VOGONS


First post, by christal87

User metadata
Rank Newbie
Rank
Newbie

Hi,

I have just started to feel a retro fever a year ago again. One of my teenagehood's card was a V2. I used to play a lot of Glide titles, before switching to a GF256DDR that had HW T&L.
Bought a cheap noname (STB?) card here in Hungary from a well known source. It could be one those late 12MB V2s with 1Meg EDO DRAM chips. I bought this card with a slightly pinkish/purplish output. After creating a video file consisting of the SMPTE color bars and individual RGB color fills and a white one, went ahead and played it back using a dshow filter video playback utility on the card. The 10+ yrs old TFT monitor was too good to direcly eye out the problem. I suspected that either red is too much or one of the other two color signals (green) are weak. So hooked up the individual color tests again to a scope. Well wha'd ya know... All the outputs were just below 700mV, excluding the green color bursts which were under 400mV. I started to suspect damage to the output selector IC (PI5c3384), since this card has unpopulated TVS diodes. After measuring the output signals directly from the RAMDAC (ICS5342) green was OK, but still not up to the VESA standard somewhere between 0.7-1.5V. Yes, one of my S3 cards have an output just under 1.5! Went ahead and ordered both the output selector and the RAMDAC from UTsource, exchanged them and voila! It's good as new!

Anyone experiencing such symptoms on a V2 can grab these and with some proper soldering skills with a cheap, but temperature controlled chinese hot air station can fix it. Just watch out for the GNDs. You will have to apply the right amount of heat. My first attempt had the same problem and original glide drivers crashed right after loading Windows. Fastvoodoo drivers tend to crash when I started loading a game. It was just a floating ground. The card only has a single quartz attached to the RAMDAC. I believe every other clock is synced to the RAMDAC, so if you have a RAMDAC/xtal main clock issue it might act like this.

On the other hand, I have two other cards with freezing up or no display output. I still have to develop a better way to debug cards then mojo.exe and enabling logs with SSTV2 envvars to a serial port or using the donut and other demos. Especially a good EDO DRAM testing technique. Also I started to reverse engineer the working card to a KiCAD schematic diagram. It's still in the early phase consisting of only the output selector, RAMDAC and parts of the clock synchronization circuitry. Unfortunately I don't have too much time for a hobby nowadays...

Good luck to you all!
Chris

Attachments

  • symptom.jpg
    Filename
    symptom.jpg
    File size
    274.12 KiB
    Views
    1972 views
    File comment
    NFS before the fix
    File license
    CC-BY-4.0
  • v2_green_weak.png
    Filename
    v2_green_weak.png
    File size
    23.89 KiB
    Views
    1972 views
    File comment
    Green output level weak before the fix
    File license
    CC-BY-4.0
  • the_fix.jpg
    Filename
    the_fix.jpg
    File size
    1.87 MiB
    Views
    1972 views
    File comment
    Replacing parts
    File license
    CC-BY-4.0
  • output_fixed.jpg
    Filename
    output_fixed.jpg
    File size
    1.32 MiB
    Views
    1972 views
    File comment
    NFS after the fix
    File license
    CC-BY-4.0
  • v2_repaired.png
    Filename
    v2_repaired.png
    File size
    31.42 KiB
    Views
    1972 views
    File comment
    Output level after fix
    File license
    CC-BY-4.0

Reply 4 of 16, by The Serpent Rider

User metadata
Rank l33t++
Rank
l33t++

Interesting. I have similar problem which only occurs if I connect pass-through cable.

I must be some kind of standard: the anonymous gangbanger of the 21st century.

Reply 5 of 16, by Thermalwrong

User metadata
Rank Oldbie
Rank
Oldbie

Those are some really interesting insights into how the card works and how to troubleshoot these cards when things aren't working right, thanks.
Nice clean work replacing the RAMDAC too 😀

Reply 6 of 16, by christal87

User metadata
Rank Newbie
Rank
Newbie
The Serpent Rider wrote on 2020-03-15, 10:35:

Interesting. I have similar problem which only occurs if I connect pass-through cable.

That could be just the PI5c3384 gone bad or the pass-through cable shorting out somewhere etc.
My pass-through cable from China for example isn't made in a very quality controlled environment.
The cable I ordered first has a contact issue, turning the screen yellow when I poke it.
At least it was cheap!

That output switching IC is just working like a switch. Picture 4+4 MOSFETs switched by two enable signals.
If one of those internal switches catch electrostatic discharge, they can fail.
Does your main non-3DFx video card work okay hooked up just by itself?

Reply 7 of 16, by christal87

User metadata
Rank Newbie
Rank
Newbie
Thermalwrong wrote on 2020-03-15, 10:59:

Those are some really interesting insights into how the card works and how to troubleshoot these cards when things aren't working right, thanks.
Nice clean work replacing the RAMDAC too 😀

Thanks! 😁 My pleasure!

