You could check for impacts on the chip's pins that shorted them.

I've looked at all the little pins connected to the main S3 chip, and they all appear to be OK - none of them touch each other.

ohhh...almost look like it has those small ram chips I need for my AST machine 😁

Yeah, those look like them to me.

Hello retro and everyone here at vogons.
I just registered to post on this topic.

Recently I had some similar issues with a handful of S3 Trio64 cards..
The mainboard booted up fine (no beeps), but the screen remained black.

I solved the problem by entering the BIOS and setting the "PCI Latency Timer" to 64.
Maybe some higher values are needed for your card.

Bye, Jo22

I removed the "no POST" STB Trio64v+ card and replaced it with a working STB Trio64v+ card. As seen in the photo below, it's not exactly the same card as in the O.P, but it's probably good enough. This replacement card works OK. I then went in to the AMI BIOS set up area to look for PCI latency timing settings, but there weren't any. (The BIOS set up area options did look fairly comprehensive.) Perhaps PCI latency timing settings were introduced a bit later on, during the Windows 98 era? (I'm currently testing socket 3 and 7 boards for DOS and Windows 3.11.)

Edit: This is just a general observation, and not connected to the current problem.

Card in O.P. = (c) 1995, BIOS sticker says -60
Card below = (c) 1996, BIOS sticker says -50, RAM markings say 50

I guess that means that the card in the O.P. had RAM speed of 60, and by 1996 STB were able to "increase" that to a faster speed of 50.

I'm curious what's the cause for this behavior.
On my socket 3 board (80486DX2@66 with slow PCI bus) all my Trios work.

Difficulties only occured on my more modern mainboards in connection with the older Trios (VGABIOS dated before 1996).
Playing with the earlier mentioned setting always solved it.

Seams more like your motherboard doesn't like the card.
Try some BIOS update (if it's a flashprom already and not an eprom).

I really hope you get it working. 😀

Greetings, Jo22

I am testing the "no POST" STB S3 Trio64v+ card in old and slow socket 3 and 7 mobos, which are ideal for any Trio card. I notice in your screen shot, that your BIOS has a copyright date of 1985-2004. I guess the PCI latency fix is good for more modern mobos.

Please note that my other STB S3 Trio64v+ card worked OK in both of my test mobos. It's unlikely that these mobos will accept one STB S3 Trio64v+ card, but not another.

I think the card got damaged during transit.

Yes, that's right. It's from 2004 (last firmware release, Socket 939). 😅
As I mentioned, the problem occured only on my more modern boards:
Two newer Trios (dated '96,'97 with VBE2.0) worked, the other four didn't until I altered that value.

My old socket 3 mainboard (ALI chipset, '94) doesn't have that option and doesn't seam to need it.
I remember my Socket7/SuperSocket7 mainboards also worked fine with those cards.

Maybe your card is indeed broken.

Jo22

Well, I have some last idea that may help..
Some Socket7 mainboards let you choose the PCI revision (2.0 / 2.1) in the BIOS. Alter that setting. I could help, if the card ist still working.

If the card is damaged, then there's something that makes me wonder:
Why does your board even startup with a defective card? Normaly it should beep like crazy if no working card is avaible to the system -or- it shouldnt't start at all.

I'm sorry, my posts weren't that helpful. 😢

Bye, Jo22

Jo22 wrote:

I'm sorry, my posts weren't that helpful. 😢

Cheer up! 😁

I dug out the socket 7 board again, and looked for "PCI (2.0 / 2.1) support" inside the BIOS set up area. Unfortunately that option wasn't there. So I dug out a slot 1 board, an Asus P2B. Its BIOS has a PCI 2.1 Enabled/Disabled option. It also has a PCI latency setting.

Test #1: PCI 2.1 = Enabled, PCI Latency = 32. These are the default settings. Incredibly, the graphics card produced a BIOS POST message. Unfortunately, it quickly went "mad", because the characters on screen became corrupted and garbled.

