Even careful visual checks miss the systematic failures in molded production: one cavity drifting while others run clean, gate vestige snagging assembly, dimensions walking across a long run, and decoration flaking after transport vibration. These flaws hide from single-piece approvals and single-carton sampling because the checked units come from the healthy part of the pattern. This second part of the defect series shows where these quieter failures hide and how disciplined sampling, functional trials, and adhesion testing expose them before shipment.
Key takeawaysSample across cavities and run positions, because mold flaws repeat in patterns that single-carton checks miss. · Judge defect counts with sampling discipline, drawing randomly and applying acceptance numbers by class. [1] · Prove function under assembly conditions rather than judging appearance alone. [2] · Test decoration adhesion on production resin after handling simulation, since print failures appear late. · Track dimensional drift across the run instead of trusting first-article measurements alone.
Part one of this series covered the five structural defects buyers most often miss: flash, short shots, sink marks, warpage, and weak weld lines. This second part covers the quieter failures that survive even careful visual checks: cavity-to-cavity variation, gate and ejection damage, dimensional drift across a run, and decoration flaws that appear after the parts leave the factory.
These defects share a trait: they are systematic rather than random. One cavity runs hot, one gate vestige snags assembly, one color batch drifts, one print process flakes after vibration. Single-piece approvals and single-carton sampling miss systematic flaws by design, because the checked unit comes from the healthy part of the pattern.
Read this part together with part one as a single defect manual. Part one names the headline flaws; part two shows where the remaining failures hide and how inspection evidence exposes them before shipment.
Cavity-to-cavity variation
Multi-cavity tools rarely age evenly. One cavity's cooling channel scales up, one shutoff wears faster, one gate erodes wider, and from that point the cavities produce subtly different parts: one with heavier flash, one with sink the others lack, one shade darker in the same shot. Buyers comparing a single approval sample against a blended shipment never see this divergence.
The inspection answer is positional sampling. Draw samples from every cavity across multiple cartons, record the cavity origin of each defect, and compare dimensions and appearance cavity by cavity. Where the tool design permits, ask the factory to map cavity numbers to the evidence; where it does not, segregated short runs per cavity still reveal the pattern. Maintenance aimed at the offending cavity costs a fraction of tool-wide rework.
Variation also crosses material lots. A resin batch change or masterbatch adjustment mid-order shifts color, shrinkage, and strength for the remainder of the run. Require the factory to declare material lot changes, and treat the first cartons after each change as a fresh sub-lot for sampling purposes rather than blending them into the earlier evidence.
Gate, ejection, and handling damage
Gate vestige, the remnant where the runner was separated, causes more assembly trouble than its small size suggests. A vestige a fraction over tolerance snags sliding fits, prevents flat seating, and scratches mating surfaces during assembly. Specify maximum vestige height and whether flush cutting or hot-tip gating applies, then gauge it rather than eyeballing it. Degating method changes mid-order are a classic silent substitution worth an explicit contractual lock.
Ejection damage shows as drag marks along deep ribs, whitening or stress marks at ejector-pin points, and bent thin features that were stripped before full cooling. Light witness marks on hidden faces are usually acceptable; whitening on visible faces or deformation that alters fit crosses into major territory. Check thin snap arms and tall slender bosses with particular care, flexing them through their designed range to reveal stress damage that static inspection misses.
Handling between press and packing adds the third layer: parts tumbled loose in bins arrive scuffed before any carton test begins. Verify that the factory's in-process handling matches the specification, lined bins, separated layers, no dragging across concrete, because no downstream packaging compensates for parts already damaged. When the evidence for these handling-stage flaws needs neutral documentation, plastic mold inspection plastic mold inspection records them with photos and counts tied to run positions so the factory can trace each mark to its station.
Drift across the production run
| Drift source | What shifts | When it appears | How inspection catches it |
|---|---|---|---|
| Tool heating and cooling balance | Dimensions, warpage, gloss | First hour and after breaks | Samples from start, middle, end |
| Material lot or regrind change | Color, shrinkage, strength | After the changeover carton | Sub-lot sampling per material lot |
| Process adjustment by operators | Weight, fill, flash level | Night shift, rush periods | Weight trending across cartons |
| Tool wear within the order | Flash growth, venting loss | Long runs progressively | Flash measurement trending |
| Decoration process drift | Print position, adhesion | Across print batches | Adhesion tests per print batch |
Decoration flaws that appear late
Print and decoration fail on a delay that fools factory-gate inspection. Pad print looks crisp at the packing table and flakes after vibration; hot-foil stamps adhere on day one and lift at edges after temperature swings; texture differences between cavities show only under the customer's directional lighting. Testing decoration at the factory in its just-finished state systematically overestimates its survival.
Counter this with handling simulation before judging. Rub adhesion tests on production resin, cross-hatch with tape rather than fingertip opinion, flex the substrate near the print, and run a packed-carton vibration proxy such as a shake and drop sequence before re-examining edges. Specify the adhesion grade required and the test method in the criteria so results are comparable across orders.
Position and legibility need the same rigor. Measure logo position against its tolerance, check minimum print size for legibility at realistic reading distance, and verify contrast on the actual resin color rather than the artwork proof. Multilingual markings and warning symbols deserve character-by-character comparison against approved files, because one wrong character in a safety message outweighs a hundred correct dimensions.
Prove function, count defects properly
Every defect in this series ends at the same question: does the part work in its intended setup. Assemble the product with its mating components, operate every moving element, fasten every joint to its specified torque, and seal every joint that claims sealing. This acceptance logic, verifying performance under defined conditions rather than admiring appearance, is what separates a shippable lot from a photogenic one. [2] Write the functional trials into the criteria with sample sizes and pass thresholds so they happen on every visit.
Count what the trials and checks find with sampling discipline. Acceptance sampling theory draws units at random and judges each defect class against its acceptance number, which is why positional sampling across cavities and run segments matters as much as raw sample size. [1] A finding concentrated in one cavity or one run segment triggers expanded sampling of that sub-lot, not a shrugged average across the whole.
File both parts of this series with the specification and update them from experience. Each order's findings should sharpen a threshold, add a checkpoint, or reclassify a defect whose field cost is now known. Defect lists that learn compound in value; static ones slowly drift out of alignment with what customers actually punish.
Sources
- Wikipedia: Acceptance sampling —— Overview of acceptance sampling including AQL concepts used in product inspection.
- Wikipedia: Factory acceptance test —— Overview of factory acceptance testing performed before shipment or installation.
Frequently asked questions
How do inspectors trace a defect to one cavity?
Mark or map cavities where the tool allows, sample across all of them, and record which cavity each defect came from. When one cavity produces the finding repeatedly, maintenance can target that cavity instead of disturbing a tool that is mostly healthy.
Are ejector marks a real defect?
Check for drag marks, whitening at ejector points, and deformation of thin features, then confirm the parts still assemble and fasten correctly. Minor witness marks on hidden faces are usually acceptable; deformation that changes fit or leaves stress whitening on visible faces is not.
Why does sampling position matter as much as sample size?
Sampling statistics assume randomly drawn units judged against defined acceptance numbers, so expanding the draw across cavities and run positions is what makes the verdict representative. Single-carton sampling can miss entire defect patterns.
When is a functional trial mandatory rather than optional?
Test whether the assembled product meets its agreed setup under defined conditions, which is the acceptance logic that matters for functional characteristics. Appearance judgments alone cannot confirm fit, movement, or sealing.