PC dead. Fans and pumps don't ramp up.

[yoadknux]

I was playing Overwatch today, and all of a sudden, the PC just shut down out of nowhere. I power it on, try to go into Overwatch, shuts down again. OK, I open the case, shut the PSU down for 30 seconds, try again. Something is making noise. Either one of my two pumps or the fans or all of them.

I monitor GPU temps - normal, 30c. Pump RPM is also normal.
I monitor CPU temps- 70c spikes, looks high, but when running CPU-Z stress for a few minutes, I get 70ish - looks about normal. I cannot monitor pump RPM unfortunately.

OK, I shut down the PC and check the cables - They look OK, but cannot get a good look. I try to power it on again - it powers for a split second and shuts down (fans turn on and die within the same second) RAM RGB lights up and stays on, also the keyboard and USB devices, but that's it, no fans, no pumps, no display.

Anyway, PC is currently dead. I'm trying to diagnose the problem. I think it's either:

a) psu issue
b) motherboard issue
c) bad connection

Is there any way I can diagnose the problem without changing everything?

I think it's not any of the following components: Bad CPU should give POST problems, bad RAM should give POST problems, bad GPU should give display problems, bad SSD should give boot problems -> All of those things should not stop the fans and pumps from running.

Regarding the fans and pumps:
Corsair H75 - Connected directly to PSU via molex adapter
CPU Fans - Connected to motherboard via PWM splitter (both are connected to CPU fan header)
Kraken X41 - Connected directly to PSU via molex adapter
GPU Fans - Connected to Kraken X41 fan controller, which connects to internal USB connection to the motherboard.

In my opinion the PSU is busted, but it's hard to tell. Anyone have experience with those things? Rig available in my signature.

EDIT:
I got the PC to work for about an hour by removing some extra cables. I did the following tests:
- Superposition & FireStrike Ultra : Those are GPU intensive applications, got normal scores, passed stress test, temperatures are good.
- Ran RealBench for an hour: Crashed during the test, issue happened again. Shut down, attempted a reboot, failed. It was NOT because of temperatures - CPU cores were at 70-80, which is normal for 8700k.

Final EDIT:
issue is resolved. I removed the PSU, unplugged all the cables, re-seated GPU, re-seated PSU, plugged all the connections back - everything works. This was probably a loose connection. I think it was caused by me when I switched to a new case and worked on the cable management. Most likely the back panel which required force to be held in place, strained one of the connectors. When working with a new case always make sure you route the cables properly.

 

[trents]

Sounds like either a PSU failure or CPU hydro cooler pump failure. Can you check CPU core temps in bios? Will it run that long?

 

[yoadknux]

Sounds like either a PSU failure or CPU hydro cooler pump failure. Can you check CPU core temps in bios? Will it run that long?

The PC still sometimes works, when it does, I can run Cinebench and temps hit about 70c. I think the CPU pump works for the most part.

 

[maxfly]

How do your psu rails look underload with a multi? Are all of the standoffs screwed down? Any of them misaligned?
Id start with the easiest swap first, the psu. If that doesnt do it, its time for a teardown and partial rebuild on the mb box to test the mb

 

[yoadknux]

How do your psu rails look underload with a multi? Are all of the standoffs screwed down? Any of them misaligned?
Id start with the easiest swap first, the psu. If that doesnt do it, its time for a teardown and partial rebuild on the mb box to test the mb

Thanks for the comment. I don't have a multi on me but it would be a good test. I'll try to borrow one from the lab.
Last night I took out the PSU, cleaned the connections and reconnected all the cables. Also re-seated the GPU. Things look fine for now, I managed to power on, boot, and pass RealBench for 2 hours and 3 stress test runs of FireStrike Ultra. Time will tell, but the PC is operational at the moment.
It makes sense that this might be a loose connection, because recently I swapped cases. I didn't think a loose connection would cause this issue though. This unit should be relatively reliable and it's only 2 years old. Anyway, if it happens again, I'm going for SeaSonic FOCUS Plus 850W 80+ Platinum. As you said, PSU is the easiest thing to solve.

 

[Bill_Bright]

Last night I took out the PSU, cleaned the connections and reconnected all the cables.
Time will tell, but the PC is operational at the moment.
It makes sense that this might be a loose connection, because recently I swapped cases. I didn't think a loose connection would cause this issue though. This unit should be relatively reliable and it's only 2 years old. Anyway, if it happens again, I'm going for SeaSonic FOCUS Plus 850W 80+ Platinum. As you said, PSU is the easiest thing to solve.

No sure why you don't think a loose connection could cause such issues. Of course, without knowing which connector and more specifically, which pin or pins were not making good contact, there's no way to tell what the effect might be.

Is there any way I can diagnose the problem without changing everything?

You've made a good start by reseating all your power connectors. If your problems returns after reseating all the connectors, swapping in a known good PSU is the next logical troubleshooting step - especially before spending money on anything else.

