DISCUSSION: PC shuts down without warning, afterwards boots only mysteriously after performing weird fix.

Toggle sidebar Toggle sidebar
A

AirClown

Reputable
Aug 23, 2014
11
0
4,520
Hey guys,
So my pc is acting really strange the past couple days and I would love to hear what you might think is going on.
In the last 3-4 days my build suddenly turned off without warning 3 times. Its almost like there's a power outage, but its not an power outage, it just turns off. After it turns off, it tries to start back up again but fails to do so.

2 of the 3 times it happened while playing a game. 1 time it happened when I slightly touched the pcie16x extension/riser cable for my GPU (i thought it looked like it wasnt plugged in properly (it was plugged in just fine)).

When the computer tries to boot up again after the sudden shut down this is what happens:
- cpu fan (water pump) spins slightly, but not as fast as it would normally do (when I unplug the motherboard cable and try to boot up again the speed doesnt change, so this is just the standard speed it spins from the psu connector when its not receiving any signal from the motherboard)
- hard drives dont spin, after a couple tries of booting, clearing CMOS, unplugging and reconnecting all the hard drive cables, they work again (im actually not entirely sure what exactly fixes the hard drives because they always start spinning after just a couple tries of booting, so im not really paying attention to them)
-gpu fan, no idea what it does, i have a custom waterblock on it, but the led lights do turn on, so it does receive power.
-motherboard cpu phase LED's tell me the cpu is in phase 1 power mode when it fails to boot (when it succesfully boots all the LED's turn on which means its in 6 phase power mode)
-psu fan spins normally

When it fails to boot it will stay in this "dead" state of slow spinning pump, phase 1 power mode forever (super quiet, wish it would always be like this.. but then in a working state)

I somehow managed to "fix" the issue 3 times. The first time this happened I dont remember exactly what I did to fix it. I did a lot of things, but it all basically comes down to clearing the CMOS a lot of times, unplugging and replugging everything alot of times (including ram), removing the CMOS battery for a while, wait and put it back in. After that it still didnt work, I waited longer, unplugged and replugged everything again (if i remember with one ram stick) and suddenly it booted up again (with an notification that the hardware chaged because of the single ram stick).

The next time I did the exact same thing like the first time I magically fixed it.. but it didnt really work.. so I became desperate, thinking it was probably something with the motherboard, i wanted to check if the system fan headers on the motherboard still have power (normally I dont have any fans plugged into the motherboard). So i wanted to check that but didnt have a normal fan laying around. I ghetto rigged a small electric motor to a fan header connector thingy (which i did have laying around for some reason). Just to see if there was power on the fan headers, I plugged in the motor and tried to boot up the pc. Of course the tiny little motor begins to spin like crazy while making a lot of noise. I didnt want to break the motor so quickly pulled it out of the header.. And you know what.. the pc suddenly booted up succesfully right there... I have no idea how that fixed it.. but I was just happy it worked again at that point after 1,5 hours of trying everything i could think of.

(I know this post is getting way too long)

Just today it suddenly turned off again while playing a game. I started to get experienced with putting it back to life again, so I unplugged everything, replugged everything, cleared cmos, tried to boot.. didnt work.
Unplugged everything, removed CMOS battery, waited for half an hour, connected everything again, tried to boot.. didnt work.. So guess what.. after the second try of booting it with that stupid motor in the fan header, the computer boots up normally again (again with the warning that hardware changed because I had a single stick of ram here too).

Now im not sure if that motor somehow keeps magically fixing my computer or if its just a coincidence.

Some background info which for some reason might help with finding out what is going on..
I have quite an unusual build.
It started as an usual desktop pc about 5 years ago.
2500k @stock speed (i know)
r9 280x
8gb of kingston memory
corsair cx600m (600W)
ocz agility3 ssd as boot drive and 2 random hard drives for storage

a year ago i decided to wall mount my pc and water cool everything in the most ghetto way ever.
- The gpu is extended using 2 cheap extenders, using aluminum tape for electromagnetic interference (it didnt boot without that aluminum tape (after all you dont need super expensive riser cables))
- The gpu and cpu are both cooled using the same type of cpu water cooling block (i thought it looked cooler this way (and also a cpu block was 100 euro cheaper than a dedicated gpu block))
- Gpu vram has got some heatsinks dont worry
- for the radiator i used some old weird radiator i had laying around (all copper from the inside so no problem there)

And this worked fine up until now, I dont think the cooling is a problem though. Because of the radiator, i dont have system fans which cool the water, but i have to manually turn on the radiator fan when the pc gets too hot. (to help me with that i wrote some software to give me an alert when the cpu/gpu/vram/motherboard temps get above 60 degrees (and automatically turns of if it rises above 75 for when im not in the same room (planning on just using a simple relay to turn the fan on automatically)))

So... big wall of text later.. I think I have given all the information that I can think of about the problem and my computer.
I hope you guys have some slight idea what is going on.. Personally i think the motherboard is failing, but im not sure.