Reply 9 of 16, by christal87

User metadata
Rank Newbie
Rank
Newbie
maxtherabbit wrote on 2020-03-15, 14:14:

V2s have very noisy video output - I wonder if there a is a pin-compatible DAC which would yield better output?

Not sure. The other question is: are they still available like the ICS? The ICS has an RSET pin to set DAC output levels on each channel by resistors to GND. Like one would have it on an RGB LED controller's sink current setup. In theory levels can be raised even further. If that S3 card outputs 1.5V this should cause no harm. Well at least for the monitor itself. I've seen some cards on the net using a Texas RAMDAC. Seems pin compatible enough. Maybe it was just TI acquiring ICS GENDAC. I'm not very good at RAMDACs, but if someone could mimic the behaviour of these DACs with like a CPLD plus a modern RAMDAC it could be possible. Put it on a castellated PCB and done, pin compatible. When it comes to manufacturing of course 4+ layers are not cheap for a small,but complex PCB.

Also the noisy output has to do with the long signal path involving an active component (selector IC) in it too. It looks odd now, but this blending filtering 3DFx used on the small scale pixelated textures of that time, to be smooth was quite useful. Nowadays it's all about fast AND crisp output. Times sure have changed.

Attachments

Reply 10 of 16, by The Serpent Rider

User metadata
Rank l33t++
Rank
l33t++
christal87 wrote on 2020-03-15, 13:50:

That could be just the PI5c3384 gone bad or the pass-through cable shorting out somewhere etc.

That's not ordinary Voodoo 2. It's Canopus 3D card with custom cable.

I must be some kind of standard: the anonymous gangbanger of the 21st century.

Reply 11 of 16, by christal87

User metadata
Rank Newbie
Rank
Newbie
The Serpent Rider wrote on 2020-03-15, 16:47:
christal87 wrote on 2020-03-15, 13:50:

That could be just the PI5c3384 gone bad or the pass-through cable shorting out somewhere etc.

That's not ordinary Voodoo 2. It's Canopus 3D card with custom cable.

It looks basically the same, except for the additional Scalable VGA to NTSC/PAL Encoder. It has a similar QS3L384QG output selector IC. Some other cards use it too. The only other big difference is: it has only one DSUB connector. The other one is routed to an IDC connector from the selector IC to hook it up to other cards supporting it or hack a DSUB connector to it on an empty, separate PCI slot cover. So it's basically mixing/switching VGA signals with other supported cards. I have read a few minutes ago that they called it a WitchDoctor config. Interesting! I've never seen it before. We learn something new (old) every day. 😀

Reply 12 of 16, by kolderman

User metadata
Rank l33t
Rank
l33t

I was going to ask this question separately, but now there's a thread for it...

I have a S3 Savage4 passing through a voodoo2 SLI combo, and sometimes the color comes out green. In DOS the text will be green, and when playing a games, the colors have been green-washed but still distinguishable.

Before I pull apart this whole system, where is the problem likely to occur...the Savage4, the V2s, or the passthrough cable?

Reply 13 of 16, by maxtherabbit

User metadata
Rank l33t
Rank
l33t
kolderman wrote on 2020-03-15, 18:33:

I was going to ask this question separately, but now there's a thread for it...

I have a S3 Savage4 passing through a voodoo2 SLI combo, and sometimes the color comes out green. In DOS the text will be green, and when playing a games, the colors have been green-washed but still distinguishable.

Before I pull apart this whole system, where is the problem likely to occur...the Savage4, the V2s, or the passthrough cable?

you can very easily narrow that down by just removing the pass through cable from the equation and plugging the monitor directly into the S3 or directly into the V2 (while it is active of course)
the pass through cable is not required for the voodoo to function

Reply 14 of 16, by christal87

User metadata
Rank Newbie
Rank
Newbie
kolderman wrote on 2020-03-15, 18:33:

I was going to ask this question separately, but now there's a thread for it...

I have a S3 Savage4 passing through a voodoo2 SLI combo, and sometimes the color comes out green. In DOS the text will be green, and when playing a games, the colors have been green-washed but still distinguishable.

Before I pull apart this whole system, where is the problem likely to occur...the Savage4, the V2s, or the passthrough cable?

If your simple DOS command prompt is also green, as you say, then there could be a contact issue involving all these in the signal path:

  • your S3 card's DSUB connector and/or
  • pass-through cable and/or
  • voodoo2 DSUBs and/or
  • monitor cable and/or
  • could be the output selector IC.

Try fiddling around with the cables to see if it's intermittent. My issue was constant and limited to glide usage (gameplay).

Reply 16 of 16, by christal87

User metadata
Rank Newbie
Rank
Newbie
kolderman wrote on 2020-03-15, 20:15:

It's very intermittent. As in it flashes while playing a game between green-shaded to normal. I will try and take the V2SLI out of the equation first and see what effect that has.

I think the SLI has nothing to do with it, because it doesn't pass any analog video signals between cards. The card I repaired was used as a slave card before. It didn't have the color issue that way. It's the DSUB connectors and cables you should check.