RX 470 OC No voltage on the card and zero ohm on vcore

stonehencs

n00b
Joined
Dec 23, 2021
Messages
6
Hi

I have a Sapphire RX 470 4GB OC which is behaving quite strangely. No voltage is present on the card except 12V 3.3V and 5V.
Another strange thing is that the vcore has 0.3 Ohms to ground. And I tried to apply power here but nothing got hot and I don't want to disassemble the whole card.

Any ideas?
 
I've not had one of these 470s pass across my workbench, but as I recall, you should have about 1200 milliohms resistance to ground on the core rail. If you've got less than that, you've got a short somewhere.

There is a pretty good chance that the short is in the core itself, or in the PCB, but it's not guaranteed. Take the heatsink off and look for visible damage. A cracked MLC capacitor could very easily cause this, as could a failed low side FET in one the core power phases. You'll need to be more specific WRT "I tried to apply power here." Where exactly, did you inject current? How many amps? What voltage? What did you use as a ground? Did you check things like the VRM controller IC, phase controllers, et cetera? Details matter.
 
Hi

I am sending you a picture of the card.

20220130_150814.jpg
20220130_150821.jpg

I have measured the coils and marked them in the picture.
My meter reads 0.2 Ohm at idle, I have already subtracted it from the values.
20220130_150814a.jpg

red 1: 277 Ohm
red 2: 2.45 KOhm
green: 4.3 Ohm
blue: 0 Ohm
Orange: 40.9 Ohm
pink: 277 Ohm
yellow: 21.8 Ohm
 
What is the brown stuff on R509 (back of the board, first populated phase from the PCI-E slot connector)? Flux? Corrosion?

This being a discrete FET design, you could conceivably try removing the low side FETs (USE A PREHEATHER!!) and see if that clears your short. If so, replace them with new ones. Failing that, if current injection doesn't provide any close as to where the short is, it's probably either the board itself or the GPU.
 
What is the brown stuff on R509 (back of the board, first populated phase from the PCI-E slot connector)? Flux? Corrosion?

This being a discrete FET design, you could conceivably try removing the low side FETs (USE A PREHEATHER!!) and see if that clears your short. If so, replace them with new ones. Failing that, if current injection doesn't provide any close as to where the short is, it's probably either the board itself or the GPU.
That's flux there. I'm going to take the low side fets off and try to inject the current that way. I did it with 0.8V last time.
 
Hi

I removed the low side mosfets and tried to inject current (0.8V). But it did not work, so I looked around and found a capacitor with a very low resistance. I removed it and the pci 12V to ground is now in kOhm.
So any other ideas where the short could be?
 
Back
Top