PC blackouts and can't find the cause

[p1ns]

Hi, long time reader, first time poster looking for some help.

I've filled my PC specs but I'll post them here too for ease of viewing:
Processor Intel i5-4670k
Motherboard Gigabyte GA-Z87X-D3H
Cooling Noctua NH-L12
Memory G.Skill 8GB DDR3 2133Mhz
Video Card(s) KFA2 1660 GTX Super
Storage Samsung 850 Evo 500 GB SSD & various HDD's
Power Supply Corsair CS550M
Software Windows 7
No brand old OEM case.

This has been my main PC since I built it 2013 and I upgraded the SSD in 2017 and graphics card in 2020, but it's mostly the same as it was at the beginning, I've kept the same win 7 installation since then.
On May or June 2023 I ripped off the PCIe 16x slot when removing the graphics card. I trimmed the power and ground pins to avoid short circuits and moved the graphics to the PCIe 8x slot. PC kept working without problem during many months.

Around 6 or 8 months ago I had some sudden blackouts, everything stopped, all fans including PSU and it restarted again a couple seconds later. It didn't happen on processor load, just browsing some websites. I thought that perhaps was that the PSU was aging and at some moments the voltages were off or whatever. It was quite unusual so I didn't care a lot about it.

But since last month, these blackouts started to happen more and more frequently, and whenever the first happened I had to turn off the PSU, wait and turn it on to try to boot again. Blackouts may happen in rapid succesion, just in BIOS splash screen and before booting windows.

So, I got another new PSU (cheap 550W model, just to try), replaced it and to increase my confusion, I got the same symptoms, so I ruled out my previous PSU.

Then I started to disconnect peripherals, disks, front panel wires and so, until just booted to BIOS config screen with integrated graphics and keyboard. It kept shutting by itself

To continue, I got all the board off the case, reset BIOS, replaced board battery, removed cooler and CPU, reaseated CPU and new thermal paste.
I booted the board off the case, with same minimal peripherals to BIOS and the blackouts stopped.

It was quite weird and at that moment I started blaming the case. Perhaps there was some contact between the board and the case wall. This case is like 20 years old and have been in use continuously since then holding several PC's, but who knows.
I examined the case I could find some lines in the wall, like it some lead was grinding against the wall, but nothing else.

So, I kept working with all parts on the ground, did some power testing and when running Prime95 it went off again after a minute or so.
Looking at monitor software I could see that the power reported by the CPU is very low:

It doesn't go over 10 or 12 Watts. But apparently I can reach max freq of this CPU (3'8 Ghz).

So, I am almost clueless. Have I ruined my PC for letting it work while the occasional blackout happened ?
Why does getting it out of the case seems to alleviate the problem ? I've written this whole message with all parts laying on my room mat.
What about that ridiculous power reports ? I can't find anything similar to this on google. I've worked on many PC's and solved or at least diagnosed many problems but this one is beating me. It seems that there is some power delivery problems with this CPU or board, but how can I further diagnose them ?

I can post some pictures of the current "build" if requested.
Any hints will be welcome, thank you.

 

[Bill_Bright]

I would have liked to have seen a little bigger PSU - in part because of your vagueness regarding the number of hard drives. But since you get the same results with 2 PSUs, then I agree, it is not likely your old PSU is bad.

If you breadboarded (ran the computer outside the case on a large bread/cutting board) and it worked fine, that points to three possible problems, (1) heat or (2) electrical short.

Addressing the possible short first, note cases are designed to support 1000s of motherboards of different sizes. For example, most standard mid-tower cases support standard ATX motherboards as well as smaller µATX and often larger EATX motherboards too. Even among standard size ATX boards, different manufacturers may, or may not put a motherboard mounting hole in a designated location. Until just recently, the ATX standard did not dictate where mounting holes "must" go, only where they "may" go.

Therefore, cases tend to have more locations for standoffs than many motherboards have mounting screw locations.

A common mistake by the less experienced and distracted pros alike is to insert one or more extra standoff in the case under the motherboard. Any extra standoff creates the potential for an electrical “short” in one or more circuits. The results range from "nothing" (everything works perfectly) to a wide assortment of odd problems, to "nothing" (as in nothing works at all ).

