Dell G5 5505 Code 43 Problem

[sect-sin]

Hello,

I have a Dell G5 5505 notebook with an AMD RX 5600M dedicated graphics card, and the machine is throwing me a Code 43 error when I am trying to install drivers, and although GPU-Z sees the GPU, it shows 0 mb of GDDR6 memory. So I guess one or more of the 6 memory chips, or perhaps the memory controller is defective.

I tried to pinpoint the faulty chip by running dmgg.py on Linux, but I ran into some issues. Lspci says that the starting memory address of the dedicated GPU is fce0000000, so I added this as a parameter for dmgg.py (see attached screenshot). Despite doing this, dmgg.py detects the integrated GPU (see green text) instead of the dedicated GPU, and then finds no faulty chips.

Please note that I cannot set the integrated GPU as primary display on this notebook from BIOS.

I also tried memtest.py, it gives a ton of errors, but I cannot use that to pinpoint the faulty chip(s).

Any help from anybody on how I could properly diagnose the issue would be much appreciated.

Cheers,

sect-sin

[sect-sin]

Re: Dell G5 5505 Code 43 Problem

Hi all,

I just noticed that I forgot to add the screenshot. Also, I have attached the schematics for this board, should anybody be interested in it (for reference and search engines, the board number is SELEK MLK AMD 19802-1 FGRDN$LA).

Since then I have tried to reprogram the VBIOS. The programming was successful, but Code 43 still remains, so we know that it is not a corrupted BIOS issue.

Also, I think I forgot to add some important details. The notebook originally arrived to me with a dead short (no power, adapter in short circuit mode). I found an unrecognizably burnt-up 1R0 inductor and a dead SM7321 Dual MOSFET. After replacing both, the Notebook turned on and booted fine, but I had the initially mentioned Code 43 error.

According to the schematics, this is a 1,35V rail, so that's why I think I have a memory/memory controller issue. I do not think that I have a dead GPU, as the GPU gets detected, the GPU power rails all show voltage, and the GPU is getting warm.

It is possible that during the catastrophic breakdown of the MOSFET some nasty high voltage reached the memory chips and/or the memory controller, or some other component down the line. Nevertheless, I need to be sure before ordering costly replacements.

Could anybody help me in diagnosing this matter? Which chip is the memory controller? Could another component down the line be defective? Is it even worthwhile to work on this?

Cheers,

sect-sin

[piernov]

Re: Dell G5 5505 Code 43 Problem

MOSFET short on VRAM power rail kills both VRAM and GPU often enough. First thing would be to confirm resistance to ground to see if there's possibly a partial short to ground.

[sect-sin]

Re: Dell G5 5505 Code 43 Problem

Hi piernov,

Thanks for picking this up.

The resistance between S1/D2 of the previously fried MOSFET (which leads through the previously also fried Inductor to WR_VGA_1PD35V, although I think this might be a typo in the schematics and should be WR_VGA_1D35V) to ground is 23,3 Ohms. VRAM chips are Samsung, I do not know whether this is OK or not.

Any ideas on where to go from here?

Cheers,

sect-sin

[sect-sin]

Re: Dell G5 5505 Code 43 Problem

I have perhaps one more question.

If I remove all memory chips, can I "test" that the GPU is all right? If the Code 43 error vanishes after removal of all chips, would that mean that I have just to replace the memory chips?