Random power cuts, no BSOD

[StaticShadow]

I've been having a problem with my desktop randomly restarting when under any load (when just on my desktop or programming I haven't experienced any crashes). The crash involves an immediate shutdown of the computer, as if all power was cut. The fans and lights all go out. I've tried continuously to find any logs, however it appears none are produced, which with the way it shuts down, doesn't surprise me. It seems if I leave the computer unplugged for a day or two I can go for around 6 hours or more without crashing, however if I continue using the computer without doing this the crashes become more and more frequent. Running benchmarks like furmark and aida64 can have a wide range before the system crashes, anywhere from 5 minutes to an hour or more. These symptoms make me feel like the issue has something to do with power or temps, but voltages and temps seem to be fine in aida64 under load. My motherboard is relatively new, and this issue was happening before it was replaced. I'm not sure what could be causing this. My guess is the PSU, but I don't have any others to test with and want to try and diagnose the issue without buying anything. All suggestions are appreciated, thanks

OS: Windows 10 64-bit
Specs:

• GTX 970
• Intel i7-4770
• 2x 8gb DDR3
• MSI z97 gaming 5
• EVGA 550 G2 (psu calculators suggest ~400w, so this should be sufficient)

[Skiiwee29]

Almost seems like the PSU might be overheating. Is the fan running on the PSU?

[StaticShadow]

2 minutes ago, legacy99 said:

Almost seems like the PSU might be overheating. Is the fan running on the PSU?

I checked and it seems to be, however not very strong. Should I try putting it on top of my bottom intake fan? (Shouldn't hurt the overall heat too much)

[Skiiwee29]

5 minutes ago, StaticShadow said:

I checked and it seems to be, however not very strong. Should I try putting it on top of my bottom intake fan? (Shouldn't hurt the overall heat too much)

No I dont think that will be needed. By not very strong, do you mean you can see the individual blades as its spinning, or does it seem to be spinning like a normal fan would?

[StaticShadow]

1 minute ago, legacy99 said:

No I dont think that will be needed. By not very strong, do you mean you can see the individual blades as its spinning, or does it seem to be spinning like a normal fan would?

It runs like a normal fan, however a 1"x1" piece of tissue will barely stick to the fan, and a slightly smaller piece of printer paper falls off. Not sure how powerful a PSU fan needs to be, but my intake fans can hold the piece of paper against the case. Should I try feeling the outside of the PSU after a crash? Would it be hot to the touch?

[Skiiwee29]

Just now, StaticShadow said:

It runs like a normal fan, however a 1"x1" piece of tissue will barely stick to the fan, and a slightly smaller piece of printer paper falls off. Not sure how powerful a PSU fan needs to be, but my intake fans can hold the piece of paper against the case. Should I try feeling the outside of the PSU after a crash? Would it be hot to the touch?

not Necessarily, its more the components get to hot and it triggers a fail safe to shut down. @STRMfrmXMN would be able to help you more on this if it is truly the PSU as he is a PSU guru around these parts. 

[STRMfrmXMN]

4 minutes ago, legacy99 said:

not Necessarily, its more the components get to hot and it triggers a fail safe to shut down. @STRMfrmXMN would be able to help you more on this if it is truly the PSU as he is a PSU guru around these parts. 

If a PSU fails the system usually BSODs.

 

35 minutes ago, StaticShadow said:

I've been having a problem with my desktop randomly restarting when under any load (when just on my desktop or programming I haven't experienced any crashes). The crash involves an immediate shutdown of the computer, as if all power was cut. The fans and lights all go out. I've tried continuously to find any logs, however it appears none are produced, which with the way it shuts down, doesn't surprise me. It seems if I leave the computer unplugged for a day or two I can go for around 6 hours or more without crashing, however if I continue using the computer without doing this the crashes become more and more frequent. Running benchmarks like furmark and aida64 can have a wide range before the system crashes, anywhere from 5 minutes to an hour or more. These symptoms make me feel like the issue has something to do with power or temps, but voltages and temps seem to be fine in aida64 under load. My motherboard is relatively new, and this issue was happening before it was replaced. I'm not sure what could be causing this. My guess is the PSU, but I don't have any others to test with and want to try and diagnose the issue without buying anything. All suggestions are appreciated, thanks

OS: Windows 10 64-bit
Specs:

• GTX 970
• Intel i7-4770
• 2x 8gb DDR3
• MSI z97 gaming 5
• EVGA 550 G2 (psu calculators suggest ~400w, so this should be sufficient)

Is your hardware overclocked? What temperatures for your GPU and CPU are shown in RealTemp?
https://www.techpowerup.com/download/techpowerup-real-temp/

[StaticShadow]

6 minutes ago, STRMfrmXMN said:

If a PSU fails the system usually BSODs.