To add to the confusion, these issues may be intermittent, depending on heat, expansion/contraction of materials, as well as continuity/resistance through the contact point. This is why it can work fine for extended periods, then start to fail. Therefore, you need to ensure the case only has an inserted standoff where there is a corresponding motherboard mounting hole.

If you have the correct number and placement of standoffs, install back in the case but do NOT connect the motherboard to the case's front panel connections - yet. That is do NOT connect the power and reset button switches or any audio or USB connections - yet. Only the keyboard and mouse should be connected, for now.

And when reassembling back in the case, pay special attention to other possible locations of a short. The rear panel I/O area can be a problem because of the 100s of tiny "fingers" - grounding tines around each connector than can easily get bent, then short something out.

Ensure all cables are securely fastened to the proper connectors. Then boot. If it starts to fail now, I would suspect heat. Blast a desk fan into the side of the open case. If it works with the fan, you need to look at your case cooling. Ideally, you typically want lots of front to back flow of air through the case.

If it boots fine, then shutdown, unplug from the wall, and connect the case's cables to the motherboard. Double check to make sure connections are right, and secure, and that you did not knock anything loose while in there. Connect power and boot.

If it fails now, that would point to a case connector/switch/cable.

Having said all that, this is clearly an old system. There could be other components intermittently failing that just didn't show up when breadboarding - like the motherboard itself, or your old, DDR3 RAM. Eventually, all electronics will fail - if not replaced before then.

 

[FreedomEclipse]

You can also rule out if its your 1660.

Your 4670k has an iGPu and MB has both an HDMi, DVi and analogue port. Take your 1660 out and run the system off the intel iGPu for a while and see what happens.

 

[Bill_Bright]

You can also rule out if its your 1660.

Except he said it all ran fine outside the case. That would seem to rule out the 1660.

 

[FreedomEclipse]

Except he said it all ran fine outside the case. That would seem to rule out the 1660.

I dont see where he said it ran fine. I can see where the OP said he ripped the 16x slot off the MB when taking out the GPu.

I read that he replaced the PSU with another but problems still persisted. Nothing about testing the system without a D-GPu.

 

[Bill_Bright]

I may have misunderstood what he meant, but in his opening post he said,

To continue, I got all the board off the case, reset BIOS, replaced board battery, removed cooler and CPU, reaseated CPU and new thermal paste.
I booted the board off the case, with same minimal peripherals to BIOS and the blackouts stopped.

 

[jaggerwild]

You riped off the PCI slot and can’t figure why the computer is shutting off randomly?

Take it apart and trim back the PCI pins that are exposed, hopefully you didn’t hurt anything….

 

[FreedomEclipse]

@p1ns - If your system is a regular tower unit standing vertically. Lay it flat horizontally and see what happens.

 

[p1ns]

You riped off the PCI slot and can’t figure why the computer is shutting off randomly?

Take it apart and trim back the PCI pins that are exposed, hopefully you didn’t hurt anything….

Hi, that happened a long time ago (around may or june 2023) and after trimming the power pins of the PCIe slot, the computer ran fine for many months. I agree that I can try to trim all other signal leads.

@p1ns - If your system is a regular tower unit standing vertically. Lay it flat horizontally and see what happens.

It was actually a regular vertical tower. Now the board is laying on the ground.

Let me explain again the part where I removed peripherals, including the graphics card.
First, with the board in the case, I booted with integrated graphics and only keyboard and monitor. And it continued to shutdown.
So after this test, I ruled out the graphics card because it didn't solve the problem.

I am going to reply to the first longer message, thank you for your detailed answer.

I would have liked to have seen a little bigger PSU - in part because of your vagueness regarding the number of hard drives. But since you get the same results with 2 PSUs, then I agree, it is not likely your old PSU is bad.

I am using 2 7200 rpm HDD, one 1TB Seagate, one 2TB and one 5400 rpm 6TB from Toshiba. Total power consumption by these drives and the SSD shouldn't go over 40W in worst case I'd say.

