[SOLVED] Desktop PC shutting off during use—help needed with precise diagnosis

[Nicknimh]

CPU: i5-6500
CPU cooler: Thermaltake Gravity I2
GPU: RX 480, stock cooler
MB: MSI B450M Mortar
PSU: Rosewill 550w bronze, semi-modular; I'm not exactly sure of the precise model, as I don't have a receipt and there's no further designation on the currently visible surfaces
RAM: 16 GB DDR4
2 SSDs, 1 HDD, and 1 BD-RW drive
2 case fans, one in front and one in back
OS: dual boot Ubuntu and Windows 10
PC was built by me summer 2016; about two months ago I thoroughly cleaned it and got a new CPU cooler when the stock one wouldn't go back on; I properly cleaned off the old thermal paste with alcohol and used the pre-applied paste on the new cooler. Shortly after, I added some new ram (identical to the first pair) and an SSD I salvaged from a broken laptop. It isn't overclocked in any way.

Over the past year or so, I had occasional incidents where the PC would not boot up, which could be fixed by re-plugging the power cord or switching the PSU back switch on and off a couple times. This was a bit concerning, but not all that pressing. I don't recall it happening since I cleaned up a couple months ago, so I chalked it up to some loose cord or dust or something. Three nights ago, while playing Fallout 4, my computer suddenly powered down after twenty or thirty minutes of gameplay. No BSOD or anything, just completely shut off as if the power had gone out. The monitor, which is plugged into the same surge protector, was still on. It wouldn't turn on again initially, so I did the power cord and PSU switch routine, and it did, but the fan (unsure which one) was louder than I'd ever heard, so I turned it back off while on grub for fear of damaging something.

Tried again an hour or two later, and again the machine shut off after less than an hour of gameplay. The next night, I replaced the power cord and surge protector and tried again. I played for an hour or so, keeping an eye on the thermals and such, and everything seemed normal. As far as I could tell, the CPU never went above the 40s celcius, and the CPU fan never went above its lowest speed (it never does, even when I was playing Cyberpunk). So I figured problem solved, and stopped monitoring the metrics. But after I'd been playing three or four hours, the problem happened again. Today, I used it again a couple times with no issue, then today I tried again and it shut off after just a few seconds, but strangely, I noticed that the light on my usb bluetooth dongle was still on. Does this suggest a MB problem rather than a PSU problem?

So, what exactly is my problem here? PSU seems the most obvious, but how do I know for sure? I would hate to spend $70 and spend hours setting up a new one only for it not to fix the problem. A thermal issue (at least outside of the PSU itself) seems unlikely, but I'm no expert. Those who are knowledgeable about this, what do you think the problem is?

 

[kerberos_20]

if psu turns off and u cant power it on, unless u unplug it from wall outlet or switchin button on psu..then some power protection is kickin in
u may want get some more reliable PSU first, tier B or better

 

[Nicknimh]

if psu turns off and u cant power it on, unless u unplug it from wall outlet or switchin button on psu..then some power protection is kickin in
u may want get some more reliable PSU first, tier B or better

So, is it your opinion that this is in fact a power supply problem which would be resolved by getting a new PSU?

 

[DSzymborski]

So, is it your opinion that this is in fact a power supply problem which would be resolved by getting a new PSU?

It may be, it may not be. You really ought to look at the PSU to see what we're dealing with here. There's a lot of variability in the quality of Rosewill PSUs.

The light does not necessarily suggest a motherboard problem. There are different rails in a PSU and USB is basically interest in +5V power, while your CPU and motherboard require +12V power. PSUs can fail in many different ways beyond just not turning on.

 

[Phillip Corcoran]

It may be, it may not be. You really ought to look at the PSU to see what we're dealing with here. There's a lot of variability in the quality of Rosewill PSUs.

The light does not necessarily suggest a motherboard problem. There are different rails in a PSU and USB is basically interest in +5V power, while your CPU and motherboard require +12V power. PSUs can fail in many different ways beyond just not turning on.

Very true, that's why it's nigh impossible to give a straight "Yes" when someone asks "Are you really certain a new PSU will fix the problem?"

 

[Nicknimh]

It may be, it may not be. You really ought to look at the PSU to see what we're dealing with here. There's a lot of variability in the quality of Rosewill PSUs.

Alright, I looked through my order history and found the exact model. It's a HIVE series 550w model.

 

[DSzymborski]

Alright, I looked through my order history and found the exact model. It's a HIVE series 550w model.

Not their worst -- in the really old days, they had some pretty awful ones -- but far from their best model. Given that it's an old, group-regulated model and past its warranty, I would be inclined to replace it anyway, so it's not a bad first thing to address.

 

[Nicknimh]

Not their worst -- in the really old days, they had some pretty awful ones -- but far from their best model. Given that it's an old, group-regulated model and past its warranty, I would be inclined to replace it anyway, so it's not a bad first thing to address.

Alright, I suppose I should go ahead and order a replacement (is EVGA's BQ series a good choice, or should I shell out an extra $20 for a seasonic focus?). In the event that that doesn't work, what would be other likely causes of the problem?

 

[DSzymborski]

Alright, I suppose I should go ahead and order a replacement (is EVGA's BQ series a good choice, or should I shell out an extra $20 for a seasonic focus?). In the event that that doesn't work, what would be other likely causes of the problem?

Shell out the extra $20 for sure. Get something you can use worry-free for a decade.

Sudden power-offs are usually temperature or power delivery and you've already eliminated temperature. Next I'd look at trying it out with the GPU removed and running on integrated graphics. Failing that, then a single RAM stick (preferably one of the old ones). Then I'd probably suspect motherboard after that.

Hopefully the PSU replacement will resolve the issue. Since it's the part you really ought to replace anyway, it would be highly convenient if it turns out to be the direct issue! I'll cross my fingers.

 

[Nicknimh]

A little late, but thanks to all who replied to this thread. I replaced the PSU, and everything has been working fine since then.