As for C213, the missing pad was connected to the via directly below (or above, if the card is not upside down). It's very clear to see in the picture.

Are you planning to play games on this card? Because it's very weak.

I have the version with support for two monitors and no TV tuner. I don't think it's damaged but it also produced screen corruption running 3DMark2000 in Windows XP. I don't think the drivers for these cards are good with games.

Errius wrote:

Are you planning to play games on this card? Because it's very weak.

I have the version with support for two monitors and no TV tuner. I don't think it's damaged but it also produced screen corruption running 3DMark2000 in Windows XP. I don't think the drivers for these cards are good with games.

I am aware it's not the best gaming card in the world... I bought it to mess around with making a Win98-powered video wall, which is absolutely a productive use my my time & money in 2019, and it's literally what these cards were made for. I have lots of better gaming cards but none with quad-head support.

derSammler wrote:

As for C213, the missing pad was connected to the via directly below (or above, if the card is not upside down). It's very clear to see in the picture.

Do you mean here? If so, that's not too bad... There are little contact points like that for the other cap too. They're TINY, but doable.

^^ this pic is about 4x actual size on a 13" laptop screen BTW.

Can I replace these SMD caps with a through-hole tantalum of the same rating, or are they "special" somehow? I think it'd be easier if I had legs to solder to.

Yes, that's the spot I mean. Also yes, you can replace that cap with a tantalum one.

Update: well, I managed to replace the missing caps - sort of. For C319 (at the front of the card) I replaced the original one, but on C213 (under GPU #2), I used an appropriately-rated electrolytic since my local shop didn't have any tantalums.

Incidentally working on these was pretty hard, this stuff is TINY. I was able to determine with a multimeter that my solder joints are "good" and both sides of each cap connect to where I think they're supposed to, but I didn't try to test their capacitance in-situ (my meter can't do that.)

Results:
- the texture corruption (see Drakan screenshot, above) in 3D games is fixed. To be honest, that wasn't the problem I was expecting to fix, but I'll take it.
- the vertical banding is still present in 3D games (not on the desktop as far as I can tell)
- GPU #2 still just shows a black screen 🙁

The weird thing with GPU #2 is the sync signals are apparently present, the monitor on that display shows the right resolution & refresh, and responds when you change it independently of the others. Just the picture shows up black.

My "hypothesis" is that C213 is part of an amplifier circuit for the RGB signals (only!) on that GPU? Maybe the electrolytic I used can't charge and discharge fast enough to cope with a video signal??

I would love to hear some other opinions on this!

The technical reference for the G200 is available, but I'm not an engineer, and without a schematic for this specific card, I haven't been able to figure out what those components actually do...

I did harvest some 10µF tantalums from another board (through hole - the yellow blob type), but 20+ year old recycled tantalums tend to go 'pop', so I didn't wanna use them. Soldering & de-soldering on this spot was a big enough pain I don't want to do it unnecessarily. Is it worth trying one?

Capacitor type isn't gonna have such an effect, or shouldn't anyway. There ripped pad might connect to more than one place and things won't work if all those got connected. The card has 4x copies so you could desolder parts on the other sections to see exactly where things end up and restore things on the faulty part.
There's also a missing part on the repair near connector end, a small capacitor it looks like.