The bigger caps die of heat near the CPU if they aren't of top quality and durability, and sometimes near RAM or the first PCIe slot. The small caps and the more distant from CPU rarely fail.

Bad ones can typically be replaced on boards that have good reason to salvage. I've saved at least a couple of dozen boards that suffered the bad cap plague that began escalating around 20 years ago, and put Abit and Soyo out of business.

Odds of finding bad ones in the PSU are considerably higher than even the older motherboards without polys. Lots of PSUs got made with custom caps squished close together that when they go bad, no equivalent replacements that fit can be found. IIRC, all my Antec Earthwatts, 4 or maybe 5, had to have caps replaced, except the one that quit so soon I got Newegg to replace it during the no questions asked return period.

2 members found this post helpful.

Yes, it was the antec earthwatts that failed, but only that one series, they quickly remedied the situation, all mine were replaced no questions asked on warranty, and replaced with better units. That was a while ago though.

As an aside, I've over the past year plus been experiencing similar failures with usb 3 controller cards, all replaced on warranty until I told them to stop sending me replacements because it was obvious that an entire production run of usb 3 circuits were bad, those failures did destabilize the system too, but not quite in the way the OP has experienced.

1 members found this post helpful.

As an ex-hardware head who is a little out of touch, which 'poly' are you referring to?

1 members found this post helpful.

1 members found this post helpful.

Oh, yeah, Tantalum I knew; Nobium I had met, but didn't have much detail on.

1 members found this post helpful.

I’ll be replacing my own PSU any day now; my stability issues have gotten better, but they’re still there. And the crashes always happen when, say, DOOM Eternal gets really fun.

The PSU’s as old as the Haswell CPU it’s powering.

1 members found this post helpful.

Sorry for such a delay, but I wanted to give an update since I am 99% sure I found the source of the issue: Failing USB 3.0 controller/bus on the motherboard.

1. I examined the motherboard and didn't see any swollen capacitors.
2. I tried replacing the PSU because it's the cheap option and a 10 year old PSU is just asking to lose the entire system when it finally croaks.
3. While replacing the PSU, I noticed that I wasn't using one of the available USB 3.0 headers so I wired up some extra ports. This was the final clue, since I noticed that a few minutes after boot the xhci modules would complain to dmesg about the controller failing to respond.
4. I unplugged the extra USB 3 ports from the header and carried on with the new PSU, only to have yet another series of crashes just like before.
5. I then removed all USB 3 devices and only used USB 2 just to see if one of the devices didn't play nice with USB 3: More crashes.

After disabling the USB 3 controller in the BIOS, I haven't had a single crash, so I'm thinking that the controller is the culprit.

Good to hear that you found the problem. With quality tanking these days on most things I recommend replacing it with a Gygabite motherboard unless you want problems in the near future.

1 members found this post helpful.

On a tangent here. but due to design considerations, there's no real need to fear a psu taking out a motherboard.

The old, standard single-line design could do that all right. But pc supplies have a transformer acting as inductor/transformer allowing several ± lines at the same time. The failure mode of that design is to go to 0V.

1 members found this post helpful.

Make sure your power bar is supplying enough power. They’re not all the same (they have different joules ratings), and plugging one power bar into another is bad.

A UPS is another (more expensive) way to address this.