Question PC power cycles seemingly randomly, please help!

[snefan]

Hi all,

This is my current setup:

Mobo: ASUS ROG STRIX B450-I GAMING (BIOS 3103 x64, 6/17/20)
CPU: AMD Ryzen 5 3600X
Memory: Corsair Vengeance RGB PRO 32GB (2x16GB), CMW32GX4M2C3200C16, running at 2133 MHz
GPU: ASUS GeForce RTX 2070 8GB Dual MINI OC
NVMe: Samsung 970 EVO Plus 500GB (with Windows 10 Pro, fully updated according to Windows update)
SSD#1: Samsung 860 EVO 2TB (with Ubuntu 20.10 installed on one partition)
SSD#2: Samsung 860 EVO 2TB
PSU: Corsair RM750X V2 750W (see below though)

Other peripherals are:
Plexgear webcam connected directly to mobo USB port
Sennheiser headset connected to case from USB port
Plexgear USB hub (with separate power supply) connected directly to mobo USB port
Apple wired keyboard A1243 connected to hub
Microsoft Intellimouse Explorer 3.0 connected to hub

Neither CPU nor memory is overclocked.

The issue I'm having is that at my system power cycles at random. It's not my OS rebooting, and there's no BSOD, power just cuts out and then comes back again half a second later and then system boots fine. I'm seeing no correlation between what I'm doing and this happening. It's happened while I was gaming, while I was on a Teams video call, but it will also happen if I just start the PC and leave it doing "nothing".

Windows Event Viewer shows no useful information, only a a kernel power error about system being restarted unexpectedly. Nothing before that to indicate what has happened.

All temperatures seem fine, the only exception is the NVMe drive where Samsung Magician claims the temperature of 61° C is too high although specs say operating temperature is 0-70° C.

I have tried different stress tests (Heaven, OCCT, Prime95) to see if I can trigger the problem with high load or high temperatures but so far nothing. System runs through all tests without issues (see below for Memtest).

I'm aware software voltage readings are unreliable, but looking at them from BIOS or within Windows they seem fine.

Here are the things I've done so far:

1. Connected computer (with new cable) to another outlet in a different part of my house.
2. Replaced PSU with a Kolink Core 850W PSU. After replacing PSU, computer ran flawlessly for about 48 hours before power cycling. Once the first power cycle happened, it's gone back to the old behavior of doing it several times a day.
3. Tested memory with Memtest86+. This showed some weird behavior, with both sticks installed I ran the test for 24 hours without issue, same with stick A which tested fine for 24 hours in either slot, however with only stick B installed (in either slot), computer wouldn't boot at all so this stick hasn't been tested by itself. Please note that the power cycles occur when Windows is started with only stick A installed.
4. Installed Ubuntu 20.10 on one of the SSDs to see if the problem was related to the NVMe disk somehow. I have yet to experience a power cycle, however Ubuntu has frozen completely (screen locks up and system no longer reachable through SSH) on at least one occasion, but that shouldn't be the same problem manifesting itself differently right?
5. Booted into BIOS (with everything connected) and left computer running. It's now been sitting at BIOS screen (still responsive, I've checked!) for 38 hours straight.

Anyone have any ideas on how I can identify what's causing this? I appreciate any help!

 

[Fix_that_Glitch]

Try pulling one memory stick out at a time, see if it restarts randomly, if so repeat with the other memory stick and see if it stops.

 

[snefan]

Thanks, but as I mentioned (somewhat inexplicitly perhaps, original post edited) I get the restarts with only memory stick A installed as well. With only stick B installed computer doesn't post.