I had a 2007 era Corsair HX 620 slowly kill two boards, but not in the way you'd expect a PSU to take out a motherboard.

In both cases, the motherboards slowly started failing, piece by piece, in exactly the same way. It around two months for each board to completely fail after the first symptoms started appearing. First, the NIC would go, followed by a few of the USB ports. Then the PCIe x1 slot that hosted the sound card would stop working. Eventually, the boards would stop POSTing.

The first time this happened, I didn't even suspect the PSU (I mean, who would? The way it failed, I assumed it was some chipset-related fault)! So I replaced the motherboard--an old Asus LGA1155 model of some sort--with a Gigabyte 1155 board. The Gigabyte worked well for around 6 months...and then the same series of failures started occurring. And get this: a few days before it completely stopped working, I started getting Windows activation errors.

I finally tested the power supply after that second board stopped POSTing, and indeed, one of the output rails was just all over the map. Curious, I hooked up the Corsair to an old S478 board that I didn't give a damn about. That board wouldn't even start immediately; the CPU fan would merely twitch. I think I had to jab the power button 5 or 6 times before it finally roared to life. Once it did get going, it had no problem booting into Windows, though who knows how long that would've lasted. Hooked up to a known good PSU, the S478 board would always fire up on the first try.

On the same note, I've once debugged a client's computer which apparently had a failing HDD (MHDD would show countless bad and slow sectors during the surface scan).
So, naturally, I replaced the HDD with a new one, expecting a quick Windows XP reinstall. You could imagine my surprise when I started experiencing the same problems that were indicative of a bad HDD.
So I ran a surface scan on the new disk and had the same results (bad & very slow sectors). IIRC, I think I even tried a third disk with the same symptoms. At that point I started suspecting the motherboard's IDE controller.
Anyway, long story short, after I basically eliminated all possible causes, I finally replaced the PSU and... lo and behold, all disks (even the client's original disk) worked perfectly, passed the surface scan without a hitch.
This was probably the strangest PSU related problem that I ever encountered...

Yep, a relative's (at least) 10-year old no-name 450W PSU died and took out an Asus Socket AM2+ motherboard but everything else, CPU, RAM, storage, survived as they worked after replacing the PSU and motherboard. Apparently they could smell "something burning" but there was no visible smoke or flames.
Previously I've witnessed the magic smoke being released from the PSU of a ~6 year old IBM PC 5150 and my own ~2 year old no-name 300W PSU used with a Slot 1 motherboard when it was around 8-9 years old and no longer in daily use. In both cases, everything survived, including the PSU after a re-cap.

TheMobRules wrote on 2021-03-16, 18:05:

The only "common" cases that I know of power supplies killing PC hardware is due to some dodgy ATX units where the +5VSB line wo […]
Show full quote

The only "common" cases that I know of power supplies killing PC hardware is due to some dodgy ATX units where the +5VSB line would fail and kill components that remain powered during standby (such as RAM and some other chips in the motherboard). That's why I'm always puzzled about the usual perception on this forum that AT units are "much more dangerous" or "ticking time bombs" even though they don't even have a standby line.

Rogue +5VSB is the #1 reason of PC hardware getting killed on PSU failure, and other than that I feel the emphasis should be put on "well built" vs "cheap garbage" rather than "new" vs "old". Any PSU can fail, but quality ones have the proper protections to avoid damage to other components, whether they are new or old.

I've heard the concern about AT power supplies being "time bombs" many times in Phil's Computer Lab videos, so given how popular those are I think it kind of propagates to forums like this one. The problem is that he never goes into any technical depth as to why they are inherently more unsafe than ATX other than "they're old".

