Tips for getting a first board right the first time

Congrats on the clean first spin, that feeling never gets old. My last board, a pool pump timer, also came up working on the first try and I credit the same habit of powering the supply section by itself. In my case I caught a reversed footprint on the regulator before anything else was soldered, which would have been a miserable rework job with the whole board populated.

Where I differ from you is the breadboard stage. I skip it for the microcontroller side and just tack the chip onto the board with solder paste and hot air, then debug in place. Breadboarding a modern chip with all its decoupling always behaved differently for me than the real board anyway, so I stopped trusting the breadboard results years ago. A stencil and a cheap hot plate paid for themselves after two boards.

The connector thing hits hard though. My pump board has one silkscreen label that says 'VALVE' where it should say 'PUMP', and every time I look at the enclosure I remember that I rushed the layout on a Friday night. Next revision is allready on my desk with corrected labels and a slightly bigger spot for the wire clearance, and I am honestly tempted to leave it there forever since the current one just works.

The ones that work first spin always make me nervous, half the time the real bug is still sleeping in there waiting for a hot day or a brownout. I learned that back when we wire-wrapped prototypes, a board could look perfect on the bench and still fall over in the field. These days I let rev A run on the bench for a good month with the supply cranked high and low before I even think about a second version. Nothing worse than finding the gremlin after you've already ordered ten boards.

On the firmware room point, that's saved me more times than layout changes have. My greenhouse vent controller from a few years back had spare flash and two unused pins broken out to pads, and I've pushed four firmware updates to it since without touching the hardware. Last one added a slow startup ramp I never would have thought of on day one. These days I won't order a board unless there's at least one spare GPIO soldered to a pad somewhere, sounds wasteful until the day it isn't.

One thing I started doing after a lucky first spin like that: write down every voltage you measured during bring-up and tape it to the inside of the enclosure. A fan board of mine ran fine for two years, then started cycling wierd on hot afternoons. Pulled it out, measured the 5V rail, compared to my old notes, and it had drifted low. Swapped the regulator and it was back to normal in ten minutes. Without those numbers I'd have been guessing at the firmware first.

Update: the poster who said first-spin boards hide a bug til a hot day was bang on. Ran the controller in the attic box last week during a warm spell and the fan ramped in little jumps instead of smooth steps. Turned out my sensor averaging was too slow. Fixing it in firmware only, so the copper stays clean, but yeah, never trusting a quiet bench again.