[SOLVED] PC power dies during gaming - struggling to troubleshoot

Nibbles48

Reputable
Aug 17, 2019
26
0
4,530
Hi,

I am having an increasingly frustrating issue with my PC. Essentially, during gaming (most notably Assetto Corsa, though others have triggered it aswell!), the power dies completely. The power buttons wont work to turn it back on. When it does eventually fire up again, normally once completely cooled down etc., i notice that the date and time remain as they were when it died (i.e. several hours behind current time).

Troubleshooting so far

  1. I replaced the battery on the motherboard - that got it to kick back in to life the first time, but has made no difference since
  2. i thought the CPU might be running hot (i noticed that the PC was running slow, like it was being throttled). Sure enough, Speedfan tells me the temperatures were in the high 70s/low 80s (this for a CPU which apparently has a maximum temperature of 72 degrees - see https://safetemp.blogspot.com/2016/06/intel-core-i5-650-max-temp.html). I decided to remove the heatsink and give it a clean, and whilst at it replaced the thermal compound, which immediately brought the temperature back to a sensible level. But still, the PC kept crashing during gaming.
  3. but, when remounting the heatsink one of the plastic push pins holding it in place snapped, meaning a not quite perfect fit. Within a few months, i was back seeing similar temperatures to previously and then, during a heatwave a few weeks back, the PC shut off (as per the symptoms above) whilst browsing the internet (not under load). I decided to install a new cpu fan (this one with screws) which has dropped the temperatures to a perfect range. I've been running a burn in test on the intel diagnostic tool for the last few hours, and saw a max temperature of 64 degrees - so i think this has worked. But, despite that, it still shuts off during gaming.
  4. i read around this forum and found others suggesting this sounded like a PSU issue (such as https://forums.tomshardware.com/threads/pc-randomly-shutting-off.3498715/). That sounded logical, and my previous PSU had a pretty poor rating online. I replaced with a corsair vs550. Lo and behold, running a test yesterday and it shut down once again.
I've run the intel diagnostics tool many times, and that says the motherboard is absolutely fine. I did notice that, when it had died and refused to power up even when pressing the power button, if i unplugged the 24 pin plug from the motherboard and connected the green cable to a black cable, the PSU started fine (my old one did too). But then when i plugged back into the motherboard, it still wouldnt fire up. The second time i tried that i let the fan run for a couple of minutes and when i plugged it back in it worked as normal. I've tested a few other bits, i think the HDDs (via a seagate app) and the GPU (using Unigine Heaven), all seem to work fine (i didnt run the Unigine test for long, though perhaps will now). Re the GPU - i dont have any of the issues highlighted on https://www.maketecheasier.com/faulty-graphics-card-and-the-fix/ .

Does anyone have any ideas, or other troubleshooting tips?

System details;

Gigabyte H55M-UD2H
Intel i5 650 @ 3.2GHz
Nvidia GeForce GTS 450

Thanks in advance
Steve
 
Solution
Sounds like mobo is going out. Any way for you to test known good parts: mobo, cpu, ram? Is the PSU plugged directly into the wall or is it on a strip?
Welcome to the forums my friend!

Firstly, the link for the Intel temperature you have is incorrect. This site is citing the Intel TCase temperature, which is NOT the same as maximum core temperature. The Tcase is a factory only measurement at the integrated heat spreader. TJunction is what you need. As an extract from Computronix Intel temperature guide:

Tcase has always been a confusing specification. Here's why:

When users of 6th Generation and earlier processors see their Thermal Specification on Intel’s Product Specifications website, most don’t realize what Tcase actually means. Since there are numerous software utilities for monitoring Core temperature, users assume Tcase must be maximum Core temperature. This is a basic misconception which has persisted since 2006.

Tcase is not Core temperature.
https://forums.tomshardware.com/threads/intel-temperature-guide.1488337/

Secondly what is your entire system spec including PSU make and model? Is it the VS?
What is the maximum individual CPU core temperature under load you are reaching? Is this the 64 degree you are referring to?
What is the maximum GPU temperature under load you're reaching?

I replaced with a corsair vs550
The VS is also a poor quality PSU, so can't remove it from the table just yet.

Makes no difference, common symptoms are just that, common, but there are plenty of other symptoms of a dying GPU. One of which is random restarts.

Does the crash only occur under load?
 
Hi,

I don't have another computer that I can try parts in.

Yes the PSU is the VS550. Noted that it's not a great one, but gets better reviews than my previous one.

