Troubleshooting Desktop Failure, Some Moral Support?

Zarathustra[H]

Extremely [H]
Joined
Oct 29, 2000
Messages
38,741
Hey all,

So I am no spring chicken when it comes to this stuff, but it has been a while since I had to troubleshoot a mystery problem where a system just refuses to work, so I figured I'd access the [H] hivemind for some suggestions and to validate my approach.

Last Thursday I came home after work, pressed the power button on my my Desktop and it is behaving like a paper weight, and refusing to turn on. The system had been working just fine since my Threadripper upgrade in December. The case had not been opened and nothing apparent changed since it was in a functional state.

Initially thinking it might have been a motherboard issue, I started discussing it here and got some good feedback from Mega6 and pendragon1

I opened it up on my workbench and did an initial examination. Nothing visual apparently wrong. No water leaks anywhere, in general looks pretty good, apart from its refusal to power on. I then plugged in the PSU, the main RGB LED on the motherboard briefly flashed on, and then turned off immediately. The on-motherboard power button remained lit up until I pressed it, at which point it went out, and the power never came on.

Based on this, I suspected the PSU. I removed the PSU from the case, and hooked it up to a known working PC. This PC also failed to start up. I thought I had found my culprit.

I RMA'd the PSU to Seasonic. Last night the new one came in.

Just to make sure I was on the right path, I first hooked up the replacement PSU to the same known good PC which had failed to boot last time. It powered on normally. Great! I now spent the time to re-route the cables and install everything in my main desktop. Plugged it in, flipped the rocker switch from 0 to 1, and the same thing as before happened. Main RGB LED flashed on, and then turned off. Power button remained lit until I pressed it and then went out. System refuses to power on.

Damn. Looks like the PSU was a misdiagnosis. It looks like my system has a short somehwere and the PSU's short protection circuitry is kicking in. I wonder if I tested the old PSU on the known good PC too quickly, before the short protection circuit reset after tripping.

Alright. So now what? First suspect. I have a hand soldered board for manual PWM fan control which has worked well, but could be a source of error, and I don't really need it anymore now that I have an Aquaero. So I pulled it out (and damaged one of the clips that holds the front panel to my case in the process :( )

Still no bueno.

Next steps:

1.) Reseat RAM and test?

2.) Remove all PCIe boards except GPU and see if it will start without them?

3.) Remove GPU and replace with known good GPU and see if it will start?

4.) If none of the above work, RMA the motherboard?


What a pain in the butt. I haven't had a non-functional system like this in YEARS.

I'd appreciate any suggestions, or input in my troubleshooting!
 
I read the thread title and hopped in to offer my moral support:

"You got this, Sara!"
 
Youve likely already done this but make sure all of your cables are plugged in nice and secure specifically the power and reset buttons.
Have you tried powering up using the power button on the mb?
You may also try running another leak test with your pump running at 100%. You may have a slow leak that dripped onto your mb.
Have you tried reseating your cpu?
Thats all ive got.

I hate to say it but it really seems like a mb failure. Hopefully it isnt and turns out to be something easy. GL!
 
Swap out the GPU and test, after unplugging everrything else. I had a G92 card go dead short before and took out the PSU with it. Barring that, probably the mobo
 
try removing the power button from the mobo header and jump start it. maybe the switch is faulty.
 
So, just to be a bit of a drag here, you may not have had a misdiagnosis. It is not impossible that when a power supply goes boom that it nukes some other components on the way out the door. If the PSU didn't power on your other system, it *was* dead, and *when it died* it was connected to your main system.

There is a non-zero chance that your motherboard, CPU, RAM, GPU, and disk drives may be dead. Or none of them. Or just some of them. Electricity is fucking weird and when shit's not working the way it's supposed to, predicting what will actually happen is damn near impossible.

My recommendation is to start from a position of assuming zero trust in all your remaining components. If it turns out your motherboard is dead, then I'd start looking pretty damn suspiciously at both your CPU and RAM. They may work, they may not. They may work *now* and then stop working in the near future.

