WORK ORDER — Maintenance and Improvement Record
Work order no: WO-A-2026-08 Date issued: 2026-09-11 Category: Marking and labeling
Machine / cell: Inline labeling and laser marking station LM-4 on the finished goods line
Status: Closed Engineer: P. Novak (automation) Total labour: 5.5 hours Parts cost: EUR 61.00
Reason for work: applied labels drift along the container axis and the laser code lands off the label centre
1. Fault description as logged by the shift team
Thursday 09:40, operator record: the labeling module applies a 40 x 25 mm product label to the container side, then the laser marking head burns the batch code inside the label field, and over the last week the label has drifted up to 3 mm along the container axis, with the laser code then sitting 1.5 mm off the label centre, and the quality office rejected one whole batch because the code clipped the label edge. The team had re-tensioned the label web, cleaned the peel plate and re-set the laser home register twice, and the drift returned within a day, which said the two positioning elements, the container stop and the label applicator carriage, were not holding the register they claimed.
The station is a fixed labeling head with a pneumatically extended applicator foot, and the container is stopped against a spring-loaded front stop while a side clamp squeezes it against a fixed datum rail. The laser head is mounted on the same carriage as the applicator, so the two are rigid relative to each other, which means a label drift and a code drift that grow together point at the container datum, not at the laser or the label web. I went to the cell with a gauge block, a dial indicator and the product drawing rather than the label roll, because a pair of registers that both walk usually share one datum that has started to lie.
Evening note to the file: when the label and the code both move, stop blaming the printer and the web, the two are bolted to the same carriage, and a bolt that loosens moves one story that both of them tell together.
2. On-site inspection record (Day 1)
I stopped the line, put the labelling carriage in its apply position, and clamped the dial indicator against the fixed datum rail, then pushed the side clamp against a gauge block the width of the product. The datum rail face was 0.6 mm out of true along the 400 mm clamp length, worn to a gentle bow by years of containers sliding against it, and the side clamp jaw had worn a 1.1 mm step at the height where the container wall sat, so the container did not seat against the datum plane the drawing believed in. I checked the front stop next: its wear pad had lost 0.7 mm from the stopping impact, so the container over-ran 0.7 mm before the spring caught it, and the two worn regisisters added up to roughly the 3 mm the line saw in label drift, spread across the clamp step, the datum bow and the stop wear.
I measured the laser-to-label relationship separately to prove the carriage was innocent, and the code offset against the label die-cut was within 0.2 mm, dead on register, because the two heads were bolted rigid to the same carriage as built. The moving part of the station was not the label and not the code, it was the container, which claimed a datum that had worn away. The quality office’s 1.5 mm code-off-centre reading was simply the container landing 1.5 mm off the datum on that batch, and every batch landed somewhere in the 3 mm slop of the three worn features, so no software offset could ever hold, because the software corrected an imaginary position to a datum that was no longer there.
Evening note: the station had two registers doing three sums that were all drifting, a datum rail bowed 0.6 mm, a clamp jaw worn 1.1 mm and a stop pad worn 0.7 mm, and the three added to the 3 mm the whole line had been chasing with label tension and laser home registers.
3. Corrective action taken
I resurfaced the datum rail flat and true down its full 400 mm length, brought it back to the drawing straightness of 0.02 mm over the span, and faced the side clamp jaw back to a clean datum, removing the 1.1 mm step and re-establishing the container seating plane. I replaced the front stop wear pad with the polyurethane replacement, restoring the stop surface by the full 0.7 mm, and I re-set the clamp force to the maker’s 180 N so the container is squeezed against the rail and held there for every apply cycle, rather than trusting a toothless clamp to hold the register. With the three datums back on the drawing, I re-ran the station with the same label tension, the same laser register and no software offset change, and the only thing I changed was the hardware truth under the containers.
The result on the first twenty parts was label placement within 0.3 mm of centre on every container, code dead centre inside the label at 0.15 mm average, and a full quality pass with the parts inspector confirming the code sat inside the die-cut with a clean 2 mm margin all round on the worst part. I cycled another hundred parts across the afternoon to be sure the register held under production temperature, and the drift stayed inside 0.3 mm, against the 3 mm the line had been reporting, with no software offsets in the recipe at all.
| Item | Before | After | Spec |
|---|---|---|---|
| Datum rail straightness / 400 mm | 0.6 mm bow | 0.02 mm | ≤0.05 |
| Clamp jaw step at container height | 1.1 mm wear | resurfaced datum | 0 |
| Front stop pad | 0.7 mm lost | new polyurethane | as drawn |
| Label placement scatter | up to 3 mm | ±0.3 mm | ≤0.5 |
| Code offset from label centre | up to 1.5 mm | 0.15 mm avg | ≤0.5 |
The duty sheet now carries the datum rail and clamp jaw at the yearly machine audit, and the stop pad on a quarterly wear check, with the gauge-block-and-indicator method written down so the next audit measures the same three numbers instead of trusting the label tension to lie about them.
