Jackal1983 wrote on 2022-12-21, 08:08:

Was wondering if someone could point me towards it considering I've basically confirmed the bios chip is borked (flashed a Diamond Riva 128 bios to a EPROM and the card worked, it didn't give a signal with the stock chip) "Velocity 128 PCI NTSC 1.82" is what I need.

I just checked and I have a card with that exact BIOS.

What is the easiest way to get the BIOS from this without pulling the chip and reading it? I remember there was a program that did that but I can't remember which one it was.

Ozzuneoj wrote on 2022-12-22, 00:19:

Jackal1983 wrote on 2022-12-21, 08:08:

Was wondering if someone could point me towards it considering I've basically confirmed the bios chip is borked (flashed a Diamond Riva 128 bios to a EPROM and the card worked, it didn't give a signal with the stock chip) "Velocity 128 PCI NTSC 1.82" is what I need.

I just checked and I have a card with that exact BIOS.

What is the easiest way to get the BIOS from this without pulling the chip and reading it? I remember there was a program that did that but I can't remember which one it was.

Actually I had a copy stashed on my HDD from back when I first got my TL866. Burned it to a 27c512 (copied the contents into the upper half) and it boots! Posting the original 256K copy and a doubled image for used in a 27c512 eeprom (it is compatible with the card).
Hopefully they'll be of some use to someone else in the future. I'd be curious to know what happened with the original chip: I re read it and I did note some segments where it seemed like the lines had been changed as chunks. Are these "Low Voltage" OTP PROMs vulnerable to a form of bit rot? The card itself was NOS and did initially work when I bought it and also worked fine after I recapped it (the caps looked fine but they were no name brand, swapped them for quality Nichicons and and Rubycons I think). Some months later I tried using it as an output card after recapping a board and the card seemed dead (though it was detected in win 98se).

Jackal1983 wrote on 2022-12-22, 02:18:

Ozzuneoj wrote on 2022-12-22, 00:19:

Jackal1983 wrote on 2022-12-21, 08:08:

Was wondering if someone could point me towards it considering I've basically confirmed the bios chip is borked (flashed a Diamond Riva 128 bios to a EPROM and the card worked, it didn't give a signal with the stock chip) "Velocity 128 PCI NTSC 1.82" is what I need.

I just checked and I have a card with that exact BIOS.

What is the easiest way to get the BIOS from this without pulling the chip and reading it? I remember there was a program that did that but I can't remember which one it was.

Actually I had a copy stashed on my HDD from back when I first got my TL866. Burned it to a 27c512 (copied the contents into the upper half) and it boots! Posting the original 256K copy and a doubled image for used in a 27c512 eeprom (it is compatible with the card).
Hopefully they'll be of some use to someone else in the future. I'd be curious to know what happened with the original chip: I re read it and I did note some segments where it seemed like the lines had been changed as chunks. Are these "Low Voltage" OTP PROMs vulnerable to a form of bit rot? The card itself was NOS and did initially work when I bought it and also worked fine after I recapped it (the caps looked fine but they were no name brand, swapped them for quality Nichicons and and Rubycons I think). Some months later I tried using it as an output card after recapping a board and the card seemed dead (though it was detected in win 98se).

Awesome! Glad you got it working!

I have no idea what would cause bit rot or otherwise corrupt a PROM like that though. Just curious... was your card one of the ones with the through-hole caps with the legs left super long, like this one? I thought they were SMD pads, but they are in fact going through the board...

(not my picture)

That is how my card looks and it is... really baffling. In my experience, STB always produced rather high quality devices. This is such an incredibly shoddy looking way to populate a PCB, I have no idea what they were thinking unless there was a problem sourcing the correct caps at the time? It could be why a lot of cards in the late 90s used SMD tantalums instead. I know the AGP versions of this card use tantalums... I always assumed there was some difference in voltage between the PCI and AGP versions of cards, which is why the passive components are often different between them.

Anyway, it wouldn't surprise me if there was some connection between this crazy capacitor setup and a corrupt PROM.

Ozzuneoj wrote on 2022-12-22, 12:06:

Awesome! Glad you got it working! […]
Show full quote

Jackal1983 wrote on 2022-12-22, 02:18:

Ozzuneoj wrote on 2022-12-22, 00:19:

I just checked and I have a card with that exact BIOS.

What is the easiest way to get the BIOS from this without pulling the chip and reading it? I remember there was a program that did that but I can't remember which one it was.

