3dfx Voodoo5 glitches

[Justin1091]

Re: 3dfx Voodoo5 glitches

I did find something with Q1-Q4 (IR IRF7811A = MOSFET N-CH 28V 11.4A 8-SOIC)

Bad card Q1-Q4 to gnd
Q1 0.5v 1 side 3.0v other side
Q2 1.75v to 3.0v
Q3 3.5v to 3.0v
Q4 12v to 3.0v

Good card Q1-Q4 to gnd
Q1 0.5v to 2.8v
Q2 1.9v to 2.8v
Q3 3.5 to 2.8v
Q4 12v to 2.8v

No idea if that tells anything, what do you think? I've very little experience with this stuff. Learned a lot from the guy at vogons.

I would like to try your option, can you tell me how I would accomplish it?

It is possible that you could have a bad gpu BGA issue. I think the front chip runs all the time and the rear chip only kicks in during 3D detection rendering.

I hope not, but the card works fine without guardband clipping, checked in 3Dmark2000 against the good card with single chip and dual chip performance. Single chip and dual chip had the same scores with both cards (double with dual chip mode that is).

[Sparkey55]

Re: 3dfx Voodoo5 glitches

I did find something with Q1-Q4 (IR IRF7811A = MOSFET N-CH 28V 11.4A 8-SOIC)

Bad card Q1-Q4 to gnd
Q1 0.5v 1 side 3.0v other side
Q2 1.75v to 3.0v
Q3 3.5v to 3.0v
Q4 12v to 3.0v

Good card Q1-Q4 to gnd
Q1 0.5v to 2.8v
Q2 1.9v to 2.8v
Q3 3.5 to 2.8v
Q4 12v to 2.8v

No idea if that tells anything, what do you think? I've very little experience with this stuff. Learned a lot from the guy at vogons.

I would like to try your option, can you tell me how I would accomplish it?



I hope not, but the card works fine without guardband clipping, checked in 3Dmark2000 against the good card with single chip and dual chip performance. Single chip and dual chip had the same scores with both cards (double with dual chip mode that is).

Well as far as the DC voltage readings are concerned they look OK compared to each other. Some deviation most likely because of parts tolerances. But are these Test Points for measuring both voltages and AC waveforms or what? Here is where a scope comes in handy. would probably need a steady test pattern to know.

[Justin1091]

Re: 3dfx Voodoo5 glitches

No, I took the voltages directly from the pins of Q1-Q4 to GND (molex connector).

As far as the TP's go (this is taken from a post of me in the vogons forum topic).

Bad card TP to GND
TP1: nothing
TP2: updated: 1.66-1.70v (3d same)
TP3: nothing
TP4: nothing
TP5: 0.01V (3D the same)
TP6: 1.66-1.70v (3D same)
TP7 through 11: nothing

Good card TP to GND
TP1: nothing
TP2: 1.62v(3D same)
TP3: nothing
TP4: nothing
TP5: 0.01 to 0.03 (3D same)
TP6: 1.62v(3D same)
TP7 through 11: nothing

Downclocking the card from 166mhz to 133mhz did not show a difference in voltage on those Tp's (my guess was that the 1.66v was linked to 166mhz, it isn't).

Update: also did TP's to 5v and 12v (card's molex connector)

Bad card:
TP2 to 5v: 3.20v
TP2 to 12v: 9.88v

TP5 to 5v: 5V
TP5 to 12v: 12.24V

TP6 to 5v: 3.22v
TP6 to 12V: 9.82
3d no big difference (when it crashes that is). Single chip or dual chip no difference

Good card
TP2 to 5v: 3.4v
TP2 to 12v: 10.55v

TP5 to 5v: 5v
TP5 to 12v: 12.18v

TP6 to 5v: 3.4v
TP6 to 12v: 10.55v

Now the resistance on the TP's
TP1: nothing
TP2: 11.86 for good card, 11.90 for bad card
TP3: nothing
TP4: nothing
TP5: nothing
TP6: 12.42 for good card, 11.16 for bad card
TP7 through 11: nothing

Metal plate near Q5 (labeled Q6) reads 3.32v for both cards. Both chips use 3.3v by the way (as written in 3dfx's VSA-100 pinout document). VSA-100 is the name of the chips on the card. Don't know if it's possible to measure the voltages the chips get.
I don't have a oscope :/. Wouldn't know how to use one either.

