It's been a while since I have written one of these computer troubleshooting posts. In fact, I think this is only the second and the last one was years ago (and still unresolved). These things can be so frustrating...
I have this machine that I built several years ago that, among other things, includes a Ryzen 5 3600 CPU and an nVidia RTX 3060 GPU. It is in an aluminum Lian-Li case that has been recycled for several builds over the years. I believe it started life as part of an Athlon XP based build so it is pretty old.
Anyway, while this machine worked well for a few years, it developed a very annoying problem. It would just randomly shut off at random times. Sometimes it would shut off before the boot process even completed (or sometimes even before post completed). Other times, it might be on for several hours before turning off. The problem got worse over time.
There are a few problems that might cause this. The first, and probably most likely, is temperature. Most remotely modern machines will shut down if they detect a CPU temp above safe limits. This is a nice feature that didn't always exist. I remember destroying a Duron CPU I bought at a computer show because the heatsink/fan they provided was not meant for that CPU. At least they provided a replacement.
Then next most likely cause is a bad power supply. A failing power supply can cause various instability issues so it is always suspect when your computer starts behaving strangely. In extreme cases (and especially with cheap power supplies), they can also take out other components when they fail.
Likewise, a bad motherboard or a motherboard with a failing component can cause random instability issues (Athlon XP/Pentium 4 era motherboard capacitor issues were notorious for causing stability problems). These are the worst because often you have to find them by eliminating all other possibilities first. The truth is, just about any failing piece of hardware can cause random stability issues but these seem to be the most common.
I eliminated temperature as a likely culprit pretty early on. When the machine was on long enough to check, temperatures seemed to be normal. All fans also seemed to be operating normally (well, the top blow-hole fan is kind of noisy but it was still spinning well).
Other problems are harder to eliminate unless you have spare parts to swap in. However, I tried reseating components, moving RAM around, unplugging drives, etc. None of these thing helped. Actually, after reseating components, things seemed to improve for a while but I think this was a false positive so to speak. The improvement didn't last.
Things finally got bad enough to make the machine unusable. It never stayed on for more than a few minutes anymore. Given the troubleshooting I had gone through, I had pretty much narrowed things down to either the power supply or the motherboard. I tested the power supply with a cheap power supply tester that I have and it tested ok (though sometimes things can be different under load). I didn't really think this was a power supply problem though. In my experience, power supplies usually fail in a way that either makes your computer freeze and/or prevents it from powering on at all and that wasn't what was happening here. Plus, it is a fairly high quality power supply that wasn't really that old.
Sadly, this meant the most likely culprit was the motherboard. However, I wasn't really ready to upgrade at this point. I have other machines though this is the newest and fastest.
I decided to take this machine out for one more round of troubleshooting. I was going to try swapping the GPU and the power supply even though I didn't really think that either of these things were the likely problem. At least I would be sure. It was at this time, before I have swapped anything, that I noticed a screen pop-up in Windows for a split second just before the machine turned itself off. It appeared to be the same screen that pops-up on your screen (depending on how you have things configured) that prompts you if you want to turn off the computer when you press the power button.
You can probably see where this is going. To make a long story a bit shorter, the problem ended up being the power switch itself. I tried reseating the switch into the button, unplugging and replugging the connector, etc. but wasn't able to improve things much. Until I can find a replacement, I have just unplugged the power switch from the motherboard header and moved the reset switch connector so that it now operates as the power button (most modern cases don't even seem to have reset switches anymore so no great loss). Problem solved.
I guess you learn something new every day. While I've experienced many failures over the years using and building computers, I had never seen a power switch go bad so it just didn't occur to me as something to check. However, it's something you should keep in mind if you are having random power off problems, especially if your case is older.
The above photo isn't mine but my case is identical except that mine has red lighting (primarily from LED fans) and I only have one drive in the bays, a single optical (DVD-RW) drive. I believe the model is the Lian-Li PC65B. I'll probably repurpose this case in the future for a retro Athlon XP, Athlon 64, or Pentium 4 machine. Except for the faulty power switch, mine is still in very good condition.
