Do the PSUs have lights on them? Do those lights turn off? How is the power quality, clean or noisy? Are you using two power leads, or a Y-cable? If two leads, to the same circuit or separate?
I would agree that it's probably the backplane, and you can probably find one on eBay or Kijiji for relatively cheap.
And of course, if you haven't already, check the idrac event logs for anything interesting, and try the diagnostics.
The PSUs have lights on in the normal manner, they both have green lights when the machine is unresponsive. The power is probably OK, it works well enough for many other systems. The PSUs are configured in a redundant configuration, I've tried it running each of the PSUs separately and running both. Tried running both from a Y cable and from separate cables. I've tried switching the PSUs between bays. None of this changes it. But all on the same circuit.
There's nothing in the iDRAC logs related to it or at the same times.
Thanks for the suggestions. I'll try it on a different circuit but I have to put it back together again.
Yeah if you've eliminated all of those, it's probably the board.
I replaced the Power Backplane board J14R7 0J14R7 and it's been running for just over 24 hours. As it was an intermittent fault I don't count this as proof that it is fixed (I'll wait another couple of days to be sure) but previously it never lasted 8 hours so it seems most likely.
The old Power Backplane had no puffy capacitors, no scorch marks, and no other indication that it was faulty. I also used all the same cables so it doesn't look like a cable issue.
Thanks for your comment it helped give me confidence to spend the money on a new board. The other option I had considered was to just buy another system that would take the CPUs I bought.
I had a similar issue with an r730xd. Spent ages troubleshooting and replacing PSUs. Ended up being a fan I had plugged into the internal USB to cool a GPU. Removed that and never had any power issues since. Just incase you had done something similar
Thanks for the information, that's an unusual situation that you got yourself into. ;)
I had followed the advice of Dell support and unplugged everything including very unlikely things like NVMe devices and RAM and it still happened. It turned out to be the power distribution board.