A customer returned a pallet of plated parts, and three inspectors gave three verdicts: pass, rework and scrap. The parts had not changed; the lighting and the sampling method had.
This case walks through inspection lighting setup, AQL sampling basics and the record keeping that gives a junior inspector one consistent answer for every lot.
Common Mistakes and How to Avoid Them
| Mistake | Why It Happens | Practical Fix |
|---|---|---|
| 1. Inspection under one ceiling light | Shadows hide real defects | Add 500-1000 lux task lighting |
| 2. Sampling by habit instead of the table | Wrong lots accepted or rejected | Read the AQL table for every lot size |
| 3. Judging defects at arm's length | Small defects get missed | Hold parts 30-40 cm from the eye |
| 4. No agreed defect dictionary | Inspectors disagree on the same part | Agree photos for each defect class |
| 5. Reject decision made by the loudest voice | Outcomes change with the shift | Follow the written acceptance criteria |
| 6. No retention sample kept | Disputes cannot be settled later | Keep one reference part per lot |
| 7. Glare from bare lamps | Reflections hide surface faults | Use diffused lighting over the bench |
| 8. Inspecting the same spot twice | Sampling becomes biased | Rotate parts and change positions |
| 9. Decision not logged | The same dispute returns | Log lot, sample size, defects and decision |
Best Practices
- Set task lighting to 1000 lux with daylight color temperature
- Read the sample size from the AQL table, never from memory
- Hold parts at a fixed 30-40 cm distance from the eye
- Keep a photo defect dictionary on the inspection bench
- Retain one reference part from every inspected lot
- Log every inspection with a decision, date and inspector name
Implementation Roadmap
| Step | Frequency |
|---|---|
| Check lighting and bench setup | Weekly |
| Cross-check inspectors on one identical lot | Weekly |
| Review sampling against the AQL table | Monthly |
| Update the defect dictionary with new photos | Monthly |
| Review reject records for repeating patterns | Monthly |
Inspection decision flow
- Light - 1000 lux daylight at the bench. (Glare hides surface defects.)
- Sample - Size from the AQL table. (Habit sampling skews results.)
- Judge - Against the photo dictionary. (Personal opinion causes disputes.)
- Log - Decision with lot details. (Unlogged calls cannot be defended.)
Lighting must be identical for every shift and every inspector. | The sample size follows the contract's AQL code letter.
Inspection setup reference data
Reference values for visual inspection of plated parts.
| Parameter | Reference | Why It Matters |
|---|---|---|
| Lighting | 500-1000 lux at the part surface | Below 500 lux hides defects |
| Lamp color | 5000-6500 K daylight | Warm light masks yellowing |
| Inspection distance | 30-40 cm from the eye | Greater distance reduces resolution |
| Sample size | AQL code letter, e.g. n=32 at level II | Wrong letter changes acceptance risk |
| Accept numbers | Ac/Re from table, e.g. 1/2 at AQL 1.0 | The contract decides the AQL |
| Viewing time per part | 3-5 seconds | Rushed checks miss defects |
| Retention | 1 reference part per lot | Evidence for later disputes |
Inspection bench setup card
| Check | Target | Action |
|---|---|---|
| Lighting | 1000 lux at bench | Measure with lux meter weekly |
| Lamp color | 5000-6500 K | Check lamp specification |
| Distance | 30-40 cm | Mark the bench position |
| Sample size | Per AQL code letter | Keep the table on the bench |
AQL quick reference, level II, AQL 1.0
| Lot size | Sample n | Accept/Reject |
|---|---|---|
| 2-8 | 2 | 0/1 |
| 9-15 | 3 | 0/1 |
| 16-25 | 5 | 0/1 |
| 26-50 | 8 | 0/1 |
| 51-90 | 13 | 1/2 |
| 91-150 | 20 | 1/2 |
| 151-280 | 32 | 2/3 |
Case 1 - the pallet that drew three verdicts
Scenario. A plating plant's customer returned a pallet of zinc-plated brackets, and three inspectors judged it pass, rework and scrap under different lamps and sampling habits.
Action. The QC bench was re-lit to 1000 lux daylight, the AQL level II table was pinned to the bench, and a photo defect dictionary was agreed for every defect class.
Result. The same pallet was re-judged with one answer, the reject rate settled at 2.3 percent, and cross-inspector disputes stopped.
Case 2 - samples drawn only from the top layer
Scenario. A fastener customer in Southeast Asia rejected a shipment even though every top-layer sample passed receiving inspection at the plant.
Action. Sampling was spread across cartons, layers and positions using the AQL code letter, and the plan was written into the inspection work instruction.
Result. The next three shipments passed customer receiving inspection with zero claims, and the sampling plan became part of the contract review.
Frequently Asked Questions
How much light do I need?
Aim for 1000 lux at the part surface with daylight color; below 500 lux defects disappear.
What color lamp is best?
5000-6500 K daylight; warm lamps mask yellowing and tarnish on plated surfaces.
How far should I hold the part?
30-40 cm from the eye, rotating the part to see every face.
Which AQL do I use?
The one in the contract or drawing; 1.0 to 2.5 is typical for plated finishes.
How do I read the sample size?
Find the lot size and code letter on the table, then read sample n and the accept and reject numbers.
What if inspectors disagree?
Judge against the defect dictionary and the lighting standard, not against the loudest opinion.
Do I keep samples?
Retain one part per lot until customer receiving inspection confirms; it settles disputes.
How do I log a rejection?
Record date, lot number, sample size, defect type and decision; the log is the evidence.
Can I use a magnifier?
Magnifiers help small defects but must not replace the standard 30-40 cm visual check.
What Would You Like to Solve?
Describe your part type, defect patterns and current sampling method, and share the specific rejection disputes your line sees.