Seller Note “Dead”
Summary
- Rare card, bought cheap, have never seen a GTX 760 ti before, but then it is an OEM. Would be great to repair this.
- 12v PCI BUS short detected
- Resistances
- Vcore 6.2Ω
- Vmem 84Ω
- PEX
- 12v PCI Shorted
- 12v x 2 PCI-E OK
Moving to check VRM power stages
| Phase | ESR Ohms | |
|---|---|---|
| 1 | Vcore | 1.16Ω |
| 2 | Vcore | 1.14Ω |
| 3 | Vcore | 1.04Ω |
| 4 | Vcore | 1.17Ω |
| 1 | Vmem | 1.13Ω |
| 2 | Vmem | 1.14Ω |
Phase 3 appears to be the problem. Next, I would like to get so data on the MOSFETS.
| Phase | High-Side MOSFET Gate Resistance To GND |
|---|---|
| H1 Q22 | 42 KΩ |
| H2 Q15 | 42 KΩ |
| H3 Q18 | 15 Ω |
| H4 Q9 | 42 KΩ |
Q18 also shows an abnormal gate resistance, further confirmation that phase 3 and high-side MOSFET Q18 are at least partly responsible for the 12v short.
In this type of configuration, there will be 2 x low-side MOSFETs and 1 x high-side in each phase, we can work out which high-side is in which phase and then know which one to replace (see photo).
High-side MOSFET is 4927N https://www.onsemi.com/pdf/datasheet/ntmfs4927n-d.pdf
Low-side MOSFET is 4936N https://www.onsemi.com/pdf/datasheet/ntmfs4936n-d.pdf
After removing Q18, the short is resolved. Better still, I am absolutely delighted to see that the card is able to boot into Windows 10, load drivers and even benchmark briefly! I really expected the GPU to be dead, as early in my graphics card repair days, I thought that I might have accidentally injected 12v and that it might have reached the core. Happy to get away with that.
More than just the MOSFET…
After trying to replace just the faulty MOSFET, I found that the driver was also compromised and failed on startup (small amount of smoke!).
The MOSFET Driver is https://www.onsemi.com/pdf/datasheet/ncp81062-d.pdf, I have replacements now. So can go ahead and try to replace both the MOSFET and driver.
Full Repair
OK, so I have replaced the MOSFET and the driver U4.
TODO Add picture of driver from microscope
However, I am not happy with the resistance measurements from the driver to MOSFET circuit. The gate resistance at first measured OL (other phases measure 40K). At first, I thought I had mis-soldered the driver U4, so I re-soldered it, but no change. It could be what is actually wrong is that the resistor shown R57 has gone OL? As, there is no continuity from the gate of the replaced MOSFET to the other end of R57, whereas there is on all other phases. I am not sure what value R57 should be, but I think it could be near 0 Ohms, I could take another off to measure it. Also, pin 8 of U4 is DRVH and should measure 40K and connect to the gate of the replaced high-side MOSFET, it instead measures ~151K.
I need to take some time to get these measurements right.
Fix Cooler / Missing Screws
TODO Fix the VRM cooler, missing screws/bolts.