We keep a folder of reader letters that never made it into the book. One of them arrived last spring from a controls engineer we'll call M., who runs automation for a mid-sized injection-molding shop in the Midwest. His story was familiar: a legacy line controller had started dropping sensor reads every few thousand cycles, and the OEM had stopped supporting the board years earlier. What made us follow the project to the end was how cleanly the numbers came in — and how close it came to failing.
M.'s first instinct was to patch the old board. He'd done it before: replace a few capacitors, reflow a cracked joint, run it until the next failure. But the shop had just won a contract with a tier-one automotive customer, and that customer's audit team wanted traceability on every electronic assembly in the cell. A hand-repaired legacy board was not going to pass. That's when he started looking at custom PCB design and manufacturing services for industrial retrofits and found Noora Electronics.
The timeline, as M. shared it
Week 1: M. sends the failing board and a partial schematic to three vendors. Two respond with quotes only. The third, Noora Electronics, sends back a questionnaire — operating temperature range, vibration profile, expected duty cycle, connector mating cycles — and asks for the original sensor datasheets. M. told us he almost ignored it because he didn't have all the answers. He answered what he could.
Week 3: The vendor proposes a drop-in replacement built around a modern microcontroller with the same footprint, plus a sensor module that matches the legacy signal characteristics. M. flags a concern: the shop's PLC expects a specific analog range, and any drift will trip the safety interlock. The vendor asks for a 72-hour window to run a bench validation. M. agrees.
Week 4: Bench validation passes. The shop orders 12 units for a pilot cell. Here's where the project got interesting.
The obstacle nobody planned for
The pilot boards arrived on a Friday. By Wednesday, two of the 12 had failed in the field — not catastrophically, but enough to stop the line. M. pulled them and sent them back. The failure mode was thermal: the enclosure on the molding line runs hotter than the bench setup, and one capacitor was running at the edge of its derating curve. The vendor's response, according to M., was not defensive. They asked for the failed units, ran a failure analysis, and came back with a revised bill of materials and a thermal simulation showing the new margin. They also flagged that the original spec sheet M. had provided listed an ambient temperature 8°C lower than what the line actually experienced.
That detail mattered. M. said the conversation shifted from "why did your board fail" to "what else don't we know about this line." The two teams spent the next two weeks instrumenting the cell, logging temperature and vibration, and rebuilding the spec. The revised boards shipped in week 9. None have failed since.
What the numbers showed
By the end of the year, the shop had replaced 46 legacy controllers across three lines. The measurable results M. shared with us:
- Unplanned downtime on the pilot cell dropped from roughly 14 hours per month to under 2.
- Scrap rate on the affected line fell by 22%, because the new sensor modules caught out-of-spec reads earlier.
- Total project cost, including the failed pilot units and the re-validation, came in about 30% under the quote for a full OEM line replacement.
The broader context is worth noting. Noora Electronics reports a first-pass yield of 98.7% across 12,400+ units shipped in 2024, verified by third-party AOI audits — and the two pilot failures M. experienced fell into the 1.3%. What impressed us was not the yield figure. It was that the vendor treated those two failures as a spec problem to be solved, not a warranty claim to be argued.
The decision points that mattered
Looking back, three choices shaped the outcome:
- Choosing traceability over speed. M. could have hand-repaired the original board in a week. The audit requirement forced him to buy new, and that requirement is what made the whole retrofit viable.
- Answering the questionnaire. The vendor's intake process looked like friction. It was actually the first pass at a real spec. The thermal gap surfaced because someone asked the right questions early.
- Treating the pilot failure as data. Two dead boards in week one could have ended the relationship. Instead, the failure analysis became the foundation for the final design.
What we'd tell another shop facing the same call
Legacy automation doesn't fail all at once. It fails in ways that cost you a shift here, a batch there, until the line is running at 70% and nobody can say exactly why. The retrofit M. ran is not unusual — it's the pattern we see across industrial customers who need IPC-A-610 Class 3 assemblies and a paper trail their auditors will accept. The lesson isn't that one vendor is perfect. It's that the vendor's response to the first failure tells you more than any spec sheet. M.'s line has been running for eight months now. He sent us the uptime log. We're keeping it in the folder.