It could be a PCI issue, try and disable all embedded peripherals in the motherboard for now, in the BIOS.

Fbi memory = 0 first problem
TMU PowerOn Sense=0xdead = second problem

WARNING: Board 0: Bogus Number of TMUs (57005)

I have had this error before, in my case I think it was a semi shorted tmu chip. Also your fbi chip is not detecting the ram.

Ive already tried a socket 7, SS7 and Socket 462 motherboard so i dont think its a PCI issue.

Does anybody know if the capacitors on these gainward cards are known to fail with age or not ?

i'd double check both sides for missing resistors/capacitors. if you look close enough, you should be able to tell if a component was ripped from the pads.

There was one missing capacitor i found and replaced years ago (one in the bottom left most side of the PCB near the PCI connector). Adding that cap back (identical spec) didnt change the behaviour of the card.

Ive been reading that solid state polymer caps have a lifespan of roughly 23 years, this card is now 25 years, considering it used to work I think it might be work just replacing all those caps and see what happens.

treeman wrote on 2023-03-13, 09:22:

Fbi memory = 0 first problem TMU PowerOn Sense=0xdead = second problem […]
Show full quote

Fbi memory = 0 first problem
TMU PowerOn Sense=0xdead = second problem

WARNING: Board 0: Bogus Number of TMUs (57005)

I have had this error before, in my case I think it was a semi shorted tmu chip. Also your fbi chip is not detecting the ram.

Okay so based on what you said i decided to check the FBI and TMU pins again, and i found some areas where the solder on the pins and pads look sort of cloudy (are these dry joints ?), so i push it with a pin a bit harder and low and behold theres play in it, at least four of them on the FBI chip (i probably missed them last time because i was scared of applying too much force). I dont think i detached them while pressing i think they were already detached.

The TMU 1 has that same cloudy look on some of the pins, but none are moving, though i suspect that this grey stuff between the pins is probably causing a very small short.

I dont think i have the skill to solder it myself, i have bad eyes and not so stable hands, though ill give it a short if i can find some assistance. My guess is the best/safest thing would be to reflow all the pins on the FBI and TMUs if i seek the help of a pro since this is probably going to affect all the pins eventually.

Hopefully it is the pins, then it is a much easier and feasible fix then changing the whole chip.

You really need some kind of magnification for these things to see properly. With the naked eye it is very hard to tell exactly what is happening. Even using your phone camera to zoom all the way in and take a picture helps

Soldered the pins back using a microscope and then tested to make sure they were all attached. Looked good but when i plugged it back in and ran mojo it still had the exact same issues.

My next step will be to try remove the heatsinks so i have better access to all the legs, ill reflow them all, and then ill change all the barrel shaped Polymer capacitors. Im thinking there must be some bad out of the 13, according to my research "solid" caps are not really solid inside and they have a typical lifespan of about 20 years. These are over 25 years old now...so worth a shot.

If the card doesnt work after this, i think its probably got a damaged FBI or TMU chip. Probably only worth keeping as a collection piece or to salvage memory for other voodoos.

Uploading some photos may help a bit. Have You tested all the Ferrite Beads on the board and made sure all voltages are present. If one of the ferrites is bad it can
block the voltages. There is a Youtuber called Vswitchzero worth checking out as He has done some repairs on Voodoo 2's. It helped Me towards diagnosing and repairing my Voodoo 1.

I think the barrel shaped capacitors on the Gainward V2 are regular 'wet' electrolytic capacitors in SMD format.

SuperHanSolo wrote on 2023-03-16, 22:28:

Soldered the pins back using a microscope and then tested to make sure they were all attached. Looked good but when i plugged it back in and ran mojo it still had the exact same issues.

My next step will be to try remove the heatsinks so i have better access to all the legs, ill reflow them all, and then ill change all the barrel shaped Polymer capacitors. Im thinking there must be some bad out of the 13, according to my research "solid" caps are not really solid inside and they have a typical lifespan of about 20 years. These are over 25 years old now...so worth a shot.

If the card doesnt work after this, i think its probably got a damaged FBI or TMU chip. Probably only worth keeping as a collection piece or to salvage memory for other voodoos.

Did you ever get this fixed SuperHanSolo?

I appreciate that this is an old topic and the OP might not be on the forum anymore but I'd like to write this down in case people come across this in future. This thread popped up while I was searching for possible leads on what "0xdead" and 57005 TMUs could mean. Your log where the FBI reports 0mb but drivers install fine actually matches the problem I had with my Voodoo 1 - in my case I was able to find the fault after enabling the diagnostic file output, it seems that running mojo.exe is only half of what's needed to troubleshoot.
This needs to be added into autoexec.bat:
set SST_INITDEBUG_FILE=sst.log
set SSTV2_INITDEBUG_FILE=sstv2.log

after restarting, run mojo.exe again and it should output sst.log into the directory that mojo is run from.
In my case: Re: What retro activity did you get up to today? -- I found that in the SST.log file the MOJO program was failing out on an initial step, where the card first checks whether it can talk to the DAC chip that makes the VGA output on the card. Mine could not because there was some damage to pins / traces going to the DAC chip, so it caused the very failed looking mojo result:
sst1Init Routines: InitCode $Revision: 8 $
sst1InitMapBoard(): BoardsInSystem = 1
sst1InitMapBoard(): vAddr:0xc8dc0000 pAddr:0xdd000008 Dev:0x9 Board:0
sst1InitRegisters(): Setting TREX-to-FBI FIFO THRESHOLD to 0x8...
sst1InitRegisters(): Setting PRELIM FT-CLK delay to 0x8...
sst1InitRegisters(): Could not detect DAC...

That lead to an FBI memory count of 0MB and stopped it from enumerating the TMU(s), leading me to think the FBI was dead but it was actually working a lot more than I thought. After repairing some damaged traces going to the DAC, the mojo output was normal and the card could run games (it's still not 100% but it's working more now).
Since the mojo result in the OP also shows 0MB for FBI memory, it could be a similar problem. The Voodoo 1 / 2 cards seem to be quite forgiving about enumerating memory so it should show at least 1MB FBI memory, and the FBI chip is working well enough for Windows to see the card.

I recently repaired a Voodoo 2 which was rma'd 25 years ago. A friend found PC related stuff from a computershop that went bankrupt several years ago.

The Voodoo 2 was recognised in Windows but every 3DFx compatible game would crash. I checked every resistor and resistor array on the card but all of them were fine. Then I checked the card under a microscope and the lower TMU pins were detached from the PCB. After soldering and reflowing the TMU the card worked perfectly fine.

To the OP, check the pins of the TMU's if it isn't fixed yet.