I just noticed this thread:

Voodoo 5 5500 Repair (project thread)

Seeing as the boot window of win98 is the same as mine:

Im beginning to loose all hope...

I would really not suspect capacitors to be the issue here.
Maybe the EEPROM on the card is bad?
How did you go about flashing the video BIOS?

Edit: Oh, with the last picture this could be a whole lot of problems, maybe even related to capacitors after all. But I would start by examining the video BIOS EEPROM because it all started with the flash.

I flashed the bios according to these instructions:
https://3dfxbios.stantoworld.co.uk/html/faq.html#5

How can I make sure the bios EEPROM is OK? I flashed it successfully multiple times, according to the output of FLASH.EXE ( it said it succeeded and I needed to turn of power)
I booted in msdos by pressing F8 and using the files I prepared on my C drive.

There are multiple instructions on that site. I would always recommend using a PCI graphics card with another vendor's chipset as the primary display adaptor when flashing an AGP card. Did you do this or did you try it blind?
You can make sure the EEPROM is ok by reading its contents and comparing them to the file you flashed onto it.
I'm not familiar with the tool so I can't really tell you how to.

I tried both. First try was on the TK333 system with only the V5 attached. that succeeded but began the problems with the screwed output when I power cycled the machine.
last 3 tries was on the P3 system with a S3 virge dx/gx PCI card as my main output.

I see there is also a dump tool on that site, I will try to make one.

Ensure your PSU has loads of amps on the 5v rail.

texterted wrote on 2020-09-26, 22:03:

Ensure your PSU has loads of amps on the 5v rail.

The card worked fine before I flashed it. I doubt its both of my PSU's that are now insufficiant in both the P3 and the Athlon system?

the P3 has a brand new PSU with 16A 5V, the athlon system has a brand new PSU with 24A 5V.
Again, card worked fine on both systems (albeit it not working with 3d apps on the athlon system) before flashing.

I tried to make a dump of the bios:

left is the supplied 1.06 bios from that site, right is the original saved bios when I first flashed it.

When I flash back the bios that got saved the first flash (the 1.06 from above) I get this:

Seems there is something wrong with the flash indeed.

Could the caps maybe screw up the flashing voltage of the EEPROM? Although it doesnt make sense to me that the program says it succeeds everytime I do it:

Oh wow, I suddenly realised that the dump I made was from the S3 card... haha. The text "S3 incorporated" in the diff with the original bios kinda gave it away.

Im kinda wondering now if the flash tool even targets the voodoo then...

I will try to see what I can do to make a new dump by changing the targeted card or something...

So I made a new dump, while going in blind into dos mode, after using the recovery disk found here https://www.3dfxzone.it/dir/3dfx/utilities/?objid=543 with bios M55K118.ROM:

there are differences, but im not sure if the bios also gets altered while using the card.

Next I will try to flash the 1.06 orignal dump back into the card and try getting a new dump while going into dos blind again. I guess that must be 100% the same.

Wait....to be clear did you just brick a voodoo5 by flashing a s3 bios on it?

kolderman wrote on 2020-09-27, 07:49:

Wait....to be clear did you just brick a voodoo5 by flashing a s3 bios on it?

I thought that for a second too, but it just seems the dump tool makes a bios copy of the wrong card. The first flash I did, was not with other cards in the system. When I got the crap image output, I tried to reflash it in other system with the S3 as primary card to see what I am doing.

I now flashed the voodoo 5 again while going in blind again. I flashed the original bios back, rebooted and then made a dump again blind:

Seems the flashing goes fine for the most part, but there are again differences... looks a bit as if some areas of the EEPROM cant be written or something, but the differences are only a few bytes which makes this weird...

diving a bit deeper in the differences, it looks like the bytes 1A 12 02 00 are written in the original bios at the very end of the flash, while in the dump, these bytes are located halfway in the flash:

While looking a bit at the datasheet of the EEPROM (SST39VF512) https://datasheetspdf.com/pdf-file/511519/Sil … gy/SST39VF512/1 it could be that one of the adres lines are not working maybe?

data 1A 12 02 00 location offsets:

original bios:
0x00007FF8 = 0111111111111000b
dump:
0x00008FF8 = 1111111111111000b

Seeing that the location in flash is adres wise only 1 bit (MSB), it seems to me that when writing the bios, this adres line is at the wrong logical voltage.
I will recap the board this week and try to flash it again. If that doesnt work, I might need to replace the EEPROM.

It is possible the BIOS chip is defective, it happened to me on a Banshee card. I guess the only way to know for sure is to desolder the the old BIOS, buy a new chip, flash it by external programmer and put back on the V5. If you want to go this route, then I recommend to solder PLCC socket on the Voodoo, it makes debugging much easier.

havli wrote on 2020-09-27, 08:37:

It is possible the BIOS chip is defective, it happened to me on a Banshee card. I guess the only way to know for sure is to desolder the the old BIOS, buy a new chip, flash it by external programmer and put back on the V5. If you want to go this route, then I recommend to solder PLCC socket on the Voodoo, it makes debugging much easier.

Any tips on a programmer? I found this one: https://www.ebay.co.uk/itm/NANO-USB-Programme … g-/271313593344

You should be able to put a blank EEPROM and just flash it like usual, without the need of an external programmer.

sdz wrote on 2020-09-27, 10:52:

You should be able to put a blank EEPROM and just flash it like usual, without the need of an external programmer.

that is possibly a cheaper first to try. but if it is something like a disconnected adress pin from the VSA-100 chip to the EEPROM chip, that would not work.
I think I will order a socket for it like havli mentioned and a blank chip first and try that (after the recapping of the card, I already ordered caps anyway)

Can always a programmer later 😀

If it's a broken address trace between the VSA and the EEPROM, an external programmer would not help. The EEPROM content will be good, but the VSA won't read it properly. Highly doubt that is the case though, most likely faulty EEPROM.