Serialisation, marking, and record-keeping decided at design time cost almost nothing. Retrofitted, they cost a line.
The retrofit problem
Traceability tends to arrive as a customer requirement partway through a programme, sometimes after the first production run. At that point it is not a feature; it is a process rebuild.
Adding a serial number to a finished design means finding board area, choosing a marking method the existing line can apply, re-qualifying the process around it, and reconciling records for units already shipped.
What to decide at design time
First, the unit of traceability: the board, the sub-assembly, or the finished product. Each implies a different marking point and a different record structure, and choosing late means choosing twice.
Second, the marking method and its location — laser, label, or 2D code — placed where it survives assembly and stays readable through the enclosure. Third, what gets recorded at each station and where those records live.
Records you cannot reconstruct are the ones you will be asked for.
Make the mark machine-readable
A serial number a human can read is a compromise, not a system. If AOI or a scanner at each station cannot read the mark reliably, traceability depends on operator diligence, which means it depends on a good day.
Specify contrast, size, and quiet zone against the actual marking process and verify on real boards, not renderings.
The payoff is asymmetric
Done at design time, this costs a few millimetres, one process decision, and a schema. Done afterwards, it costs a line requalification and a gap in the record that no amount of engineering can close retrospectively.
The asymmetry is the whole argument.
Written by the manufacturing delivery team. Published articles carry a named author once attribution is confirmed.
