ATX computer power supply repair

I bought a ATX computer power supply last year to make a power supply for hobby electronics. And while using it, I got curious of how much -12V rail would push. It said 0.3A, but I tested it with a multimeter anyways, and got around 6~8A. It wasn't stable and after 2~3 seconds, it made some popping sound and shut down.
The symptoms are : When you give power to the power supply and plug the power cable, you get power for the 5V Standby rail(which should be on all the time when the PSU gets power).
When you turn on the PSU, you get voltages for all the rails(+12V, -12V, +5V, +3.3V) for less than a second and everything goes down to 0V except for the 5V stand by, 5V power good(it powers on when PSU is switched on), and 5V

Now, I just took it apart and took a closer look inside it. None of the components seem burnt or anything horrible. They all seem good on my eyes. There is a NTC 5D-9 power thermistor which looks good on my eyes, but is currently the most suspicious component for now.(I get 0 ohms with my multimeter, but that reading is when it's connected to the PCB)

What do you think is the problem?? I haven't given the PSU and power since last year so the capacitors are fully discharged and I checked it again by shorting them with a screwdriver. It's not a expensive or premium power supply. I bought it for 15 bucks and while I can just make a new one, I'd really like to repair the one I have now. Any ideas??

Almost certainly you have burned out the rectifier on the -12V rail. It has probably gone short circuit and the extra load on the supply it is creating is shutting the rest of it down.

Follow the -12V wire back to the PSU. In most cases it will lead to a capacitor to ground and a wire-wound choke. On the other side of the choke will be a second capacitor and a diode. It's that diode that has probably died. Both the capacitors will be electrolytic types in upright cans and the diode will probably be either a surface mount one or in a standard black wire ended package. The capacitors have probably survived but don't keep switching on as the shorted diode will cause them to be overloaded. If you do have to change them, use ones with the same value and voltage rating and ensure they are 'low ESR' types.

Try to use the same type of diode. A normal silicon rectifier (1N400x type) will NOT work in that position, it has to be a Schottky or fast recovery type or it will overheat even with no load.

Brian.

If the PS died with a popping sound, there will be "some" visible evidences. On this side (delivery) of the PS, there are only diodes and capacitors and inductors. As pointed out above (#2), the PS has failed with a great short and you need to debug the -12V part only.

What type of connector and no of pins your PSU have for DC?

One of them may be 'PWR GOOD' signal.
If so , you must appropriately keep it in the required level.

srizbf said:

What type of connector and no of pins your PSU have for DC?

One of them may be 'PWR GOOD' signal.
If so , you must appropriately keep it in the required level.

Click to expand...

Ummm, I don't get what you're saying, sorry.
The PWR GOOD signal is okay. It spits out 5V when I turn the switch on but even when the PSU isn't outputting any current due to the short.

- - - Updated - - -

c_mitra said:

If the PS died with a popping sound, there will be "some" visible evidences. On this side (delivery) of the PS, there are only diodes and capacitors and inductors. As pointed out above (#2), the PS has failed with a great short and you need to debug the -12V part only.

Click to expand...

Popping sounds, but no visible evidence.. That was a bit suspicous to me too. I just took a look again and notice a diode is broken. I didn't notice that since it was hidden under a big inductor. I think there will be more components broken inside but I'll have to desolder all 4 enameled copper wires in order to remove the inductor. Damn it

PS_On is typed(wrongly) as PWR_GOOD.

As the damaged components surface now , it is not a point to check.