[Sparkey55]

Re: 3dfx Voodoo5 glitches

Too bad these are not socketed chips. Anyway to isolate the power rails get some jumper cables and only connect either 5V or 12V with the jumpers from the PSU 4 pin molex going to the cards power plug. Like I said I have never done this myself. WARNING THIS HAS THE POTENTIAL RISK TO CAUSE DAMAGE. ATTEMPT AT YOUR OWN RISK.

Personally I would not want to try that method on a Voodoo5. I read about that method many years ago in a galaxy far far away. This was back when Voodoo5 cards were more plentiful and cheaper as compared to today.

[Justin1091]

Re: 3dfx Voodoo5 glitches

Ok, not sure yet if the risk is high. But, I would connect the 5v wire and 1 ground to the card? Then measure what on the card? Same for 12v, take the 12v wire and it's ground and connect it?

Going to wait a little more to see if more tips come in. Maybe there are components (that are perhaps known to fail after ~18 years like the caps) I can replace cheaply? Some resistor networks (labelled 472) measure only 4.60. I can't find their tolerance online. Do you think replacing them might help? This is just an example, there might be more components.

Can we rule out a bad bga chip yet? Because they work without guardband clipping. Suppose one is bad after all, it's done right? Nothing I can do to fix it (did use the oven trick with a few modern cards which worked very well. Don't think that works with bga chips).

Thanks so far by the way, I really appreciate your suggestions and tips!

[Per Hansson]

Re: 3dfx Voodoo5 glitches

I hope not, but the card works fine without guardband clipping, checked in 3Dmark2000 against the good card with single chip and dual chip performance. Single chip and dual chip had the same scores with both cards (double with dual chip mode that is).

Uhh what?
Am I being thick here, I thought you said SLI did not work with Guardband clipping?
If the card works fine in SLI with Guardband clipping disabled why not just run like that?
Both Guardband clipping and HSR (hidden surface removal) are quite advanced features.
Would not surprise me if there are quite a few situations where they don't work due to 3DFX going bankrupt during the development of the Voodoo5

The final thing to keep in mind is that Guardband clipping is only really useful with a weak CPU, like the slow FPU on an AMD K6 series CPU.
It should not give a very noticeable performance difference if you have some CPU power.
Say a decent Pentium III setup.

[Sparkey55]

Re: 3dfx Voodoo5 glitches

Well we need to know what wattage the card pulls at max, Then figure half that watts divides up between the two power rails. by isolating the rails on at a time that means the card will want to pull all of the watts from the connect rail. This in effect will double the current thru that rails regulator; not a good thing to happen. Will need to limit the current using a resistor to half of whatever it normally uses; an inline resistor to the power rail in use at the time of testing.

[Sparkey55]

Re: 3dfx Voodoo5 glitches

Uhh what?
Am I being thick here, I thought you said SLI did not work with Guardband clipping?
If the card works fine in SLI with Guardband clipping disabled why not just run like that?
Both Guardband clipping and HSR (hidden surface removal) are quite advanced features.
Would not surprise me if there are quite a few situations where they don't work due to 3DFX going bankrupt during the development of the Voodoo5

The final thing to keep in mind is that Guardband clipping is only really useful with a weak CPU, like the slow FPU on an AMD K6 series CPU.
It should not give a very noticeable performance difference if you have some CPU power.
Say a decent Pentium III setup.

Glad you spotted that. What is he trying to accomplish here?

