mold temperature, and total cycle time.
The supplier should define an approved operating range instead of depending on one exact set of values. If parts pass only when holding pressure is 72 MPa but fail at 70 or 74 MPa, the manufacturing window may be too narrow for comfortable long-run production.
Changes made during production should be recorded and tied to a reason. Frequent undocumented adjustments make later quality analysis difficult because two lots carrying the same part number may have been molded under noticeably different conditions.
Material preparation is the next place to check. Hygroscopic engineering resins can change during processing when moisture is not controlled. Review resin storage, dryer type, drying temperature, drying time, hopper residence time, conveying, material identification, regrind rules, and the method used to prevent one resin from entering another production lot.
ASTM D618-21 addresses conditioning plastics for testing and notes that temperature and relative humidity can materially affect measured properties. It also notes that some materials may require 20–100 days or more to approach substantial equilibrium, depending on material, thickness, and prior exposure. Production inspection therefore needs a defined measurement condition rather than an informal rule such as “measure after molding.”
For moisture-sensitive materials, ask for actual dryer records from previous production. A written procedure may specify four hours of drying, but production records show whether the requirement was followed at 2:00 a.m. during a weekend shift as consistently as during a customer audit.
Traceability should connect those material records to finished parts. A normal lot record may include:
-
resin manufacturer and resin lot;
-
colorant or additive lot;
-
mold number and cavity identification;
-
molding machine;
-
production date and shift;
-
approved process revision;
-
inspection results;
-
packing and shipment lot.
A useful audit test is to select one carton from a completed lot and ask the supplier to reconstruct its manufacturing history. If the records can be retrieved within the quality system and linked without guessing, the traceability process is more useful than a certificate hanging in the reception area.
Measurement capability should be reviewed with the same level of detail. A supplier working with ±0.02 mm dimensions may need a CMM, optical measurement system, calibrated micrometers, pin gauges, or purpose-built fixtures, depending on geometry.
Equipment resolution alone is not enough. Ask how the fixture locates the plastic part, how much operator technique affects the result, whether the part is measured under controlled conditions, and whether the measurement method has been evaluated for repeatability and reproducibility.
For a new dimension, 30 parts provide more information than three first-off samples, while 50–100 measurements can show the distribution more clearly when the feature is important to assembly. Sampling plans should also cover time. Fifty consecutive pieces molded within two minutes do not test the same variation as samples collected across an eight-hour run.
Statistical capability can help compare suppliers, but Cpk should not be read without context. A Cpk of 1.33 is often used by manufacturers as a practical internal capability target, while some programs request 1.67 or another agreed level for selected characteristics. The customer drawing, quality agreement, or program requirement should define the actual acceptance rule.
| Review item | More useful supplier evidence | Weak evidence |
|---|---|---|
| Tight dimension | 30–100 measured parts with cavity IDs | 3 hand-picked samples |
| Multi-cavity mold | Data separated by every cavity | All cavities mixed together |
| Process control | Recorded parameter ranges by lot | Operator memory |
| Resin handling | Lot and dryer history | Dryer visible on factory floor |
| Metrology | Calibration plus measurement study | Equipment list |
| Tool maintenance | Cycle-based service history | Repairs only after failure |
Capability numbers also need stable input data. If cavity 1 averages 10.01 mm and cavity 8 averages 10.07 mm, combining both populations can make the overall result difficult to interpret. Cavity-level reporting shows whether a steel correction, cooling change, or processing adjustment is required.
Tool maintenance becomes more important after that initial capability study. Ask whether mold service is scheduled by shot count, observed condition, elapsed time, or a combination of the three. Gates, vents, slides, lifters, ejector pins, leader components, hot-runner tips, seals, and cooling passages do not wear at identical rates.
A program expected to produce 3 million parts should therefore include a maintenance plan before the tool enters high-volume production. If an eight-cavity mold runs a 20-second cycle at full availability, theoretical production exceeds 11,000 parts per eight-hour period; real production will be lower after planned stops, maintenance, quality checks, and other losses are included.
Capacity quotations should use realistic production hours rather than 100% theoretical availability. Ask the supplier to show how annual demand compares with available press hours, planned utilization, mold maintenance, changeovers, and expected cycle time.
If projected annual demand is 2.4 million parts, an Industrial plastic molding manufacturer should be able to show the number of press hours required, the production schedule, and the alternative press that can accept the mold if the assigned machine becomes unavailable.
Backup capacity should also be technically qualified. Moving a mold from one press to another can change screw diameter, pressure response, plasticizing behavior, nozzle geometry, and process settings. A backup machine listed on a spreadsheet is less useful if no transfer study has been completed.
Quality-system certification belongs later in the review. ISO 9001 certification can confirm that a documented quality-management framework is in place, while automotive programs may require IATF 16949 and medical manufacturing may operate under ISO 13485 requirements. Certification does not replace part-specific production data.
Ask to review one previous nonconformance. Look at the complaint description, affected quantity, containment, dimensional records, tool or machine findings, corrective work, and follow-up verification. A record involving 12,000 inspected parts and 84 rejected pieces provides more manufacturing information than a presentation stating that quality is a priority.
Supplier performance should also be checked through recent operating data. Request monthly rejection, on-time delivery, customer-return, and production-scrap records covering at least 6–12 months where the supplier is willing and permitted to share them.
A claimed 99% delivery rate requires context: 99% across 10 shipments carries less information than 99% across 500 shipments. Scrap reporting needs similar context because a 1.5% plant average can contain one difficult molded component running at 7% rejection.
Commercial comparison should therefore include the cost of making acceptable parts, not only the quoted molding price. A difference of $0.03 per unit creates a $30,000 difference over 1 million parts, but additional sorting, freight, assembly interruption, or mold correction can exceed that amount.
The final supplier review should use a production run long enough to expose conditions that short sampling misses. Run every cavity, allow the mold to reach normal thermal conditions, record process parameters, and collect parts from several points during the run.
For an eight-cavity tool, a study of 80 parts gives 10 pieces per cavity if sampled evenly. A stronger validation can include samples from the beginning, middle, and end of a four- or eight-hour run, with the same dimensions measured under the same conditioning method.
Compare those results with the approved drawing, cavity averages, overall range, process settings, material lot, mold temperature, and inspection method. Repeatability across time, cavities, material lots, and normal operating conditions provides a much stronger basis for supplier approval than a polished set of T1 samples.