Seller Note “Faulty graphics card. It doesn’t power up. Died without any smoke or other artifacts. Everything else worked fine even power supply.”
Summary
- Resistances
- Vcore 2.4
- Vmem 15.6
- PEX 283
- 5v 1 KOhm
- 12v PCI OK
- 12v 6pin OK
- 12v 8pin Short
VRM ESR Analysis
| Phase | Type | ESR Ohms |
|---|---|---|
| 1 (U8) | Vcore | 1.26 |
| 2 (U9) | Vcore | 1.26 |
| 3 (U10) | Vcore | 1.28 |
| 4 (U11) | Vcore | 1.28 |
| 5 (U12) | Vcore | 1.28 |
| 6 (U13) | Vcore | 1.15 |
| 1 () | Vmem | 1.35 |
| 2 | Vmem | 1.32 |
This makes U13 a suspect. It is a combined power stage SIC 789 https://www.vishay.com/docs/62972/sic789a.pdf
Well, after removing U13, the good news is that the short is gone and the 8 pin PCI-E has a healthy resistance. The bad news was that removing it was damn hard! Even using a preheater and 440 degrees heat (normally I use 400 degrees C no preheater), the MOSFET was seemingly stuck down. Finally, it was removed, but a portion of the MOSFET is remaining on the pad and a small nearby capacitor was dislodged by all the heat and hot air.
Power testing actually went very well (at first). The card looked very smart, with fans and lights etc, and Windows 10 was useable. Unfortunately, I seemingly pushed things too far by attempting to run Heaven benchmark for a few seconds, which lead to a crackling noise, the screen freezing up and a nasty burning smell! On inspection, it seems the MOSFET next to the removed one must have been compromised (perhaps already, or by the excess hot air I had to use) and broke down under load. Probably, not all the phases were used until I subjected the card to load.
Hopefully, this one will be cleaner to remove than U13! I haven’t attempted to retest the card, I think I need to get these areas cleaned up first.
Next Steps
- Remove the remains of U13, people have suggested using a Dremmel.
My friend the Dremmel, seems to solve one problem and leave another! Or, at least I think I need to work on my Dremmel skills, it seems quite hard to see exactly what is being ground away.
TODO Add picture
There is still a short, possibly secondary to the DRMOS remains. Voltage injection very quickly shows a the small nearby capacitor is to blame.
This board doesn’t seem very lucky, the capacitor resists all attempts to desolder it, it is fused on. Back to Mr Dremel, capacitor ground away and the short is gone.
Risky power-on test, only 6/8 phases present. Encouragingly, the card still manages a picture! Gives me some hope of a possible solution, not going to be easy though fixing up the carnage on the PCB, but all learning hopefully.