Ok the last thing to try now would be using the old psu with the bad hashboard unplugged ( completely removed ) and there will be 2 outcomes

1 - The temp error disappears and the kernel log looks very different to the first time  = psu is good but the board causes the errors sometimes.

2-  Kernel log goes back to the unusal temp and voltage readings = psu is bad, the bad hash board made things even worse.

And regardless of what happens, it is now safe to assume the hashboard is bad, maybe not completely dead but bad, the fact it shows 22 asics (if that doesn't change) it means either chip 22 or 23 is bad, after the psu swap you can put it back, perform a few restarts and see if it still shows exactly 22 chips, i will guide you on how to troubleshoot the bad chip and if you are willing to invest in some tools, i will tell you where to start with replacing the chips and where to buy them.