Tests #2, #3 and #4 were simply combinations of PCI 2.1 = Enabled / Disabled, and also PCI Latency = 32 / 64. For these 3 tests, I got no BIOS POST message. I then tried setting the general BIOS settings to "default" (and not "set up"), and that set the PCI Latency value to 0. Unfortunately, that didn't work either. (I then set the general BIOS settings back to "set up" again, which set the PCI Latency value back to 32.)

Also, returning to test #1 again, I now get no BIOS POST message. Please also note: I have attached a little speaker to the mobo's speaker pins. Now I can listen out for any special error code beeps. For every test, I just hear one "healthy" beep. Also, for every test, the NUM & CAPS keys operate correctly: when they are pressed, the keyboard's lights operate correctly.

The more I think about this, the more I think that the area around the video cable socket could be damaged. The motherboard would not be "clever" enough to realise that this section of the graphics card is somehow damaged, and that is why no special error beep codes are being made.

Unfortunately, the card was shipped to me "loosely" inside a box, and I think it got "bashed about" during transit. I've spent several hours on this video card, and I think I've tried most things. I think you should be able to just stick this card in to an old socket 3 or 7 board without any problem.

Edit: I keep looking at this card, to see if I can spot any signs of physical damage. Looking at the photo in the O.P., you see the 2nd RAM chip from the left. If you look underneath the letters "V3", 2 of the "data lines" are scratched. Also, the solder points immediately to the right of this area have cut marks on them.

Test #1: PCI 2.1 = Enabled, PCI Latency = 32. These are the default settings. Incredibly, the graphics card produced a BIOS POST message. Unfortunately, it quickly went "mad", because the characters on screen became corrupted and garbled.

Hmm.. I recommend PCI 2.1=Enabled, PCI Latency = 64 / 96 / 128.
32 was the default setting my four old Trios (pre '96) didn't work with.
According to some of my PC books, the PCI 2.1 was bugfixed and backwards compatible to PCI 2.0. PCI 2.2 is another story.

Edit: I keep looking at this card, to see if I can spot any signs of physical damage. Looking at the photo in the O.P., you see the 2nd RAM chip from the left. If you look underneath the letters "V3", 2 of the "data lines" are scratched. Also, the solder points immediately to the right of this area have cut marks on them.

You're right about that. This card has some chips layout I never saw before. It has so much unused soldering points..
Looks almost as somebody had made changes to the card after it was produced. 😕
Perhaps the card maker (or user) wanted to use cheaper chips or install more memory.

If it's true that this card is modified (I hope it isn't), then it would be safer for your classic mobos to
put that card into the waste bin (some people here might want the RAM chips, though).
The VGABIOS chip however could still be useful as replacement for other cards with an Trio64V+ chip.

Cheer up!

I will try to do so. 😊

Bye, Jo22

I just tried PCI Latency = 96, but unfortunately that didn't work.

BTW, thanks for your suggestions - it was interesting to learn about PCI Latency - do you know if altering the PCI latency value can help reduce or remove "crackles" heard on poor quality PCI sound cards, such as early Creative PCI soundcards?

It could help, just try it.
Some webpages state, the setting defines how long (>cycles<) the currently active PCI device is allowed to use the bus before another one takes over.
Note, a to high value like 255, also slows down the PCI-ISA bridge.
The 64 seams to be alright most of the time.

So yes, the DMA transfer from the soundcard could benefit from slightly more cycles.

The crackles could also come from bad shielding.
Put the audiocard far away from the CPU, the PSU and graphics card.
Use ferrite clips for the cables from mainboard to the front panel.

To be honest, I always used ISA soundcards (CT1740 with DSP 4.05 / CT1350B; everything set by jumpers) in DOS PCs (286,386). 😀

BTW, thanks for your suggestions - it was interesting to learn about PCI Latency

That's nice to hear (read). I almost feared I wrote to much useless stuff onto your thread.. 😅

Bye, Jo22

Thanks very much for the advice.