Components that I've seen typically fail and cause a short in a PSU are caps, transistors, and diodes too. There are a few YT videos which explain how to investigate PSU shorts. Needless to say, it's dangerous and you could cause a fire or zap yourself there with 110v. If your PSU is causing breakers to trip, then the short is in the high-voltage area of the board. A cap or a diode on the bridge rectifier maybe.

I don't think any of us would be able to pinpoint exactly which component to check for an issue like this. Best to post some pictures here so we can see what you see.

In general, though, the only path that can exist between ground and Neutral is through the Y2 safety-rated ceramic EMI/RFI suppressing caps. Typical configuration for those is one between Live and Ground and another between Neutral and Ground. There should also be a 3rd one, between negative rectified bus on the primary (connected to the negative lead of the primary cap) and ground/case/secondary-side ground (all the same thing, really.)

Now, proper safety-approved Y2 ceramic caps should go open-circuit in the event of over-voltage... but it's not 100% guaranteed. Still, if the one between Neutral and Ground does go short-circuit, this shouldn't be causing the PSU to shut-down or trip breakers (even GFCI ones.)

So I suspect you're measuring a fluke somewhere. The real problem, if any, is probably elsewhere. Thus, again, I think it would be best to post some pictures here and do some basic tests.

Basic test #1: plug PSU into the wall and measure the 5VSB (standby) voltage. Should be close to 5V +/-5% tops (i.e. 4.75 t0 5.25V).
Basic test #2 if #1 FAILS above: disconnect PSU from wall power, let it sit for a few minutes (for the HV caps to discharge), and check the resistance across the fuse on the input:
If test #2 passes above, check high voltage DC across the big cap(s) on the primary side (~170V for PSUs with 2x 200V caps or about 170V or 340V for PSUs with one large 400/420/450V capacitor.) If test #2 FAILs, check bridge rectifier and MOVs (if equipped, either across the 2x 200V caps or across Live-Neutral lines after the fuse typically.)

That's about as far as I can go for advice without any pictures and a little more information (do at least test #1 above).

Got some time to work on it.

momaka wrote on 2024-07-11, 21:42:

Basic test #1: plug PSU into the wall and measure the 5VSB (standby) voltage. Should be close to 5V +/-5% tops (i.e. 4.75 t0 5.2 […]
Show full quote

Basic test #1: plug PSU into the wall and measure the 5VSB (standby) voltage. Should be close to 5V +/-5% tops (i.e. 4.75 t0 5.25V).
Basic test #2 if #1 FAILS above: disconnect PSU from wall power, let it sit for a few minutes (for the HV caps to discharge), and check the resistance across the fuse on the input:
If test #2 passes above, check high voltage DC across the big cap(s) on the primary side (~170V for PSUs with 2x 200V caps or about 170V or 340V for PSUs with one large 400/420/450V capacitor.) If test #2 FAILs, check bridge rectifier and MOVs (if equipped, either across the 2x 200V caps or across Live-Neutral lines after the fuse typically.)

That's about as far as I can go for advice without any pictures and a little more information (do at least test #1 above).

Test 1 fails. The PSU doesn't generate any voltage (no PS_ON, 5VSB). The short has also went away after disassembly. Fuse seems okay (o.03 ohms). In the first picture it looks like there is a burnt trace, but there is still copper underneath. I am skeptical of measuring inside the PSU while theres live voltage going through it. I have read about common faults on this PSU, but I don't think the short would have been caused by the common faults. Any ideas?

You're gonna have to measure the diode bridge on input and the switching transistors on primary side, a short usually means some silicon has given up. The burned trace will help to find where the problem is.
Sometimes the filter capacitors go bad too but you have to desolder them to test them.
There is no need to do live tests, and without isolation transformer and actually knowing what you're doing it can be very dangerous.

momaka wrote on 2024-07-11, 21:42:

Still, if the one between Neutral and Ground does go short-circuit, this shouldn't be causing the PSU to shut-down or trip breakers (even GFCI ones.)

