Confused. In device manager, where is the PCI card that you used to display the device manager?

When you open the dialog on the Radeon 9600XT, what is the error code? When was the 9600XT last working?

MattRocks wrote on 2026-05-09, 22:18:

Confused. In device manager, where is the PCI card that you used to display the device manager?

When you open the dialog on the Radeon 9600XT, what is the error code? When was the 9600XT last working?

The PCI is under unknown devices, as I haven’t installed it’s drivers (it is an NVIDIA GT610).
The error code I get is 10.

I see your onboard audio drivers are not installed. Have you installed stable AGP GART drivers for your motherboard chipset?

Thanks for your help Matt!

The unrecognized audio interface is also from the GT610, it is the HDMI audio output.
Chipset drivers are up to date, if that is what you mean? The other AGP card I have (9550) works perfectly.

The 9600xt does not show any video output (VGA or DVI) at any stage, even during POST before the OS is loaded. I am not sure this card has ever been working. I purchased it used with a “working” claim.

qtunnel wrote on 2026-05-10, 08:46:

The 9600xt does not show any video output (VGA or DVI) at any stage, even during POST before the OS is loaded. I am not sure this card has ever been working. I purchased it used with a “working” claim.

I've had a some of those. I feel like starting a name and shame thread, and maybe a masked vigilante group! 😉

There is a thread on using the ATI R3MEMID/R5MEMID/R6MEMID DOS utilities to test a Radeon, which might still work if booting from NVIDIA PCI card:
(SOLVED) R3MEMID - "Error: SUBID 7C20174B not found in r3memid.cfg"

Also consider returning "not as described" because you can't get your time back, and if you only half fix it then you trap yourself by having modified it.

MattRocks wrote on 2026-05-10, 09:25:

I've had a some of those. I feel like starting a name and shame thread, and maybe a masked vigilante group! ;) […]
Show full quote

qtunnel wrote on 2026-05-10, 08:46:

The 9600xt does not show any video output (VGA or DVI) at any stage, even during POST before the OS is loaded. I am not sure this card has ever been working. I purchased it used with a “working” claim.

I've had a some of those. I feel like starting a name and shame thread, and maybe a masked vigilante group! 😉

There is a thread on using the ATI R3MEMID/R5MEMID/R6MEMID DOS utilities to test a Radeon, which might still work if booting from NVIDIA PCI card:
(SOLVED) R3MEMID - "Error: SUBID 7C20174B not found in r3memid.cfg"

Also consider returning "not as described" because you can't get your time back, and if you only half fix it then you trap yourself by having modified it.

It does not work if the Radeon is not the primary card.
Before you go at it with heat, could you do some voltage readings around all the large caps the card has?

tehsiggi wrote on 2026-05-10, 10:04:

Before you go at it with heat, could you do some voltage readings around all the large caps the card has?

Hi tehsiggi, readouts attached.

I don’t know enough to understand if they are good or not.

Those voltages look alright, the memory ones are a bit on the low side for the -33 variant of memory (specced at 2.8V) - but nothing I'd deem causing the card issues to be properly recognized. I had a 9600XT with BGA and it had the same issue, never got it working so far.

tehsiggi wrote on 2026-05-10, 14:53:

Those voltages look alright, the memory ones are a bit on the low side for the -33 variant of memory (specced at 2.8V) - but nothing I'd deem causing the card issues to be properly recognized. I had a 9600XT with BGA and it had the same issue, never got it working so far.

Thanks.
I am planning to have a look at the BIOS anyway just to rule it out entirely (the moment I get a hold of a programmer).

And in the meantime, I had a second look at the card and found a missing cap (C261) close to one memory module (U31). I am skeptical this could be the culprit, but I definitely will give it a shot 😀

In the same area, pads for supposedly unpopulated components (R395, Q27, Q28 - see Retroweb ID: 12304) are a bit messed up. Probably as a result of physical damage - I don't think anyone tried to solder something there.

I don't have a schematic for this card. Do you know if C261 is a typical 100nF decoupling cap?

Thanks!

New info.
Trying to read the bios IC causes the programmer to disconnect from my PC and throw an "USB not recognized" error, something I have seen before in cases of shorts.
I have checked the resistance of the VCC-GND legs of the EEPROM and I get 86 ohm, confirmed by measuring the 3.3V rail from the AGP slot.

Isn't that low? I get 500-600 ohm on my 9200SE and 9200Pro.

I finally managed to check the BIOS by taking off the IC from the board.
The BIOS image is perfect, matching 008.017D.016.000 found in TechPowerUp.
Without the BIOS IC, the resistance on the 3.3V rail is still 86 ohm.
I also tested the card with those combined AGP-PCIe testers and all light lighted up - no detected shorts or open circuits.

