Hmmmm, a shot in the darke here, but maybe try the FastVoodoo drivers from 3dfxzone.it, as they supposedly work even on mismatched voodoo2s in SLI?

I did originally run the FastVoodoo drivers when I originally built the system but wasn't a fan of the overclock they put on the cards.

I've just tested the cards in another system (the P200 box I mentioned). Same output from mojo.exe and testsli.exe under DOS. I installed GTA (also under DOS) and copied over the 2.54 glide2x.ovl file and the game runs perfectly, no colour issues etc. It's still not detecting SLI of course but that does seem to point to a Glide issue for the colours etc on the first system (hopefully).

So the question is - what's going on with SLI?!

Are you sure the SLI cable is correct, and no bad solders on the pins of the cards ?

The ribbon cable worked when I first go the cards so it's correct in that sense. I swapped out to a PCB version and that was fine too. I think swapping between them rules them both out as failure points, unless the PCB decided to die and the ribbon died sat in the draw. Seems unlikely to me.

Both cards look ok to me for solder joints and the SLI connector in general, and the edge connector looks ok too? I've included photos.

Additional image

There is a demo of "Donut", this application uses Glide. If you leave only one card in the computer at a time, and run "Donut" - does each card work well? The screen properties look the same?

Thanks for the donut suggestion. I've given that a try which involved installing drivers for the cards in the P200 test system they're in at the moment. It had a fresh Win95 install and installing the 3.02.02 drivers immediately showed 'SLI detected'. But running mojo (Windows version) and it's now showing issues. Re-testing in pure DOS and mojo is happy (but no SLI).

The donut test runs fine with the cards in SLI (rendering in 1024x768 but the same FPS as a single card). I tested each card individually after that (they both tested exactly the same) and mojo was again confused under Windows but ok in DOS, and the donut scene was fine too.

Not sure what to make of the mojo results under Windows - driver issue?

Additional image

I have never used mojo under Windows, only under DOS. So I can't say anything about the behavior of this application under Windows.
So, we have two Voodoo2 cards that work well separately. They also work fine in SLI in DirectX applications. They also work well in SLI in the Donut demo, these are the Glide instructions.
Well, you can also check the work in SLI in the game Quake2 1024x768.
Then you need to think about the results.

Time for a long overdue update.

I came back to this a few days ago and did lots of testing in the Socket 7/430TX/S3 Virge system. Each card works correctly on its own and there is nothing unusual with the mojo results, in Windows or DOS. When configured in SLI I sometimes saw some odd output from mojo, like 'bogus number of TMUs' and sometimes it would mention a power issue. This was intermittent though. Regardless of what mojo said, the cards worked perfectly in games with SLI enabled or disabled.

I transplanted the cards back to the original P3/440BX/Matrox G200 system and now SLI is showing up as 'Detected!' in display properties and, everything works as expected. I tried a few different driver revisions, all totally fine. Same occasional odd messages in mojo, and of course it never said SLI detected (neither did it in the S7 system). I'm beginning to think mojo can't detect SLI?

The only weird issue now - if I disable SLI in the software and run a Glide game the output has pink artifacts at 75Hz, and is very pink at 85Hz. But at 60Hz and 120Hz it's fine. Performance is worse at 120Hz compared to 60Hz though.. which was odd, I thought. V-sync disabled of course. So although this is still weird, maybe it's a compatibility issue with the G200? At any rate, for any games that don't like SLI I do now know how to get it working with SLI disabled.

I think there's a high potential that something isn't 100% right with one or both of my cards, but they do seem to be working properly and gaming well now so... 🤷‍♂️

Maybe using a different PCI slot could help.
And resetting the bios to "default" for Irq allocation.
I don't think Voodoo2 use an Irq, but some strange things happen sometime 😀
Or not related to Irq, but to PCI primary/secondary.
Without any other card, especially network card.

Shadzilla wrote on 2025-12-03, 18:36:

Time for a long overdue update. […]
Show full quote

Time for a long overdue update.

