Random reboot on old PC - what is dying?

[qbazdz]

Hi,

Wonder if anyone else encountered a similar problem and can speed up my troubleshooting please.
I've got an older PC (~10y old) with custom WC setup so a bit harder to strip down.
I had a replacement PSU from Corsair under warranty back in 2016 and since then the PC had had very occasional use (certainly fewer than 8hrs in a month, don't ask, it's complicated haha)

I started to use it more recently, did a few upgrades that suit me and done a fresh install of Win10.

Now, the actual problem is, that randomly the system will simply reboot, no warning, nothing simply instant power down and attempt to reboot. This very occasionally will succeed, but most of the time it will be stuck in the attempt to boot, fans will spin up and than it will die again and repeat.
Turning PSU off, de-energising the mobo and power up doesn't usually help (the moment you flick the psu switch it normally tries to boot without case power button being pressed), unless you give it 5-10min.
My Z68 Extreme 7 Rev 3 board, shows code corresponding with 'PEI Core is started' before it crashed after first one.

I suspect the PSU may be dying but it had so little use I am amazed this would be the case?

Any other ideas?
SSD is brand new, with all tests passing no problem.
I checked the battery voltage on the board - seems ok.
I didn't have any problems getting windows up and running, a number of time, had been using it for extended periods of time and this problem seems to be coming and going but its incredibly annoying!

Thanks

 

[gavinh87]

Run a rem test program

 

[Tetras]

If you have a spare battery I'd replace it anyway.

Is it overclocked? Memory voltages and timings all okay?

What upgrades did you do?

Have you tried another SSD?

 

[qbazdz]

If you have a spare battery I'd replace it anyway.

Is it overclocked? Memory voltages and timings all okay?

What upgrades did you do?

Have you tried another SSD?

Its an old 2500K that run a small o/c when I needed it (say 8-9 years ago). Once I stopped using it regularly it was dialed back to stock.
DDR3 Memory has been running with its XMP profile since the built.
Most recently I had replaced the GPU (however the problem was present before hand) and prior to that I swapped 128GB old SSD for a new 1TB 860 Evo (this coincides with my increased usage and its hard to say if the problem was present before the SSD swap as I almost immediately went to set it up from scratch).

Just kicked off the Memory Diagnostic Tool from Windows, need to find a spare USB to run MemTest.

 

[HangTime]

Check in eventvwr to see what errors have been logged when it reboots.

Another thing to consider with an old CPU that has been overclocked is it could be decaying somewhat, in my experience some CPUs can require more and more voltage to run stable if they are old and have been overclocked. For example I have a 3570k that I've dialed back from 4.5ghz to 4.4ghz and needs a bit more voltage than it used to. I know you are back at stock speed but you could try bumping the vcore a bit to see if that helps.

 

[qbazdz]

Quick update - Memtest86 run 4 loops, ~2hrs and found no issues.
I had a spare battery so swapped it again, was amazed to see that the replaced one already was down to 2.7V, a bit low but apparently functional for all the functionality safely.

Now I've got to deal with the stupid windows update which failed repeatedly (due to the unexpected reboots) and hasn't yet installed 21H2 properly an it keeps coming up with an error..

 

[Major774]

Maybe PSU again for you.

My new Seasonic SFX 650W couldn’t handle my undervolted 5800X and undervolted 3080 FE (odd random reboots).

A Silverstone SFX 1000W resolved my issue, no more random reboots.

 

[Ormy1]

Few things to try:
Reset BIOS to defaults
Remove and reseat every component and power cable (if you can't remove something because it's water cooled, just give it a wiggle and make sure it's seated firmly).
Try a different/backup PSU
Unplug all non-essential peripherals
Check temps are all within normal ranges when PC is running, poor temps could be a sign of degraded TIM which could be causing shutdowns.

 

[Trickle]

I've got an older PC (~10y old) with custom WC setup so a bit harder to strip down.

It shouldn't be too hard to get the power supply out of it.