Had your house suffered a power surge, my suggestion would be to start talking about an insurance claim and claim the entire system.
 
  • Like
Reactions: N4CR
like this
shouldn't be that hard to figure out...get your act together! you have the right ideas....got to get it as minimal as possible and see whats triggering the new supply to power off. You have all that water cooling stuff to deal with so im sure it doesn't help when trouble shooting..now if you had access to another working motherboard AND cpu it would be super right? No stores that take returns near you?
 
Rip everything out and assemble in a cardboard box/on a test bench to isolate a ground causing teh initial issue
if i recall op has a pretty major water cooled setup which might make this a bit of a pain:eek: but of course i agree
 
So, just to be a bit of a drag here, you may not have had a misdiagnosis. It is not impossible that when a power supply goes boom that it nukes some other components on the way out the door. If the PSU didn't power on your other system, it *was* dead, and *when it died* it was connected to your main system.

There is a non-zero chance that your motherboard, CPU, RAM, GPU, and disk drives may be dead. Or none of them. Or just some of them. Electricity is fucking weird and when shit's not working the way it's supposed to, predicting what will actually happen is damn near impossible.

My recommendation is to start from a position of assuming zero trust in all your remaining components. If it turns out your motherboard is dead, then I'd start looking pretty damn suspiciously at both your CPU and RAM. They may work, they may not. They may work *now* and then stop working in the near future.

Had your house suffered a power surge, my suggestion would be to start talking about an insurance claim and claim the entire system.

This.

Had a similar situation with two dead Mobos and it turned out to be PSU. PSU took 2 weeks to kill one board and 3 months on the other.
Could have easily porked something on the mobo on the way out. Also had one kill a GPU before.
 
if i recall op has a pretty major water cooled setup which might make this a bit of a pain:eek: but of course i agree

Aye. But it's soft tubing so he may be able to accomplish it without draining...It ain't easy, and in many cases it can't be done but it might be able so it's something to leave on the table until the root cause is identified and fixed

I believe he installed the new stuff into the old case and something about that may have led to the issue... maybe a loose screw from the original build back in the day finally jostled loose/around and maybe created a short or something like that...or could be a cable from back in the day that wore through the lining over time...
 
Appreciate all the recommendations guys.

I had my big fat birthday party this weekend (turned 40) so I didn't have time to do any troubleshooting. The week is not good, so I'll be doing some more troubleshooting next weekend.

For what it's worth there have been no known outages or spikes in the power to my knowledge. Nothing shows up in the logs on my servers UPS units.

May just be a good old "bathtub curve" issue for some newly acquired component.

Will post back here as I learn more.
 
Last edited:
It could be a short in the case -> motherboard, it had happened me a couple of times. I would remove the components from the case and build it somewhere else like over the motherboard box, with minimal components (cpu, 1 ram module, gp, etc).
 
Go jerk everything out and test.if you find a dead component you are that much further ahead. Probably not what you want to hear but maybe the lonely way to be sure
 
Go jerk everything out and test.if you find a dead component you are that much further ahead. Probably not what you want to hear but maybe the lonely way to be sure

Hehe

This type of troubleshooting used to be so much easier before I installed a custom water loop :p
 
Hehe

This type of troubleshooting used to be so much easier before I installed a custom water loop :p
Not to mention you have TERRIBLE luck wit this stuff lol:) (remembering your older threads)


ANYWAYS......So no luck getting getting any kind of post even with just the board plugged in? with everything else unplugged wiring wise? Usually when the supply wont even power the fans its a bad supply. Kinda figured you could do try this with out touching/undoing all the water stuff
 
Not to mention you have TERRIBLE luck wit this stuff lol:) (remembering your older threads)


ANYWAYS......So no luck getting getting any kind of post even with just the board plugged in? with everything else unplugged wiring wise? Usually when the supply wont even power the fans its a bad supply. Kinda figured you could do try this with out touching/undoing all the water stuff

Yeah, I unplugged everything except ATX 24pin, EPS power and PCIe power to the GPU. No luck there. Then I removed the PCIe power to see if that would do it. Nothing.