Personally I also prefer using (somewhat) period accurate PSUs for several reasons:

• Cost (I'm not going to spend on a new Corsair 600W monster + AT/ATX adapter + SATA-Molex adapter(s) each time I want to build a 486 or a Pentium 1)
• Compatibility (enough amps in +5V/+3.3V for hungry Athlons and no cross-loading issues due to low 12V consumption of older builds)
• Looks (this is personal, but IMO black PSUs look like ass in old cases, also the big fan on the bottom can end up blocked by flat cable spaghetti in small cases)

Refurbishing old power supplies is a great skill to have in the retro world, and it's usually a very simple job to do: replace fan + replace caps.

Regarding ESD, it's very real, I think that fact cannot be argued. That said, I usually don't do anything to prevent it other than storing hardware in anti-static bags.

Sums it up nicely

Not a single case of damaged hardware by using old PSUs here, but I had fried at least a few weaker PSUs by overloading them in the early days, when video cards began to require an external power source, but those failures were minor and incapable of damaging anything further. I don't know how a PSU has to fail spontaneusly that it would bring down anything more with it, though there was at least one case on Vogons where a faulty standby circuitry in ATX PSU had fried a motherboard. Cheap PSUs is the real threat to hardware IMHO, not just old PSUs in general, I have one super high tech XT era Mitsubishi PSU (1985) that in quality rivals modern ATX PSUs, so an old PSU doesn't automatically equate to being worse, compared with modern, cheaper-built units.

by the time video cards began to require external power pretty much any PSU should have had proper overcurrent protection though and just shut down ^^... so I guess you were using some of those "dodgy" ATX PSUs after all.

Pretty much, it was an older 300W PSU with a weak 12V line, IIRC the graphics card I was using back then required to have at least a 350W unit.

kixs wrote on 2021-03-16, 13:30:

I had a few LC Green 550W bang up and didn't break anything else. Then I bought Antec PSU - it still works fine 15 years later...

That is actually surprising as not so many years ago I was only using Antec power supplies until they all started having caps go bad in them. I recapped a few of them which were still working last I knew but I won't buy Antec PSUs anymore because of it.

After 3-4 different Antec PSUs all having the same problem, I decided it wasn't worth it.

TheMobRules wrote on 2021-03-16, 18:05:

The only "common" cases that I know of power supplies killing PC hardware is due to some dodgy ATX units where the +5VSB line wo […]
Show full quote

The only "common" cases that I know of power supplies killing PC hardware is due to some dodgy ATX units where the +5VSB line would fail and kill components that remain powered during standby (such as RAM and some other chips in the motherboard). That's why I'm always puzzled about the usual perception on this forum that AT units are "much more dangerous" or "ticking time bombs" even though they don't even have a standby line.

Rogue +5VSB is the #1 reason of PC hardware getting killed on PSU failure, and other than that I feel the emphasis should be put on "well built" vs "cheap garbage" rather than "new" vs "old". Any PSU can fail, but quality ones have the proper protections to avoid damage to other components, whether they are new or old.

I've heard the concern about AT power supplies being "time bombs" many times in Phil's Computer Lab videos, so given how popular those are I think it kind of propagates to forums like this one. The problem is that he never goes into any technical depth as to why they are inherently more unsafe than ATX other than "they're old".

Personally I also prefer using (somewhat) period accurate PSUs for several reasons:

• Cost (I'm not going to spend on a new Corsair 600W monster + AT/ATX adapter + SATA-Molex adapter(s) each time I want to build a 486 or a Pentium 1)
• Compatibility (enough amps in +5V/+3.3V for hungry Athlons and no cross-loading issues due to low 12V consumption of older builds)
• Looks (this is personal, but IMO black PSUs look like ass in old cases, also the big fan on the bottom can end up blocked by flat cable spaghetti in small cases)

Refurbishing old power supplies is a great skill to have in the retro world, and it's usually a very simple job to do: replace fan + replace caps.

Regarding ESD, it's very real, I think that fact cannot be argued. That said, I usually don't do anything to prevent it other than storing hardware in anti-static bags.

I agree with what you say. Especially when it comes to Phil's statements. I also noticed that everyone who had a PSU fail and cause damage in this thread was of an ATX variant (often low-end). Are we kind of concluding that the risk with an AT power supply is alot less?

cyclone3d wrote on 2021-03-17, 03:43:

kixs wrote on 2021-03-16, 13:30:

I had a few LC Green 550W bang up and didn't break anything else. Then I bought Antec PSU - it still works fine 15 years later...

That is actually surprising as not so many years ago I was only using Antec power supplies until they all started having caps go bad in them. I recapped a few of them which were still working last I knew but I won't buy Antec PSUs anymore because of it.

After 3-4 different Antec PSUs all having the same problem, I decided it wasn't worth it.

Cheers for the headup on Antec PSU's. I have one from about 2003 installed in my Slot 1 machine. Definitely need to open it up and check the caps!

bloodem wrote on 2021-03-16, 19:17:

On the same note, I've once debugged a client's computer which apparently had a failing HDD (MHDD would show countless bad and s […]
Show full quote

On the same note, I've once debugged a client's computer which apparently had a failing HDD (MHDD would show countless bad and slow sectors during the surface scan).
So, naturally, I replaced the HDD with a new one, expecting a quick Windows XP reinstall. You could imagine my surprise when I started experiencing the same problems that were indicative of a bad HDD.
So I ran a surface scan on the new disk and had the same results (bad & very slow sectors). IIRC, I think I even tried a third disk with the same symptoms. At that point I started suspecting the motherboard's IDE controller.
Anyway, long story short, after I basically eliminated all possible causes, I finally replaced the PSU and... lo and behold, all disks (even the client's original disk) worked perfectly, passed the surface scan without a hitch.
This was probably the strangest PSU related problem that I ever encountered...

I had a similar occurrence with a hard drive failing after a week in use on an older pentium computer, AT PSU. returned under warranty, got a new one, failed again, same procedure, another failure.
Eventually whole computer gone to the shop for a week, store wanted to verify nothing was wrong with the computer, and had run out of the quantum 3.5 hdd's so installed a quantum bigfoot. it ran for a week in the store.
When I got it back, I added another powersupply since that was the only thing I hadn't ruled out myself. So now I had one psu for the mainboard and io cards and one for the drives 😀
served me well for a few more years until I upgraded to a Pentium II. This was 25years ago.
Both psus still work today, though they're hardly being used, maybe once or twice a year and not for long and haven't yet been recapped.

Only once. I bought a desktop from recycling where someone had switched the voltage selector from 230->110 and I forgot to check, so it was bang and smoke. Can't remember if it actually damaged the board or if it was otherwise busted, but I don't think I ever got it working again.

Sure different PSU were gone in the past computer usage and test and some with bad effects like flames or lights or sounds. Only in one case I remember a 550W low-brand low-end PSU powering something like an Athlon probably I can't remember but I do remember measuring the voltages when it was brand new, it gave +17 volts on the +12 volts line and something like +7 volts on the +5 volts line, never investigated why or how.. first boot the mechanical hard disk was gone with a bad sound effect and the vga smelled badly (I don't remember which one it was.. probably a Radeon 9500 with external floppy connector).
The seller in the shop said they tested it and it was ok.. I think it was not and I doubt I didn't set up a voltage tester correctly..

cyclone3d wrote on 2021-03-17, 03:43:

kixs wrote on 2021-03-16, 13:30:

I had a few LC Green 550W bang up and didn't break anything else. Then I bought Antec PSU - it still works fine 15 years later...

That is actually surprising as not so many years ago I was only using Antec power supplies until they all started having caps go bad in them. I recapped a few of them which were still working last I knew but I won't buy Antec PSUs anymore because of it.

After 3-4 different Antec PSUs all having the same problem, I decided it wasn't worth it.

I could be lucky - it's Antec HX620W.

Interesting thread. I had only one instance of a failed PSU, it was in 2006 shortly after I upgraded my S939 Opteron 165 rig with a Radeon x1900xt. The PSU that failed was an OCZ Powerstream 420W. Luckily there was no fireworks and no other component died, simply one day the PSU just refused to power on.