 

[qbazdz]

It shouldn't be too hard to get the power supply out of it.

Yes, except for all those cables that are neatly (*at least to some extent) organised in a tightly packed case

Ormy1
Defaults flashed, cables already re-checked twice, I even had the PSU once fully out to check with my external PSU checker by Antec
Temps had been absolutely fine, I even had done several loops of 3d Mark and Furmark once I got new GPU installed.

 

[Lee Monk]

Hey.

I had a similar problem two years back..... My PC is about as old as yours.

I noticed it was struggling with some games and then over a one week period it would restart and in some cases .... not.

I took it all apart and gave it a 'general' clean. It didnt work. Upon closer inspection the graphics card was litteraly packed and I do mean packed with dust.

I took it apart (be careful, not because it was expensive but beacause it was 10 years old). Gave it a damned good clean and it all worked.

Mine was easy to do. The case of the GPU was only held by 4/5 tiny screws.

 

[qbazdz]

Reviving this again as I had been monitoring the situation over the few weeks with little progress really.

Had another look at the CMOS battery and thought it was a bit low and had a new one so replaced it just in case - CMOS clear and re-configured back to defaults.
@HangTime thanks for reminding me about event viewer - this showed almost nothing - certainly no events in minutes leading up to power being cut off, except for one particular process (think intel ISA) that was trying do do an update for it and failed to connect with the server.
After disabling it, I thought that would be it, however had another event last night while kicking some ass in MK11.. likely coincidental, but about the same time as my previous reboot. This time however nothing in the even viewer, so definitely not the process that had been the problem before..

Currently still can't for some reason successfully complete Windows Update to 22H2 as 21H2 fails.. I tried all possible fixes and will have to probably bite the bullet and do a clean install of 22H2 right off.

@Lee Monk, everything is clean - machine had a rebuild and a clean recently and new GPU has a water block and had also been cleaned before fitting.

Think I may try the PSU next, unless some good ideas come up?

 

[darael]

Get a multimeter and test the voltages, both at idle and when fully loaded. FurMark and Prime95 might be ideal candidates for this, I just run both at the same time if I really want to stress test my system.

 

[qbazdz]

@darael CMOS battery? or CPU Vdd? not sure if mobo has got test points for that

 

[darael]

@darael CMOS battery? or CPU Vdd? not sure if mobo has got test points for that

I test using the 4-pin hard drive molex connector, as it has 12v and 5v rails. If you're careful, you can probe the 3.3v on ATX connector too.

Most PSU's seem to have all their wiring in black these days, but normally orange wires are 3.3v, red 5.0v and yellow is 12.0v. You're looking for the voltages to, ideally, not exceed 5% tolerance values.

 

[erazor51]

Sounds like a faulty motherboard.

- YouTube

Enjoy the videos and music you love, upload original content, and share it all with friends, family, and the world on YouTube.

youtu.be

 

[qbazdz]

Sounds like a faulty motherboard.

- YouTube

Enjoy the videos and music you love, upload original content, and share it all with friends, family, and the world on YouTube.

youtu.be

That is the type of behavior I get after it crashes, however why would it boot up after a while, as well as crashed after random times?

My current routine is normally to boot up - start up MK11U and be able to play for at least 30min before anything happens. Sometimes it takes longer.
I will still look around for leaking caps or anything on the mobo.

 

[Had3s]

Perhaps you nudged the CPU cooler. Do you know the state of your thermal paste, or last time it was done? Thermal cut-out could be your problem, if it's no bulging/failing capacitors.

 

[qbazdz]

Perhaps you nudged the CPU cooler. Do you know the state of your thermal paste, or last time it was done? Thermal cut-out could be your problem, if it's no bulging/failing capacitors.

Its under WC, so unlikely, besides temps don't move past 40-45c under load. Ill try with a different PSU first.

 

[Had3s]

Reluctance due to WC loop is understandable but after 10 years you have no idea what the paste is like under that block. Is your plan to never remove the cooling pipes?