Multi-cavity mould first article checklist

Why one good part is not an approved tool

A multi-cavity mould is not one tool making one part. It is a set of cavities fed by a common runner, each with its own gate, its own cooling and its own small differences in steel and fit. The parts that come out share a drawing, not a history. A first article that measures one cavity, or a hand-picked part from the middle of a good run, proves that one cavity can make a good part. It does not prove the tool.

The question is not whether a part can meet the drawing. It is whether every cavity can meet the drawing at the same production settings, across the run, with enough margin left for the material and the machine to vary. A sample that passes by a hair in one cavity is a warning, not an approval.

Use this checklist as a sequence. Each step produces a record, and the records together are what the approval rests on.

Set the run state before taking a sample

A trial that is slowed down to protect the tool can hide a cavity that fills late at production speed. A trial that is pushed hard can hide a cavity that shorts at a normal cycle. Either way the sample is not the production part.

Agree the process window with the tool designer and the moulder before the trial: melt temperature, injection speed, hold pressure and time, cooling time and ejection. Then run at the nominal production setting and record the actual values, not the set values. Record the shot number, the time from start-up and any interruption. A part taken after a stop is a different part from one taken mid-run.

Dry the material to the grade requirement and record the dryer temperature and time. A wet engineering resin can produce splay, weak weld lines and dimensions that move after moulding, and the first article then blames the tool for a material problem.

Sample every cavity, more than once

Every cavity needs its own sample, and one part per cavity is a weak basis for a decision. Take at least three consecutive shots per cavity once the process is stable, so the repeatability within a cavity is visible. Then take a set at the start, the middle and the end of the run to show drift.

Keep the first shots as well, but label them separately. They show how the tool behaves from cold and whether the start-up procedure is repeatable. They should not be the parts that decide approval unless the customer procedure says otherwise.

For illustration, an eight-cavity tool sampled over three consecutive shots produces 24 parts, and a start, middle and end set adds another 24. That is a manageable number for a laboratory and enough to expose a cavity that is drifting.

Record the shot layout. On a family tool, or a tool with an uneven layout, the position of each cavity in the shot matters when a short shot or a weld line appears. A photograph of the open tool with the cavities numbered is a useful page in the report.

Mark and trace every sample

A first article without cavity traceability cannot support a decision. If a dimension fails, the report has to say which cavity failed and whether the failure repeats. If the tool has no cavity marks, that is a finding in its own right, and the review should agree a marking plan before the samples are measured.

The preferred mark is a small engraved cavity number on a non-functional face, applied at tool build and repeated on the tool drawing. Where the part is too small or the face is visible, use the shot position and a photographed layout, and bag each sample with its cavity and shot number.

Label the bag, not just the report. Samples get mixed on a bench. A bag with the cavity number, shot number, date, material batch and moulding machine is a record. A bag with a part in it is a rumour.

Measure against the drawing datums

The measurement plan follows the drawing, not the convenience of the bench. Locate the part on the drawing datums, in a fixture that supports it without clamping out the warp the drawing is trying to control. Record the fixture and the gauge used for each critical feature.

Plastic parts need a stated condition. Record the time, temperature and humidity between moulding and measurement, and measure all cavities in the same condition. A part measured warm and a part measured after a week are not comparable, and the difference can be larger than the tolerance.

The report should cover the critical features, the features the tool controls directly, and the function. It should not be a list of every dimension with no view of which ones decide the fit.

Read the spread, not the average

An average hides the cavity that sits close to the limit. In an illustrative example, a feature is nominally 20.00 mm with a tolerance of plus or minus 0.10 mm, and the cavities measure between 19.94 mm and 20.08 mm. Every part passes. The spread is 0.14 mm, or 70 per cent of the tolerance band, before the material lot or the machine has varied at all. That result should send the tool back for a gate or cooling correction rather than to the customer for approval.

Compare two spreads. The within-cavity spread over consecutive shots shows process stability. The across-cavity spread shows tool balance. If the within-cavity spread is as large as the across-cavity spread, the problem is the process, not the cavities, and a tool correction will not fix it.

A capability study on a multi-cavity tool should be calculated per cavity, or with the cavity treated as a stratifying factor. Pooling the cavities into one distribution can produce a capability number that no single cavity can achieve.

Prove the function on the worst cavity

Dimensions are necessary and not sufficient. Assemble the part with its mating parts, test the insert pull-out or torque, check the snap fit retention, run the leak or pressure test, and confirm the electrical clearance if the part is a housing. Use the cavity with the least margin, not the best-looking sample.

Record which cavity supplied each functional sample. A functional pass from cavity three says nothing about cavity six. If the drawing calls for a test at a stated condition, record the condition and the equipment.

Where the part goes into a subassembly, check the stack. A cavity that is inside its own tolerance can still fail when it meets a cover or a terminal at the other end of the stack.

Decide, set conditions and requalify

The review has three outcomes. Approve the tool for production, approve with conditions, or reject and correct. A conditional approval must name the cavity, the feature, the action and the date, and it must include a re-sample plan. An approval with an open condition and no owner is a rejection with better wording.

Keep a cavity-marked master sample set, the shot record, the measurement report, the material certificate and the tool maintenance state with the approval. When the tool is repaired, a cavity insert is replaced, the steel is changed or the tool moves to another press, the sample is no longer proof. Re-run the cavity sampling, at least on the cavities affected.

The checklist is finished when a reader who was not at the trial can see which cavity made each sample, what the tool did at production settings, and why the decision was reached.