Marek @marek.onl · 31/03/2025After some more drilling and cleaning, the s/c disappeared. I connected some signal paths on the PCB by making some educated and some completely random guesses, and I used the thermal camera again to check that at least the last ASIC still worked. And it did! 020
Marek @marek.onl · 31/03/2025After more drilling, the mf still wouldn't go away, so I took another video. Btw, drilling a PCB is no fun. I was essentially burning the material away and subsequently inhaling it instead of drilling it. It seemed to be a four-layer PCB. 120
Marek @marek.onl · 31/03/2025However, the s/c wouldn't go away, so it was time to take out my new thermal camera. I connected the board to a regulated power supply, set the voltage to 12V, and limited the current to 10A. Here's what I got. 100
Marek @marek.onl · 31/03/2025I removed the SMDs related to the affected ASIC and drilled out the burned part to get rid of the s/c. 100
Marek @marek.onl · 31/03/2025The same failure happened again on the same board. This is what it looked like on the other side of the board. 100
Marek @marek.onl · 31/03/2025I recently managed to fix yet another failure of ASIC mining hw. The Z15 machine has three hashboards, each of which has three ASICs. This is what the hashboard looks like. The hole in the PCB is due to a previous failure in the power circuitry. 151
Marek @marek.onl · 03/01/2024This fella has been going strong for a decade now. It went through many short circuits and occasional flame but still in service. 010
Marek @marek.onl · 09/09/2023What the OS gets when you keep mallocing, but don't match it with free. 000
Marek @marek.onl · 05/08/2023Searching for 'error "Invalid search bound (wrong side of point)"' for the past month in Google doesn't return anything relevant. ddg.gg returns an exact match to a static html page in its first result. 000
