Do sampling rounds always need three passes?

A: No. A colour-only change with no tool work closes in two rounds and 17 days. Anything touching wall or ribs is safer at three rounds and 29 days; the 12 days saved return as rework.

Are 2, 3 and 5 samples enough for destructive testing?

A: Yes, because the rounds do not repeat a test. The two T1 samples go to assembly and thickness, one T2 sample to a colour re-check, one T3 sample to the archive.

What if T2 finds the wall frozen at T1 too thin?

A: Roll it back through a change note: the wall value, the freeze record and the remaining tolerance bands, then one fresh 7-day round. A margin note instead of a rollback leaves the same rib 0.05 mm thinner in production.

Is ΔE 1.2 worth an extra round over 1.5?

A: At 1.5 most people see nothing under D65 but a boundary shows in raking light; 1.2 is invisible. Side-by-side shipments justify 1.2, single-case shipments do not.

How long should review records be kept?

A: For the life of the tool, at least 3 years of production. Lose one of the five fields and a review two years later cannot tell which sample froze which value.

Sampling Rounds: Three-Round Sampling

Quick answer: Three sampling rounds hold only when each answers one question: structure, colour, then packaging consistency, on 2, 3 and 5 samples.

Why sampling rounds are split by question
Round one: structure and assembly only
Round two: appearance and colour only
Round three: packaging and production consistency
Freeze points that make three rounds enough
FAQ

1. Why sampling rounds are split by question

A 45-litre case takes three passes, and each may answer one class of question: does the lid close on the body, does the surface survive raking light, is the moulded part the one that was signed off. Run all three in one meeting and the readings contaminate each other.

Tighten the lid gap from 0.5 mm to 0.3 mm and the grain stretches deeper. Push colour one shade darker and melt temperature climbs 10 °C, shrinkage follows, dimensions slide back two weeks. Freeze structure first and the tool needs one gate and one vent change; change both and it sees three revisions, turning 29 days into 41.

Round	The only question it answers	Frozen at the end	Release test
T1	Does the case close and assemble	Wall 2.6 mm and rib layout	Both samples assemble, no interference
T2	Does the surface match the colour chips	Three colours and grain depth	ΔE ≤1.2 on all three samples
T3	Is the moulded part the same part	Tool and process parameter sheet	Five samples inside the first-off band
Extra	A frozen item was overruled	Freeze value rolled back and logged	Only after a change note is approved

One release test per round leaves nowhere to hide a maybe.

2. Round one: structure and assembly only

T1 takes 2 samples and 7 days, the only pass where structure may move. What gets measured is closure: lid gap, latch effort, hinge play, cavity against drawing.

Sampling rounds night scene of a black case on wet rock, telescopic handle raised, side latches closed, a helmet with green night-vision optics resting nearby

Lid gap is checked with feeler gauges at four corners: a 0.25 mm blade enters, a 0.30 mm blade does not. A corner that refuses means look at lid warp before touching the tool. Latch effort sits between 45 N and 70 N; below it the latch opens in transit vibration, above it one hand cannot work it and the crew reaches for a screwdriver.

Wall freezes here at 2.6 mm, tolerance ±0.08 mm, so T2 and T3 read the same section. Move the wall and shrinkage plus flatness move with it, voiding every number collected so far.

The cavity is checked at 20 mm above the base, mid-height and 15 mm below the rim, two readings each, spread capped at 0.6 mm. A wide spot sits opposite the gate. Hinge play is allowed 0.15 mm, beyond which the lid shifts sideways once closed.

3. Round two: appearance and colour only

T2 takes 3 samples and 10 days and accepts no structural complaints. The variables are surface: colour, gloss, grain, weld lines, drag marks.

Three colours are fixed — body, accent, black hardware. Three chips go on the table, master in the middle and one limit either side, compared in a D65 cabinet at 45 cm and 45 degrees. The line is ΔE ≤1.2; anything above returns to the machine.

Gloss is read at 60 degrees, target 12 to 18 GU, on front, side and lid top. A spread above 3 GU points at uneven mould temperature, so check the water circuit first; a new resin lot only moves the deviation to another face.

Grain is 0.08 mm, five points, and a 0.02 mm difference is visible in raking light. At the end of T2 the colours freeze and the three chips are signed into the record.

4. Round three: packaging and production consistency

T3 takes 5 samples and 12 days and tests only whether the moulded part reproduces what the first two rounds froze.

Sampling rounds final sample case standing in an alpine meadow, handle fully raised, side latches shut, purple and orange wildflowers in front, snow-streaked ridges behind

First-off against last-off means the same tool and the same shift: part one and part 500, both measured for lid gap and wall. A difference above 0.1 mm says mould temperature drifted inside the shift. Packaging is trialled at 6 cases per carton, 5-ply board, on a 1200 mm × 1000 mm pallet stacked 3 tiers and kept under 1.5 m.

The tool freezes at T3. A later rib or vent change needs a change note and a fresh round outside the schedule.

5. Freeze points that make three rounds enough

Three rounds freeze three things: wall, colour, tool. They go into one record with five fields — sample ID, frozen value, open issues, owner, release signature.

Open issues are capped at 3 per round. A carried issue changes shape: a gap left at T1 becomes an unreadable colour reading at T2 and a first-off deviation at T3.

Round	Samples	Duration	Open issues	Cumulative tool changes
T1	2	7 days	3	1
T2	3	10 days	3	2
T3	5	12 days	3	3
Total	10	29 days	9	3

Two root causes explain most four-round projects. A drawing arrives at T1 with 4 undecided dimensions, so the sample exposes a structural change nobody could freeze. Colour is raised at T3, after the tool has frozen, turning a colour change into a tool change.

Extra rounds are a sequencing failure, not a skill failure. The ceiling is one gate, one vent, one rib; if three revisions do not converge, the missing document is a signed freeze list.

Conclusion

Three rounds work because each is narrow: T1 owns structure and wall, T2 owns appearance and colour, T3 owns packaging and production consistency, and only after all three freeze does the tool become fixed. Ten samples, 29 days and a three-change ceiling are not predictions; they mark where a project can still turn back cheaply. Write the freeze values into five record fields and the process stops depending on who remembers.

This article covers case-side specifications only and does not address any handling, loading or use of the contents.