Cavity-to-Cavity Variation: What to Measure, and When

A multi-cavity tool rarely fails at T1. It fails in week three, when one cavity starts running light and nobody is weighing the parts individually.

We see this pattern often enough that it is worth writing down: the tool is signed off, the first samples look right, production starts, and three weeks later the customer's incoming inspection finds a dimension drifting on one cavity out of sixteen. The tool is not scrap. But nobody can say which cavity, or since when.

Why cavity variation appears late

At T1 the tool is new. Cavity surfaces are at nominal, gates are as cut, and the press is being watched by the person who built the tool. Every cavity gets the same attention.

Production is different. Gates wear at different rates depending on fill balance. Cooling channels in the middle of the block run hotter than the ones at the edge. Core shifts settle in after a few thousand cycles. None of this shows up in the first fifty shots.

The problem is not that variation exists. It is that nobody measured the baseline when the tool was new, so there is nothing to compare against later.

Measure at T1, per cavity

At sampling, weigh parts cavity by cavity — not the shot weight, the individual part weight. Twenty consecutive shots per cavity is enough to see a trend; five is not.

Record four things per cavity:

• Part weight, mean and range

• Critical dimension at the feature the drawing actually controls

• Cavity pressure peak, if the press logs it

• Visual condition at the gate and at the weld line

Keep the record with the tool. It takes an hour at sampling and it is the only way to answer "when did this start" three weeks later.

What the numbers tell you

A spread of more than about one percent in part weight across cavities usually points to fill imbalance — gate size, runner layout or a blocked cooling line. If weight is even but a dimension is not, the problem is more likely mechanical: core shift, ejector deflection, or a slide not closing to the same position every cycle.

If only one cavity drifts over time, look at the gate first. Gate erosion is progressive and it is the most common cause of a single cavity running light.

The production habit that prevents it

Once the tool is running, weigh parts per cavity at a fixed interval — weekly is usually enough for a stable part, daily for a tight-tolerance one. Plot the means. A trend is visible long before a part goes out of tolerance, and correcting a gate or a cooling circuit during a planned stop costs a fraction of correcting it after a containment.

The measurement nobody takes is the one that turns into a recall.

If you are about to sign off a multi-cavity tool, ask for cavity-by-cavity weight data at T1. Not the shot weight. The cavity weight.