Regarding the case, this is it:

As you can see, it has some bumps instead of the usual standoffs you were thinking about.
If look at the middle of the board wall, you can see some scratches, those worried me because I think that some lead at the back of the board may have been in contact with the metal wall.

The current "build" is like this, I am currenly typing this post from it:

And for those curious off the ripped slot:

To additionally explain the current status of the PC.

Outside the case, as you can see it, it seems to prevent random blackouts, for example just idling or browsing a site. Those happened when the board was inside the case.
BUT, if I apply load on the CPU, for example using Prime95 or playing CS2, it will eventually blackout.

¿Any hint about the lower power readings? As I said I am quite confused about those.

I know it's an old build, but I have been using it for so long... anyway I have the feeling I won't be able to fix this, but at least I wanted to learn something from it and write a post about in case somebody may find it useful in the future.

A friendly coworker lent me a Z97 and and a i5-4690k. I know they are working fine, I can swap parts to test if another CPU runs fine, although I don't want to ruin good parts. I will think about what to test the following days. Not a lot to time to work on this during the week.

Anyway, thank you for your replies.

 

[FreedomEclipse]

Im wondering if there is hidden damage to the MB when you ripped the PCIe slot out. Maybe there are some hidden traces somewhere that are damaged and pull apart when the MB warms up and flexes?

 

[Dr. Dro]

Im wondering if there is hidden damage to the MB when you ripped the PCIe slot out. Maybe there are some hidden traces somewhere that are damaged and pull apart when the MB warms up and flexes?

That would be my hypothesis as well, especially considering that it works when laid flat. This motherboard clearly has severe physical damage done to it. Given the state of that hardware, and factoring wear from age, a short could very well have developed there.

I'm gonna be honest with you OP, this system isn't worth putting a minute further in it. This platform is well over 12 years old: if it still suits your needs, buy a cheap socket 1150 board from China or something on the used market and rebuild it. Otherwise, if your financial reality allows, seriously consider an upgrade to a more modern platform, you've skipped the entire DDR4 generation there, after all.

 

[eidairaman1]

Hi, that happened a long time ago (around may or june 2023) and after trimming the power pins of the PCIe slot, the computer ran fine for many months. I agree that I can try to trim all other signal leads.


It was actually a regular vertical tower. Now the board is laying on the ground.

Let me explain again the part where I removed peripherals, including the graphics card.
First, with the board in the case, I booted with integrated graphics and only keyboard and monitor. And it continued to shutdown.
So after this test, I ruled out the graphics card because it didn't solve the problem.

I am going to reply to the first longer message, thank you for your detailed answer.

I am using 2 7200 rpm HDD, one 1TB Seagate, one 2TB and one 5400 rpm 6TB from Toshiba. Total power consumption by these drives and the SSD shouldn't go over 40W in worst case I'd say.

Regarding the case, this is it:

As you can see, it has some bumps instead of the usual standoffs you were thinking about.
If look at the middle of the board wall, you can see some scratches, those worried me because I think that some lead at the back of the board may have been in contact with the metal wall.

The current "build" is like this, I am currenly typing this post from it:

And for those curious off the ripped slot:

To additionally explain the current status of the PC.

Outside the case, as you can see it, it seems to prevent random blackouts, for example just idling or browsing a site. Those happened when the board was inside the case.
BUT, if I apply load on the CPU, for example using Prime95 or playing CS2, it will eventually blackout.

¿Any hint about the lower power readings? As I said I am quite confused about those.

I know it's an old build, but I have been using it for so long... anyway I have the feeling I won't be able to fix this, but at least I wanted to learn something from it and write a post about in case somebody may find it useful in the future.

A friendly coworker lent me a Z97 and and a i5-4690k. I know they are working fine, I can swap parts to test if another CPU runs fine, although I don't want to ruin good parts. I will think about what to test the following days. Not a lot to time to work on this during the week.

Anyway, thank you for your replies.

Time for an upgrade

 

[Bill_Bright]

I am using 2 7200 rpm HDD, one 1TB Seagate, one 2TB and one 5400 rpm 6TB from Toshiba. Total power consumption by these drives and the SSD shouldn't go over 40W in worst case I'd say.

