Sapphire HD 7850 OC - Detected but No Display

Collapse
X
 
  • Time
  • Show
Clear All
new posts
  • прямо
    Badcaps Veteran
    • Sep 2022
    • 261
    • Indonesia

    #1

    Sapphire HD 7850 OC - Detected but No Display

    Friends,

    The card is missing a memory rail. Enable transistor is ok, but I'm only getting about 0.48V at the enable pin.
    There's a compensation RC network on this pin and any of them could be the reason why the buck controller isn't working properly.

    Can't tell for sure because my DMM can't measure caps, and I don't have ESR meter.
    I also don't have a replacement for the buck controller, yet.

    What I plan to do is temporarily inject 1.4V into the memory rail just to see if the card will at least display a picture.

    Now, this is where I need a little bit of information, or rather a confirmation from any of you who has done this before.


    Which one of the following approach is safer or better? obviously I want to avoid a catastrophic situation where upon powering up, the high side mosfet becomes active. letting 12V pass through and destroy everything.

    1. Remove only the buck controller, ground the mosfet gates, then inject voltage.
    2. Remove the buck controller and mosfets, then inject voltage.


    Here are boardview SS of the memory rail section.
    Click image for larger version  Name:	low side.jpg Views:	0 Size:	78.0 KB ID:	3169782 Click image for larger version  Name:	high side.jpg Views:	0 Size:	140.2 KB ID:	3169783 Click image for larger version  Name:	sapphire-7850.jpg Views:	0 Size:	191.1 KB ID:	3169784

    Q701 = high side mosfet
    Q702 = low side mosfet
    U701 = 5V/12V sync buck controller
  • прямо
    Badcaps Veteran
    • Sep 2022
    • 261
    • Indonesia

    #2
    Nevermind. It was a stupid idea. I think I will just strip it for parts, at least the fan can be sold for a decent amount of money. Why did I even bother asking for help in a forum... 🤦

    Comment

    • прямо
      Badcaps Veteran
      • Sep 2022
      • 261
      • Indonesia

      #3
      After I replaced the buck controller with APW7165C salvaged from a dead nVidia card and changed one feedback resistor because the chip has a different VRef, 1.45V memory rail returns and the card is now working normally. Tested with Tserver and Furmark, no artifacts.

      Edit: sorry uploaded a wrong GPU-Z screenshot, but the edit photos won't let me delete.
      Last edited by прямо; 01-08-2024, 04:29 AM.

      Comment

      Related Topics

      Collapse

      Working...