Engineering notes · 3 published
Short engineering notes written from programmes running on this floor, not from a marketing calendar. Each one takes a single decision that goes wrong often at a contract manufacturer, states the numbers behind it and shows what the process has to look like to avoid it. They are written on a floor with 6 SMT lines and 4,100 m² of ESD-controlled production area, which is why they stay specific about process instead of general about products; where a figure is an industry band rather than a result from this plant, the sentence says so.
3 published · 5 in preparation · reviewed February 2026 · written by the programme engineering desk, Bayan Lepas
Published notes
Each note stands alone and links to no product page. Read the one that matches the decision in front of you: starting a build, buying the parts, or deciding how much inspection is enough.
Note 01 / New product introduction
Most bad first builds are not process failures; they are decisions taken before the data package was finished. This note lists the gates worth closing before a single panel is ordered, from the land patterns and the panelisation to the paste volume and the test coverage, and shows how a 24-line checklist maps onto a 3-5 day prototype turn.
Read the note
Note 02 / Component sourcing
A counterfeit rarely arrives as an obvious fake. It arrives as a recovered part with a fresh top mark, in a reel that looks correct, during a shortage that has already pushed the lead time past the customer's launch date. This note follows the routes a part takes into a build, the screening that catches each one, and the six lines on a bill of materials that carry most of the risk, including the date-code and traceability checks that a franchised purchase makes routine.
Read the note
Note 03 / Inspection
Two inspection methods are often treated as alternatives when they answer different questions. AOI sees the surface and cannot look under a package; X-Ray sees the joint and cannot see whether the circuit works. This note sets out the failure each method catches, the failure each one misses, and how to choose coverage from the escape risk of a specific board rather than from habit.
Read the noteNote index
The index is maintained as a pipeline rather than a list of past posts, so a note that is in preparation can be requested and will be sent before it appears here.
| Note | Sector | Reading time | Status |
|---|---|---|---|
| The NPI checklist that prevents a bad first build | New product introduction | 9 min | Published |
| BOM sourcing risk: how counterfeits enter a build | Component sourcing | 7 min | Published |
| What AOI misses and what X-Ray cannot see | Inspection and test | 8 min | Published |
| Choosing a conformal coating for an outdoor enclosure | Materials | 6 min | In preparation |
| Design rules that decide 01005 placement yield | Design for manufacture | 7 min | In preparation |
| Thermal vias and heavy-copper stackups for power boards | Power electronics | 8 min | In preparation |
| Planning test coverage by escape risk | Test engineering | 6 min | In preparation |
| Reading an ionic contamination report | Cleanliness | 5 min | In preparation |
Editorial process
Three steps, applied to every note on this page, which is why a note carries measured values rather than general advice.
The first draft comes from the engineer who ran the build, and it carries the programme reference, the board class and the measured values from the lot. Where a figure is an industry band rather than a result from this plant, the note says so in the sentence, because a buyer who later discovers the difference stops trusting the whole set.
A second engineer verifies every number against the standard revision cited and against the record it came from, which is typically a first-article report, an X-Ray void measurement against the 10% Class 3 limit or an ionic contamination result. If a claim cannot be supported by a record we can produce on request, it is removed rather than softened.
Each note carries the date it was last reviewed, and any note affected by a standard change is revisited within 30 days of the revision being published. The affected note is updated with a dated change line so a reader who cited the old value can see what moved and why. Notes are added to the index when they are written, which is why the five in preparation appear above without a link.
Seven fields. A process engineer replies within one working day.
Seven fields. A process engineer replies within one working day.
Seven fields. A process engineer replies within one working day.