[Sparkey55]

Re: 3dfx Voodoo5 glitches

For a few weeks I've been trying to fix a voodoo5 5500 agp video card. The machine locks up when running a 3d application in dual chip mode. I found out that it happens when a driver feature called 'guardband clipping' is enabled.

Disable guardband clipping. Now what happens?

[Justin1091]

Re: 3dfx Voodoo5 glitches

Yes, guardband clipping doesn't work with SLI. The good card doesn't have this problem ( I said this in my first post) and if I want to for example play a DOS game I can't because it freezes and afaik you can't turn it off there (set env doesn't work). I didn't know for certain guardband clipping was the issue until some time ago. My point is the card obviously has some problem because it won't run with the original drivers and I have to use third party control panels to disable guardband clipping. Besides, the driver turns in on again sometimes. Default setting is enabled.

I tried a bios editor to disable it but couldn't find a way how, besides it would be nice to just fix the card so it can run like it's supposed to

[Sparkey55]

Re: 3dfx Voodoo5 glitches

I am out of ideas for now for checking the card for defective parts. The most clear results would be to remove a suspected chip and replace with a known good one until it works like you want it to. In the meantime enjoy what you have working with more advanced drivers.

[brethin]

Re: 3dfx Voodoo5 glitches

I want to see pics of after you replaced the caps, I'm %100 sure you broke the card putting in a new cap but I want to see it so I can show you.

[Justin1091]

Re: 3dfx Voodoo5 glitches

I want to see pics of after you replaced the caps, I'm %100 sure you broke the card putting in a new cap but I want to see it so I can show you.

Sure, here they are: https://imgur.com/a/crso8IS
I apologize for the quality of the photo's. The black spots you see are because of the bad shape the lens of my phone's camera is in. I can provide a clear picture soon.

You say I broke the card, but it still runs as before replacing the caps. Nothing has changed. I've been using it for three weeks now after replacing the caps.

[momaka]

Re: 3dfx Voodoo5 glitches

Can we rule out a bad bga chip yet? Because they work without guardband clipping. Suppose one is bad after all, it's done right? Nothing I can do to fix it (did use the oven trick with a few modern cards which worked very well. Don't think that works with bga chips).

Unfortunately, I have seen VooDoo cards listed with showing artifacts on eBay before, so just because the GPUs on these VooDoo cards are not the flip-chip type, doesn't mean they can't go bad either. On that note, I also have a defective GeForce 4 TI 4600 card that shows a ton of artifacts, despite using the older non-flip-chip technology. So I think it is still possible to have a bad GPU chip on yours.

The oven "trick" indeed might not work with these cards, but I don't know for sure. I tried reflowing my defective TI 4600 mentioned above, and it worked for maybe 30 minutes without artifacts. But I don't think I heated it enough, so that could be the reason why. Thus, consider the reflow experimental at best.

Other than that, I'm not too sure what else could be the problem either.
As Sparkey says, you can either use the card as is with the 3rd part drivers to get partial functionality or try and reflow it. But be very careful if doing the latter. After all, these cards are worth quite a bit, so a partially working card is still better than none at all. You could make a decent chunk of change from it even in its partially-working condition.

[Justin1091]

Re: 3dfx Voodoo5 glitches

Unfortunately, I have seen VooDoo cards listed with showing artifacts on eBay before, so just because the GPUs on these VooDoo cards are not the flip-chip type, doesn't mean they can't go bad either. On that note, I also have a defective GeForce 4 TI 4600 card that shows a ton of artifacts, despite using the older non-flip-chip technology. So I think it is still possible to have a bad GPU chip on yours.

The oven "trick" indeed might not work with these cards, but I don't know for sure. I tried reflowing my defective TI 4600 mentioned above, and it worked for maybe 30 minutes without artifacts. But I don't think I heated it enough, so that could be the reason why. Thus, consider the reflow experimental at best.

