troubleshooting GTX 1060 3GB +12 and GND Shorted

[LazyDev]

Following up on my folding at home thread, where my 1060 died. Currently, it has a dead short on the +12 pins, as shown below.

 

 

The card was fully powered and loaded when the short happened.

 

 

Some pictures of the card.

 

 

 

I have a multimeter on hand, and willing to test to see if this card can be salvaged. Advise would be very appreciated.

[Windows7ge]

It's hard to tell if the short stayed a part of the power delivery side of if it leaked into the data transmission lines because it almost looks like it friend your PCI_e slot and that could go back to the CPU.

 

There's no immediate signs of a blown component. If you're up for component level repair it's going to be a bit of a hunt. Do you have a hot air rework station?

[LazyDev]

1 hour ago, Windows7ge said:

It's hard to tell if the short stayed a part of the power delivery side of if it leaked into the data transmission lines because it almost looks like it friend your PCI_e slot and that could go back to the CPU.

 

There's no immediate signs of a blown component. If you're up for component level repair it's going to be a bit of a hunt. Do you have a hot air rework station?

Only have a basic heat wand and soldering iron.

[Windows7ge]

6 minutes ago, LazyDev said:

Only have a basic heat wand and soldering iron.

If you don't have the proper equipment you're not going to have a fun time here since what failed isn't obvious besides a short to ground.

 

Only thing you could try is start picking pieces off the PCB where the 12V go and see what makes the short go away. My first suspect would be the VRMs. Inspect each one closely. Do any appear as if they were getting hot (like really hot) or does any one have a pit or hole that none of the others have?

[LazyDev]

3 hours ago, Windows7ge said:

If you don't have the proper equipment you're not going to have a fun time here since what failed isn't obvious besides a short to ground.

 

Only thing you could try is start picking pieces off the PCB where the 12V go and see what makes the short go away. My first suspect would be the VRMs. Inspect each one closely. Do any appear as if they were getting hot (like really hot) or does any one have a pit or hole that none of the others have?

Unable to see anything. I did get a good tip from a youtube video, where you spray isopropyl onto the board, and power up the card via a bench power supply. The dead short could show up, as the damaged components will warm up the fastest. When I get some time, I may test this theory, as I don't own a thermal camera.

 

Update.

Been probing the damaged PCIe cable. It looks like the +12 volts shorted to the 3.3 volts. Source: https://pinouts.ru/Slots/pci_express_pinout.shtml, Side A. Pins 1 and 11 are accessible in the hole I shoved my probe into, aswell. It's a mess, and lucky that it didn't take my motherboard and CPU out. Now I'll need to check the 3.3 volts on the 1060.

[Windows7ge]

If you use the isopropyl method cooling the die might get tricky. Leaving it exposed may cause it to heat up exponentially quickly.

 

Did you test with another card if it isn't normal for those sets of 12+ & GND to appear shorted? Perhaps the riser cable just failed.

[LazyDev]

3 hours ago, Windows7ge said:

If you use the isopropyl method cooling the die might get tricky. Leaving it exposed may cause it to heat up exponentially quickly.

 

Did you test with another card if it isn't normal for those sets of 12+ & GND to appear shorted? Perhaps the riser cable just failed.

Installation of the card into the system, the PSU won't even power on. The 1060 was the fault.

[LazyDev]

Just an update.

 

I've since found the short circuit and removed the defective mosfet. The card now powers on.

 

Measured voltages and notes,

 

0.8 volts on the core

1.8 volts on the bios chip

3.3 volts present

Fan is spinning at the correct speed

 

Card is not currently been detected.