My partial system upgrade

pillainp, A WHEA error is basically a hardware error. It is in many cases an unstable overclock. If you are overclocking, set everything back to defaults and see if you still get them. If you do, you have a hardware problem. If you don't, it is your overclock. Re do the overclock and test the crap out of each one and see if you get the WHEA errors. If it is your Overclock, in most cases it means increase your Vcore. But first you need to find out if you are having a hardware problem by setting everything back to defaults.

WHEA means a Windows Hardware Error ?. (3 out of 4 isn't too bad)

All overclocks disabled. Only RAM (TridentX - F3-2400C10D-16GTX - G.SKILL DDR3 Memory) running at XMP Profile #2.

No BlueScreens in two days now with moderate use.

I am beginning to wonder whether it might be my PSU.

When it was in my old system, I had a problem beginning a couple months ago with the death of an HDD. Following that, whenever I shut down the system, Windows would shut down but the hardware would remain powered up (CPU and case fans spinning, all LED's on the board and case receiving power and lit, etc.). The only way to totally power off the system was to lean on the power button or to cycle the PSU manually using its power switch.

This particular issue is not being repeated in the new build. 5VSB seems to be fine since the STRIX LED and the ROG logo and power button are lit, as they should be.

This is the PSU : TX Series™ TX850 — 80 PLUS® Bronze Certified 850 Watt High Performance Power Supply

It was purchased in 2013, and was part of an almost-always-on system (kids in the house - urgently need to play games at all hours), but was never a problem until a few months ago.

Wait and watch a few days and then go RMA? Advice?

I do note a tendency to sometimes freeze when running AIDA64's System Stability Test.

So I ran an overnight stress test (~4hr30min) on AIDA64 and no crashes, peak temps 68C.

Seems to be running OK so far.

Will do an extended stress test (>6hours) this weekend and report with logs.

It could have just been an unstable overclock. Sometimes, an overclock will 'degrade', for lack of a better word. That's when I go back to stock, like you are and test for stability. If it's stable, redo the overclock and make sure it's stable.