No idea how to fix my PC -- black screen and 100% fans

[repoman0]

First, specs ...

Asrock X370 Taichi
Ryzen 3700X
2x16GB Crucial 3200 MHz RAM
eVGA 1070
Seasonic G-series 550W (bought new in 2015)
Latest Windows 10, Ryzen drivers, nVidia drivers
... all running with no overclock of any kind

The problem is that it crashes with a black screen and the CPU fan spins up while playing Overwatch, and I can only shut down by switching off the PSU (power button and reset unresponsive). I can't get anything to fail stability tests -- Prime95, Linpack, Memtest86 and Memtest for Windows, Furmark. I even ran three instances of 3GB memtest for Windows, Linpack, and Furmark all at the same time overnight for 8 hours with no errors. After a hard crash, I have a really hard time booting back up. I can sit there and try 20 times in a row now, switching off the PSU in between each try, and it won't boot until I leave it alone for 30 minutes, after which everything is fine until I try to play video games again.

Worth mentioning that when I do boot back up, BIOS settings are all at their old settings, which leads me to believe the board thinks everything is fine. When I was trying to run my B-die with first-gen Ryzen and it crashed due to memory, it would fail to boot a few times and then go into "safety mode" by putting RAM back to 2400 MHz.

Things I have tried:

- New motherboard, twice (ASUS and ASrock X570 boards, returned both)
- New GPU, eVGA 2060 Super (returned)
- This has happened with the CPU and RAM that I upgraded from ... 2700X and 16 GB B-die. It used to be significantly less frequent but I upgraded in part because I thought it would make my problems go away. This one might be a different issue -- I had a lot of resets due to memory for a while, but it would reboot fine.
- Heatsink on my NVMe drive -- I reproduced the issue with Linpack+memtest+furmark+CrystalDiskMark running at the same time, leading me to believe it was high disk temps ("temp 2" = 100-110C max). It now hits 50C max so $13 well spent but no dice re: black screen crashes.
- Measured PSU voltages at multiple pins with my Fluke DMM -- rock solid 3.38, 5.09 and 12.20V
- Finally reinstalled Windows even though I had my doubts. Sure enough made no difference

The boot routine of the ASRock X570 board was different enough that when it would black screen, the PC would try to boot a bunch of times rather than just get locked. My X370 boot LEDs do not light up when I'm trying and failing to boot, meaning it doesn't even get to the point where it starts the boot routine. Whatever the problem is, it definitely seems to be getting worse. I used to be able to play for an hour or two, now I can't even make it three minutes.

Any ideas are appreciated. I live close to a Microcenter but I'm getting tired of going there and returning things that don't fix my problems. Soon I will just replace every part and throw the rest out the window.

 

[UsandThem]

Since you had crashes with the previous CPU / build, and you seem to have replaced everything besides the RAM, what about the PSU? A failing PSU can cause weird crashes / instability like you are experiencing.

 

[repoman0]

Since you had crashes with the previous CPU / build, and you seem to have replaced everything besides the RAM, what about the PSU? A failing PSU can cause weird crashes / instability like you are experiencing.

I think that's next -- just was thrown off that voltages are rock solid to within 1/100V. The Fluke obviously doesn't have the resolution of an oscilloscope to catch dips but it does indicate that the PSU is good, and it's a quality brand. Will try it anyway because I can't think of anything else.

For clarity, I did replace the RAM at the same time as the CPU

 

[VirtualLarry]

Yeah, in this order, I would try: 1) A new PSU, with a higher wattage, and 2) RMA the CPU.

Maybe number 0, try setting "Idle Control" to "Normal" or "Typical" or the opposite setting. Play around with it. Also consider disabling "Global C-States".

I had similar black-screen crashes, but I was overclocking my R5 1600 at the time.

 

[UsandThem]

I think that's next -- just was thrown off that voltages are rock solid to within 1/100V. The Fluke obviously doesn't have the resolution of an oscilloscope to catch dips but it does indicate that the PSU is good, and it's a quality brand. Will try it anyway because I can't think of anything else.

For clarity, I did replace the RAM at the same time as the CPU

Just your description of your issues had me thinking it was the PSU before even reading half of your post. While the Seasonic G series is a good entry-level unit, even some of the top-of-the line titanium units can fail early, or start having issues. If it was my system, the PSU would be what I would be focused on at the moment.

 

[repoman0]

Another trip to Microcenter later today then ... will update the thread. They're going to ban me if I keep returning things.

 

[repoman0]

30 minutes in and so far so good with the new PSU. Figured I should probably jinx it now before I get my hopes up too high ...

 

[UsandThem]

30 minutes in and so far so good with the new PSU. Figured I should probably jinx it now before I get my hopes up too high ...

What model did you end up buying?

 

[repoman0]

What model did you end up buying?

Seasonic Focus 650W. Two hours in so far and all is well. So strange, never would have expected that this was a PSU issue given the DMM measurements ...

edit: it's still under warranty and discontinued. Maybe I'll end up with something decent I can sell and end up with a near $0 problem.

 

[UsandThem]

Seasonic Focus 650W. Two hours in so far and all is well. So strange, never would have expected that this was a PSU issue given the DMM measurements ...

Yeah, PSU problems are tricky to figure out sometimes, since the symptoms (and crashes) can vary. But when I saw you say this:

I can sit there and try 20 times in a row now, switching off the PSU in between each try, and it won't boot until I leave it alone for 30 minutes, after which everything is fine until I try to play video games again.

I was pretty sure it was your PSU, as this is one of the most common results when a PSU still works for the most part, but something is beginning to fail inside the PSU.....be it a capacitor, switch, controller, etc.

 

[repoman0]

Peeking inside the old PSU last night, there were exploded capacitors everywhere. At least five and I didn't look that carefully. Must have had a lot of ripple and really bad transient response. Full load at constant power level was okay (Furmark+Linpack+Memtest overnight) but as soon as the power demand changed rapidly (Overwatch locked to 162 fps) it must have started providing really poor quality power that my multimeter couldn't pick up.

 

[DrMrLordX]

Damn, a Seasonic went up like that? Hmm:

https://www.techpowerup.com/review/seasonic-g550/4.html

Was that your unit?

 

[repoman0]

Damn, a Seasonic went up like that? Hmm:

https://www.techpowerup.com/review/seasonic-g550/4.html

Was that your unit?

Yep, that's the one. My RMA was accepted so will see what they send back once I get the motivation to actually send it out ...

 

[DrMrLordX]

That series does not seem to be in production anymore so . . . hmm. Yeah we'll see what you get back. Probably a 550W Focus gold?