RX 470 mining short

[Noovas]

Hello friends,
I have RX 470 4G mining card which is not working. After removing the mosfets there is still short between drain and source pads of the PCB.
I think that the problem could be a bad cap, but without schematic it is not easy, I just dont know where to start from. I am uploading a photo of the PCB with removed mosfets. Any idea will be helpfull!! Thanks for the support !!

[didi79]

Re: RX 470 mining short

So we are talking about the low side MOSFETs. This means you have a short between your load (the GPU) and ground.
You could try isolating the VRMs by removing the chokes. Not easy to desolder tho.

Another option is voltage injection and check for heat. But since we are on the GPU side, make sure to keep voltage low enough. I find this technique quite hard because there is not that much heat generated with low voltage. And heat dissipation in the pcb makes it quite hard to detect the actual heatspot.

[Noovas]

Re: RX 470 mining short

Thank you for your reply didi79.
I also was thinking to apply 0,5volts between drain and source and see if there is something heating.

[Noovas]

Re: RX 470 mining short

If I don't detect any heat is that mean that the GPU is dead?
Also if I remove all 4 chokes and there is no shortage, this means that the GPU is dead?

[didi79]

Re: RX 470 mining short

If I don't detect any heat is that mean that the GPU is dead?
Also if I remove all 4 chokes and there is no shortage, this means that the GPU is dead?

Answer to both questions: not necessarily

The chance of the chokes creating a short is extremely low. You remove them to separate the VRM phases. Would make finding a short in one of the VRMs easier.

[Noovas]

Re: RX 470 mining short

I applied 0,8V, 2A and after 10 minutes I still can find a warm component.
Now I will remove chokes one by one

[Noovas]

Re: RX 470 mining short

I think I stucked.... After removing the first choke there was no short for the corresponding MOSFET. Then I removed the second and the short for corresponding MOSFET also was gone. This is the same and for the other two. Now I don't have chokes and no short. It looks they are parallel to the MOSFET? One side to source and the other to drain ��

[didi79]

Re: RX 470 mining short

Each MOSFET has its corresponding coil. But on the GPU side all coils come together: you can see they are all connected to the same big trace on the pcb.

I'm not sure how and where you are measuring. Do you have a short from GPU (the big trace where all the coils connect to) to ground?

You should also be able to measure the high side MOSFETs easily (since your low sides are gone): none of the contacts should be shorted to ground. If one is, you have found the faulty one.
In my experience it's usually the high side MOSFETs that fail.

But don't mistake a switched MOSFET as shorted (measured from source to drain). If you put your multimeter positive probe on the gate, you will briefly switch the MOSFET.

[Noovas]

Re: RX 470 mining short

Yes, on the GPU side all coils are together and grounded!

[didi79]

Re: RX 470 mining short

Yes, on the GPU side all coils are together and grounded!

Low resistance over the GPU is normal. Depending on the (lack of) precision of your multimeter, you may not even see the difference between short or low resistance.

[Noovas]

Re: RX 470 mining short

Thank you Didi,
Here are some more tests I made but unfortunately they didn't show where the problem could be...
Till now with MOSFETs on high side (all low side removed) and coils motherboard want run. Then I removed all coils and motherboard runs. I put one coil and motherboard want run. After that I removed this coil and put another one to other place but again motherboard want run.
Then I decided to remove all low side MOSFETs and to put the coils. Motherboard runs. After installing just a single MOSFET on low side motherboard want run.
Is that means that with removing all MOSFETs I am disconnecting the short from one side and removing all coils I am disconnecting the short from the other side?
I am uploading two photos just for visualization (motherboard want run in both situation).
Thanks,

[Noovas]

Re: RX 470 mining short

I just found that with coil and MOSFET on high side burns the MOSFET.

[Noovas]

Re: RX 470 mining short

I just turn it on to measure voltages but smoke come out from the cap

[didi79]

Re: RX 470 mining short

Thank you Didi,
Here are some more tests I made but unfortunately they didn't show where the problem could be...
Till now with MOSFETs on high side (all low side removed) and coils motherboard want run. Then I removed all coils and motherboard runs. I put one coil and motherboard want run. After that I removed this coil and put another one to other place but again motherboard want run.
Then I decided to remove all low side MOSFETs and to put the coils. Motherboard runs. After installing just a single MOSFET on low side motherboard want run.
Is that means that with removing all MOSFETs I am disconnecting the short from one side and removing all coils I am disconnecting the short from the other side?
I am uploading two photos just for visualization (motherboard want run in both situation).
Thanks,

In those photos, are these new MOSFETs? If you put back a shorted one, it would make sense the short is still there and card won't power on.

I just found that with coil and MOSFET on high side burns the MOSFET.

Yeah well, shorted MOSFET...

I just turn it on to measure voltages but smoke come out from the cap

I already saw in the above photos that those caps got some heat damage. You should really protect them. Better replace them all: toxic fumes and risk of popping.

[Noovas]

Re: RX 470 mining short

Hello,
When I find a MOSFET dead I don't use it anymore
Caps were ok but the heat from the gun melted this plastic under.
Whe I have time I will remove the cap and measure it....

[didi79]

Re: RX 470 mining short

Whe I have time I will remove the cap and measure it....

You said smoke came out: DON'T reuse it!

[SiO2]

Re: RX 470 mining short

Does the cap show any sign of damage? If smoke came out, that should be a catastrophic failure and hence clearly visible. Also check the neighboring components: MLCCs, the MOSFET driver chip, resistors before applying power again.
Things like his can easily cause a chain reaction of popping components if something is overlooked in a rush.

[Noovas]

Re: RX 470 mining short

Ok, to be honest I am not 100% sure that the smoke was from the cap, it could be around. I was in hurry to switch off the power .
But I will post the result if there is any progress....