That's a common misconception. Shorts between neutral and ground do trip GFCI breakers. They don't cause the PSU to shut down or trip regular breakers, though. The GFCI is meant to monitor that all current that "enters the circuit" via the live wire also "leaves the circuit" via the neutral wire. As soon as there is a short between the neutral wire and the ground wire, some of the current that is supposed to "leave the circuit" via the neutral wire actually "leaves the circuit" via the earth wire. So there will be an imbalance between live and neutral which is supposed to trip the GFCI. The GFCI doesn't know the difference between current diverting to the ground wire and current diverting through your body into some other ground path, so it is entirely correct to trip in that case.

momaka wrote on 2024-07-11, 21:42:

So I suspect you're measuring a fluke somewhere. The real problem, if any, is probably elsewhere. Thus, again, I think it would be best to post some pictures here and do some basic tests.
If test #2 passes above, check high voltage DC across the big cap(s) on the primary side (~170V for PSUs with 2x 200V caps or about 170V or 340V for PSUs with one large 400/420/450V capacitor.) If test #2 FAILs, check bridge rectifier and MOVs (if equipped, either across the 2x 200V caps or across Live-Neutral lines after the fuse typically.)

Supplies with a 400V capacitor should usually have around 320-340V (bridge rectified 220-240V in Europe or voltage-doubled rectified 110V-120V in the US). If the capacitor is rated 420 or 450V, oftentimes you have a supply with active power-factor correction, and in that case, the target voltage on that cap is 380-400V.

kotel wrote on 2024-08-21, 08:19:

The short has also went away after disassembly.

Assuming there actually was a short (see last paragraph), this sounds like some part connected to neutral touching the grounded case. Disassembling would remove that short circuit. Check the neutral wire for puncture or for excessively long component lead ends on the solder side that might touch the case.

kotel wrote on 2024-08-21, 08:19:

Fuse seems okay (o.03 ohms).

This makes the hypothesis of failed semiconductors on the primary side unlikely. Failed semiconductors usually blow the fuse. OTOH, if the main breaker of your circuit is acting way faster than the fuse, there might be a chance of the fuse not being blown even if a primary semiconductor (bridge rectifier, high-voltage MOSFET) failed.

kotel wrote on 2024-08-21, 08:19:

In the first picture it looks like there is a burnt trace, but there is still copper underneath.

In case of doubt, measure. Are you talking about the trace in the top left of that picture? I don't think it looks "burnt", but the solder resist (the green stuff) has been scratched away. In a closed box, there shouldn't be anything scratching away solder resist, so should investigate whether the case is bent or something like that. As that trace is on the secondary side, damaged solder resist should not be able to trip a breaker (including GFCI ones). Anyway, if solder resist is required for the isolation to be sufficient, there is some root fault. Solder resist is not regarded as sufficiently robust to provide saftety-critical isolation.

kotel wrote on 2024-07-08, 09:21:

While back it would shutdown after 1 second when I shorted PS_ON with GND.

This might be an indication that the 5VSB supply is marginal and the extra load caused by shorting PS_ON (supplied from 5VSB) to GND causes it to fail.

kotel wrote on 2024-08-21, 08:19:

Test 1 fails. The PSU doesn't generate any voltage (no PS_ON, 5VSB).

This might be an indication that the marginal 5VSB supply is now entirely dead. The 5VSB supply is on the vertical PCB on the left in your second picture. On the mains side (the lower side), I see a 50V, 47µF capacitor that is used to supply the control chip. Failure of that capacitor is a standard cause for a supply to degrade gradually until it stops working completely.

kotel wrote on 2024-07-08, 09:21:

Measuring in the AC input, I found that the live wire had around 100mV to earth

Please explain what you mean by that. If you measured in voltage mode, this is completely fine and does not indicate a short. The meter indicates how much voltage is currently in the circuit. You measured residual charge of an interference suppression cap. On the other hand, if you measure in "diode check" or "continuity test" mode, the meter also displays a voltage, but in this case, it is the amount of voltage required to push a test current of about 1 to 10mA from one end to the other. The intended use case for that value is testing the voltage required to "turn on" diodes, which is about 0.6-0.7V for silicon diodes. If you measure 100mV "forward voltage" between live and earth (aka ground), something is very fishy and needs to be fixed for safe operation.

mkarcher wrote on 2024-08-21, 08:49:

Assuming there actually was a short (see last paragraph), this sounds like some part connected to neutral touching the grounded case. Disassembling would remove that short circuit. Check the neutral wire for puncture or for excessively long component lead ends on the solder side that might touch the case.

No punctures or leads that touch the case. The short also seems to be gone after I reassemble it fully.

mkarcher wrote on 2024-08-21, 08:49:

This makes the hypothesis of failed semiconductors on the primary side unlikely. Failed semiconductors usually blow the fuse. OTOH, if the main breaker of your circuit is acting way faster than the fuse, there might be a chance of the fuse not being blown even if a primary semiconductor (bridge rectifier, high-voltage MOSFET) failed.

