Most plants know their rework headcount. Few know what rework actually costs, because the cost is spread across labor, materials, capacity, schedule and reliability, and only the first of those shows up under the word “rework”.
The 10x rule
The 10x rule is a long-standing manufacturing rule of thumb: a defect costs roughly ten times more to deal with at each later stage where it is found. It is a guide and not a law, but it describes electronics assembly well.
| Where the defect is found | What fixing it involves |
|---|---|
| At the printer | Wipe and reprint the board |
| After placement | Remove parts, clean, reprint, replace |
| After reflow | Rework station, skilled operator, local reheating, re-inspection |
| At functional test | Fault diagnosis, then all of the above, then retest |
| In the field | Return, failure analysis, replacement, possible containment action |
The costs you can see
- Rework operator time at a fully loaded labor rate.
- Replacement components, flux, solder and cleaning materials.
- Rework stations, tooling and their floor space.
- Retest and re-inspection time.
The costs you cannot see as easily
Work in process. Boards waiting for rework are inventory that is paid for and not shippable. They also get lost, damaged and mixed up.
Schedule. A lot that is 97% complete cannot ship. Rework is the most common reason an order that was built on time is delivered late.
Reliability. A reworked joint has seen extra thermal cycles. Pads lift, laminates are stressed and neighboring components are heated. A reworked board passes test, but it is not the same board as one that was built right the first time. For high-reliability products, some customers limit or prohibit rework for this reason.
Scrap. Some boards cannot be recovered, and they are scrapped with every component already on them.
Engineering attention. Every hour spent diagnosing an intermittent short is an hour not spent improving the process.
Why rework persists
Rework is comfortable. It is an established department with skilled people who solve visible problems every day. The alternative, preventing the defects, requires knowing where they come from, and without measurement at the source that is guesswork.
A practical plan to reduce it
- Classify a month of rework by defect type and location. A handful of causes will account for most of it.
- Trace each major defect type back to the process step that creates it.
- Measure at that step. For solder defects this usually means measuring paste at the printer.
- Set control limits and react to trends, not only to failures.
- Track rework rate monthly and confirm it is falling.
The cheapest defect is the one you find while it is still wet paste on a bare board. Everything after that is the 10x rule at work.