Before a machine leaves our workshop, we test it like a real job.

The material was correct. The mould was correct. The machine had been running for years. But every shift produced a stack of rejects — boxes with a thin white corner, boxes that would not sit flat, boxes with a sheet of plastic webbing joining them together like a fan.

The instinct in that situation is to blame the sheet supplier, or to start shopping for a new machine. Both are expensive. Neither was the problem.

This is what we found when we looked at the line, and what changed.

Not from "bad material". From four specific conditions that can be seen, measured and corrected.

1. Webbing between cavities. When the sheet is too hot, it stretches like chewing gum instead of forming. The plastic bridges the gap between two cavities and cures there, joining the boxes into one piece. The operator pulls them apart by hand — which tears the rim, which creates a reject two stations later.

2. Corner thinning. Every draw pulls material from somewhere. On a deep punnet, the corners take the worst of it. If the sheet is not hot enough, or the pressure comes too late, the corner goes white and thin — it will crack in the customer's cold chain, not on your floor. This is the most expensive defect in the whole list, because it fails after delivery.

3. Warping after demoulding. PP has a narrow softening window. If the top and bottom of the part cool at different rates, the box twists as it leaves the mould. It looks acceptable on the stacker and turns into a reject when the case will not close.

4. Trim misalignment. Sheet creeps in the chain. A tenth of a millimetre per cycle becomes five millimetres by lunchtime. The knife then cuts outside the rim, and every part after that is scrap until someone notices — which is often at the end of the shift, in the reject bin.

Notice what these four have in common. None of them is a hardware failure. All four are settings, timing and attention.

We did not sell a new machine for this job. The HX-6171 platform was already there, and it was capable. What changed was how it was set and how it was watched.

Heat was profiled, not set flat. The oven went from one common setpoint to zones matched to what the sheet actually needs at each position. First zone warm, middle zone working, last zone recovering after the draw. Webbing at the cavity edges dropped first — the sheet stopped arriving at the mould half-melted.

Pressure timing moved later in the cycle. Corner thinning is usually a timing problem, not a pressure problem. Releasing the forming pressure a fraction later, after the sheet has time to relax into the corner, thickens the weak point without touching the machine's capability.

Cooling was balanced top against bottom. Warping comes from a temperature difference, not from a total cooling time that is too short. Extending the cycle only hides it. Balancing the two sides removed the twist without slowing the line down.

Chain tension and trim registration were put on a check sheet. Once per shift, not once per year. A five-minute check that stops a whole shift of rejects.

Scrap was counted every hour, not every month. This is the change that mattered most. A monthly scrap figure tells you that you have a problem. An hourly one tells you which setting caused it.

Every supplier will show you a machine running well. The question is what happens on the Tuesday in month nine, when the sheet lot changes and the parts start warping.

Three things worth asking:

1. "Show me the recipe screen, not the sample part." A machine that stores settings per mould is a machine that can be recovered. A machine tuned by one experienced operator and memorised by nobody is a machine that produces scrap the week he takes leave.

2. "Is the oven zoned, or is it one temperature?" Zoned heating is what makes thickness control possible on a deep punnet. One setpoint means one compromise, and the compromise is always the corner.

3. "How do I measure the weakest point?" If the answer is "we check the sample", ask which dimension and with what instrument. A supplier who measures the thinnest point of the part, every shift, is a supplier who has already had this argument with a customer.

Scrap is never brought to zero, and any supplier who promises that has not spent time on a production floor. There is a floor level for every mould and every sheet lot, and the job is to sit on it rather than above it.

What we can say for this job is which four things were causing the rejects, and that fixing them did not require replacing the machine. The customer's own scrap figures belong to the customer — ask them, not us.

If you are running PP punnets and the reject bin is filling, send us the part drawing, the sheet thickness and your cycle time. We will tell you which of the four is most likely, before anyone talks about new equipment.