A couple of weeks ago, as I was using the Convex to debug the SMD disk emulator, the Convex started failing in its memory initialization. 

Memory interleave set to 4
Initializing memory   0:00:00
Hard error reported by ATU
Micro-sequencer forced error
Source: fault during fault store
ROM address: 1f6   Base: 24   Index: 24

mminit: Unable to initialize main memory
Initialization aborted

During my attempts to debug this new failure, there was a fairly loud bang, as one of the smoothing capacitors in one of the LH Research Super Mite power supplies exploded. I immediately hit one of the emergency power-off buttons I have installed in my lab, fearing what damage to the cpu the power supply failure might have done. These same power supplies checked out just fine when I tested them with a dummy load, multimeter and scope a few weeks earlier. Not longer trusting these power supplies, and given the choice between risking further damage to some rare, irreplaceable CPU boards or replacing the power supplies with non-original newer parts, I decided to do the latter. As it happens, I had a few Cosel 5V @ 200 Amp power supplies lying around that are brand new in box (bought at a liquidation sale some time ago). Using these, and some smaller power supplies for the -5V, +12V and -12V lines, I put together a replacement for the three power supplies in the C1.

After installing the new power supplies, (and checking all the CPU boards for any visible damage), I apprehensively threw the switch to supplies power to the supplies, then turned the key to turn on the C1. As fortune would have it, it appears that the catastrophic power supply failure did no damage whatsoever to the rest of the machine. The mminit error went away, so probably this power supply started acting up some time before it blew up. Lesson learned: re-check the power supplies when sudden problems develop.

Oh, the power supply space in the C1 now looks kind of empty. It used to look like this: