All footprints are estimates.
Front:

C3 - 100µF 16V electrolytic capacitor

C41 - 1µF 0805 (for PVDD, if possible 10V or higher)

Y3 - 27MHz Oscillator (pinout: 1 - EN, 2 - GND, 3 - OUT, 4 - VCC)

Back
:
C619 - 10nF 0402 (for MVDDQ, take 25V or higher)

Various memory power de-coupling caps. Unfortunately I don’t have the R420 reference schematic.
They’re either 100nF or 1nF, as that’s the same on the R480 - however the layout and identifiers are different.

C94 - 
C275 - 
C276 - 
C298 - 
C261

The larger C261 is still 1nF or 100 nF?

Any good approach to this scenario? Should I alternate 1nF with 100nF ?
Or maybe should I take some of the components from the nearby ram chip and measure them?
Though none are in the same pattern.

As I don't have my X800XT with me right now, can you make a picture of another memory cap section? then perhaps we can match which ones are which and need to be measured by you.

For the larger C261, sorry, missed that one. Much likely to be a 10µF or 22µF part.

Pictures with all 4 ram chips on the back with neighboring capacitor formation.

I installed the missing components except from the caps near the ram chip. I did however installed the large cap near the ram chip with a 10 uF one. Hopefully it's the correct value.
The card doesn't display anything on any output.
The card is detected in ATI flash and I did a bios flash. No improvement.
The card is detect in windows (when using a secondary PCI card). I installed video drivers and now windows hangs at loading screen.

Are those caps that important? To me there should be artifacts in the worst scenario, not no-display.
I have no idea which values to choose. I measured the resistance to ground of those pads and 3 caps measure 17 Ohm and one measures 338 ohm.
Any ideas?

Could it be that the oscillator is not working? The card does get detected in windows, so it would weird if it didn't work. It does receive 3.3V, but I can’t seem to measure any waveform coming out of it. I’m referring to that corroded Y3 component.

That said, I have to admit I’m still pretty awkward with the oscilloscope and I’m never fully confident that I’m measuring things correctly. I did manage to probe the motherboard crystal oscillator and got a clean waveform there, so I think the scope is set up properly.

However, when probing all the pins on the video card’s oscillator, I get absolutely nothing — no voltage reading on the scope, no detectable frequency, nothing at all. The only thing I can confirm is that one of the pins measures 3.3V with a multimeter.

Retronerd878 wrote on 2026-05-16, 22:50:

Could it be that the oscillator is not working? The card does get detected in windows, so it would weird if it didn't work. It does receive 3.3V, but I can’t seem to measure any waveform coming out of it. I’m referring to that corroded Y3 component.

That said, I have to admit I’m still pretty awkward with the oscilloscope and I’m never fully confident that I’m measuring things correctly. I did manage to probe the motherboard crystal oscillator and got a clean waveform there, so I think the scope is set up properly.

However, when probing all the pins on the video card’s oscillator, I get absolutely nothing — no voltage reading on the scope, no detectable frequency, nothing at all. The only thing I can confirm is that one of the pins measures 3.3V with a multimeter.

r480 schematic
https://drive.google.com/file/d/1T2U3JKfLgIvo … QJcJBD2kVK/view
from
https://www.schematic-expert.com/amd/

Y3 is not quartz, its oscillator.
So it should produce a very clear signal.

Retronerd878 wrote on 2026-05-16, 22:50:

Could it be that the oscillator is not working? The card does get detected in windows, so it would weird if it didn't work. It does receive 3.3V, but I can’t seem to measure any waveform coming out of it. I’m referring to that corroded Y3 component.

That said, I have to admit I’m still pretty awkward with the oscilloscope and I’m never fully confident that I’m measuring things correctly. I did manage to probe the motherboard crystal oscillator and got a clean waveform there, so I think the scope is set up properly.

However, when probing all the pins on the video card’s oscillator, I get absolutely nothing — no voltage reading on the scope, no detectable frequency, nothing at all. The only thing I can confirm is that one of the pins measures 3.3V with a multimeter.

If you have oxidation on the pads, you need to pierce through the oxide layer in order to measure anything meaningful.
I would expect that card not to show up at all without it's main clock. Unless the AGP interface etc. would work with AGP clock alone, which I somehow doubt.

What is also suspicious is that to measure the oscillator i need to take down the cooler and with the card powered on, the GPU goes up to like 50 C and pretty much stays there. I can put my finger on it even after 5 minutes.

I've been poking around the corroded area and I found a 0402 resistor that wasn't properly attached.
I tried to solder it back but couldn't as one of the pads of the resistor is eaten away or broken off.

I need a new resistor so I tried to measure it. Applied force with the probes on the 2 poles of the resistor and I do get a reading of 680 Ohm. Can I trust this reading, even though the resistor is missing some or all of the pad on one side?

680Ohms is correct.