Thanks if you took the time to read through all this!!

(Currently running a virus scan just to be sure)

EDIT: woops forgot the most important detail, Z68MA-ED55 (B3) motherboard
 
I had similar problems with RAM where it`s pins basically corroded or burnt I couldn`t determine it;
Also had similar problems with power supply.

I would check these 2 for now;
Borrow PSU from friend, and try your RAM sticks in his PC.
 
Thanks for your answers, heres an update on the situation:

(Virus scanner (antimalwarebytes) found nothing)

I ran the prime95 test, temperatures were normal (around 50 celsius), also a big plus that the computer didnt crash during this test.
I wanted to test the gpu too so i ran furmark.. here things started to get kind of weird, so im starting to suspect gpu problems.

The first time I ran furmark, it was getting around 7 fps and the temps were constant at 35 celsius..
Maybe I didnt configure it right, because the second time I ran furmark (using the 1080p preset test) I got normal performance for a while, and then the video output froze.
I force shut down the computer by holding down the power button. After that I had the problem again where it didnt boot up anymore.

This time I wrote down exactly what I did before it started working again in chronological order:

- Before I did anything, the pump was spinning at normal speed like it would when booting up, harddrives were spinning, it didnt sound like anything was wrong.
Only the cpu was in power phase 1 according to the mb leds.

- I cleared the CMOS, after that the pump was not spinning as fast anymore, cpu still in power phase 1.

- I removed the CMOS battery and ram, waited for about half an hour, put the battery and a single ram stick back back in.
This changed nothing, the pump was still spinning slowly and the cpu was still in power phase 1

- I tried booting it three times with the same result each time.

- Connected the electric motor to the fan header, booted it up and pulled out the motor, cpu got in power phase 6 (all mb leds were turned on)
When I pressed the power button now, the computer turned instantly off, I didnt have to long press the power button.

- Tried to boot without motor. The pump was spinning normal speed again, but no post. Cpu power phase 1 and had to long press the power button to turn it off.

- Tried the different ram stick, no change.

- Cleared the CMOS, pump was spinning slowly again.

- Removed CMOS battery, waited a while and put it back in.. nothing changed

- Connected the electric motor, booted and pulled the motor out, the computer got in power phase 6 and booted normally (with the warning of changed hardware)

- Turned the pc off to install the other ram stick.

- Tried to boot again, no succesful boot, cpu in power phase 1 and pump is spinning at normal speed.

- Connected the electric motor, booted and pulled the motor out, and it succesfully booted.



What I noticed is that the electric motor somehow helps to get the CPU in power phase 6, thats when all the motherboard led's are turned on.
If I wait until all the leds are turned on, and then pull out the motor, it boots succesfully.
It only boots succesfully after i pull out the motor, one time I had the motor plugged in for a little longer, the led's were all on for 5 seconds but nothing happened.
Only after i pulled out the motor, the monitor picks up a signal.

At least I kind of know how to get the computer to boot, but its still a mystery to me.
Apparently that motor somehow gives the computer some kind of kickstart.
Is it possible that it has anything to do with a slight voltage drop/spike which helps the cpu reach power phase 6 and post?



To check if it was a problem with the gpu, I tried to install an "old" gtx 570 which my brother is not using anymore.
It booted succesfully with the gtx 570, but i had troubles with installing the nvidia driver.
I used DDU to uninstall all the current graphics drivers, yet when I try to install the nvidia drivers I get a BSOD with the following error:
SYSTEM_THREAD_EXCEPTION_NOT_HANDLED (nvlddmkm.sys)

Tried to install it a couple times, each time resulted in the same error, so I aborted that idea to try to run furmark on the gtx 570.

I didnt experience any problems while installing the latest catalyst drivers afterwards.. maybe that gtx 570 is just dead.

