Question PC randomly shuts off

[lazo1999]

Hello,
I have a very weird problem with my pc, which i have been trying to solve for months now with no result. It shuts off abruptly when gaming or if I run some stress tests. Its like some sort of protection kicking in, but mobo LEDs still light up, not the fan led tho.
Some observations;

• It has to do with temperature, BUT IT IS NOT NEAR THERMAL LIMITS
• It has not a pattern that i can think of (it can happen 5 minutes into the game, or 30, no correlation)
• When shutting off, temps are about 50 for cpu, 55 for gpu, 65 for gpu hotspot, 41 for mobo, 38 for ram, 55 for m.2
• In the past I had it as an open testbench and had no issue.
• If i leave the side panel off, it happens less frequently
• If i stress components individually, they can achieve higher temps, no problem
• Stressing gpu, panel on, still happens
• It never shows any sign of thermal throttling, just emergency shut off
• Tested gpu on other pc, works fine
• Tested pc with other PSU, still shuts down
• Tested with fans on max speed, still happens
• Upgraded and downgraded and upgraded again ALL bios versions that I can
• Formatted Windows
• Tried to have a fan blownig on certain components like mobo, ram m.2 etc to controll temps, while others climb, but still happens
• Used to have a 1080ti, no problems (so its not a power issue)

At this point I dont know what else to try, I need ideas and recommendations
Specs;

• AMD ryzen 3600, with 280mm arctic water cooler
• 16GB Corsair Vengeance RGB Pro DDR4
• Asrock B450M Steel Legend
• MSI MECH X2 RX 6650XT
• Samsung 960 EVO
• 2x WD 500GB HDD
• Corsair CX550F RGB
• 2x 120mm fans (tested with and w/o, they do not cause the problem)
• The case has RGB (tested with and w/o, they do not cause the problem)

I need help, I cant even play one game. Ask any information you need.

 

[Ralston18]

Look in Reliability History/Monitor and Event Viewer for any error codes, warnings, or informational events being captured just before or at the time the PC shuts off.

Reliability History/Monitor is end user friendly and presents a timeline that can reveal patterns.

Event Viewer is more difficult to navigate and understand. To help:

How To - How to use Windows 10 Event Viewer | Tom's Hardware Forum (tomshardware.com)

In both tools any given entry can be clicked for more details. The details may or may not be helpful.

Note reoccurring Event ID's.

Also check Update History for any failed or problem updates.

If you find anything of concern via either tool, then take a full screenshot so all can be seen.

Post the screenshots here via imgur (www.imgur.com).

 

[rickypicky5]

What program are you using to monitor temperatures? I believe HWiNFO (I'll check this evening) allows you to monitor MOFSET/VRM temperature, assuming your motherboard has a sensor for it.

 

[lazo1999]

Look in Reliability History/Monitor and Event Viewer for any error codes, warnings, or informational events being captured just before or at the time the PC shuts off.

Reliability History/Monitor is end user friendly and presents a timeline that can reveal patterns.

Event Viewer is more difficult to navigate and understand. To help:

How To - How to use Windows 10 Event Viewer | Tom's Hardware Forum (tomshardware.com)

In both tools any given entry can be clicked for more details. The details may or may not be helpful.

Note reoccurring Event ID's.

Also check Update History for any failed or problem updates.

If you find anything of concern via either tool, then take a full screenshot so all can be seen.

Post the screenshots here via imgur (www.imgur.com).

Event viewer gives kernel power 41 error. It could be anything triggering emergency shut down. Updates are all fine, also i did a full format.

 

[lazo1999]

What program are you using to monitor temperatures? I believe HWiNFO (I'll check this evening) allows you to monitor MOFSET/VRM temperature, assuming your motherboard has a sensor for it.

I have used OCCT, HWINFO, and i use msi afterburner for checking in game, but there are just 2-3 mobo sensors, that there is no way of knowing where are they located. No sensor reaches dangerous temps. Also i have tried shoving a fan directly over mobo VRMs but no luck. Also i have arctic liquid freezer with a mini fan blowing on the vrms, so I doubt its that.

 

[Ralston18]

PSU = Corsair CX550F RGB - correct?

Noted that you tested with another PSU. What was the make, model, and wattage, and age?

Did you use only the cables that came with the test PSU?

CPU: thermal paste. What was used and how applied?

 

[lazo1999]

PSU = Corsair CX550F RGB - correct?

Noted that you tested with another PSU. What was the make, model, and wattage, and age?

Did you use only the cables that came with the test PSU?

CPU: thermal paste. What was used and how applied?