I even tried eithout the EPS connectors to see if it would stay on, but no luck.

I don't think my luck has been too terrible. I did have my XSPC swivels go bad on me and start dripping but they never got anything so there was no damage back then.

I also had my fire and flames incident where a bad EPS extension turned my desktop into a smoke machine, but that never did any damage, once replacing the cable and power supply that build returned to working normally again.

Only thing in common between this build and the one with previous issues is the GPU as I am not quite ready to replace the Pascal Titan yet (holding out for the 3080ti or whatever the next badboy is called. Heck maybe even big Navi if it surprises me and is really good).

So, I'm hoping the GPU is not the issue.

If I am going to go through this trouble I may wind up buying a new case with more space and a couple of new radiators while I am at it. No idea what is good these days. I like full towers, but most of the good ones are $500+ which is nuts for some painted sheet metal.

I'll have to make up my mind.
 
not really any quick easy way to pull the gpu out either? even though you already unplugged the power to it? really starting to look like a bad board at his point...suck to cause thats the most work to change out.....i find it very unlikely a bad cpu could have any effect like this...time to pick up another board it looks:(
 
not really any quick easy way to pull the gpu out either? even though you already unplugged the power to it? really starting to look like a bad board at his point...suck to cause thats the most work to change out.....i find it very unlikely a bad cpu could have any effect like this...time to pick up another board it looks:(

Well, it isn't pretty, but just for a momentary test with another knowm working GPU it works.

IMG_20200202_195251.jpg


I removed all my PCIe cards (Pascal Titan which is dangling, Sound Blaster X-Fi Titanium HD, Dual Intel x520 10Gig Ethernet, and old Intel 750 PCIe SSD.)

I disconnected power to everything except the 24 pin motherboard ATX connector and the two EPS connectors.

I temporarily installed an old Nvidia Quadro 2000 known to work from another system.

Plugged it in, and exact same thing as before. Motherboard LED briefly lights up. On board power button stays red until pressed when it goes out and nothing further happens.

The only thing I haven't been able to test is the RAM. This is the only DDR4 system I have accesss to right now, so I have nothing I can steal RAM from even for a short test.

I guess the next step is to disassemble everything out of the case and see if I can get the motherboard to respond out of the case, and if not, a motherboard RMA is in order.
 
Well, it isn't pretty, but just for a momentary test with another knowm working GPU it works.

View attachment 220356

I removed all my PCIe cards (Pascal Titan which is dangling, Sound Blaster X-Fi Titanium HD, Dual Intel x520 10Gig Ethernet, and old Intel 750 PCIe SSD.)

I disconnected power to everything except the 24 pin motherboard ATX connector and the two EPS connectors.

I temporarily installed an old Nvidia Quadro 2000 known to work from another system.

Plugged it in, and exact same thing as before. Motherboard LED briefly lights up. On board power button stays red until pressed when it goes out and nothing further happens.

The only thing I haven't been able to test is the RAM. This is the only DDR4 system I have accesss to right now, so I have nothing I can steal RAM from even for a short test.

I guess the next step is to disassemble everything out of the case and see if I can get the motherboard to respond out of the case, and if not, a motherboard RMA is in order.
think i would have already ordered a new board from prime....easy enough to return if not needed for sure
 
Given what you just said, I'm going to quote myself lol.

If it turns out your motherboard is dead, then I'd start looking pretty damn suspiciously at both your CPU and RAM. They may work, they may not. They may work *now* and then stop working in the near future.

Had your house suffered a power surge, my suggestion would be to start talking about an insurance claim and claim the entire system.
 
Given what you just said, I'm going to quote myself lol.

Yeah, I couldn't point to any outage or any voltage spikes. The logs from my two UPS units in the basement don't indicate anything from the time around when this happened.

I think I'd have a difficult time making any claims.

I'm going to do some thorough stability testing when I get it back up and running again. If it catches any issues at all, I'll deal with it then.
 
Last edited:
Memtest for sure. Burn-in testing in general would be good.

