PC seemingly cuts power after 15 minutes

[TLK1340]

Fair warning, posting from mobile. 

 

Specs:

Processor: AMD 3 Ryzen 2200g (1.4v stock) 

Graphics card: AMD RX 570 6gb 

RAM: Gskill TridentZ RGB 16gb (2 stick kit) 2133MHz

Motherboard: ROG Strix B350-f

PSU: Cooler Master MWE Gold 750w (Fully modular)

HDD: Seagate Barracuda 1tb (Drive which Windows is installed on)

SSD: Kingston SA400S7240G (240GB)

BIOS version: 5204 x64 build date 7/27/2019

OS: Windows 10 Pro x64 (Activated) 

 

Problem:

For the last few days I've been an issue with my PC shutting completely off entirely after about 15-45 minutes of usage (after boot) and unable to be turned back on for 30 minutes (sometimes sooner sometimes later) strictly after resetting the PSU. 

 

Notes about the problem:

Power is still driven to the board after the hard shut off (LED's are still on), and an audible clicking sound as if you held the power button down is heard when the shut off happens. If the PSU is reset and the power button is attempted to be used to turn on the PC before waiting, the PC clicks on for less than a second, all fans spin on including the PSU fan, the LED's kick on for the same duration, and the PC immediately powers off again and will not start until the PSU is reset again. Though the LED's on the board do not go out at all until the switch is flipped on the PSU. 

 

Troubleshooting methods attempted:

Under clocking CPU to 1.34v (No affect on problem). 

Reinstalling/updating GPU drivers. Tried 3 different versions (No affect on problem). 

Tried a different Graphics Card entirely (No affect on problem). 

Checked thermals (High 44C low 36C under slight load or idle) and replaced thermal compound (No affect on problem). 

Cranked fan speed up to max on every fan in the case including the Graphics Card fan up to 75% (No affect on problem) 

Changed placement/attempted to remove 1 stick of RAM with every RAM slot tested with both sticks separated and in at the same time (No affect on problem). 

Checked if the PSU fan was still spinning while running (It does). 

Attempted booting windows from a different HDD (No affect on problem). 

Freshly cleaned Windows from the initial HDD and reinstalled it completely (No affect on problem). 

Updated the BIOS from stock build version 4024 to version 5220 (No affect), rolled BIOS version back to 5204 (No affect on problem). 

Updated Chipset drivers to version 1.7.29.0115 (No affect on problem). 

Changed power settings (None had an affect on the problem including changing sleep mode timing to never). 

 

I'm stumped at this point. Though I did take a mental note on the time it took for windows to clean the initial HDD and reinstall (9 hours), and during that time the PC didn't shut off. But after booting up into windows shut off after 15 minutes of idle/downloading Google Chrome/downloading drivers for the GPU. The PC was built in March, all new parts. If no other possible solutions are out there, then I'm just going to purchase a new CPU/Motherboard/PSU I wouldn't say the issue lies with the other parts but I may be wrong. (I'm hoping I am). If I missed anything spec wise please let me know. I've tried scouring the web for a solution and couldn't find one I haven't tried yet. 

[Juular]

Definitely looks like smth related either to power delivery or thermals. Since thermals on the CPU are fine (assuming your reading are right) the only suspects are either motherboard VRM or PSU, since PSU still powers 5 & 3.3v rail i'd assume it's fine so my guess is the motherboard VRM. Does it shut down any faster if you run CPU stress test ?

[Juular]

Yeah, looks like this motherboard has pretty bad VRM design, clocking 120° C on OC'd R7 2700. Do you have VRM thermal sensor readable by HWiNFO ?

[TLK1340]

2 minutes ago, Juular said:

Definitely looks like smth related either to power delivery or thermals. Since thermals on the CPU are fine (assuming your reading are right) the only suspects are either motherboard VRM or PSU, since PSU still powers 5 & 3.3v rail i'd assume it's fine so my guess is the motherboard VRM. Does it shut down any faster if you run CPU stress test ?

I actually don't have any proper stress tests on it since it happens idling, however I have ran stream tests. 720p 60fps 4k bitrate runs for about 5-15 minutes on Ori and the Blind Forest (Not a demanding game at all). Usually about the same times as idle. But installing windows though took 9 hours as I said and it didn't shut off then. I have ran temps while streaming and barely peaked 45C. For GPU and CPU. 

[TLK1340]

3 minutes ago, Juular said:

Yeah, looks like this motherboard has pretty bad VRM design, clocking 120° C on OC'd R7 2700. Do you have VRM thermal sensor readable by HWiNFO ?

Would that usually be found in the bios? Haven't heard of that honestly. 

[Juular]

It's unclear by the pictures i can find of this board if VRM heatinks are screwed or push-pin. If they're push-pin they can get loose overtime and wouldn'e be capable of dissipating the heat efficiently, check them.