That PSU got replaced with a Corsair HX520, which powered my main rig until late 2020, seeing almost daily use, and did not give me any problems at all. I will probably continue to use it in some retro build, since it has a decent 5v output.

Many years ago I ran a dual Opteron rig with a HD4870x2 card. The Power supply was some no name gamer model purchased from Compgeeks and was supposed to be 800W+. The GPU needed 1x6 pin and 1x8 pin power connectors but the PS only had 2x6 pin so I purchased a 6 to 8 pin adapter. System was running doing a benchmark and the PS literally blew and the computer turned off. I looked inside the PS and some chips had exploded. I purchased a Corsair 750W model and popped it into that machine and to my surprise everything worked fine.

I think most any modern design will probably shut off before taking out components unless it is something made in China that the Chinese won't even use and ship to North Korea.

Sometimes I snag a rig somebody found and gave to me with a dead PS and all it takes is a capacitor or two and it runs again. Some vintage HP/Compaq and DELL OEM supplies like to die for some reason and they are proprietary.

I understand the points people made about ESD but I’ve fixed to many things that were literally left out in the rain or at a dump to really be concerned with it personally. I also have and will probably always sit in my carpet putting my stuff together. The only time I sit at my workbench is for soldering and sometimes I still sit legs crossed on the floor 🤣! I also vacuum my PCs with a straight shop vac and blow things out with a normal compressor that has a filter tip on it. Some of this stuff can be filled with various bugs, leaves etc. I don’t want that in my house! Into the sink dirty parts go and I wash them with dawn and water then clean with rubbing alcohol. I know I’m pretty abusive to stuff but most of the things I buy and fix are in really bad shape

I can't remember ever having had a working PSU fail on me or break other components. I have had parts breaking but all of these would usually have a (sometimes clearly) external cause.

Having said that, I have had PSUs that were defective when I got them. I always do a visual inspection of any non-brand new PSU before I put it to use and have found plenty with some kind of defect (usually the caps were clearly swollen or leaking and in a few cases something had actually burned before I got that PSU).
I always tended to use PSUs of at least decent quality, even from what are considered questionable brands or questionably build PSUs like a brand new Tagan I used for several years in my Barton 3200+ system or having used some kind of HEC unit(?) which I ghettomodded a new exhaust fan as the original fan didn't work anymore (I routed the power for that fan from outside the case actually and didn't use the internal PSU fan connector for some reason).

Only when my hobby was very very new, I briefly used a gutless wonder but only for a short time and it was also brand new. And it weigh like half of any contemporary FSP unit weighted back then.

EDIT: A friend of mine has had 2 PSUs fail. One was some OCZ PSU which had (iirc) failed 5v rail (and I don't mean the 5vsb, but the real 5v rail. Mind you I never got to test this unit personally but this friend is reasonably proficient at building PCs. Only ever time I heard of any PSU having a 5v rail fail). The other was an old Antec PSU which had faithfully served for many years until it failed due to its dust filter having clogged up which caused the unit to starve of fresh air and probably the heat killed it. Iirc both these PSUs did not kill any other components.
Otoh, another friend of mine had used some cheaper PSU for his Athlon XP 2600+ and when it died, it fried the mobo (it had visual burn marks on some of the power connector pins at the motherboard).

Anecdotally it seems like having a PSU fail and definitively take out other components is rather uncommon. What I'd like to know is how to tell the difference between a well built unit and a garbage one, regardless of era - with a high level of certainty - without becoming an actual electrical engineer. I know you can look for bad caps, and I've heard weight can be an indicator. Some brands are trusted more than others in general, but all of them have their hits and misses. What else is there to look for?

Take a look at the big PSU build quality thread (I cant remember the name) and check out some posts. Super informative and in depth, then check out some Johnny Guru psu reviews, especially some older PSU reviews with internal breakdowns and hardware tests are really handy to get a decent idea of components and what they are rated for.