3 hard drives plus 1 SSD? 40W is worst case? Sorry but no way.

Of course all drives are different but you should base your possible (worst case) consumption scenario on all drives demanding maximum power at the same point in time. Unlikely you say? Sorry, but again that is incorrect. Consumption is maxed out all at once when powering up the computer. Estimates vary depending on your source of information but as seen here,

A 7200 RPM hard drive would normally consume 20-25W for 5-10 seconds during startup. It will then idle at around 6-8 watts. The average power consumption during read/write tasks will be around 8-12 watts, and the maximum can go up to 16 watts. In sleep or standby mode, it will consume 0.6-1.8 watts.

A 5400 RPM hard drive can consume 15-20W for 5-10 seconds during startup. It will then idle at around 4-6 watts. The average power consumption during read/write tasks will be around 6-9 watts, and the maximum can go up to 12 watts.

Assuming maximum, your 3 hard drives could demand 70W when powering up. A SSD could add another ~10W.

Im wondering if there is hidden damage to the MB when you ripped the PCIe slot out. Maybe there are some hidden traces somewhere that are damaged and pull apart when the MB warms up and flexes?

It did work fine for months after the rip-out but I agree, over time and many expansion/contraction (heat-up/cool-down) cycles, possible damage could have expanded. Just not sure how likely that is - again, since it did work for months.

I just think this could easily be simple age related. The board uses DDR3 RAM. The most recent BIOS update for that Gigabyte GA-Z87X-D3H is from nearly 11 years ago. Any component, at this point, could be failing.

 

[Shrek]

Test the RAM?

 

[Deleted member 242843]

used z87 motherboards are very cheap and easy to come by these days. That motherboard has had it.

 

[p1ns]

Hello again guys, replying to put an end to this story, in case it is useful for somebody who may struggle with similar issues and this threads pops-up in google.
I've not reply for many days because I was working on it.
I got myself with the following hardware.

- Previous HW: Gigabyte GA-Z87X-D3H and i5-4670k
- Lent from coworker: Gigabyte GA-Z97-HD3 and i5-4690k
- Another 1150 board and processor I managed to grab: Asus H81M-K and i3-4170

I cleaned up all the shit from the boards and old thermal paste from all boards and fired up the Z97 board and 4690k. To my suprise, similar effect appeared, blackouts and no video signal.
I put a cooler on the H81M with the i3 and booted a Kubuntu from USB fine. I downloaded prime95 and had it running for a few minutes.
I did the same test on the same board with both i5's. No issues, they ran prime95 without blackouts. I know it's not the best board for the high power chips but they can run for a while.

So, it was clear the chips were fine because both my own i5 and the other i5 ran prime95 in kubuntu without any blackouts.
And the following conclusion is that both Z87 and Z97 were toasted. The Z97 was stored in working condition according to my coworker, but I didn't manage to even get video signal from it.

3 hard drives plus 1 SSD? 40W is worst case? Sorry but no way.

Thank you for the link, I was thinking about steady power usemore than spin-up. We were on the magnitude order at least.

It did work fine for months after the rip-out but I agree, over time and many expansion/contraction (heat-up/cool-down) cycles, possible damage could have expanded. Just not sure how likely that is - again, since it did work for months.

I'd say this is the more reasonable conclusion. No case shorts or whatever other idea I could think of. However, I find supsicious in some way that another Gigabyte from the same time was showing similar issues.

Test the RAM?

In theory it was something to test, but I don't think a RAM can cause a blackout unless it's shorted inside or something like that.

So, I am running the Asus H81M-K and 4690k right now and after installing drivers it runs the win7 as before. Yes, I know, worse board, DDR3 limited to 1600 MHz, no over clock, no VRM heatsinks... I am not going to squeeze this build anyway. And that weird low power consumption reports of the chip is fixed of course, the i5-4690k goes up to 80W running prime95.
I've bought some extra time to upgrade to something newer but now I can do a smoother migration.

Thank you everybody !