At this stage I am running out of ideas! I'll probably give the card a proper clean, replace the missing cap and try it again.
I found one schematic for the 9600XT, but it's for the BGA version. Does anyone know how valid that is, for my non-BGA card?
I'd like to troubleshoot the 3.3V rail and a schematic would be of great help.

You may find some schematics on elektrotanya.com.
I would download some Gigabyte GV-R95*** Radeon 9550 or GV-R9600NF ones - they are with tsop memory.
+3.3V_BUS rail is connected to many components including GPU.
I took Radeon 9600 and 9550 which were at hand - they have the resistance about 188 Ohm.

qtunnel wrote on 2026-05-22, 20:58:

I finally managed to check the BIOS by taking off the IC from the board. The BIOS image is perfect, matching 008.017D.016.000 fo […]
Show full quote

I finally managed to check the BIOS by taking off the IC from the board.
The BIOS image is perfect, matching 008.017D.016.000 found in TechPowerUp.
Without the BIOS IC, the resistance on the 3.3V rail is still 86 ohm.
I also tested the card with those combined AGP-PCIe testers and all light lighted up - no detected shorts or open circuits.

At this stage I am running out of ideas! I'll probably give the card a proper clean, replace the missing cap and try it again.
I found one schematic for the 9600XT, but it's for the BGA version. Does anyone know how valid that is, for my non-BGA card?
I'd like to troubleshoot the 3.3V rail and a schematic would be of great help.

The 86 Ohms on the 3.3V rail wouldn't concern me to be honest. That would mean 38mA if it was a linear resistor, nothing close to a shorted component.

Roman555 wrote on 2026-05-23, 12:44:

You may find some schematics on elektrotanya.com. I would download some Gigabyte GV-R95*** Radeon 9550 or GV-R9600NF ones - they […]
Show full quote

You may find some schematics on elektrotanya.com.
I would download some Gigabyte GV-R95*** Radeon 9550 or GV-R9600NF ones - they are with tsop memory.
+3.3V_BUS rail is connected to many components including GPU.
I took Radeon 9600 and 9550 which were at hand - they have the resistance about 188 Ohm.

Hey Roman, thanks a lot!
I indeed found the GV-R9600NF-NF2 schematic that matches pretty well my card!
And thanks for taking the time to measure the cards, it's very valuable!

tehsiggi wrote on 2026-05-23, 17:49:

The 86 Ohms on the 3.3V rail wouldn't concern me to be honest. That would mean 38mA if it was a linear resistor, nothing close to a shorted component.

Agree that it's not a full short, but so far it's the lowest readout from what I could find online. Maybe a failing component somewhere, or just the specific Sapphire design 😒

I then tried:
- Washed the card and tried it again. Same behavior.
- Put the BIOS IC back and measured voltages at its pins (attached). The IC is powered correctly from the 3.3V bus and when the card is running the Chip select pin is not pulled down. Do you know if this is normal? I know it needs to be pulled down during initialization, but I am not sure in what state it should be after the card has been powered for a while. I unfortunately do not have a scope to record the power up sequence. I also measured a 2.9V signal on the GPIO9 line to the ASIC, but I don't know if this is compatible with the logic levels of the RV350.

I am insisting investigating the communication with the BIOS IC because I still find interesting that the card can enumerate its model to the OS but ATIFLASH cannot detect it. But maybe at this stage I am chasing ghosts and this case is a classic cracked of cracked ASIC solder balls.

Next in line is a complete recap of the electrolytics. They look OK and voltages are OK-ish but perhaps the ESR is so bad something is failing.
If not, it will be a good practice 😀

qtunnel wrote on 2026-05-17, 22:22:

New info. Trying to read the bios IC causes the programmer to disconnect from my PC and throw an "USB not recognized" error, som […]
Show full quote

New info.
Trying to read the bios IC causes the programmer to disconnect from my PC and throw an "USB not recognized" error, something I have seen before in cases of shorts.
I have checked the resistance of the VCC-GND legs of the EEPROM and I get 86 ohm, confirmed by measuring the 3.3V rail from the AGP slot.

Isn't that low? I get 500-600 ohm on my 9200SE and 9200Pro.

Hi. I found 9600XT copy of yours and measured 3.3V rail to GND. It is the same 86-87 Ohm. I may try to upload some photos if it helps.
And yes, I also found 9550 128/128 of HIS. Its 3.3v rail to GND is about 640 (690) Ohm