A sample test only answers a question if you decide the question first. Most trials fail because they compare two powders without fixing the process, then draw a conclusion from three parts and a fresh pad.
Before The Sample Arrives
- Write the question down. “Does poly reduce scratch count at W3 on our pad” is testable. “Is this powder better” is not.
- Decide the pass criteria. A number, a defect count, or a tool life figure, agreed before the trial starts.
- Reserve machine time. A trial squeezed between production runs measures the interruptions, not the powder.
- Keep a reference. Current production powder, same size, tested in the same session.
The Ten Checks
| # | Check | What you are looking for |
|---|---|---|
| 1 | Lot number and grade on the pack | Identifiable material, matching the certificate |
| 2 | Net weight | Confirms no loss in transit |
| 3 | Appearance and colour | Yellow or black family as specified, no visible contamination |
| 4 | PSD report | D10, D50, D90, D97 with method named |
| 5 | Morphology image | Shape at the size you are testing |
| 6 | Ash value | Relevant for plating, slurry and thermal work |
| 7 | Coating weight gain | For coated product, plus an image showing continuity |
| 8 | Dispersion test | Does it wet out in your carrier without lumps |
| 9 | Ten-part run | Removal, Ra and defect count at fixed intervals |
| 10 | Repeat run | Same result on a second session, to rule out pad break-in |
Designing The Run
Change one variable. If you change size and crystal at the same time, the result tells you that something changed but not which one. If you must test several variants, run them in sequence on the same pad condition and randomise the order, because pad state drifts through the day.
Measure at part 1, 5 and 10 rather than only at the end. The trend is often more informative than the final number: a powder that starts well and degrades may be agglomerating, while one that improves may simply need a longer bed-in.
Comparing Against A Certificate
Your trial result and the supplier certificate answer different questions. The certificate says what was shipped; the trial says how it behaves. Keep both, and if they disagree, re-check the dispersion step before blaming the supplier.
Recording Results
Record date, operator, machine, pad or plate condition, carrier, solids loading, pressure, speed and the measured values. Without the process record, a result cannot be reproduced, and an unreproducible result cannot be scaled to production.
A Recording Sheet That Works
Write the result down at the machine, not at the end of the day. The columns below are enough to reproduce a trial months later.
| Field | Example |
|---|---|
| Date and operator | 2026-10-19, shift B |
| Powder code and lot | HDY W3, lot reference from label |
| Pad or plate | Felt pad, new / used, hours in service |
| Carrier and loading | Oil-based, stated solids percentage |
| Pressure and speed | As set on the machine, recorded not assumed |
| Parts 1, 5, 10 results | Removal, Ra, defect count under fixed lighting |
| Observations | Colour change, foam, unusual noise, slurry settling |
Turning Results Into A Decision
Three outcomes are possible, and each leads somewhere different. If the new sample is clearly better and the repeat run confirms it, scale it. If the two are indistinguishable, keep the cheaper one and stop testing. If results are mixed across parts, the process is not stable enough to compare powders yet, and the next step is process work rather than a third supplier.
That third case is the one buyers most often misread. Mixed results are information about the process, not about the powder.
Common Mistakes
- Three parts and a conclusion. Pad break-in dominates the first few parts.
- Testing on a different machine. Results do not transfer, and the customer’s machine is the only one that matters.
- Not keeping the reference sample. Six months later, nobody can prove what the trial powder was.
How To Verify
Repeat the winning condition on a fresh batch before placing a production order. If the second run matches, the result is a property of the powder. If it does not, the first result was a property of the day.
Next Step
Related reading: quality control and batch records.