Preventing skipped steps in DIP: insertion, post-soldering and touch-up
Steps get skipped in DIP not because operators forget, but because nothing enforces the order. Routing checks must hold three things: station sequence, no-skip rules and the rework return path.
Most skipped operations at the DIP stage are not caused by operators forgetting; the process allows the skip. If there is no enforced order between insertion, wave soldering, touch-up, cleaning and test, a board can go from insertion straight to test with a step missing and nothing shows it.
Three ways an operation gets skipped
- Skipping a station: the next station does not check the previous one, so it accepts the board regardless
- Merging stations: one person performs two operations and scans once, saving the action and the record with it
- Rework returning: a defective board is repaired and goes straight back to test, bypassing its original operations. What was repaired and which part was replaced cannot be found in the system
In reports all three look like a complete routing. Only an investigation on the floor reveals them. Closing them depends on station-pass rules, not on more patrols.
Three checks a station-pass rule has to enforce
| Check | Rule | Error it blocks |
|---|---|---|
| Station order | A station accepts only SNs that passed the previous one | Skipped stations, missing operations |
| Time interval | Reject if the gap from the previous pass is below the process minimum | Moving on before soldering is complete |
| Rework path | After repair the board must return along a designated path | Shipping straight after repair, no rework record |
The first is basic; the second is often left out. Where cooling is required before touch-up after wave soldering, without an interval check a rush will push uncooled boards downstream.
Rework return is the weakest link
A reworked board has to answer three questions: who repaired it, what was done, and whether it was verified afterwards. On paper only the first usually has an answer: the repair record is signed, but which part was replaced and whether it was retested is often missing.
The system approach makes rework an independent pass path: defect registration → material issue for repair (recording the replacement lot) → retest → return to the designated station. Every step on the path is scanned, and the board cannot leave if one is missing.
Reporting and station passes are recorded separately
A pass record says "this board went through here"; a production report says "this station produced this much". Mixing them causes two problems:
- When one operator runs two stations, reported quantity does not match pass quantity
- Reworked boards pass again, so output is counted twice
Kept separate, output comes from reporting and traceability from passes, each closing on its own. Piece-rate disputes also drop by half, because reported data maps to a specific person and time window.
Suggested roll-out order
Do not put the whole DIP stage live at once. A workable order:
- Station-pass rules first: block skipped stations and fix the routing order
- Then the rework path: give defective boards a trail, which is what audits ask about most
- Then test data: bind test results to the SN and route defects into repair automatically
- Piece-rate reporting last: settle output definitions before discussing pay
Doing it the other way round (piece rate first) usually fails: when pass data is inaccurate the piece-rate result is disputed, and the plant falls back to manual counting.
When the DIP stage can wait
- Products with a single model, fewer than five operations and one team working end to end, where routing order held in people's heads rarely goes wrong
- Products with no rework flow, where defects are scrapped, can skip the rework path
- Where the customer requires batch records rather than unit traceability, pass records can be limited to key operations
Frequently asked questions
FAQ
- Does station-pass checking affect capacity?
- One scan takes about one to two seconds; at three to five passes per board, the added time per shift is measured in minutes. What actually hurts capacity is a badly designed rule that blocks valid work, for example a minimum interval set too long.
- How are reworked boards prevented from passing twice?
- Give reworked boards their own status flag, and distinguish first passes from rework returns at the station. Output counts only first passes; traceability keeps both records.
- Can production reporting and station pass share one scan?
- Technically yes, but the definitions get muddled. With one operator running several machines, labour hours cannot be attributed, and reworked boards are counted twice. Record them separately and link them through the same work order.
- Can an old line without network access still check passes?
- Yes. The client buffers pass records locally, checking continues offline and records upload when the network returns. Operations that need cross-order queries, such as stock or substitute-material lookups, are unavailable offline.