yellow thing is a tantalum capacitor which may fail very often shorting through itself, may blow up very loud sometimes.
replace capacitor and most likely it will restore power for 3dfx chip

It's really that easy? I ordered a few of these and will give it a try

So i've got the capacitors today and soldered one in.
Now i get graphic output through the voodoo, but it still isn't "there". detect.com doesn't see it.
Also the mosfet still gets very hot.
After exchanging it i measured it again and there's still a short. I removed the cap again and measured the two solder points for it, and there's a short on it.
What does that mean?

MrWeedster wrote on 2025-05-09, 10:22:

So i've got the capacitors today and soldered one in. Now i get graphic output through the voodoo, but it still isn't "there". […]
Show full quote

So i've got the capacitors today and soldered one in.
Now i get graphic output through the voodoo, but it still isn't "there". detect.com doesn't see it.
Also the mosfet still gets very hot.
After exchanging it i measured it again and there's still a short. I removed the cap again and measured the two solder points for it, and there's a short on it.
What does that mean?

As far as I remember, that capacitor is directly connected to multiple VCC pins of all 3 chips (FBI + TMUs). So, if you indeed still see a short, you might be out of luck. 🙁
Then again, it's very likely that it also powers most other components on the board. A thermal camera might help to see where the short is located.

bloodem wrote on 2025-05-09, 10:39:

As far as I remember, that capacitor is directly connected to multiple VCC pins of all 3 chips (FBI + TMUs). So, if you indeed still see a short, you might be out of luck. 🙁
Then again, it's very likely that it also powers most other components on the board. A thermal camera might help to see where the short is located.

Is there a schematic where to look?
I don't understand how a thermal camera helps in this process. Care to explain?

MrWeedster wrote on 2025-05-09, 10:50:

Is there a schematic where to look?
I don't understand how a thermal camera helps in this process. Care to explain?

If a component is shorted, it tends to draw a large amount of current, which is converted into heat - hence, the thermal camera will help to see which component is quickly heating up for no apparent reason.
I don't think a (good) schematic exists in the wild, though I might be wrong. I, for one, just trace the connections manually, depending on what the issue is.

Well as i said the most obvious hot component is the mosfet besides the cap.

MrWeedster wrote on 2025-05-09, 10:57:

Well as i said the most obvious hot component is the mosfet besides the cap.

That voltage regulator tends to get hot, so it's probably expected. Also, you mentioned that you already replaced it - so you eliminated it as a potential root cause.
Plus, since there's a short, the voltage regulator is bound to get hotter than usual.

Well i didn't replace the mosfet, only the cap besides it which i marked red in the first pic. It also gets warm on my other V2, but not like this. i guess it's around 50 or 60 degrees celsius.

MrWeedster wrote on 2025-05-09, 11:04:

Well i didn't replace the mosfet, only the cap besides it which i marked red in the first pic. It also gets warm on my other V2, but not like this. i guess it's around 50 or 60 degrees celsius.

Ah, OK, then I misunderstood what you wrote here:

MrWeedster wrote on 2025-05-09, 10:22:

[...]
Also the mosfet still gets very hot.
After exchanging it i measured it again and there's still a short. [...]

In that case, you should definitely replace that MOSFET, because it should not get very hot in a short amount of time.

It just so happens that I have a Diamond Monster Voodoo 2 on by bench right now, these are the temps I'm seeing after the card has been sitting idle for ~ 10 minutes, in an open test bench (the MOSFET does not exceed 35 - 40 degrees):

The "it" in this sentence referred to the tantalum capacitor which was earlier identified as some root cause, which doesn't seem to be the case.
But why change it? You've said earlier that the mosfet is intended to get hot when there's a short somewhere.

MrWeedster wrote on 2025-05-09, 11:14:

The "it" in this sentence referred to the tantalum capacitor which was earlier identified as some root cause, which doesn't seem to be the case.
But why change it? You've said earlier that the mosfet is intended to get hot when there's a short somewhere.

Did you identify another larger heat source on the board? Did you measure the output voltage of the MOSFET? (it should be ~ 3.3 - 3.4V).
I definitely did not say that "it's intended to get hot". I said that it might get hotter than usual even if there's a short somewhere else (and depending on where that short might be). But, as far as you know right now, it doesn't seem to be the case and the MOSFET is the only component that quickly heats up. I was also writing under the assumption that the MOSFET had already been replaced.

bloodem wrote on 2025-05-09, 11:19:

Did you identify another larger heat source on the board?

