Solved! PC only starts after discharge

[jem-mosig]

I have a self-built PC with an ASRock Fatal1ty X399 Motherboard and an AMD Ryzen Threadripper CPU. It has run well for years, but recently I ran into the following problem:

* When I shut down my PC, sometimes the power LED blinks as if it is in stand by, and sometimes I think it shuts down normally, but I'm not sure.
* When I then push the power button, I hear a "click", all fans spin for a fraction of a second, and that's it. Not boot, no fans, no LEDs.
* I can now unplug the power supply and everything else, hit the Reset CMOS button, wait for half an hour, and reconnect things again. Then the motherboard slowly flashes blue light (from behind the ASRock sign). Sometimes, if I now press the power button, the PC actually starts and Windows is trying to repair itself, which works most of the time. I can then work or play games as if nothing happened.
* Next time I power down the PC, I have the same problem again.

To debug the problem, I've taken out all but one RAM stick, both GPUs and all HDs and so its just the bare MB with CPU connected to fans and power. The problem is the same! I can start it up once and wait for the Dr. Debug LEDs to turn off, so I know it has reached a steady state (I cannot connect a display without a GPU). When I turn it off and on again a few seconds later, I get the "click" and momentary spinning of the fans, but nothing happens. I tried the same thing also with another single RAM stick at another position.

Leading up to this problem, I had possibly related, but less severe issues with Windows Update. Whenever Windows tried to update and restart, it would drop into an infinite boot loop and I'd have to reset the PC using the windows repair stick. I suppressed Windows updates ever since.

Today I went through the CMOS-reset-and-wait procedure I described above to get my PC running again and then started a Memory Check. After the check was finished, it tried to restart but instead got into the Windows Repair loop. I turned off the PC at some point and updated the UEFI to Version 2.10, as it was quite outdated. The update worked, but the problem persists. That's when I started taking out RAM sticks.

What could the problem be, and what should I try next to fix it?

 

[mxnerd]

PSU. Almost 100%. Since I had almost completely identical experience.

You can try to replace CMOS battery first, but it probably will not fix the problem.

 

[jem-mosig]

PSU. Almost 100%. Since I had almost completely identical experience.

Thanks for your quick reply @mxnerd! I just took out the PSU and checked it with my Power Supply Tester, but the fan spins up, so it seems to work fine. Any other tests I could do (without buying another PSU)?

 

[mxnerd]

Never used PSU tester, so I don't know. But PSU fan is not the problem.

When my PSU was dying, it will not go into sleep mode properly. Sometimes it would, sometimes it wouldn't. Even if it seems went into sleep, CPU fan kept spinning, I wouldn't able to wake it up most of the time, however. Sometimes it went to sleep and froze or bluescreened.

USB peripherals sometimes will disconnect and reconnect randomly. And like you, if completely shutdown, it probably will not start unless unplug the power, wait 30 minutes, or unplug 24-pin power plug on the motherboard and reconnect to get it going. And many times I had to try it several times to get machine to boot up.

 

[jem-mosig]

Thank you for elaborating!

I now brought the PSU to a PC Repair shop and they tested it, saying that it seems to be fully functional. Though describing the symptoms this would also have been their first guess. I suppose if it's not the PSU, it must be the Motherboard or CPU? Can I test somehow which one?

 

[blckgrffn]

Thank you for elaborating!

I now brought the PSU to a PC Repair shop and they tested it, saying that it seems to be fully functional. Though describing the symptoms this would also have been their first guess. I suppose if it's not the PSU, it must be the Motherboard or CPU? Can I test somehow which one?

You PC is pretty high draw. If they just tested it with a tester, that wouldn’t have stressed the caps or whatever is having issues in your PSU in my opinion.

I’d buy a quality power supply and use that to test.

Worst case scenario you return it, keep it as a spare or sell it for a small loss. You’ve got nice stuff it would be a shame to keep pushing it until the PSU hurts everything.

