Two different graphics cards with continuity between the PCI-E connector pins

[SuperDuty]

Hi guys,

I have a Manli GeForce 9600GT and a Biostar HD4850 graphics cards that have both the same problem : they short the PSU as soon as the PCI-E connector is plugged.
Now upon inspection, I've noticed that there is continuity between all the pins of the PCI-E connector on both cards between the GND and +12V. I'm pretty sure it's not supposed to be this way as I have made the same test on my other video cards with a PCI-E connector and none of them show continuity between the GND and +12V pins. And that's on at least ten other video cards, including two other 9600GT and two HD4850.
There is no sign of damage anywhere.

I've checked all mosfets and caps, they're all good. Could this be internal to the board itself ? Two layers could be in contact ?
What could I check now ?

[SuperDuty]

Re: Two different graphics cards with continuity between the PCI-E connector pins

Bump !

[mariushm]

Re: Two different graphics cards with continuity between the PCI-E connector pins

Get an adjustable power supply, set it to a low voltage and limit current to 0.5-1a or so ... plug it in and see what gets hotter?

There was another method, if you have a high count multimeter, you could actually measure the resistance gradually and when resistance changes you know where the short is

See http://www.youtube.com/watch?v=JAOV_erLpYw

[SuperDuty]

Re: Two different graphics cards with continuity between the PCI-E connector pins

Thanks, it's going to be a tough one with my tools. I don't have an adjustable power supply. I'll go check if I can make one with a spare ATX PSU.

As for checking manually with a multimeter, I doubt mine is precise enough for that kind of low resistance values.

[Th3_uN1Qu3]

Re: Two different graphics cards with continuity between the PCI-E connector pins

Most likely shorted MOSFETs in the VRM. If you don't have the tools to test, you'll just have to pull all of 'em off and see which one is short.

[SuperDuty]

Re: Two different graphics cards with continuity between the PCI-E connector pins

Well I followed your advice and I've rechecked all the mosfets again...

On the 9600GT, I HAD replaced a shorted mosfet prior to creating this thread, and I was surprised it didn't change the issue. Happened I had replaced a bad mosfet with...another bad mosfet. I guess I had damaged it when soldering. Once removed, the card displayed a nice boot screen ^^

As for the HD4850, I did the same and...found a bad mosfet. I had checked them all though, but I think I just used the "beeper" of the multimeter. This time I READ what was written and found one was showing about a hundread ohms where the others next to it were showing no continuity. I removed it, plugged the card in and again, perfect boot up screen


Thanks guys !


So it makes four cards out of six from my last lot of dead cards that have been revived, the HD5750 works but the memory is faulty, I would need a BGA station to repair that, and the HD6770 won't do a thing. The GPU doesn't even heat up. The fan spins but that's it.

[momaka]

Re: Two different graphics cards with continuity between the PCI-E connector pins

and the HD6770 won't do a thing. The GPU doesn't even heat up. The fan spins but that's it.

Check GPU core voltage and RAM voltage. Post some pictures of the bare board (i.e. heat sink removed) so we can see layout.