Seller Note “This graphics card sadly stopped working randomly so I had to replace it.”
- Draft TODO add thermal camera image, add images of MOSFET and driver
- VCore – 0.4
- VMem – 38.8
- PEX – 79.9 ?
- 1.8V – 1.8K?
- 5V – 950 ?
- 3.3V – 40K?
- 12V PCI -K+
- Short on PCI-E 6 pin – 32.1 Ohms is way too low
Faulty MOSFET getting hot
I couldn’t detect the shorted MOSFET with the usual ESR approach, as the 12v rail still has 32 Ohms. Instead, the thermal imaging camera reveals that Q38 gets pretty hot when injecting 12v. I couldn’t detect much heat much below this voltage and had to be careful to measure about to make sure 12v wasn’t getting to the core or anywhere else it shouldn’t.
TODO Add thermal camera image
The MOSFET is AON6414A 6414A, which I luckily bought 10 of for another card.
To double-check, the gate resistance of Q38 is a tragically low 1.6 Ohms, whereas the other high-side gates have 8K Ohms.
After replacement, the good news is the short is gone and the 12V PCI-E has a healthy, high resistance. The not-so-good news is that the gate resistance is still 1.6 Ohms, so I expect the driver is also faulty.
Driver type is the same as my Gigabyte RX 460 2Gb, it’s marked A4U which is NCP81161 https://www.digikey.co.uk/htmldatasheets/production/1129670/0/0/1/ncp81161.html
Replacement of faulty driver
I decided to use a donor card’s driver (have a Gigabyte GTX 1060 3Gb with PCB damage, but most of the same parts), and after replacing the faulty driver, the gate resistance is now 9.34K. The other phase’s gate resistances are all about 8K, so hopefully, this isn’t a problem. I have also noticed a capacitor that looks a bit damaged but doesn’t seem to be measuring a different ESR, so I’ll leave it for now.
Well, the card’s resistances all appear normal, so time for a power-on test. Hopefully, it won’t all go up in smoke!
Well, no smoke, but no VCore either! All other rails seem to be there. Perhaps an issue with the PWM too.