Psu yes is cx550f, i tried with a cx450 about 5 years old, yes i used the original cables, paste is mx4 and 5 i tried both, temps are not high though, why do you suspect that?

 

[Ralston18]

The temps can spike very high, very suddenly and not even be captured or measured before the thermal shutdown is invoked.

The analogy I have used is a lightning bolt. Very, very hot but will not raise the air temperature per se. Caveat being that you are not close enough to feel the heat and, if so, the issue may be moot. ⚡

Research (Google) using words and phrasing such as:

"symptoms of a computer CPU thermal shutdown"

Limit search results to the last year or so.

Then, as you read and learn, simply modify the search c riteria to further filter and reduce the results as applicable based on your system and observations.

 

[lazo1999]

So you think its cpu related thermal? On a 95W cpu with 280mm cooling you think its possible? If so what would be a solution? LN2?? Jokes aside how can i test further? Small fft on stress tests reaches 85 degrees. I also never experienced throttling.

 

[lazo1999]

The temps can spike very high, very suddenly and not even be captured or measured before the thermal shutdown is invoked.

The analogy I have used is a lightning bolt. Very, very hot but will not raise the air temperature per se. Caveat being that you are not close enough to feel the heat and, if so, the issue may be moot. ⚡

Research (Google) using words and phrasing such as:

"symptoms of a computer CPU thermal shutdown"

Limit search results to the last year or so.

Then, as you read and learn, simply modify the search c riteria to further filter and reduce the results as applicable based on your system and observations.

Update:
I tried stressing just the gpu, with the side panel closed, so it heats up more, and the issue persists. Vrms and cpu should be cool (also keep in mind that I have placed the cpu cooler in the front as intake for the case).
So i dont think its the problem.

 

[Ralston18]

Just two immediate questions:

1) Stress test: Did the test create any conditions that would or could actually cause a shutdown?

As a reference (you can find other similar links):

https://techlibrary.hpe.com/docs/is...tilities screen,degrees of the critical level.

2) "Vrms and cpu should be cool" - what are the actual temperatures? Cooler placement is important but that is secondary to the actual effectiveness of the cooler and resultant airflows The various fans could be countering each other and overall air flows reduced.

I suggest taking a couple of photographs showing case, cooler, fans etc.. Indicate the air flow directions. Post the photographs here via imgur (www.imgur.com). Someone may spot some actual problem or potential issue.

Objective simply being to rule out possible culprits - one of which could a thermal shutdown.

Very sure that you do not want to run any stress tests that could actually break things. Unless that is the true intent for the running the test.

Will defer to the "cooling" community folks as to what tests are truly meaningful and useful for troubleshooting purposes.

 

[lazo1999]

Just two immediate questions:

1) Stress test: Did the test create any conditions that would or could actually cause a shutdown?

As a reference (you can find other similar links):

https://techlibrary.hpe.com/docs/iss/proliant-gen10-uefi/s_set_thermal_shutdown.html#:~:text=From the System Utilities screen,degrees of the critical level.

2) "Vrms and cpu should be cool" - what are the actual temperatures? Cooler placement is important but that is secondary to the actual effectiveness of the cooler and resultant airflows The various fans could be countering each other and overall air flows reduced.

I suggest taking a couple of photographs showing case, cooler, fans etc.. Indicate the air flow directions. Post the photographs here via imgur (www.imgur.com). Someone may spot some actual problem or potential issue.

Objective simply being to rule out possible culprits - one of which could a thermal shutdown.

Very sure that you do not want to run any stress tests that could actually break things. Unless that is the true intent for the running the test.

Will defer to the "cooling" community folks as to what tests are truly meaningful and useful for troubleshooting purposes.

1) I cant really find anything out of the ordinary. Max temp for cpu is 85 degrees and other components even less. Everything should work ok.
2) I cant be sure about vrms since there is not a dedicated sensor on them, but they are not really hot to the touch, they have proper airflow due to a small fan on the pump of the water cooler, and I even tried to install a fan just blowing air on them with no change.

My fan placement is ok, I have been working as a technician for years, I would have spotted it if I had placed a fan in the other direction by mistake.
I have the cooler in the front as intake, and two fans as exhaust; one at the back, one at the top. Its not the lack of airflow, nor the trapping of heat or smth in the case.

 

[Ralston18]

Seems to be narrowing things down somewhat.

About all I can immediately think of at this moment is to continue to monitor temperatures, what is running at any given time, and keep checking error codes, etc..

Here is a link that presents various options:

https://www.avast.com/c-how-to-check-cpu-temperature

There are other similar links and someone else may post additional ideas and suggestions.