The 180µF cap is the input bulk cap for the 12V_EXT rail. It's needed by the MVDDC regulator to properly function. Not having it means the switching regulator needs to rely on the impedance of the power rail coming from the PSU to properly function. That can work. I wouldn't expect it to cause the card to be not recognized. From the pictures some parts of it look "dirty" - like there is corrosion on pads or something?

Doesn't look that bad. I definitely seen worse. The fan is completely dead and the top part with blades is detached from the motor as the motor is filled with green corrosion. This card has definitely seen its fair share of moisture.

I actually had a 180 cap and after i installed it nothing changed. Any ideas what to do next?
Could it be that the bios is broken? I do have a bios programmer, though I have no idea which chip is the bios. Or maybe there is way to program it using another video card for display?
I also have an ultrasonic cleaner. Think that will make a difference? Maybe there is some corrosion under the GPU and it will help.
Would like to try the bios thing first, If i clean it, It's nice to clean the flux residue from the bios chip removal as well.

Bios is U80 on the backside of the card.

If you have a PCI card at hand, you can boot with that one as your primary and have the 9800Pro alongside it. Iirc "newer" versions (like 3.60) of ATIFlash should be able to detect non primary cards as well, that way you could easily get a bios dump of the card.

Check "external power" (molex) circuit detection

ATIflash detects the card but i can't flash the bios. Am I doing something wrong?

"atiflash -unlockrom 0"

https://archive.techarp.com/show26c2.html?fileID=22

https://web.archive.org/web/20050307111746/ht … amsung256MB.bin
https://web.archive.org/web/20050307112159/ht … amsung128MB.bin

Just for your awareness, your card has Samsung memory and you're attempting to flash a hynix card bios.

OK, I got a bios rom with samsung memory.
The command unlockrom doesn't work so i got a newer version of ATIflash.

With the new version the command returns:
Failed to read ROM.

Also, when I try to program the bios with the newer version the computer hangs.

I would take the chip off but it's not clear to me what adapter i need to convert SOIC-8 to DIP, so i can use the programmer.

Retronerd878 wrote on 2026-05-11, 11:33:

OK, I got a bios rom with samsung memory. The command unlockrom doesn't work so i got a newer version of ATIflash. […]
Show full quote

OK, I got a bios rom with samsung memory.
The command unlockrom doesn't work so i got a newer version of ATIflash.

With the new version the command returns:
Failed to read ROM.

Also, when I try to program the bios with the newer version the computer hangs.

I would take the chip off but it's not clear to me what adapter i need to convert SOIC-8 to DIP, so i can use the programmer.

SOIC8 SOP8 to DIP8 EZ Programmer Adapter Socket Converter Module With 150mil
https://www.amazon.in/Menton-SOP8-DIP8-SOIC-A … r/dp/B073TYXLTR
https://www.amazon.com/HiLetgo-Programmer-Ada … r/dp/B01HTC5DKS
https://www.amazon.in/SOIC8-SOP8-Programmer-A … r/dp/B0DPR32S4X
150 Mil

I actually got one of those adapters in the box. Took the chip out and reflash the chip successfully.
The card behaves the same way. No output and post code beeps.
However, if I use a PCI card, when I get to windows, the card is detected in device manager.

Retronerd878 wrote on 2026-05-11, 13:38:

I actually got one of those adapters in the box. Took the chip out and reflash the chip successfully.
The card behaves the same way. No output and post code beeps.
However, if I use a PCI card, when I get to windows, the card is detected in device manager.

Some graphics cards do not display any messages if the external Molex power connector is not connected.
Check the signal lines from the Molex connector.
It’s something like a ground pin that’s connected not to grownd but to the base of a transistor or some kind of little logic chip.
Perhaps @tehsiggi has a suitable circuit diagram.

Ohh if I only had the 9800 Pro reference schematic. The MSI one is not default, unfortunately. They're solving the matter with a PNP transistor (Q10) whereas the 9700 Pro reference schematic has U93 - neither of these can be found on a regular 9800 Pro.

I could try to trace around a little tomorrow, but no promises.

So, any ideas which IC we are talking about? I do have another Radeon 9800 pro with shorted VDDC line on the gpu core. I could transplant parts from it if needed. Just point me to whatever IC might be the culprit.

Retronerd878 wrote on 2026-05-15, 12:50:

So, any ideas which IC we are talking about? I do have another Radeon 9800 pro with shorted VDDC line on the gpu core. I could transplant parts from it if needed. Just point me to whatever IC might be the culprit.

I thought it was unlikely that the circuit would be in a very different place 😀

I measured the voltages for the components you mentioned.

Retronerd878 wrote on 2026-05-16, 17:04:

I measured the voltages for the components you mentioned.

Unfortunately, the signal is being processed correctly.

The only components you can check with a multimeter are the small linear regulators scattered across the board.
Naturally, the numbering won’t match that of the PCB, as this is a circuit for the Radeon 9700.

Regulator 1117 (located near the fan connector) is quite large, whilst the other four are tiny.

upd
If a donor unit is available, you can also swap the 14-pin logic ICs.
U2 ACT08 (near the DVI connector) and U3 AD86 (near the S-Video connector).
And oscilator fb-230B 27.000 (MHZ)

I measured the IC's on the front of the card. Anything suspicious?