Actually I had a copy stashed on my HDD from back when I first got my TL866. Burned it to a 27c512 (copied the contents into the upper half) and it boots! Posting the original 256K copy and a doubled image for used in a 27c512 eeprom (it is compatible with the card).
Hopefully they'll be of some use to someone else in the future. I'd be curious to know what happened with the original chip: I re read it and I did note some segments where it seemed like the lines had been changed as chunks. Are these "Low Voltage" OTP PROMs vulnerable to a form of bit rot? The card itself was NOS and did initially work when I bought it and also worked fine after I recapped it (the caps looked fine but they were no name brand, swapped them for quality Nichicons and and Rubycons I think). Some months later I tried using it as an output card after recapping a board and the card seemed dead (though it was detected in win 98se).

Awesome! Glad you got it working!

I have no idea what would cause bit rot or otherwise corrupt a PROM like that though. Just curious... was your card one of the ones with the through-hole caps with the legs left super long, like this one? I thought they were SMD pads, but they are in fact going through the board...

(not my picture)
velocity128pci.jpg

That is how my card looks and it is... really baffling. In my experience, STB always produced rather high quality devices. This is such an incredibly shoddy looking way to populate a PCB, I have no idea what they were thinking unless there was a problem sourcing the correct caps at the time? It could be why a lot of cards in the late 90s used SMD tantalums instead. I know the AGP versions of this card use tantalums... I always assumed there was some difference in voltage between the PCI and AGP versions of cards, which is why the passive components are often different between them.

Anyway, it wouldn't surprise me if there was some connection between this crazy capacitor setup and a corrupt PROM.

Yup, screwy as all hell, they're the same caps as on my Nitro 3D from basically the same time, however. On the Nitro, they're basically flush with the board. I pulled the PROM while I was soldering and it worked fine afterwards. The 128 seems to be an odd one all all around, because this is the same card that STB decided to thermal glue a mosfet heatsink to the package when they realized it would benefit from a heatsink: I found that utterly nuts so I popped it off and stuck a large rectangular Raspberry PI heatsink on it. If I had to guess: the difference between the AGP and PCI versions (tantalum vs. electrolytic) may be down to cost cutting, but who really knows.

Jackal1983 wrote on 2022-12-22, 13:37:

Ozzuneoj wrote on 2022-12-22, 12:06:

Awesome! Glad you got it working! […]
Show full quote

Jackal1983 wrote on 2022-12-22, 02:18:

Actually I had a copy stashed on my HDD from back when I first got my TL866. Burned it to a 27c512 (copied the contents into the upper half) and it boots! Posting the original 256K copy and a doubled image for used in a 27c512 eeprom (it is compatible with the card).
Hopefully they'll be of some use to someone else in the future. I'd be curious to know what happened with the original chip: I re read it and I did note some segments where it seemed like the lines had been changed as chunks. Are these "Low Voltage" OTP PROMs vulnerable to a form of bit rot? The card itself was NOS and did initially work when I bought it and also worked fine after I recapped it (the caps looked fine but they were no name brand, swapped them for quality Nichicons and and Rubycons I think). Some months later I tried using it as an output card after recapping a board and the card seemed dead (though it was detected in win 98se).

Awesome! Glad you got it working!

I have no idea what would cause bit rot or otherwise corrupt a PROM like that though. Just curious... was your card one of the ones with the through-hole caps with the legs left super long, like this one? I thought they were SMD pads, but they are in fact going through the board...

(not my picture)
velocity128pci.jpg

That is how my card looks and it is... really baffling. In my experience, STB always produced rather high quality devices. This is such an incredibly shoddy looking way to populate a PCB, I have no idea what they were thinking unless there was a problem sourcing the correct caps at the time? It could be why a lot of cards in the late 90s used SMD tantalums instead. I know the AGP versions of this card use tantalums... I always assumed there was some difference in voltage between the PCI and AGP versions of cards, which is why the passive components are often different between them.

Anyway, it wouldn't surprise me if there was some connection between this crazy capacitor setup and a corrupt PROM.

Yup, screwy as all hell, they're the same caps as on my Nitro 3D from basically the same time, however. On the Nitro, they're basically flush with the board. I pulled the PROM while I was soldering and it worked fine afterwards. The 128 seems to be an odd one all all around, because this is the same card that STB decided to thermal glue a mosfet heatsink to the package when they realized it would benefit from a heatsink: I found that utterly nuts so I popped it off and stuck a large rectangular Raspberry PI heatsink on it. If I had to guess: the difference between the AGP and PCI versions (tantalum vs. electrolytic) may be down to cost cutting, but who really knows.

Haha, yes, mine has that mosfet heatsink as well! It's crazy... I have seen so many of these over the years I'd just gotten used to that stupid cooler. It never occurred to me that it is possibly the only video card I've seen with such a hilariously misapplied cooling solution.

Wow... this card just got weirder, for sure.

Would genuinely like to talk to one of the people who had to approve these design choices.