 

Is your hardware overclocked? What temperatures for your GPU and CPU are shown in RealTemp?
https://www.techpowerup.com/download/techpowerup-real-temp/

Nothing is overclocked. Used to get a bit lower temps before the rebuild, but the issue still happened then. Attached are both no load and after 3 minutes on aida64 (would do longer but would rather not crash)

[STRMfrmXMN]

2 minutes ago, StaticShadow said:

Nothing is overclocked. Used to get a bit lower temps before the rebuild, but the issue still happened then. Attached are both no load and after 3 minutes on aida64 (would do longer but would rather not crash)

 

 

@Electronics Wizardy - Can you help him? I'm curious especially about this detail in the OP.

 

50 minutes ago, StaticShadow said:

 It seems if I leave the computer unplugged for a day or two I can go for around 6 hours or more without crashing, however if I continue using the computer without doing this the crashes become more and more frequent. Running benchmarks like furmark and aida64 can have a wide range before the system crashes, anywhere from 5 minutes to an hour or more.

 

[Electronics Wizardy]

Just now, STRMfrmXMN said:

If a PSU fails the system usually BSODs.

depends on how it fails, if it cuts power(like if there is too high of a power draw, or the input voltage is out of spec, out over temp) it will just power off like you unplugged it.

 

This really feels like a psu. Do you have a friend with a pile of parts you can test? Id try removing your gpu and seeing if it fixes it, the lower power draw may cause it to keep working.

[StaticShadow]

2 minutes ago, Electronics Wizardy said:

depends on how it fails, if it cuts power(like if there is too high of a power draw, or the input voltage is out of spec, out over temp) it will just power off like you unplugged it.

 

This really feels like a psu. Do you have a friend with a pile of parts you can test? Id try removing your gpu and seeing if it fixes it, the lower power draw may cause it to keep working.

Don't think I have any I could borrow from, the few friends I have that have built PCs are all on first iterations and wouldn't have extra parts. I could remove the GPU however not sure how much of a difference that'd be from just not running any games (It doesn't crash when just on the computer normally, aka not playing games or benchmarking). It did crash when I benchmarked excluding the GPU, however that was after gaming so the temps could've already been high on the psu. I can try running aida64 excluding gpu in an hour or so if you think that could help.

[STRMfrmXMN]

6 minutes ago, Electronics Wizardy said:

depends on how it fails, if it cuts power(like if there is too high of a power draw, or the input voltage is out of spec, out over temp) it will just power off like you unplugged it.

 

This really feels like a psu. Do you have a friend with a pile of parts you can test? Id try removing your gpu and seeing if it fixes it, the lower power draw may cause it to keep working.

From my experience with dying (old) PSUs, they shut down with a BSOD usually as soon as the WIndows desktop is seen, sometimes sooner.

 

I should have suggested to take out the dGPU, dangit.

[Electronics Wizardy]

Just now, StaticShadow said:

Don't think I have any I could borrow from, the few friends I have that have built PCs are all on first iterations and wouldn't have extra parts. I could remove the GPU however not sure how much of a difference that'd be from just not running any games (It doesn't crash when just on the computer normally, aka not playing games or benchmarking). It did crash when I benchmarked excluding the GPU, however that was after gaming so the temps could've already been high on the psu. I can try running aida64 excluding gpu in an hour or so if you think that could help.

 Try the normal troubleshooting, removing the dgpu, removing all ramsticks but one, reming all hdds but boot drive, umplug all uneded devices(usb, monitors)

 

Crashes like this are a pain to find what causes them if you don't have a pile of hardware.

[StaticShadow]

With all components in still, just ran a 4 hour benchmark without a crash. 2 hours without GPU, then added in GPU for the last two hours. I don't know how it crashes almost every 5 minutes sometimes but then lasts like this for hours...

I feel like (however haven't logged times to confirm this) benchmarks only crash the computer if run within a period of time of playing games that have crashed the computer. If I run benchmarks when I haven't played anything for a while, they seem to last. In my mind playing games shouldn't apply any more stress than a benchmark, so not sure what the cause could be. It would seem like software due to the variability of it and alignment with games, however I feel there would be BSOD if so.

 

[StaticShadow]

Okay, went ahead and started logging system stats via aida64, then played PUBG until the computer crashed. Here is the spreadsheet containing everything I logged. The game crashed after I died and hit exit to the lobby. It started going from the game into the lobby then crashed, which is why you can see a lot of the stats start to free up (ram and gpu usage, etc) right before the crash. (Apologize for the temps in Fahrenheit, figuring out how to fix them) Nothing in there really catches my eye, however someone else might see something:

https://docs.google.com/spreadsheets/d/1UFcqSuEjuazEJIyr6h-D592Q2YwCmuMzhoBVcJuceVY/edit?usp=sharing

[StaticShadow]

Even more data collected, this time logging at a faster rate and crashing mid game instead of going into lobby to make it easier to spot spikes, however I don't see anything. Seeming like it's either the PSU overheating or software.

https://docs.google.com/spreadsheets/d/1CH4Lzcvc2tDY72DP6Fm-7BVNu_pRdbzjh0elYwXqGQs/edit?usp=sharing

[StaticShadow]

Pretty stuck, can't think of any other tests to run. I've run the above stat tests, tried running with both my ram dimms individually, put a fan under my PSU, all with no luck. Any suggestions on other things I could test?