[Juular]

1 minute ago, TLK1340 said:

Would that usually be found in the bios? Haven't heard of that honestly. 

Yes, sometimes in BIOS too, look for VRM or MOS temperature. And try HWiNFO monitoring tool, it can read a lot of stuff.

[Juular]

As of CPU stress test, you can just download CPU-Z, it has simple benchmark & stress test feature, or you can use more sophisticated OCCT all-around stress test tool (newer 5+ version is a bit finicky tho, i use older v4.5).

[TLK1340]

5 minutes ago, Juular said:

Yes, sometimes in BIOS too, look for VRM or MOS temperature. And try HWiNFO monitoring tool, it can read a lot of stuff.

It appears that there is no sensor for it as I can't find one that says VRM or MOS in the sensor status of HWiNFO

[TLK1340]

The heatsink are screwed in through the back I'll tighten them up real quick 

[TLK1340]

I ran a stress and bench for the cpu, the first sensor read 58C high, and the gpu came back with a 59C high (for the sensor names Diode). Averages were the same at 49C.

[TLK1340]

23 minutes ago, Juular said:

It's unclear by the pictures i can find of this board if VRM heatinks are screwed or push-pin. If they're push-pin they can get loose overtime and wouldn'e be capable of dissipating the heat efficiently, check them.

 

5 minutes ago, TLK1340 said:

The heatsink are screwed in through the back I'll tighten them up real quick 

They didn't seem to be loosened at all however I tightened them as hard as I could. 

[TLK1340]

32 minutes ago, Juular said:

It's unclear by the pictures i can find of this board if VRM heatinks are screwed or push-pin. If they're push-pin they can get loose overtime and wouldn'e be capable of dissipating the heat efficiently, check them.

After tightening and booting up the pc again this happened.. The left monitor receives input via DVI. 

[TLK1340]

39 minutes ago, Juular said:

As of CPU stress test, you can just download CPU-Z, it has simple benchmark & stress test feature, or you can use more sophisticated OCCT all-around stress test tool (newer 5+ version is a bit finicky tho, i use older v4.5).

Here's some thermals after stressing (alt + prnt screen didn't cap one monitor).

[TLK1340]

After a little deeper inspection, the shut off only really happens while downloading something or playing ONLY Ori and the Blind Forest? I honestly haven't an explanation for it. 

[Juular]

7 hours ago, TLK1340 said:

After a little deeper inspection, the shut off only really happens while downloading something or playing ONLY Ori and the Blind Forest? I honestly haven't an explanation for it. 

Now that's real strange.

[Juular]

You mean, you can run CPU-Z stress test for a good 30 minutes without shutdown but if you're starting Ori or download stuff it shuts down ?

I can speculate that with the Ori it's probably the uncapped framerate that puts strain on GPU VRM & PSU so maybe it's in fact GPU or PSU indeed. But what could be possibly wrong with downloading i have no idea. At worst there may be BSOD if it's faulty network adapter or driver but shutdown with startup delay ?

[TLK1340]

11 hours ago, Juular said:

You mean, you can run CPU-Z stress test for a good 30 minutes without shutdown but if you're starting Ori or download stuff it shuts down ?

I can speculate that with the Ori it's probably the uncapped framerate that puts strain on GPU VRM & PSU so maybe it's in fact GPU or PSU indeed. But what could be possibly wrong with downloading i have no idea. At worst there may be BSOD if it's faulty network adapter or driver but shutdown with startup delay ?

I did swap Graphics cards to see if the problem was consistent without it and it was still shutting off. There's some games that don't crash it but none that put any stress on it. Ori is capped at 60fps unless Vsync is on, I don't play with it on with my 144hz panel either way. I have reasonable suspicion of the PCIE slots I use for both graphics cards at this point, so I'm thinking it's not a Graphics Card thing but a board slot issue, I'll be trying another slot lane in a little while. Another point is that the 570 is kind of heavy. 

[TLK1340]

So I learned that I can't use the 2nd slot for using my card, however since the 2200g comes with Vega 8, I could try to run the game without a GPU. Having it push 144hz and 60fps it was able to run for over 50 minutes so I believe it is safe to say that the issue isn't power, nor is it the CPU. Since the problem is consistent with different video cards, it makes me believe one of two things is happening, either since the RX 570 6gb is kind of heavy, it may have damaged the slot. Or the slot isn't set up correctly somehow. My guess is that it might be a good idea to try it on raid rather than 8x.

 

Attempting to change it to raid rn but I'm not sure it will actually help. Otherwise new board it is if I'm not missing anything. 

[TLK1340]

Alright my conclusion since with it ran raid which seems to be the only thing I can change about how anything gets read via the port still caused a hard shut off, it's the port of the motherboard for sure and more likely due to physical damage because of the weight of the gpu.

 

So new motherboard it is. Will post on this thread though if replacing the board doesn't help.