I don't know how it would be possible to have a misaligned standoff. But it certainly is not uncommon for new builders, and distracted pros alike to insert an extra standoff during assembly. However, you did not indicate this is a new build. Can we assume it is a couple years old like the PSU and has been working fine?

Rig available in my signature.

Except no drives are listed. Assuming 2 SSDs and 2 HDs, you could get by with a quality 650W - 700W PSU. Unless you find a Platinum on sale for a great price, I would go for "Gold" certified. Platinum certified PSUs come at a premium cost that literally takes years make up in energy savings to pay for the higher purchase prices over the price for a good "Gold".

It was NOT because of temperatures - CPU cores were at 70-80, which is normal for 8700k.

Ummm, while 80°C may not be considered "hot", for the 8700K, those temps are not normal - especially with 4 x 140mm, 2 x 200mm, 2 x 120mm, 1 x 92mm and the Corsair H75 in that case. Is 80°C the peak temp you saw? What are your idle temps? What are your ambient (room) temps?

 

[yoadknux]

No sure why you don't think a loose connection could cause such issues. Of course, without knowing which connector and more specifically, which pin or pins were not making good contact, there's no way to tell what the effect might be.

You've made a good start by reseating all your power connectors. If your problems returns after reseating all the connectors, swapping in a known good PSU is the next logical troubleshooting step - especially before spending money on anything else.

I don't know how it would be possible to have a misaligned standoff. But it certainly is not uncommon for new builders, and distracted pros alike to insert an extra standoff during assembly. However, you did not indicate this is a new build. Can we assume it is a couple years old like the PSU and has been working fine?

Except no drives are listed. Assuming 2 SSDs and 2 HDs, you could get by with a quality 650W - 700W PSU. Unless you find a Platinum on sale for a great price, I would go for "Gold" certified. Platinum certified PSUs come at a premium cost that literally takes years make up in energy savings to pay for the higher purchase prices over the price for a good "Gold".

Ummm, while 80°C may not be considered "hot", for the 8700K, those temps are not normal - especially with 4 x 140mm, 2 x 200mm, 2 x 120mm, 1 x 92mm and the Corsair H75 in that case. Is 80°C the peak temp you saw? What are your idle temps? What are your ambient (room) temps?

Hi, thanks for the reply.
1) As you said, depends which connection was (and if it was) loose. I'd imagine that maybe the specific component would not work, but I guess it could just as well short out the PSU.
2) The components are not new, but ~two weeks ago I swapped my PC case. I re-seated everything except the RAM, CPU, and NVME SSD, but as you said, it's very possible I was just careless and maybe didn't plug something all the way. With a new case also comes new orientation for cables so I could picture a situation where I closed the back door and maybe semi-pushed a cable out of the modular PSU or the connection.
3) The rig has 2 drives, a single NVME and a single SATA SSD. GPU power is about 375W according to GPU-Z, which is expected. Overclocked i7 8700k I guess also adds about 100W-150W, and everything else, like RAM, motherboard, two pumps, 7 fans, let's say another 150W - Agreed 650W-700W should be fine, for some reason in my local shop 650W-850W models cost about the same. Specifically this Seasonic 850W Platinum costs 177$, which is cheaper than what I pay would I order it in Amazon (where it would cost ~200$ including shipping and taxes). That's why I mentioned this model specifically, but yes, 650W-750W Gold are also good.
4) Ambient is about 24c. For a non-delidded i7 8700k @ 4.8 those are pretty normal temps. It's possible that the motherboard overvolts or something, but regardless, those are the normal temps I have been getting the last year or so.

 

[Bill_Bright]

I'd imagine that maybe the specific component would not work, but I guess it could just as well short out the PSU.

Depends on how loose the connection is. What often happens in loose connections is tiny arcs occur - dependent on the distance across the gap and the "potential" (voltage). Those arcs often produce burnt spots and/or carbon build up. A "short" is 0Ω (zero resistance) so a loose connection would not "short out" a PSU. In fact, you want 0Ω through connector contacts.

A loose connection initially creates an "open". Resistors are primarily made up of carbon so in effect, when carbon builds up, it decreases the gap between the two contacts and a resistance is inserted in that circuit. That will alter the characteristics of that circuit. Typically, with computers and 12V being the maximum voltage, no permanent damage occurs. But arcs are sparks and excessive heat can occur in rare cases. Not good. And of course, a crashing computer can result in data loss.

4) Ambient is about 24c. For a non-delidded i7 8700k @ 4.8 those are pretty normal temps. It's possible that the motherboard overvolts or something, but regardless, those are the normal temps I have been getting the last year or so.

Normal, maybe when taxed. But not at idle. That's why I asked if 80°C was your peak temp and what are your idle temps? If you are getting anywhere near 80°C just by typing your replies to this thread, for example, something is not right.