That said, I did have one motherboard one time that was the culprit like this, but it was a cold boot issue where it wouldn’t start unless I did a rain dance of power button, unplug, power button, plug in, pray, power button again and then maybe start. It always rebooted fine. Updated bios, changed battery, changed cpu for giggles and ultimately just trashed the board after testing it by replacing it.

 

[bigboxes]

I'm pretty sure we're talking about a power supply issue. What is your PSU model? Is it under warranty? I have a power supply tester, but I always troubleshoot with a known good power supply afterwards.

 

[mxnerd]

CPU is either functional or completely dead. There is a chance that it's unstable, but that almost only happen when you are overclocking it, not during boot up. Or another possibility again is that PSU is not stable.

If PSU is not the issue, then it's the motherboard. If you can't return the PSU, then keep it as a backup. DIY people always have at least 2 PSUs.

 

[VirtualLarry]

DIY people always have at least 2 PSUs.

At least one "competent spare".

 

[jem-mosig]

> I’d buy a quality power supply and use that to test.
> Worst case scenario you return it, keep it as a spare or sell it for a small loss.
Good point, @blckgrffn , I'll try this

> What is your PSU model? Is it under warranty?
@bigboxes It's a SeaSonic PRIME Ultra Titanium 1000 W 80+ Titanium Certified Fully Modular ATX Power Supply with 10 years of warranty left. Granted, the guy at the computer shop could only test the main outlet, not the other modules, as the shop turned out to be specialised on Laptops.

> chance that it's unstable, but that almost only happen when you are overclocking it,
@mxnerd Ok. I'm not overclocking. Only thing I had to tweak was to activate SVM Mode and IOMMU for full WSL2 support.

 

[mxnerd]

So your PSU is pretty new and competent or you just bought it?

I'll also try to contact ASRock and see what their engineers will say, I have had good experience with them.

 

[jem-mosig]

> So your PSU is pretty new and competent?
Well, it's just 2 years old and has an impressive name . But reading latest Amazon reviews now, Seasonic doesn't seem to be too good at quality control for this one, which is strange, given the 12 year warranty.

> Try to contact ASRock
Will do.

 

[maluckey1]

Another vote for PSU, and even further placing blame on the 5v standby. Without using an actual PSU tester, there's no way to know (most shops don't do more than use a multimeter).

Thermaltake Dr. Power II are relatively cheap, and will eliminate or prove PSU failure rather quickly

Dr.Power II (thermaltake.com)

I own one. I've used it several times, and it saved me time and effort by finding a bad PSU.

 

[jem-mosig]

I've sent the PSU to Seasonic and they gave me a new one. I've reconnected most things (except the second GPU and second HD). But now when I turn on the PC, all fans start spinning (case, CPU, GPU), and after 23 seconds or so they (and all LEDs) turn off again. After another second they turn on again, and so on. No output on the screen and the numpad LED on the keyboard doesn't turn on either.

The Dr. Debug LED shows "C0" or "CO", which is not listed in the manual.

 

[jem-mosig]

Looks like C0 stands for "OEM BDS initialization codes". But I don't know how this helps.

 

[jem-mosig]

Ok, looks like two RAM sticks are faulty. Now I at least get into the UEFI interface. I just need to get it to find the HDs again, but that shouldn't be too hard. Thanks for everyone's help!

 

[nerp]

Aha! Bad RAM - along with PSUs, often the culprit in all sorts of weird ways. Tricky to diagnose, too, so good job narrowing it down. At least you've refreshed the PSU in the process LOL!

 

[Furious_Styles]

Aha! Bad RAM - along with PSUs, often the culprit in all sorts of weird ways. Tricky to diagnose, too, so good job narrowing it down. At least you've refreshed the PSU in the process LOL!

Could easily still be the mobo. In fact once he said what PSU he had I doubted it was the culprit, those are excellent PSUs.