Not yet, i try to keep it powered on as short as possible to not get any further damage

Did you measure the output voltage of the MOSFET? (it should be ~ 3.3 - 3.4V).

I guess i don't know how to do that and watching some videos didn't enlighten me on the matter. How do i measure it exactly?
[E]
So i guess i figured it out: Take black to any ground and measure the three legs of the mosfet.
On the good card i get: 1,3V Gate, 0,06 Drain and 1,4V Source
I need to resolder the cap to the bad card and measure, takes a moment.

I definitely did not say that "it's intended to get hot". I said that it might get hotter than usual even if there's a short somewhere else (and depending on where that short might be). But, as far as you know right now, it doesn't seem to be the case and the MOSFET is the only component that quickly heats up. I was also writing under the assumption that the MOSFET had already been replaced.

Sorry, i misunderstood

So i soldered the cap back in, plugged the card in, started the system.
The mosfet got really hot again really fast (90 degrees after 30 seconds or so). I started detect.com and shut down the system again (no card found).
Let the mosfet cool a bit and started again. Then i measured voltages on the three legs of the mosfet. Source was 3,4V, but drain also? Then i guess i shorted Source and Drain after which the readings went to values like on the good card: 1,4V source, 0,04 drain and 1,3 gate. Restarted the system after which the mosfet doesn't get hot anymore, detect.com sees the card, and after another reboot mojo reports the card as OK.
It runs now for around 5 minutes, and the mosfet is at 45 degrees celsius. I checked for heat and found that the left upper 3dfx chip gets hot on the backside of the card around 45 degrees celsius.
I looked further and it seems that the marked resistor (?) is the one getting hot (51 degrees). I measured it and it has 0,000 Ohms.

I'll give windows and opengl a try

[E]
I tried the card standalone, it loaded a game but crashed as soon as the 3d pic was loaded. I dropped it back into the V2 SLI system and it works. I can do 1024x768 and have 78FPS in glquake on that resolution.
I'd say it's good enough, but if we could sort out the heat problem of that resistor i'd be glad.

That "Resistor" is a ceramic capacitor in fact, as indicated by its designation "C145". Ceramic caps can fail as well, while not as often or usual as old tantals. Often due to mechanical / thermal stress. If you feel like it, you could give it a shot to carefully remove it and see if that "hotspot" remains. Probably someone else here can provide you with the exact value it has. From the looks I would bet on something like 10µF - 22µF.

Edit: Just checked this thread: Voodoo2 component map + schematics

It would be 4.7µF - 6.3V - X7R

tehsiggi wrote on 2025-05-09, 12:52:

That "Resistor" is a ceramic capacitor in fact, as indicated by its designation "C145". Ceramic caps can fail as well, while not as often or usual as old tantals. Often due to mechanical / thermal stress. If you feel like it, you could give it a shot to carefully remove it and see if that "hotspot" remains. Probably someone else here can provide you with the exact value it has. From the looks I would bet on something like 10µF - 22µF.

Edit: Just checked this thread: Voodoo2 component map + schematics

It would be 4.7µF - 6.3V - X7R

Thanks, pretty obvious it's a cap with the indication. Measuring that cap on a good card also has 0,000 Ohms, so i guess that's no hint about a bad cap.
I'd guess that this cap is also a symptom, but i'll order a bunch of these and exchange it.

AS2830 is not a MOSFET. It is a low power adjustable voltage regulator with a current of 3A.
Сapacitor C145 should never heat up, its capacity is 2.2µF

Thanks for clarification. So i have to get a 2.2µF 6.3V - X7R?
Are these numbers in the picture supposed to be voltages? If so my cards have pretty different readings compared to that.
I did some testing and yes, SLI gets detected and 1024x768 works, but my performance isn't what it's supposed to be. For example i now have 148fps in glquake 640x480 compared to 184 in phils test sheet about the V2 and 63 in 1024x768 compared to 82.
At least it's working for now. Will order those caps if those are the right ones and see what exchanging it does.

I think you are measuring the voltage incorrectly. The black conductor from the power supply is the ground, common, and all measurements are taken relative to it. Capacitors can have a capacity of 2.2 to 100 μF, the voltage is better with a reserve, for example, from 10 volts and higher. The main thing is that its dimensions are the same and it is convenient to install it in its place. This is a filter capacitor in the power supply circuit, even if you remove it, the card will still work. But you will be able to measure its parameters, check it for a short circuit. Maybe it is working, and something else is heating up.