After installing the catalyst drivers I restarted my pc, but ofcourse that didnt work in one try.
The electric motor power phase trick thingy did work in one try now.
So this is the third time the electric motor helped the motherboard reach power phase 6 and boot.
That trick is doing something, but I have no idea what.

Just now I ran furmark again and it froze again. At least now I know how to get the pc back up on its feet within a minute, but you cant even call this a temporary solution.

What do you guys think, motherboard issue? Or a GPU issue? Maybe both?
 
Sometimes a PSU gets hot and fails. When it cools it works again. Any chance the fixes were just waiting for the PSU to cool down ?

The fan spinning more slowly than usual makes we think voltage was low. Voltage could be low because of the PSU, or it could be low because some part is partially failed and is pulling a lot of current making the voltage go low.

If me, I'd swap in a spare PSU if you have it. the corsair cx600m gets mixed reviews, some are excellent from trusted sources, other talk about high rates of capacitor failures.

Or you can try removing all the other components until just the MB, CPU fan are connected to the PSU and boot that a few times hoping for a fail (a fail rules out other components). Intermittent fails are the hardest to debug.
 


I don't really think this is the cause since after the last furmark test when it failed, I got it back up and running really quickly (around 1 minute). Besides that, it failed instantly after the furmark test started, so theres not much time for the psu to heat up.
I will try to do the furmark test with my brother his psu tomorrow though.




The fan (pump) is connected to the motherboard only with the blue and green wires, there are no power wires connected to the pump from the motherboard, only a sensor and control wire. The PSU delivers the actual power to run the pump. I guess the motherboard needs to give the pump a control signal to start spinning faster.. Or the PSU is not always giving the same amount of power, but I think its more likely that the motherboard is not giving any control signal because its not getting past a certain point in the boot process.


I can make it fail consistently by running furmark, I can try to run furmark on the integrated graphics to rule out the gpu.
 


Why do you think that the problem is the motherboard, not the GPU?
 
Ok, do one thing, post a pic of your rig... i want to see the exact details like how everything is wired and all... It seems to me somehow the power phase design of the motherboard has gotten faulty... The pump may be not working properly causing the system to shut and even not start, and as you say its connected to the mobo, so i suspect the mobo.
Do try with a different spare psu; Try your graphics card in a different system; But i think its very unlikely.
 


WI0c997.jpg

108CVv0.jpg

1JuBwDa.jpg

PNxZ85P.jpg


The cable management is kind of chaotic so im not sure if this is easy to follow.

I've been running the OCCT GPU burn test for 20 minutes now, so I dont think the gpu is the problem anymore.. So its probably a software related issue for why Furmark makes the computer crash.

I am not able to test with a different psu today unfortunately.

EDIT: When I run OCCT power supply test, the system eventually shut down, indicating that this is probably a power supply issue? Or it can still be the a faulty power phase on the motherboard.. At at least we can say that this is probably a power delivery issue somewhere in the system, correct?

EDIT 2: OCCT says the +12V rail is doing 8.27V, but HWiNFO says 12.056V.. I guess the value of HWiNFO is correct since the pc probably wont boot with 8.27 volts lol

EDIT 3: The reason why the computer didnt shut down during the OCCT GPU test, is because the FPS are capped at 60 by default. Which doesnt draw as much power.
I did some couple runs using different FPS caps and I think i can say with some certainty that the PSU is failing.
Running the test at 60 FPS resulted in a power draw of around 70 to 130 watts according to HWiNFO (GPU VRM Power in)
70 FPS resulted in around 80 to 150 watts, ran the test for 40 seconds before aborting
80 FPS was around 90 to 165 watts, also ran this one for 40 seconds
90 FPS resulted in an instant crash
85 FPS was around 100 to 165 watts and the pc crashed after like 30 seconds.
 

The 12 V line if reported 8 is way off the line... its definitely shows a problem with the psu.... But to rule out the probability of the psu being faulty is try testing with a different psu... Looking at the setup, overheating is out of the question... So its either the psu or the mobo...
Another question is why would you use vga with this setup...?
Have you changed the coolant in between?

 


Yea ill have to ask someone soon if I can borrow his pc soon..
Im using vga out of laziness basically, I should definitely buy some displayport to dual hdmi adapter 😀
I changed it once, but it has been some time ago, probably something like 8 months or longer.. so its also definitely time to change that again.
But I'm not seeing any problems with cooling, so I'm kind of procrastinating that.

Anyway, thanks for your help with troubleshooting the problem!
 
You must log in or register to reply here.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom