Sometimes the BIOS can go bad. If you have an external programmer such as TL-866 II with the proper adapter, you could remove the EEPROM, dump its contents and compare its md5sum against all other known BIOSes you can find for this card. Then after making a backup, try flashing a different one and see how it goes.
Maybe these could be of help https://3dfxbios.cl-rahden.de/index.php?title=Flash-Tools

Small caps like that don't tend to fail so I don't think it's worth swapping those.
I agree with the suggestion of reprogram the BIOS - do you get the "no video card found" PC speaker beeps or not?

Also, if you can check voltages while the card's running, it would be worth probing the voltage regulator at Q1 - I think that should be outputting something like 3.5v

tauro wrote on 2024-01-01, 19:55:

Sometimes the BIOS can go bad. If you have an external programmer such as TL-866 II with the proper adapter, you could remove the EEPROM, dump its contents and compare its md5sum against all other known BIOSes you can find for this card. Then after making a backup, try flashing a different one and see how it goes.
Maybe these could be of help https://3dfxbios.cl-rahden.de/index.php?title=Flash-Tools

Thanks, I will do that first. I need to get the programmer first and the adapter for it. The adapter should be PLCC32 I think..

Thermalwrong wrote on 2024-01-01, 20:25:

Small caps like that don't tend to fail so I don't think it's worth swapping those.
I agree with the suggestion of reprogram the BIOS - do you get the "no video card found" PC speaker beeps or not?

Also, if you can check voltages while the card's running, it would be worth probing the voltage regulator at Q1 - I think that should be outputting something like 3.5v

I didn't connect a PC Speaker actually, so I don't know for sure. I will check the voltages, I need to check the positive output voltage correct? So that would be the first pin positive, the second ground - negative. I will not bother with the caps. Thanks.

It is outputting 1.511V.. not even close to 3.5V. The positive input was even lower at 1.2V.

Slot1 wrote on 2024-01-01, 21:06:

It is outputting 1.511V.. not even close to 3.5V. The positive input was even lower at 1.2V.

This AMS 1085CM chip (voltage regulator?) looks suspicous. What is the brown thing on the side? Dirt or did it burn?

Agree with CoffeeOne, that 1085CM is likely dead.
That brown thing doesn't look like flux at all, but rather a burn mark.

It actually looks like a thin layer of burned flux. That thing was running hot for years probably…
Searching on the internet I found out that there are two versions of this VRM, a 1.5 and 3.3v one. On the Banshee it isn’t specified which one is it.

That AMS 1085CM regulator is the adjustable version since it has no suffix - there's also R8 and R9 making up a resistor divider pair that set the operating voltage. Let's leave that for now and look at other factors.

Please do test with the PC speaker hooked up to see if it gives beeps or not.

On my 2 functional banshee cards the core voltage is ~3.5v in operation so it should be around there really and not 1.5v. Make sure that you're measuring the correct points - I think the AMS1085CM's power input is the 5v line and since it outputs 3.5v normally perhaps your measurement was between the Vout and Vin pins rather than Vout and GND/Adjust. Measure between the tab and the leg that connects to R8/R9, that should be Vout and GND respectively.

What motherboard are you testing the card in?

Thermalwrong wrote on 2024-01-01, 23:37:

That AMS 1085CM regulator is the adjustable version since it has no suffix - there's also R8 and R9 making up a resistor divider […]
Show full quote

That AMS 1085CM regulator is the adjustable version since it has no suffix - there's also R8 and R9 making up a resistor divider pair that set the operating voltage. Let's leave that for now and look at other factors.

Please do test with the PC speaker hooked up to see if it gives beeps or not.

On my 2 functional banshee cards the core voltage is ~3.5v in operation so it should be around there really and not 1.5v. Make sure that you're measuring the correct points - I think the AMS1085CM's power input is the 5v line and since it outputs 3.5v normally perhaps your measurement was between the Vout and Vin pins rather than Vout and GND/Adjust. Measure between the tab and the leg that connects to R8/R9, that should be Vout and GND respectively.

What motherboard are you testing the card in?

I was measuring Vin+GND and Vout+GND. The motherboard is fine, it is actually Gigabyte GA-5AX ATX SuperSocket 7 board. Every other Banshee card works fine on it.
I will test it with the speaker and report back the results as soon as possible. I think it is most likely that regulator, it even has a brown mark around it from the heat.

I measured correctly now, the input voltage is 2.78V and the output 1.25V.

Slot1 wrote on 2024-01-04, 20:19:

I measured correctly now, the input voltage is 2.78V and the output 1.25V.

I can believe in output low because something on the card is shorted. Input comes directly from AGP slot pins B2/B3 in what looks like internal layer, for it to sag those internal traces must be fried, and for that to happen it would imo be more visible, but you never know. With computer turned off measure resistance between regulator pin 3 (the left leg) and 5V coming from power supply.

rasz_pl wrote on 2024-01-05, 03:49:

Slot1 wrote on 2024-01-04, 20:19:

I measured correctly now, the input voltage is 2.78V and the output 1.25V.

I can believe in output low because something on the card is shorted. Input comes directly from AGP slot pins B2/B3 in what looks like internal layer, for it to sag those internal traces must be fried, and for that to happen it would imo be more visible, but you never know. With computer turned off measure resistance between regulator pin 3 (the left leg) and 5V coming from power supply.

The resistence is 000.4Ω.

I'm 90% sure that the BIOS needs a reflash on this card..

Slot1 wrote on 2024-01-05, 13:14:

rasz_pl wrote on 2024-01-05, 03:49:

Slot1 wrote on 2024-01-04, 20:19:

I measured correctly now, the input voltage is 2.78V and the output 1.25V.

I can believe in output low because something on the card is shorted. Input comes directly from AGP slot pins B2/B3 in what looks like internal layer, for it to sag those internal traces must be fried, and for that to happen it would imo be more visible, but you never know. With computer turned off measure resistance between regulator pin 3 (the left leg) and 5V coming from power supply.

The resistence is 000.4Ω.

that means you couldnt have measured 2.78V

Not sure if anyone is still looking at this thread, but I got the card from Slot1.
I have flashed the VBIOS using a flash utility and now the card is working. When booted in Windows and showing the desktop I measured the voltage between the Vout pin (closest to R8/R9) and GND (the mounting bracket). The voltage I measure is 2.138V. Is it advisable to replace something to prevent damage?
The brown spot looks like flux to me; no deformation of the regulator or off color to the PCB.

JWW wrote on 2025-08-06, 14:30:

Not sure if anyone is still looking at this thread, but I got the card from Slot1.
I have flashed the VBIOS using a flash utility and now the card is working. When booted in Windows and showing the desktop I measured the voltage between the Vout pin (closest to R8/R9) and GND (the mounting bracket). The voltage I measure is 2.138V. Is it advisable to replace something to prevent damage?
The brown spot looks like flux to me; no deformation of the regulator or off color to the PCB.

Oh cool, nice that it's fixed 😀 These flash memory chips are getting forgetful but now it's reflashed it should stay working for a long time if it's used.

I think that voltage is kinda weird but maybe my suggestion was incorrect, really if the card works in games it's most likely outputting its correct voltage. It probably would not work at 2.1 volts at stock clocks so is probably running at ~3.5 volts as-is, try some games on it and see how it goes.

I am lucky that the flashing worked indeed. I will see when I have time to try some games and if the voltage will increase when the card gets more load.
I also believe that the voltage is off, that is why I am curious if it is not harmful to the card.

Oddly enough, the measured voltage doesn't really change when running games. I tried NFS2SE and 3. Both in Glide.