4. Acceptance result after the repair
The station ran a full week without a single label or code rejection, roughly 8,000 parts through the applicator, and the quality office confirmed the code to label centre stayed within the 0.5 mm spec on every sampled part, with the batch that used to be rejected on the code-edge risk now sailing through first time. The line takt is unchanged, because the fix was hardware truth, not a slower recipe, and the operators no longer nudge the label tension between batches to chase a drift that had nothing to do with the web. I closed the order with the station released and the three datum features logged in the audit, and pinned a one-page drawing note to the cell showing the container datum plane and the three wear points, so the next engineer measures the same places I measured.
5. Hours and parts list
- Labour — 5.5 h total: datum survey with gauge block and indicator 1.5 h, datum rail and clamp resurface 1.5 h, stop pad and clamp force set 1.0 h, production verification 1.5 h
- Parts — front stop wear pad EUR 18.00, clamp jaw facing tooling EUR 12.00, special grease and fasteners EUR 8.00, calibration block set EUR 23.00; total EUR 61.00
- Rejections avoided — one full batch returned to re-label before repair; zero label-related rejections in the acceptance week
6. Retrospective notes for the next engineer
When a label and a code on the same carriage both drift, measure the container datum, not the web, because the carriage holds the two printing registers rigid and the only register that can walk is the part. The three worn numbers, the 0.6 mm datum bow, the 1.1 mm clamp step and the 0.7 mm stop pad, added up to the 3 mm the line was seeing, and no amount of label tension or laser home re-setting could hold a register against a datum that was gone. A labeling station is a metrology fixture with a printer on it; the labels land where the container seats, and the container seats on the datum you maintain. Resurface the rail, re-face the clamp, watch the stop pad on the quarterly list, and the code will sit dead centre for years, because a station that hands the part a true datum is a station that never needs the operator to chase its drift with a trembling hand on the tension knob.
Attachments: datum rail and clamp measurement record, stop pad wear log, label and code placement verification sheets, one-week rejection log (filed with WO-A-2026-08).
Glossary — datum rail: the fixed surface the container is clamped against; side clamp: the jaw squeezing the container to the datum; front stop: the stop positioning the container along its axis; peel plate: the edge that separates the label from the liner; register: the reference position a feature must land at.
Field note — the datum discipline a labeling station owes itself
The general rule worth writing down is this: any station that prints or applies something onto a moving part is only as accurate as the datum the part stands on, and the printing head, the label carriage and the laser never lie as much as the fixture quietly does. On this cell the three registers, the datum rail, the side clamp and the front stop, together located the container in three directions, and each had worn by a fraction of a millimetre over years of contact, yet the applied result drifted three full millimetres, because the three errors added in the same direction and the part never had a true corner to sit against. The fix was not clever, it was disciplined: resurface the rail, re-face the clamp, replace the stop pad, and give the container back the datum plane the drawing assumed. Two habits keep this from coming back. First, measure the fixture at the audit with the same gauge-block-and-indicator method, because the numbers in the sheet are the only way to catch 0.6 mm of bow before it becomes 3 mm of rejected labels. Second, quote the datum features on the preventive list in the same units as the part tolerance, so a part that allows 0.5 mm of placement error gets a fixture that is held to 0.05 mm, and the station keeps a ten-to-one budget instead of a coin flip. A machine that locates its parts honestly is cheaper than a machine that apologizes with software offsets.
A closing habit from the handover: keep a set of master containers at the cell, a few parts marked DATUM that live in a drawer and are never run through production, and use one as the check piece at the start of every shift. Slide it in, apply one label, measure with the placement gauge, and you will see 0.2 mm of fixture drift days before it grows to the rejection line. The line that owns this habit spends thirty seconds a shift and never learns the lesson the way this cell did, with a whole batch sitting in quarantine while somebody re-set the laser home register that was never the problem.