Yes the highest individual core is the 64 degree temp

Interestingly (maybe) I ran the Unigine GPU test after posting the original message. I noticed the GPU temperature started to increase. I left the room, 5 minutes later went back and found the PC in darkness. Which is leading me to suspect that it's a GPU problem. I suppose first check would be to remove the fan and heatsink and give it a good clean.

Also, for info, I now can't get the PC to fire up at all (even after the paperclip trick!). Any tips?

Finally, PC is plugged in to a strip. I didn't replace the power cable when i installed the new PSU, just used the old one. Perhaps thats foolish?

Thanks
Steve
 
Yes the PSU is the VS550. Noted that it's not a great one, but gets better reviews than my previous one.
I understand - and yes I have no doubt it was better, yet still poor quality as it is their budget line not really made for higher load rigs. Cheap PSUs = Cheap Components = Lower quality.

Not saying it is the cause, as I probably would expect it to run if the PSU was swapped, but I've had countless issues with the VS series in the past (as many others have).

Finally, PC is plugged in to a strip. I didn't replace the power cable when i installed the new PSU, just used the old one. Perhaps thats foolish?
Depends on your location, but usually they're pretty interchangeable, it's the modular internal cables that are more important.

Interestingly (maybe) I ran the Unigine GPU test after posting the original message. I noticed the GPU temperature started to increase. I left the room, 5 minutes later went back and found the PC in darkness.
This could be a good indicator. I'd be tempted to remove the GPU entirely and run on integrated and see if any random reboots occur.
 
Ok il give that a go when i can get it fires up.

For info, if the GPU triggered a shutdown, does it 'tell' the mobo not to allow start-up for a period of time? Because I can't get it to power up at all at the moment (it's been off about an hour!). If i remove the GPU from the power and from the slot on the mobo is it likely to start straight away?

I was hoping to run the Unigine test again but monitor the temperature until it fails...!
 
For info, if the GPU triggered a shutdown, does it 'tell' the mobo not to allow start-up for a period of time? Because I can't get it to power up at all at the moment (it's been off about an hour!). If i remove the GPU from the power and from the slot on the mobo is it likely to start straight away?
Less that it tells it to not power up, more than the GPU will cause it to fail to POST.

Does literally nothing happen when you press the power button?
 
I haven't, nor am I completely sure how I do that. Im sure i can figure it out. Would that mean power is on when the switch on the PSU is on?

FYI, I've headed out for the evening so will have a play with it soon and come back to you.
 
OK, i'll have a play with that later. I dont think i have started it using the motherboard previously. I had tested the PSU by putting a paperclip between the green and black wires on the motherboard connector.
Ah I understand now by what you meant earlier.
Well this tests if the PSU can power up, but not if it is actually fully working or not. As it doesn't identify the voltage outputs from each rail as to whether it is correct or stable. You only need 5% fluctuation for there to be a problem.

And yes, you'll need to know which is your power pins on the motherboard and short them with a jumper of using a screwdriver (obviously one with an insulated handle so you don't electrocute yourself...)
 
So an update. When I got home from work the power button on the front of the case powered it on straight away.

I then shut down, and removed the GPU. But then nothing. I read on another thread that removing the battery for 5 minutes or so can reset the BIOS and it should then work fine, which il check shortly.

https://www.quora.com/Why-wont-my-computer-boot-after-installing-graphics-card

Assuming it starts, il run the Unigine test again and see what happens.

Whilst i think about it, I have a feeling that I removed the heatsink on it a while ago, but possibly didn't put new thermal compound on (i may be imagining things). That won't help. But if I did remove the heatsink it was an attempt to sort out the random power off issue.
 
I then shut down, and removed the GPU. But then nothing. I read on another thread that removing the battery for 5 minutes or so can reset the BIOS and it should then work fine, which il check shortly.
Also just clicking Reset to default in the BIOS does the same. Also doesn't need to be 5 minutes. Technically 10 seconds usually works, but leaving it for 1 minute is enough.
 
A further update.

I removed the graphics card, loaded on integrated, worked ok though obviously slow when running unigine. I removed the heat sink from the gpu, replaced thermal paste, gave it all a good de-dusting, reinserted it, but now it once again won't boot at all. I tried removing the cmos battery for a few minutes and that didn't help. Also tried power on via the motherboard.