I don't know if you depend on the PC for your livelihood or if it's just for fun/hobby or whatever, but if it's the former I would consider aggressive replacement of all components, depending upon your tolerance for downtime.

Either way, sounds like you're on the right path. And it sucks, lol, I've done the very same thing before. Hopefully yours ends at the motherboard - the last time I had a power supply pop (surge related, in my case) - it took out everything but the hard drives with it.
 
Not to mention you have TERRIBLE luck wit this stuff lol:) (remembering your older threads)

Maybe you are right after all.

The package with my RMA motherboard was supposed to be delivered by USPS to Gigabyte Support yesterday by 6pm in California.

It never received a "delivered" scan, and the date was moved up to 6pm California time today. It didn't get listed as delivered today either :/

The silliest part is, I didn't even think of paying for insurance. When I was at the counter I was thinking "why would I insure a dead motherboard?".

Hoping this one is just strangely delayed, and arrives tomorrow, but I'm not sure what to expect right now.
 
Maybe you are right after all.

The package with my RMA motherboard was supposed to be delivered by USPS to Gigabyte Support yesterday by 6pm in California.

It never received a "delivered" scan, and the date was moved up to 6pm California time today. It didn't get listed as delivered today either :/

The silliest part is, I didn't even think of paying for insurance. When I was at the counter I was thinking "why would I insure a dead motherboard?".

Hoping this one is just strangely delayed, and arrives tomorrow, but I'm not sure what to expect right now.
i ALWAYS insure packages cause it so cheap...but i suspect it just hadn't disappeared. That motherboard is worth exactly what it would sell for unopened new rma right here of all places....something tells me it would fetch around 200 plus the first night alone
 
Looks like it's just USPS being silly.

I just got a notification that Gigabyte has received, opened and confirmed the receipt of the serial number of my motherboard.

My USPS tracking still says that the package hasn't been delivered. Maybe they just missed scanning it before dropping it off?

I'm not sure how quickly they turn this around after receiving a package,

i ALWAYS insure packages cause it so cheap...but i suspect it just hadn't disappeared. That motherboard is worth exactly what it would sell for unopened new rma right here of all places....something tells me it would fetch around 200 plus the first night alone

I'm not sure I'm convinced anything is cheap in the shipping industry.

Back when I was in college in 1999-2003 I used to do a hell of a lot more buying, and selling of parts and a lot of shipping, often with UPS ground. At that point I could send a small box across the country for $6. A larger box cost me maybe $12.

I get sticker shock every time I try to send something. With this motherboard, first shipper I checked was UPS. For basic ground service without even adding insurance they wanted $43. I instead took it to the post office. The motherboard box was too large to fit in a large flat rate box so I had to use my own packaging. Priority Mail cost me $29.

If I had added insurance it would have been even more crazy.

I feel like pricing in that industry has increased many times faster than what can be explained through inflation.

And this whole Warranty process puts a sour taste in my mouth for the following reasons.

1.) I am claiming a warranty, something Gigabyte owes me because their process/product messed up. Why am I responsible for shipping again? This should all be on them.

2.) Why exactly is it you need to pay extra for insurance? I mean seriously. It is their job to take custody of my box and transport it safely to its destination, and they want me to pay extra to make sure they don't squash it or lose it in the process? This is absolutely insane. It's like a protection racket. Nice box you've got there. Would be a shame if something were to happen to it! If my package gets damaged or doesn't get where it is supposed to be going they should be repsonsible for it without any additional insurance, provided we aren't talking about damage due to inadequate customer provided packaging...
 
Aww crap.

Gigabyte just concluded that they could not find any problems with the motherboard... They say they tested it in multiple configurations without finding any issues.

I am at a loss now.

I tested with a different power supply, with different RAM, and a different GPU, removing all other devices.

Only thing I haven't tested another one of is the CPU...

CPU's usually don't fail. At least I've never had one fail...
 
Aww crap.

Gigabyte just concluded that they could not find any problems with the motherboard... They say they tested it in multiple configurations without finding any issues.

