I had a problem with my exhibit at The National Museum of Computing (UK) it is a transputer array connected to a PC running DOS that runs a slideshow interspersed with OCCAM code doing Mandelbrot and Ray Tracing on 1-4-16-64 transputers. The PC has four transputers internally, each with 4MB RAM and the base PC has 16MB and an 80387sx FPU, the remaining transputers are in their own powered enclosure. The HDD is an Innodisk EDC4000 4GB IDE SSD with a Seagate Ontrack BIOS overlay. The video card is an ISA CL5429 1MB. The problems I was getting were loss of BIOS settings, SSD refusing to boot, monitor losing sync, corruption in random files, machine locking up and needing to be power cycled, losing connection with the root transputer. I replaced the Dallas RTC with another known working module and that seemed to cure the problem for a couple of hours but it started locking up frequently. I replaced the SSD with a spare and that worked for an hour or so and then wiped itself. I left Doom running and made a cup of tea, came back, locked up. I changed the graphics card back to onboard and it got as far as counting the memory and froze about halfway and then wouldn't turn on again.

So I checked the power rails with the scope and yeah, they weren't good at all. Taking the PSU out of the case, it was quite dusty inside, gave it a compressed air blast and a swoosh around with a paintbrush and saw the problems quite quickly, leaking capacitors. I replaced all but two and reassembled the machine and it worked perfectly. I put all the original parts (not capacitors!) back in and no further problems after leaving it running all day. Back to Bletchley it goes 😊