 

[yoadknux]

Depends on how loose the connection is. What often happens in loose connections is tiny arcs occur - dependent on the distance across the gap and the "potential" (voltage). Those arcs often produce burnt spots and/or carbon build up. A "short" is 0Ω (zero resistance) so a loose connection would not "short out" a PSU. In fact, you want 0Ω through connector contacts.

A loose connection initially creates an "open". Resistors are primarily made up of carbon so in effect, when carbon builds up, it decreases the gap between the two contacts and a resistance is inserted in that circuit. That will alter the characteristics of that circuit. Typically, with computers and 12V being the maximum voltage, no permanent damage occurs. But arcs are sparks and excessive heat can occur in rare cases. Not good. And of course, a crashing computer can result in data loss.

Normal, maybe when taxed. But not at idle. That's why I asked if 80°C was your peak temp and what are your idle temps? If you are getting anywhere near 80°C just by typing your replies to this thread, for example, something is not right.

Hmm, I see. Not sure I understand, but see... In that case, can an "open" sometimes result in the situation I describe, in which the fans run for less than a second and everything goes down upon trying to power it on?

Regarding the temperatures: When running Cinebench, temps rise to ~70c. 80c is what I get after running a few minutes of RealBench, and those are pretty much the sustained temperatures as well. Idle temps are usually 30c-40c, with occasional spikes to 60c. Right now they sit ~35c when I comment here. I agree that if they hit 80c during idle then something would be wrong, it would mean the pump probably failed, but it's not the case.

The computer runs for now, and I ran RealBench and FireStrike Ultra a few times today to push the PC. Also played about 3 hours of Overwatch. I'm trying to see if the issue returns, and if so, when.

One thing I forgot to do, the 750RMi can also be connected to the motherboard through an internal USB header. I think it might have some sensors in it, which can be available via HWinfo or Corsair Link. I'll connect it and see if I can see anything useful...

 

[Bill_Bright]

An open means no current is flowing through that circuit - light flipping a light switch to off. It is important to remember power supplies output 3 primary voltages; +12VDC, +5VDC and +3.3VDC. When the PSU is first signaled to power up, it may start to push out 12V for a second, causing the fans to start spinning. But then the system detects one of the other voltages is missing (or fails to detect their presence) and shuts down the system. That explains one example of how the fans can spin for a second then stop.

Right now they sit ~35c when I comment here.

Okay. I would be very comfortable with those temps.

I'm trying to see if the issue returns, and if so, when.

Hopefully never and it was just a loose connection.

 

[yoadknux]

An open means no current is flowing through that circuit - light flipping a light switch to off. It is important to remember power supplies output 3 primary voltages; +12VDC, +5VDC and +3.3VDC. When the PSU is first signaled to power up, it may start to push out 12V for a second, causing the fans to start spinning. But then the system detects one of the other voltages is missing (or fails to detect their presence) and shuts down the system. That explains one example of how the fans can spin for a second then stop.

Okay. I would be very comfortable with those temps.

Hopefully never and it was just a loose connection.

It has been about 2 weeks, did a lot of gaming and stress tests during this time, everything runs perfectly. As you said, it was probably a loose connection, since the issue stopped occurring after I removed the PSU and re-plugged all the cables. Thank you for your help.
Can a splitter also cause those types of issues? I noticed some fan splitters and RGB connectors can be "open", the pins can freely touch whatever then want (Not sure if it's the right term, but they don't have an "housing").

 

[Bill_Bright]

Can a splitter also cause those types of issues? I noticed some fan splitters and RGB connectors can be "open", the pins can freely touch whatever then want (Not sure if it's the right term, but they don't have an "housing").

I think "open" in this case would be like opening a cabinet door - "exposing" the contents inside. If an electrical connector has "exposed" wires or electrical contacts, those contacts could touch something else causing a "short".

If your splitter has exposed contacts, I would probably look for a different splitter to use. At the very least, I would securely tie that splitter back so those exposed contacts could not come into contact with anything else, including the bare metal of the case.

 

[trents]

If you were only using one branch of a "Y" fan cable splitter then it is possible that the bare pins of the unused branch were making contact with and shorting out against some metallic component inside the case. If you are still using that splitter I would certainly get some electrical tape and cover the bare pins. There have been times when a case fan power cable was too short and the only extender I had was a "Y" splitter so I can envision the scenario where the spitter wasn't being used as a splitter so as to leave exposed pins on the unused branch. At any rate, you seem to be up and running normally now.

 

[Bill_Bright]

I agree covering an unused connector that has exposed contacts with electrical tape would prevent shorts. But if you have unused connectors in a Y cable, I say pull the cable. If you need the extra length, then look for a simple extension, if possible. That would completely remove the possibility of shorting something out, and it would make your cable management look more tidy too.

 

[yoadknux]

This is the connector in case you're wondering. It's an RGB splitter, it comes with the fans in the case (CM H500P Mesh), it's not removable. I did however find a small cap in the accessories and I used it to close the connector when I applied the PSU fix last time. Huh, wonder if this was the problem all along... Makes sense that the back panel forced it against the cable cover, which is metallic...