Thats good to know.

mkarcher wrote on 2024-08-21, 08:49:

This might be an indication that the 5VSB supply is marginal and the extra load caused by shorting PS_ON (supplied from 5VSB) to GND causes it to fail.

Reading in the doc that was on badcaps it says that these PSU's would shutdown if there was no load after 1 sec, atleast from what i can remember since that page is now dead.

mkarcher wrote on 2024-08-21, 08:49:

kotel wrote on 2024-07-08, 09:21:

Measuring in the AC input, I found that the live wire had around 100mV to earth

Please explain what you mean by that. If you measured in voltage mode, this is completely fine and does not indicate a short. The meter indicates how much voltage is currently in the circuit. You measured residual charge of an interference suppression cap. On the other hand, if you measure in "diode check" or "continuity test" mode, the meter also displays a voltage, but in this case, it is the amount of voltage required to push a test current of about 1 to 10mA from one end to the other. The intended use case for that value is testing the voltage required to "turn on" diodes, which is about 0.6-0.7V for silicon diodes. If you measure 100mV "forward voltage" between live and earth (aka ground), something is very fishy and needs to be fixed for safe operation.

I measured in diode test. Sorry for not writing in what mode I measured.

Tiido wrote on 2024-08-21, 08:24:

You're gonna have to measure the diode bridge on input and the switching transistors on primary side, a short usually means some silicon has given up. The burned trace will help to find where the problem is.
Sometimes the filter capacitors go bad too but you have to desolder them to test them.
There is no need to do live tests, and without isolation transformer and actually knowing what you're doing it can be very dangerous.

Will try to measure and come back with the results.

Sorry for the loooooong time it took me to reply. Just got enough courage to tackle this one.
Regarding the common faults, I have tested most of the parts, besides Q2/3 and R11 and U1 (PWM chip, looks good so IG its not the culprit?(yes, I know it still can be bad)) because I couldn't find them. There look to be shorted pins on an transistor Q9 (IIRC), the two leftmost when looking from AC input to 3.3/5/12v power cables on the PG and PS_ON board. Rest of the Dxx transistors look to be fine.
Replaced those two 47uF 50v caps and one 10uF 50v cap on the 5VSB board. No bad diodes there too.
Haven't powered it on yet due to missing top cover and concerns for my safety.

Okay it looks like 3.3v and 5VSB are there. Next up find the cover (for safety), put load and power on.

Nope, still shutdowns after 1 sec even with 5 HDD's plugged in. D2/3 check out good. Any ideas?

Out of curiosity. Do you have the whole pc this psu comes from or just the psu ?

weedeewee wrote on 2024-12-19, 17:13:

Out of curiosity. Do you have the whole pc this psu comes from or just the psu ?

The whole pc. The board is also dead.

On another thought the MOSFET's on the secondary side might be shorted, although that is very unlikely...

Okay its back to breaking the main fire breaker, again. This time there was a bright spark coming out of the 230v connector. Just bloody great.

Okay I am back onto repairing this PSU. This time I have put duct tape under the primary arena of the PSU on the case and through the dim bulb tester.
After flipping the switch 5VSB (5.1v) and PS_ON (3.32v) started generating proper. Although the PSU still dies after a second when shorting PS_ON even with some load (1x 3.5" HDD)

The TO-220 and TO-264 transistors are fine, no shorts between legs. Same for the voltage rails to ground. Smaller TO-96 transistors also don't have any shorts on them.

I guess it's time to backfeed the voltage rails?

Okay just to confirm whether the polish forums statement that these PSU's need an mainboard to fuction and not just load I plugged in an sacrificial nforce board (from my first post) and.... IT LIVES!

And of course with a catch as always 🙁
It screeches worse while running than the P4SD board. No idea why. All of the filter caps were checked and ones above or near 20% were replaced, nothing was shorted.

Any ideas?

Turns out it was two shorted pins causing this on the UC3845B PWM ic. Weird since that wasn't there since the latest disassembly, but at least this didn't damage anything inside.
I will do more stress tests just as a sanity check so that this one won't blow up another board.

It might be worth cleaning the board thoroughly and checking for any weak solder joints.

Tbh I don't even think this one was screwed from the beginning. It 100% was the missing motherboard as load, but whatever. It seems to work.