Seller Note “Does not work, PC does not work with GPU installed.”
Summary
- Resistances
- VCore – 1Ω
- VMem – 370Ω (Micron 9YB77 D9XGP)
- VDDCI – 24.5Ω
- Display Rail – 22.5Ω
- 5V – 821Ω
- 1.8V – 2.66KΩ
- 3.3V – 2.053KΩ
- 12V – 3.8KΩ (slot), 8 pin PCI-E 1.5Ω!
- Clear 12V short there, consistent with what I suspected from the seller’s note.
Investigating the 12V short
First, let’s check the gate resistances on high-side MOSFETs in the 4 VCore phases.
High-sides are SM4377 http://www.sinopowersemi.com/temp/SM4377NSKP_datasheet.pdf
Low-sides are SM4512NH https://datasheetspdf.com/pdf/1244517/Sinopower/SM4512NHKP/1
OK, so 3 of the gates measure 38.84KΩ, and one 1.5Ω! So we have a suspect (Q4605):
We should also suspect the driver (that marking A4X, X is the month, this is actually NCP81161 https://www.digikey.co.uk/htmldatasheets/production/1129670/0/0/1/ncp81161.html) is possibly faulty. There is some corrosion around it, none of the others appear this way:
As a first step, I’d like to remove the high-side, check the short is gone. Then remove the driver and low-sides and try to see if I can get a picture.
Good news and bad news!
Well, the first good news was that the high-side MOSFET came off cleanly and the 12V PCI-E returned to a healthy 2.3KΩ – good!
Removing the low-sides and driver was the bad news, the driver appears to have ripped a pad and the two low-sides will need grinding off!
With this kind of damage on such a cheap card, I need to see some proof of life before I proceed. Well, after carefully checking for any shorts caused by that wreckage, I decided to risk a power-on test and …
Not only a picture, but the drivers load too – I cannot risk applying any load, but there’s some hope at least 🙂
I’ll order the parts and hope I can grind away those low-side remains without destroying the pads. I can hopefully replace the driver pad too. Not the easy fix I hoped, but hopefully possible.