When âSoftware Bugsâ Are Actually Bad Soldering A quick MLX90640 IR camera test turned into a dead-on-I2C mystery. I ran the usual checksâaddresses, pull-ups, voltages, wiringâand even fixed my forgotten UART/I2C jumper, but still nothing. A peek under the microscope revealed sketchy solder on the 3V3 LDO; a quick reflow and the sensor sprang to life. Lesson relearned: donât trust factory soldering, look early with a microscope, and stop blaming software when the hardwareâs silent. 25 January 2026
ESP32 Rainbow QA - A tale of three boards Weâve shipped a few hundred ESP32 Rainbow boards-most are perfect - but a few failed QA. Hereâs how we tracked down and fixed three real-world faults: no sound from lifted headphone jack pads, USB not enumerating due to a mis-seated ESD IC pin, and a flaky touch keyboard from poor ESP32 module joints. A bit of microscope work and reflow brought two boards back to life (oneâs now a donor), and itâs a great reminder why we do final assembly and full QA here in Scotland. 02 January 2026
PDM Microphone Mystery I was puzzled that ESP-IDFâs PDM I2S config has no LR pinâuntil I actually read the micâs datasheet. On PDM mics, LR isnât a word-select clock at all; it picks which clock edge the mic drives and tri-states on the opposite edge. Tie one micâs LR low and the other high and both share the same data line for stereoâno extra GPIO needed. I scoped it with some nice pogo probes to confirm, and it matches the timing diagrams perfectly. Net win: one fewer pin in the next design. 10 October 2025