I have been experiencing this issue ever since I built my machine about 3 years ago but I never really gave it much attention since it used to occur once every other month back then. It has gotten considerably more annoying today at the rate of at least once or twice per day.

What happens is as I am working my machine appears as if it crashes but it continues to run. Screen goes black or freezes in a random color, sometimes stripes of alternating colors. If there was audio streaming in the same time it freezes as well in a strong buzzing sound as if it repeats the last few milliseconds of whatever audio was playing. The machine does not stop or restart, fans continue to spin until power button is held at which point it turns off and does not want to start again. In attempt to do so the fans begin to spin, as well as the HDD, however nothing is displayed and there is no beeping sound.

This has been described in a few other threads I read and from what I understand this is most likely a MB or PSU issue however what is different in my case is that if left powered off for roughly 30 to 60 minutes (in rare cases 120 - 180 minutes) then it "fixes" and boots normally. If powered on in the "broken" state and left for hours nothing happens and an immediate attempt to re-start (by turning off and on) fails until it's left without power for the intervals mentioned above.

What I've tried:

  1. Replaced the CMOS battery.
  2. Removed the graphics card, just in case.
  3. Removed all memory sticks which resulted in no beeps, however I waited for it to "fix" then repeated this test and there was vicious beeping as there should be.
  4. Returned the PSU about a year and a half ago since it was in warranty in hopes the problem will go away even though it wasn't that severe back then. The retailer just gave me a new one and didn't say if there was anything wrong with the old one.

I highly suspect the PSU but I don't have a spare one at hand to test and I'm afraid to buy a used one for tests as it could be just as bad.

The fact that it needs to stay powered off for a while before it can boot again makes me think that there are some capacitors in the root of the problem.

My components are:

  • Intel i5 4460
  • MSI Radeon R9 270x 2GB GDDR5
  • 2x4GB Corsair Vengeance 1600MHz DDR3
  • MB: Gigabyte B85M-D3H
  • PSU: Corsair VS550

I'm very interested to find out what is causing this and this behavior of "self-fixing" is making me even more curious. Ultimately I would have to take it to a repair shop but there aren't any in my area who I'd trust with a motherboard issue.

EDIT: I have monitored temperatures very closely to determine if that's what is causing the issue and I see that the processing units rarely exceed 50C.


I don't think anyone here can give you a guaranteed answer, however it would seem very likely that the PSU is the issue.

Your power supply is very close to the minumum required at 550 watts (and the Radion requires 500W minimum, with the ability to put24A on the 12 volt rail).

Whats probably happening is your system has been running the power supply components close to their maximum, and they (particularly the caps) are starting to wear out, and fail more frequently.

Power supplies typically have circuitry which allows them to "heal themselves" over several hours, which again talks to the problem you are describing.

| improve this answer | |
  • I see what you have said makes lots of sense. But if that was the case wouldn't it make it fail only when there is heavy load on the 12v rail? I observe it crashing more frequently when the GPU is under load but it also happens when the machine is on idle or browsing the web. – php_nub_qq Jul 2 '18 at 0:17
  • I also remember I had worries myself when buying the PSU so I ran the whole set up through a watt meter while running both a CPU and GPU stress test and the indication on the watt meter was something in the lines of 450W and if we take into consideration that this PSU efficiency is quite low I would assume that the actual power consumed by the PUs is around 350W which is far below the PSU's capabilities. I might as well do this again just for the sake of it and will report back if results differ. – php_nub_qq Jul 2 '18 at 0:23
  • Turns out I can't even run the benchmark anymore, when I start a GPU test it crashes. Second attempt: the test begins, on low loads is okay on higher load it breaks. This is interesting. – php_nub_qq Jul 2 '18 at 1:14
  • 1
    You need to be particularly mindful of average vs peak capacity. I would expect that a watt meter would show you an averaged draw (even if only over a half a second or so), while within that time you could get very substantial spikes - the kind thatt a capacitor might filter out - if the capacitor is not damaged. – davidgo Jul 2 '18 at 1:25
  • 1
    This would not be a conclusive test, but maybe try pulling out 1 of your memory sticks and running your test (even better if you can also run it from USB and disconnect your hard drive for the test). This would reduce your load by a few watts, and may help you towards a conclusion. – davidgo Jul 2 '18 at 1:46

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.