What an SMT error-proofing system really solves: not seeing wrong, but nobody knowing it was wrong
The gap in manual checking is not eyesight but the absence of a second judgement. The value of error-proofing is turning the check into a searchable record, so mistakes are stopped before the line.
The first problem a material error-proofing system solves is not "an operator may misread", it is "nobody knows when someone does". Manual verification is a self-reported action: it was scanned, looked at and signed, but no second party verifies the result, so a mistake produces no feedback.
Where the cost of one wrong reel lands
Most of the loss from wrong material is not in the rework itself:
| Stage | Typical cost | Reversible |
|---|---|---|
| Found at self-inspection after placement | Remove, clean, replace, tens of minutes of downtime | Reversible |
| Found only at test | Whole batch functionally bad, quarantine by batch | Partly reversible |
| Found by the customer after shipment | Complaint, return, 8D report, supplier rating cut | Not reversible |
| Found as a missing record in an audit | Non-conformity, corrective action and re-audit | Reversible but slow |
The real cost sits in the last two rows: handling one customer complaint usually costs more than ten times an internal rework, and it enters the customer's supplier evaluation. An error-proofing system exists to push that probability forward, so the error surfaces at loading rather than in the customer's hands.
Four questions it must answer after go-live
Whether a system is actually running is judged by whether it answers these four questions, not by a feature list:
- Which reels went into this board, with batch and vendor for each
- Who loaded those reels onto which feeder, and when
- Whether there was any splicing during production, and which reel was spliced in
- Whether a substitute was used, and whether it was on the allowed list
If all four come back in one query, the data really is in the system. If any of them needs someone to dig out paperwork, that step is still a paper process.
It is not the same as improving quality
Error-proofing improves consistency, not process capability. It does not make the reflow profile more accurate or the print more stable; it only ensures that what goes onto the line matches the BOM. Presented as a quality improvement project, it will fail its acceptance criteria.
A workable business case states two measures: the number of wrong-material customer complaints, and the labour hours needed to prepare traceability documents. Both are measurable and comparable, and neither depends on process improvement.
Where the benefit is limited
- Lines with a single model, long runs and fewer than two changeovers a month already have a low manual error rate
- Lines with fewer than fifty part types and fixed rack zones rarely pick the wrong position
- Plants with no customer traceability requirement and no system audit pressure will not use the output
The benefit is most direct on high-mix low-volume lines with several changeovers a day under IATF 16949 customer audits.
Frequently asked questions
FAQ
- Will wrong material still happen with the system in place?
- Yes, but the probability and the nature change. The system blocks mismatches in feeder, part number and substitution; it cannot block an operator scanning under someone else's account or a wrongly printed reel label. Those need access control and incoming verification.
- How is the result measured?
- Two measures: the number of wrong-material customer complaints and the labour hours spent preparing traceability documents. The first shows interception, the second whether the data is usable. Avoid self-reported measures such as "number of errors".
- Does it suit high-mix low-volume lines?
- That is where it fits best. The more frequent the changeovers, the higher the manual error rate and the clearer the benefit. The condition is a feeder table generated per work order; otherwise every changeover needs manual maintenance.
- Does it have to go live together with MES?
- It can run alone, but pairing it at least with the traceability module is recommended. Error-proofing only produces scan pass or fail records; traceability turns them into a history that can be queried. Error-proofing without traceability leaves the data unused.