So I'm wondering whether the motherboard is functioning well now... :|
 
Hi all,

Sorry for the delay in coming back to you. I tried for about half an hour to get the PC to fire up with teh GPU installed - couldnt get it to do a thing. At one point i had taken it off to another plug (unit only, no monitor etc.), and managed to get it to start - but then when i unplugged and put it back in its normal place it wouldnt start again.

I removed teh GPU to run on integrated again, worked almost straight away (its what im typing on now!).

If the GPU is knackered (perhaps through overheating?) and was plugged in to the board, w/could that cause it not to boot?

If the GPU is knackered, i have the choice of either replacing the mobo and GPU or just the GPU. Im likely to do the latter, given the amount of gaming i do nowadays (i found a website the other day which suggested the GPU was the bottleneck rather than the CPU, and recommended a GPU that was more balanced). But I don't want to do that if the mobo is about to go pop too.

EDIT - I was about to hit send when, unexpectedly, the pc cut off (on integrated, remember!). I've tried to get it going again with absolutely no success. When I think points me back at the mobo. I don't have a way to test the GPU. Do you think I'm likely correct?

Steve
 
Hi all,

Sorry for the delay in coming back to you. I tried for about half an hour to get the PC to fire up with teh GPU installed - couldnt get it to do a thing. At one point i had taken it off to another plug (unit only, no monitor etc.), and managed to get it to start - but then when i unplugged and put it back in its normal place it wouldnt start again.

I removed teh GPU to run on integrated again, worked almost straight away (its what im typing on now!).

If the GPU is knackered (perhaps through overheating?) and was plugged in to the board, w/could that cause it not to boot?

If the GPU is knackered, i have the choice of either replacing the mobo and GPU or just the GPU. Im likely to do the latter, given the amount of gaming i do nowadays (i found a website the other day which suggested the GPU was the bottleneck rather than the CPU, and recommended a GPU that was more balanced). But I don't want to do that if the mobo is about to go pop too.

EDIT - I was about to hit send when, unexpectedly, the pc cut off (on integrated, remember!). I've tried to get it going again with absolutely no success. When I think points me back at the mobo. I don't have a way to test the GPU. Do you think I'm likely correct?

Steve
I'd be tempted to RMA the board at this point.
PSUs are still often the most likely cause, but even a VS I would expect to not immediately replicate the same issue.
 
I went through all the posts, completely on board with everything y'all tried. I've just been through the same issues with an MSI Z390 Tomahawk, tried everything (which included everything here), and my board is still completely dark. One day it just "quit" in the after working fine for a couple months. My problem now is that numerous attempts to contact MSI have gotten absolutely no response since the very first one. It's been over 30 days now. I wonder if the trade war includes their even more miserable customer service issues than normal at this point. Good luck with your RMA!
 
Hi all,

Apologies for dragging this back up.

I've not used the desktop much for a while, certainly not for gaming anyway. But today, whilst experimenting with a program, the PC went dead on me as described above. I've so far (for an hour or so) not been able to get it going again.

I strongly suspect the culprit is the motherboard, though dont have an easy way to test. Its also ~10 years old. Given that, and that i believe the CPU is fine, im tempted to replace the motherboard completely, though keeping the remaining components the same. I'd rather not have to pay to replace the whole PC if i have to, with some big (house related) expenses on the horizon.

So, can anyone advise what motherboard will fit my current (working) components. I have DDR3 ram (4x2GB), socket 1156, and the GTS450 gpu. Im assuming it would simply be a matter of unplugging everything, removing the old one then inserting the new one?

Im looking at this one - does that look ok? Any particular things to look out for (clearly i dont want it not to fit etc.)

http://cgi.ebay.co.uk/ws/eBayISAPI.dll?ViewItem&item=254468189390

Thanks
Steve
 
Morning all,

Thought i'd provide an update - took the plunge and purchased a new (used) motherboard, not the one above but a similar one, from ebay. Ordered at 3pm, arrived the following day at 12.

Removed the old one and put the new one in and, after a few minutes of fiddling with RAM errors (they werent, despite what they looked like, connected properly) and trying to find the OS i got it running again - Windows 10 was super helpful with no need for a reinstall. Ran an intel CPU check which was absolutely fine.

Then the moment of truth - running Unigine - which had caused the computer to switch itself off as described above, normally within the first 15 seconds. Success - it ran for 10 minutes with no issues at all, and im therefore, for the first time in ~6 years, confident that the issue is solved.

Thanks everyone for your help, much appreciated!

Now to find a small GPU upgrade to remove a bottleneck! 😉

Thanks
Steve