I am at a loss now.

I tested with a different power supply, with different RAM, and a different GPU, removing all other devices.

Only thing I haven't tested another one of is the CPU...

CPU's usually don't fail. At least I've never had one fail...
unfortunately....i have seen at least 3 threads here where the ryzen cpu as failed....even threadrippers....its why i asked about locating a cheap cpu to test the motherboard a while back..or at least i thought i did. You wernt even overclocking...sucks you got a bum cpu. Maybe you get lucky with a super quick cpu RMA :)
 
unfortunately....i have seen at least 3 threads here where the ryzen cpu as failed....even threadrippers....its why i asked about locating a cheap cpu to test the motherboard a while back..or at least i thought i did. You wernt even overclocking...sucks you got a bum cpu. Maybe you get lucky with a super quick cpu RMA :)

hehe. I don't think there is such a thing as a super cheap TRX40 CPU to test with :p
 
the hell with it......pick up a
Threadripper 3990X

and while your at it pick up a second one just to keep for backup :)
 
If you've tested everything else, and the CPU is the only thing left, AMD does have warranty support. Never had to use it myself, but it beats buying a new CPU.
 
the hell with it......pick up a
Threadripper 3990X

and while your at it pick up a second one just to keep for backup :)

That would actually be a downgrade for me. A very expensive downgrade.

My workloads tend to prefer higher clocks over higher core count. At least once there are 8 cores or more.

Only reason I went with Threadripper was for the PCIe lanes.
 
Aww crap.

Gigabyte just concluded that they could not find any problems with the motherboard... They say they tested it in multiple configurations without finding any issues.

I am at a loss now.

I tested with a different power supply, with different RAM, and a different GPU, removing all other devices.

Only thing I haven't tested another one of is the CPU...

CPU's usually don't fail. At least I've never had one fail...

I've RMA'd one Intel (6700K) and one AMD (x4 9850) in all the years I've been building computers. That's a pretty good record if you ask me. Sometimes shit happens. AMD will hook you up with no problems at all. Intel is just as good.
 
I've RMA'd one Intel (6700K) and one AMD (x4 9850) in all the years I've been building computers. That's a pretty good record if you ask me. Sometimes shit happens. AMD will hook you up with no problems at all. Intel is just as good.

Yeah, in my 29 years of doing this I have had many failures, but never a CPU. I pretty much assumed that as long as you didn't do something really stupid, they were more or less bulletproof at this point, which is why it wasn't one of my suspects.
 
Yeah, in my 29 years of doing this I have had many failures, but never a CPU. I pretty much assumed that as long as you didn't do something really stupid, they were more or less bulletproof at this point, which is why it wasn't one of my suspects.
I guess as tech gets smaller and smaller and millions squared more transistors the odds of failure increases? I have only lost one cpu before as well and it was an intel slot A i was trying to over clock. (seems like that one cost me 2 or 300 bucks) I know those old xeon 1366 chips were total tanks and get take anything thrown at them over and over. Them days are long gone for both camps :)
 
Well,

Just to make sure I spent some time today double checking everything on other hardware.

GPU, NVMe drives, 10G Network adapter, PSU etc. Everything works.

(Turns out at idle an empty EK water block has enough heft to keep the die under 30C in my cool basement for a surprising amount of time)

I don't have anything to test my DDR4 RAM in, but I did test the motherboard with some other RAM before sending it to Gigabyte.

The only thing left it could possibly be is thus the CPU, so I submitted an RMA claim with AMD.

Guess we'll see how this one goes... I've never used AMD's RMA service before, but I can't imagine it is anything but good.
 
Last edited:
AMD's RMA process is pretty awesome.

They received, tested my CPU and shipped a replacement CPU all in the same day. Very impressed. They put Gigabyte to shame.

Not that Gigabyte were bad or anything, but AMD really just showed up everyone in the "how you handle a product failure" category.

I had expected this to drag out longer, so now I'm likely not going to be ready for the CPU when the new one gets here :p
 
Back
Top