Other than that, I'm not too sure what else could be the problem either.
As Sparkey says, you can either use the card as is with the 3rd part drivers to get partial functionality or try and reflow it. But be very careful if doing the latter. After all, these cards are worth quite a bit, so a partially working card is still better than none at all. You could make a decent chunk of change from it even in its partially-working condition.

Heh, I saw a video on YouTube https://www.youtube.com/watch?v=1AcEt073Uds
No idea about this video or the guy, what he says makes sense though. Cards I put Into the oven last a couple of months before I have to 'bake' them again (Nvidia 9800GT and AMD 6870). I don't think I'm going to try the oven trick with this one. I probably will never find out what's wrong with it, maybe it was bad when it came out of the factory. I did found out that when setting everything in the driver to high quality it also freezes. Some '472' resistor networks read 4.50. No idea if that's within the tolerance limit of the resistor networks. That is the last thing I can think of, case closed for now .

[Justin1091]

Re: 3dfx Voodoo5 glitches

I want to see pics of after you replaced the caps, I'm %100 sure you broke the card putting in a new cap but I want to see it so I can show you.

Can you tell me how I broke the card? Since you're 100% sure, your input can most likely help me in the future when I try to diagnose cards.

[Sparkey55]

Re: 3dfx Voodoo5 glitches

Heh, I saw a video on YouTube https://www.youtube.com/watch?v=1AcEt073Uds
No idea about this video or the guy, what he says makes sense though. Cards I put Into the oven last a couple of months before I have to 'bake' them again (Nvidia 9800GT and AMD 6870). I don't think I'm going to try the oven trick with this one. I probably will never find out what's wrong with it, maybe it was bad when it came out of the factory. I did found out that when setting everything in the driver to high quality it also freezes. Some '472' resistor networks read 4.50. No idea if that's within the tolerance limit of the resistor networks. That is the last thing I can think of, case closed for now .

The previous owner could have screwed the pooch on that card and like a lot of assholes on eBay try to sell it as good when they know it is bad anyway.
A few years ago a seller on eBay had trays of the SLI-100 chips new from the factory for auction.

[Justin1091]

Re: 3dfx Voodoo5 glitches

The previous owner could have screwed the pooch on that card and like a lot of assholes on eBay try to sell it as good when they know it is bad anyway.
A few years ago a seller on eBay had trays of the SLI-100 chips new from the factory for auction.

Yeah, that's probably how it went, fucking assholes. I did find Voodoo5 6000 chips on eBay, but they're clocked at 180~ MHz, not 166 Mhz. I would replace one or both of the chips with a working one (since the card is rare and it would be cool to restore it), but it would have to be the exact same chip I think. I can't do it, maybe there are companies over here in The Netherlands who can do this for me. Doubt it though.

If I ever see a VSA-100 chip for the 5500 AGP I'll buy it and sit on it until I can find someone who can replace them.

I have another good working card so this isn't about having a working Voodoo5, but because I've been working on it for so long I don't want to give up hah, no matter how long I store it.

[Sparkey55]

Re: 3dfx Voodoo5 glitches

If you do find a seller with the VSA-100 chips buy at least 2 or more if you can get them cheap.
One other thing to try is to put the card in the freezer overnite then let it thaw to room temp for a couple of hours before trying again, it might work, if not, nothing lost.

[Per Hansson]

Re: 3dfx Voodoo5 glitches

You can use the Voodoo5 6000 VSA chips on a 5500 or 4500.
That's what's in the eBay link below, but the aliexpress link is for the regular VSA-100 chips used on the 5500.

https://www.aliexpress.com/item/100-...935691345.html

https://www.ebay.com/itm/A-TRAY-OF-2...0/273257152276

If you need help to solder it you can try contacting member Th3_uN1Qu3 and see if he'd be willing to do it for you.
Of course there is always the possibility that the issue is not with the VSA-100 chips...