Dead Palit GTX 780

[TecZ]

Hello everyone!

Forums like these are rare these days so I'm glad to see this one still going strong!

About me: I work in IT but always had an interest in electronics and although I know some of the basics I have difficulty looking at some devices and knowing where and how to start troubleshooting.
For a long time I've lacked the tools but I've been investing in some basic items and I've slowly been repairing very basic components. I've got a few products that have stopped working over the years and I'd like to have some "fun" finding out what's wrong with them and expand my knowledge, even if I can't actually fix them.

Back to topic:
I've got given a dead Palit GTX 780, the owner said it stopped working while playing a game. It's completely unresponsive when connected to a PC, no fans, no image, nothing.
After reading online, the owner decided to try the oven method which doesn't seem to have done anything apart from melting the pcie power connector (still usable).

I've found a similar thread on your forum but didn't want to hijack it (https://www.eevblog.com/forum/repair/dead-graphics-card-780ti-dead-mosfet-perhaps/) so I've been following it and did some troubleshooting.

This is the card:
https://cdn.discordapp.com/attachments/755073723513569335/755696567146643506/20200916_084331.jpg

My conclusions so far:
L16 fuse seems to have failed (no continuity).
Q16 mosfet seems to be acting weird - how can I be sure? (https://drive.google.com/file/d/1xLn71d5f2lhZg4V4PU0ccLAS4AmwwXzL/view?usp=sharing)

I've tested a few other things that seemed fine but where would you guys start and how would you approach something like this?

Many thanks in advance!

[TecZ]

I've been doing a few more tests. Most capacitors seem fine but the top two seem to produce different values, they have the same reference though, is this normal?

https://drive.google.com/file/d/1pH05O9deAghu8OjiwCW8nlvRR7IqnosE/view?usp=sharing

[Psi]

Trying to measure things in circuit can be very confusing, even to experienced engineers.

[TecZ]

Thanks! I get the feeling that without proper schematics of what you are working on, everything is a bit of reverse engineering.
I'm guessing that as you find more about a specific type of device allied with past experiences things become more clear.

[TecZ]

I've done the test as described on the old thread which leads me to conclude that "second and third transistor from the top are 100% fine, those are powered from PCIE" -Rasz





I'm now trying to test the Mosfets. However, I'm getting different behaviours between the two of them when I run the test as described:



Top Mosfet:
Red = .537  .045 .045 .045
Green =  fluctuates until 0L (disconnection?)
Blue = .609  .394 .394 .394

Bottom Mosfet:
Red = .534 .045 .045 .045
Green =  0.00 (continuity?)
Blue = .609  .045 .045 .045

I've researched to see how these operate and I've got a basic sense of the different parts, Drain, Source, Gate but I don't understand enough to know which one (if any) is faulty.
Any ideas based on these values or any other tests I can do?

[Berni]

These are mostly part of the switchmode converters turn 12V into whatever voltages it needs at very high currents.

Some failures in this area can send 12V into things like the Vcore that is only supposed to have around 1.2V. If this happened then the card is beyond economical repair and is destined for the trash. If you are lucky then the mosfets might have died in a way that shorted the rail to ground. In that case its possible it just needs a new transistor. Its also possible the root cause of the failure is the drive circuitry, in that case a new mosfet would blow again instantly.

In most cases measuring 0 Ohm across a MOSFET means its dead, but its possible you are also measuring something else in the circuit. First way to test is to swap probes around in case of phantom voltages. But to be sure you would have to unsolder it and test it out of cirucit. Also lets you mesure the cirucit itself

[TecZ]

These are mostly part of the switchmode converters turn 12V into whatever voltages it needs at very high currents.

Thanks! This may seem obvious to many but it's the kind of confirmation that is really useful to me as a beginner in electronics.

Some failures in this area can send 12V into things like the Vcore that is only supposed to have around 1.2V. If this happened then the card is beyond economical repair and is destined for the trash. If you are lucky then the mosfets might have died in a way that shorted the rail to ground. In that case its possible it just needs a new transistor. Its also possible the root cause of the failure is the drive circuitry, in that case a new mosfet would blow again instantly.

I'm confident this is what happened, judging by the blown fuse. I'm not afraid of putting it in the bin but it would be fun to fix if viable, just need to track from the fuse back to the potential causes but not sure where to start on these multi layer PCBs as I can't see the tracks.

I've just confirmed the probe swab, same result of 0 on the bottom mosfet (green line).

Would removing the mosfet help troubleshooting further? or is worth checking something else first to see if it is even worth doing that?