Dell 275W PSU all output rails are shorted even taking all MOSFETs out the PCB

[eagle1109]

Hi,

I started working on this Dell 275W PSU that is dedicated for I guess OptiPlex 780 or 7010 desktops.

1. First I measured the output rails with the continuity and all measured shorted.
2. Then I took out all output MOSFETs, then checked output rails and same problem.
3. So I though it must be the HOT side power MOSFET, so I took out the whole semiconductor group with the heatsink and still the same problem !
4. I tested the power switches on the heatsink, and all measure OK.

Now the components that measure short after I took out all the power switches:
1. These two diodes marked in red.
2. All output capacitors.

What you think guys is the problem ?

This is the PSU:

 

[eagle1109]

Update:

I took out all the output caps, only the short on 12V went away, but still on 5 and 3.3V rails.

What you think is the problem ?

 

[FvM]

If you have removed most components, you should be able to see the short. If not, it's possible to trace the location with a current source (bench top power supply) and a sensitive volt meter by measuring the voltage drop along the current path.

 

[BradtheRad]

Your tenacity is commendable. Lacking a schematic it's astounding you got this far. I imagine I'd lack hope of reassembling the entire unit.

It sounds as though all the supplies went bad, suggesting failure of a component at the heart of operation...
But then there surely are additional components downstream producing each individual supply. And it seems unlikely so many components went short circuit all together.

However a single severe over-voltage surge could travel to the output capacitor of every supply. Can you locate the 5v & 3.3v smoothing caps, remove and test them?

 

[betwixt]

I think you may be repeating a mistake made by many people. The rectifiers (marked with 'D') on the PCB will have a low resistance transformer winding across them so it is quite normal to measure them as short circuit while in the PCB. The only way to test them is out of circuit.

It is also normal to measure a low resistance, although not zero Ohms, across the 3.3V and 5V outputs because they normally have some resistance added to ensure a minimum load current is present. One or both of them will also have some components to drive the opto-couplers IC1 and IC6 which will reduce the apparent output resistance.

If removing the capacitors on the 12V output took the short away, that may be the only faulty part. Follow the 12V output wire back and you will probably find a capacitor, an inductor and another capacitor. Change both the capacitors and try it again. Note that it may not work if you simply leave the capacitors out of circuit because the ripple on the feedback path may be too much and cause a shutdown.

Brian.

 

[eagle1109]

it's possible to trace the location with a current source (bench top power supply) and a sensitive volt meter by measuring the voltage drop along the current path.

I think you mean current injection, I learned about this technique and I'm interested to use it.

I have a bench power supply, I might try to apply this method but the problem is that where I start to inject the current ? Because I don't have the schematic, so I might continue to just take off the components until I find the faulty one.

I imagine I'd lack hope of reassembling the entire unit.

I recorded them with the component value and its number on the PCB, nice thing that Dell put the parts' names on the PCB, otherwise things could be more difficult and I may not be interested to work on such unit.

suggesting failure of a component at the heart of operation

I checked the SMPS switcher but it didn't measure short. I'm thinking of the PFC.

I'm going to take off the components one by one until I find the faulty one.

Can you locate the 5v & 3.3v smoothing caps, remove and test them?

That's what I did in the picture in #2. If you check the picture in #1, the caps on the left are the output caps, in the 2nd picture I took them all off.

I tested them and they all charge up with the ohmmeter. But there's one that's a little bit bulged but when I tested it, it still works.

I think you may be repeating a mistake made by many people. The rectifiers (marked with 'D') on the PCB will have a low resistance transformer winding across them so it is quite normal to measure them as short circuit while in the PCB. The only way to test them is out of circuit.

Yep, that's what happened and I took them off, tested them and they are working.

It is also normal to measure a low resistance, although not zero Ohms, across the 3.3V and 5V outputs

That's still a not clear area for me, I just now checked my Corsair PSU that's on the shelf for months and the output rails measure very high resistance from hundreds kilo ohms to mega ohms.

And when I short them out, and start to measure them again, the resistance is increasing, meaning the output caps are charging.

Now I'm starting to connect some thoughts; the last thing I tested on the faulty PSU when I took all the output caps off, the 12V short went away and measures like 143 ohms, but the 3.3V and 5V are shorted out.

So if the output power MOSFETs are off the PCB, then I'm thinking of the HFT that is not likely to be broken but I might take it off eventually to trace the source of the shorts, but I also might take all the components before taking the main HFT.

If removing the capacitors on the 12V output took the short away, that may be the only faulty part.

Only the short on the 12V rail went away but still on the 3.3V and 5V rails.

I'm planning to take everything off until the short on the output rails is fixed, the PSU is considered out of service and going to be thrown anyway, and I like to fix them with the good protection of using the lamp limiter so I feel safe not to have explosions from failed repair attempts.

Note that it may not work if you simply leave the capacitors out of circuit because the ripple on the feedback path may be too much and cause a shutdown.

I won't do that, when I find the faulty component, I will order a replacement, and then put everything back to its place and run the PSU.

Of course I'm using the lamp limiter, in case, if I didn't do the repair process correctly.

 

[FvM]

When I try to find a non-trivial fault on a similar PCA, I start with extracting a schematic, at least partially.

I would use the output cables for current injection, e.g. 1A, voltage limited to 0.5 or 1V, presumed there's a real short.

 

[eagle1109]

Hi,

I checked the board today and marked some notes:

1. I thought the semiconductor devices attached to the heatsink are power MOSFETs but they turned to be Schottky diodes.
2. Every anode of the Schottky diodes is shorted along all the 4 Schottky diodes parts to the GND with less than 1ohm resistance, is that normal ?
3. The cathode has some resistance and with the color of the 5, 3.3 and 12 V rails I concluded which Schottky diode serves which voltage.
4. One difference with the right most Schottky diode A2 marked with yellow circle, which has a resistance like the 5V rail of 33ohms. I don't know why.
5. The left most one has a cathode resistance to the ground of 15kohms, I think it's attached to the HFT for the standby voltage I guess I'm not sure, but with the 15kohms, I don't know which voltage it serves.


Anyway, this is my check up for today, I hope next time to discover the problem.