I came back to this a few days ago and did lots of testing in the Socket 7/430TX/S3 Virge system. Each card works correctly on its own and there is nothing unusual with the mojo results, in Windows or DOS. When configured in SLI I sometimes saw some odd output from mojo, like 'bogus number of TMUs' and sometimes it would mention a power issue. This was intermittent though. Regardless of what mojo said, the cards worked perfectly in games with SLI enabled or disabled.

I transplanted the cards back to the original P3/440BX/Matrox G200 system and now SLI is showing up as 'Detected!' in display properties and, everything works as expected. I tried a few different driver revisions, all totally fine. Same occasional odd messages in mojo, and of course it never said SLI detected (neither did it in the S7 system). I'm beginning to think mojo can't detect SLI?

The only weird issue now - if I disable SLI in the software and run a Glide game the output has pink artifacts at 75Hz, and is very pink at 85Hz. But at 60Hz and 120Hz it's fine. Performance is worse at 120Hz compared to 60Hz though.. which was odd, I thought. V-sync disabled of course. So although this is still weird, maybe it's a compatibility issue with the G200? At any rate, for any games that don't like SLI I do now know how to get it working with SLI disabled.

I think there's a high potential that something isn't 100% right with one or both of my cards, but they do seem to be working properly and gaming well now so... 🤷‍♂️

Hello, would you be able to check the voltage regulator resistances to ground per Bits und Bolts video:
https://youtu.be/4oZlt9Dl43I?t=701

I have a STB single-sided 12MB board and the 3.3 and Adj pins measure like BuB's video for good resistance, but get about 230Ohm on the 5v. I have a Monster II that measures perfect 25k, 300, 200, like his video.
Just removed all of the e-caps on the STB and the resistance did not change. Just trying to determine if the deviations from reference make the 5v measurement different from BuB's video, or if the card needs more troubleshooting. (STB and PA 2000 both test fine in DOS, but intermittently show TMUs as dead under Windows. V2 control panel ALWAYS shows cards are good, and they work in games)

Shameless request aside, maybe watch some of BuB's videos on troubleshooting V2's. He has quite a few. Recently he suggested my swapping the cards for testing is possibly exposing fractures in solder joints.
If you need it, Voodoo2 component map + schematics has component values and locations.
This pdf has how it's all wired up: https://sdz-mods.com/wp-content/uploads/2024/ … SX_PCIE_A01.pdf

Hopefully someone....will redo that schematic in KiCAD....

The card in Bits und Bolts video uses a AS2830 voltage regulator
I have a STB single side, and it uses a RC1587.
They are not the same componants, even if they do the same thing.
So you can't expect to find the same values between ground and 5 or 3.3 volts.

nali wrote on 2025-12-05, 20:48:

The card in Bits und Bolts video uses a AS2830 voltage regulator I have a STB single side, and it uses a RC1587. They are not th […]
Show full quote

The card in Bits und Bolts video uses a AS2830 voltage regulator
I have a STB single side, and it uses a RC1587.
They are not the same componants, even if they do the same thing.
So you can't expect to find the same values between ground and 5 or 3.3 volts.

Would you mind measuring yours and post back what you get for the STB? I have another one on the way.
Also acquired a Monster II 3D that measures exactly the same as BuB's, works in SLI with the STB, but solo or in SLI has dots on the textures, and "checker-boxing" on the textures.

Sorry, I wanted to do it when I first answered, and forgot.
Then saw your message yesterday, and forgot again 😀
Here is what I have :

AS2830 and RC1587 have the same pinout, Adj-Out-In, like the LM317.

On my STB :
Gnd to Adj : 110 Ohms
Gnd to Output : 94 Ohms
Gnd to Input : 4.34 kOhms

Note that even if my card work perfectly, I can't garantee those values are the same as a new card.
There's the possibility some caps leak without causing any problem for now.

Datasheets, often useful :
https://media.digikey.com/pdf/Data%20Sheets/F … 587_Rev2001.pdf
https://www.alldatasheet.com/html-pdf/102968/ … 9/1/AS2830.html