On 12 November 2026, micrometer MM-014 of turning cell T3 failed its calibration. At 30 mm it read 0.004 mm too high. Its permissible error is 0.002 mm. The gauge was the smaller problem. The question was what it had measured since it was last known to be right.
ISO 9001 expects exactly this question: when measuring equipment is found unfit, earlier results have to be assessed.
Seven steps
- Block. Take the gauge out of use on the same day and label it.
- Size the error. How much, in which direction, over which range.
- Set the window. From the last good calibration or interim check to the day of the finding.
- Trace. Which characteristics and lots were measured with this gauge in the window.
- Assess. Correct the recorded values by the error and compare them with the limits.
- Contain. Sort the lots at risk with a verified gauge. Inform the customer where parts have left.
- Prevent. Find out why the gauge failed and why nobody noticed.
The window
The last good calibration of MM-014 was on 14 May 2026. Nobody knew when the error appeared, so the whole period counted: 26 weeks. The micrometer was used for one characteristic, the outer diameter of a bushing, 30.000 ±0.012 mm. The bushing runs once a week in lots of 600. That made 26 lots and 15,600 parts.
This step was only possible because the check sheets of T3 name the gauge number. Without that field, every micrometer of the cell would have been a suspect.
The assessment
The direction of the error matters. The micrometer showed more than the true size. A bushing recorded at 29.990 mm was in truth 29.986 mm, below the lower limit of 29.988 mm.
The team took the recorded minimum of every lot and subtracted 0.004 mm. The records are samples, not every part, so a margin of 0.002 mm was added: every lot with a corrected minimum below 29.990 mm was sorted.
| Lot (year, week) | Where | Corrected minimum in mm | Below the limit, of 600 |
|---|---|---|---|
| 26-30 | Shipped | 29.987 | 7 |
| 26-37 | Shipped | 29.988 | 10 |
| 26-44 | In house | 29.989 | 9 |
| 26-45 | In house | 29.986 | 19 |
| 26-46 | In house | 29.988 | 13 |
Five lots with 3,000 parts were at risk. The other 21 lots were released with a written reason.
Containment
The 1,800 parts in house were sorted with a verified gauge on the same day: 41 below the limit. The two shipped lots meant a call to the customer. Sorting there found 17 more. In total, 58 bushings were scrapped and replaced.
| Cost of the case | EUR |
|---|---|
| Sorting in house, 15 hours | 900 |
| Sorting at the customer, charged back | 1,080 |
| Scrap, 58 bushings | 522 |
| Replacement shipment and complaint handling | 350 |
| Total | 2,852 |
What made it expensive
Not the error of 0.004 mm. The window of 26 weeks. The cause was ordinary: the micrometer had been dropped, and nobody handed it in. The plant changed three things:
- A setting master of 30 mm at the cell and one reading per week, written down. The same error would now put one lot of 600 parts in doubt.
- The interval of the micrometer went from 6 to 3 months, following the interval rule from Part 2.
- The agreement about dropped gauges was repeated in every shift: hand it in, no blame.
Practice before it happens
Pick a gauge from the list and state: found 0.004 mm too high today. Give the team thirty minutes to name the lots that were measured with it. If they cannot, the gauge number is missing in the records.
The series in one sentence
Know every gauge, keep it calibrated, choose it for the tolerance, prove the measurement with a study, define the boundary for checks without numbers, and keep the window small for the day a gauge fails.