Three kinds of checks
Many plants verify their metal detector once at shift start, run a test piece through, and then leave it alone for the rest of the day. That only proves the unit was working at that moment. A more workable structure splits verification into start-up confirmation, interval spot checks and changeover confirmation, each triggered by a different event, with every result written into the same record so the chain stays traceable.
Setting frequency
Frequency should follow risk rather than habit. Lines with large batches, long continuous runs, or dusty and vibrating environments need shorter intervals between checks. For intermittent production or frequent changeovers, verifying after each changeover and at the start of each shift is often enough. For coverage, test each lane with both ferrous and non-ferrous samples and pass them at the left, centre and right of the belt, so a centre-only check does not hide a blind spot.
Tracing a failure
Handling a failed check matters more than logging it. Stop the line, trace back to the last passed verification, isolate the product made since then and re-test or downgrade it. Then find the cause: a worn test piece, debris on the belt, altered settings or nearby interference. Production resumes only after a repeat verification passes, and the whole handling process is filed alongside the record.