[SOLVED] PC power supply short circuit

[Relayer]

Hello eagle1109,
The only way you're going to find the short is to eliminate one component at a time. But
only the components connected to the +12V rail.
Any resistors associated with the 12V's can be essentially ignored. What you're really looking for,
is a semi-conductor component. i.e. Rectifier, diodes, transistors and/or MOSFET's. Capacitors
and resistors are extremely unlikely to ever go short circuit.
You need to remove each semi-conductor component entirely, until the short is no longer there.
You should also check for the short on 200 ohm scale or higher on your DMM.
As its already been stated, using the buzzer side of your multi-meter isn't accurate at all.
Regards,
Relayer

 

[eagle1109]

Yes, I just was very busy the recent days and the PSU is on my desk, but I took out some parts, I desoldered aluminum capacitors and couple coils. I desoldered the coils just to isolate their coupling function.

But I'm looking for the main lines going out of the high frequency transformer to be the different power supply outputs, but I think it's not multiple lines. I think it's one main distribution point of +12V, and from that point it goes to three different lines, two for 3.3 and 5 volts and the third one which looks the big one is the +12V.

I also discovered that the two cards with the coils which are soldered vertically to the board are the 3.3 and 5 volts power regulator system.

 

[betwixt]

I think it's one main distribution point of +12V, and from that point it goes to three different lines, two for 3.3 and 5 volts and the third one which looks the big one is the +12V.

While not impossible to do it that way, it would be extremely inefficient and I doubt that method is used. For example, for every Amp drawn from the 3.3v line it would dissipate about 8W of heat instead of virtually none if the transformer supplies it ready to use.

What is frustrating is I could probably track the fault in a couple of minutes if it was on my workbench!

Brian.

 

[eagle1109]

While not impossible to do it that way, it would be extremely inefficient and I doubt that method is used. For example, for every Amp drawn from the 3.3v line it would dissipate about 8W of heat instead of virtually none if the transformer supplies it ready to use.

How did I understood it's done in this way is that the two little vertical boards for 3.3 and 5 volts regulators, the supply line is going through two on board coils in series with each of the 3.3 and 5 volts boards.

So I desoldered the coils and the I measured the continuity from the 12v line and the entry of each board and there is no sound which means the supply line which supply these two boards is open, that's how I knew that the 12v line is the distribution point and the only line coming out of the high frequency transformer. It's a clever way I guess, I don't know much about power supplies and how current is consumed from one distribution point or multiple lines from the high frequency transformer, but this method is new to me.

 

[eagle1109]

I found the faulty device, it's one the 4 output MOSFETs.



Pin 2 (in the middle) is connected to the MOSFET back, but pin 1 (on left) is shorted to pin 2 while it shouldn't.

I ordered a lot of these MOSFETs from Aliexpress I'm so happy that I could save my ~$100 PSU.

- - - Updated - - -

Hello eagle1109,
The only way you're going to find the short is to eliminate one component at a time. But
only the components connected to the +12V rail.
Any resistors associated with the 12V's can be essentially ignored. What you're really looking for,
is a semi-conductor component. i.e. Rectifier, diodes, transistors and/or MOSFET's. Capacitors
and resistors are extremely unlikely to ever go short circuit.
You need to remove each semi-conductor component entirely, until the short is no longer there.
You should also check for the short on 200 ohm scale or higher on your DMM.
As its already been stated, using the buzzer side of your multi-meter isn't accurate at all.
Regards,
Relayer

You absolutely right! It's the semi-conductor devices. I went for the coils and caps, but they were OK. It was one of the output MOSFETs on the +12V rail, but my question here is why MOSFETs on the output? I understood is that caps are the components on the outputs for clean DC.

- - - Updated - - -

I returned the coils and the caps and the short is gone That's absolutely amazing, now I know the fault is from the MOSFETs.

 

[pjmelect]

Are you sure the device that is shorted is a MOSFET ? The devices that look like that, that feed the different supply rails are double diodes not MOSFETS. The normal cause of these diodes to fail is due to insufficient air flow across them (the fan fails).

 

[Relayer]

Hello eagle1109,
I am very happy you've found the problem of the short on the +12V rail. :-D
Unfortunately you seemed to have ripped up the through-hole vias while
removing the MOSFET/Rectifier bank. See your modified picture below:



As shown in the pic, you can see the vias still stuck to a few component legs,
as shown by the arrows.
Before you're ready to place the new components in, you will have to scrape
some of the silk-screen mask on the component side of the PCB that attaches
to each individual leg of the new devices so it can connect to the solder side of
the PSU. There will be some legs that will not require this fix.
Therefore, remove just enough of the silk-screen that you can solder between
the top track and the leg of the device. This will allow the two sides to connect to
each other. Failing to do this will result in it not working, plus it could easily
damage the new parts when you fire it up.
Please let us know how you get on.
Regards,
Relayer

 

[betwixt]

What is the part number on the circled component?

Brian.

 

[eagle1109]

Are you sure the device that is shorted is a MOSFET ? The devices that look like that, that feed the different supply rails are double diodes not MOSFETS. The normal cause of these diodes to fail is due to insufficient air flow across them (the fan fails).

Yes, they are 023N04N power MOSFETs. Well, they 4 MOSFETs attached to that heat-sink.

But, I remember when the PSU started to fail while I was playing powerful games; like, Far Cry 4 and BF4, and also I have EVGA 780 classified which would draw a lot of current.

So, the PC started to shutdown while I'm playing, then after like minutes it work again then it took more time to work again, the last time it shutdown and I smelled like something burnt. And the PSU is very hot!

you will have to scrape some of the silk-screen mask on the component side of the PCB that attaches
to each individual leg of the new devices so it can connect to the solder side of
the PSU.

Therefore, remove just enough of the silk-screen that you can solder between
the top track and the leg of the device. This will allow the two sides to connect to
each other. Failing to do this will result in it not working, plus it could easily
damage the new parts when you fire it up.
Please let us know how you get on.
Regards,
Relayer

I understood what you mean, but I can do bridges with the solder wire. I think it would be hard for me to fix those little PCB parts and get them back. Bridging would work I hope.


Thank you

- - - Updated - - -

What is the part number on the circled component?

Brian.

All the 4 MOSFETs are 023N04N.