your bios size seems correct, the 64k is just 32k & padding

edit : there's also only three bytes difference between your bios and the other one.

Sometimes hardware just fails/breaks...

weedeewee wrote on 2021-03-23, 19:28:

your bios size seems correct, the 64k is just 32k & padding

edit : there's also only three bytes difference between your bios and the other one.

Hey, thanks for letting me know! I'm not particularly great with this stuff, but that makes sense to me.

This may be a very amateurish question, but is it possible to burn that 64k BIOS to a 32kx8 EEPROM?

@kixs, you may very well be right, but I really hope that it's not dead. It seems like a nice card and I'd really hate to think that I killed it somehow. If it just as well might have failed--I suppose that it is what it is, but I'd still like to try and fix it, if possible.

WJG6260 wrote on 2021-03-23, 20:02:

This may be a very amateurish question, but is it possible to burn that 64k BIOS to a 32kx8 EEPROM?

Yes. I guess easiest would be to just edit the file down to 32768 bytes, open in hexeditor, delete all the 0xFF padding, save, program.
or if your eeprom reader/programmer supports it, just write the first 32768 bytes from the file to the chip.
Also, intermittent... might be a cold/loose solder joint on the main chip due to flexing... Just a guess.

@ WJG6260: didn't want to sound pessimistic... I hope you get it working!

Reminder to myself... I should just stfu if I don't have anything to contribute 🤣

May be an issue with the RAMDAC? Sounds a bit analogue-y. If something went wrong with the Vref (pin 31 on the ATT20C491, from a quick search, near R61 in the photo) for the DACs that might bring down RGB all together so look like a fade out?

weedeewee wrote on 2021-03-23, 20:09:

Yes. I guess easiest would be to just edit the file down to 32768 bytes, open in hexeditor, delete all the 0xFF padding, save, p […]
Show full quote

WJG6260 wrote on 2021-03-23, 20:02:

This may be a very amateurish question, but is it possible to burn that 64k BIOS to a 32kx8 EEPROM?

Yes. I guess easiest would be to just edit the file down to 32768 bytes, open in hexeditor, delete all the 0xFF padding, save, program.
or if your eeprom reader/programmer supports it, just write the first 32768 bytes from the file to the chip.
Also, intermittent... might be a cold/loose solder joint on the main chip due to flexing... Just a guess.

Oh, wow, I never realized it was that simple. I'll give that a go, then. I'm not sure that my programmer would support it otherwise. I'm using a TL866II Plus, but the software I have isn't too great. It's the Xgpro program, and it works, but it's pretty sub-par otherwise.

@kixs, no worries! I honestly did need to be reminded of that. Truthfully, I have a problem with getting overly into figuring out what I did, and usually I'm lost chasing my own tail haha!

@snufkin, I checked the pins on the RAMDAC, and they seem okay. I was considering that as a secondary problem as well. If it's a solder joint, would it perhaps be worth going for the old "bake in the oven" method to reflow the solder? I don't have a heat gun, or really any other way of doing it.

WJG6260 wrote on 2021-03-23, 20:31:

Oh, wow, I never realized it was that simple. I'll give that a go, then. I'm not sure that my programmer would support it otherwise. I'm using a TL866II Plus, but the software I have isn't too great. It's the Xgpro program, and it works, but it's pretty sub-par otherwise.

Just make sure to select the correct chip. should be written on the bios chip itself, probably underneath the stickers.
Could also be that the chip on there isn't re-programmable... but time and you checking it will reveal that.

weedeewee wrote on 2021-03-23, 20:52:

WJG6260 wrote on 2021-03-23, 20:31:

Oh, wow, I never realized it was that simple. I'll give that a go, then. I'm not sure that my programmer would support it otherwise. I'm using a TL866II Plus, but the software I have isn't too great. It's the Xgpro program, and it works, but it's pretty sub-par otherwise.

Just make sure to select the correct chip. should be written on the bios chip itself, probably underneath the stickers.
Could also be that the chip on there isn't re-programmable... but time and you checking it will reveal that.

Looks like it’s not reprogrammable! I did order pin-compatible EEPROMs, so we’ll see how that goes. Thanks again for all your help!

And thank you to everyone for all the help! I’ll report back my findings after I try a new BIOS. The only other thing I’m wondering is about the solder joints, as @snufkin suggested. I’ll check that again this evening.

If there is any data on the reverse, could you please post a high resolution pic of the reverse? In particular looking for a FCC ID and confirmation that it is a WDC manufactured card with STB branding. I'm adding it to my database for adding to gpuzoo.com, so any data will help me to provide accurate data.

debs3759 wrote on 2021-03-23, 21:58:

If there is any data on the reverse, could you please post a high resolution pic of the reverse? In particular looking for a FCC ID and confirmation that it is a WDC manufactured card with STB branding. I'm adding it to my database for adding to gpuzoo.com, so any data will help me to provide accurate data.

Sure thing! See the attached!

The FCC ID is as follows: EKSUSAC33.

Thanks

debs3759 wrote on 2021-03-23, 22:48:

Thanks

Anytime! If you need a better photo of the front, please feel free to let me know!

The picture was fine, thanks. I have all the data I need from it.

debs3759 wrote on 2021-03-23, 22:55:

The picture was fine, thanks. I have all the data I need from it.

Sounds good! I hope this helps!

By the by, I checked out your site and quite enjoyed it! I can't wait to see it grow!

Hope the work with the BIOS does fix it. If not, do you have a multimeter? I'm guessing a lot here, but it looks like pin 32 (pin 1 is the one marked on the slikscreen by R63, count anti-clockwise from there) should connect to C51, which goes to ground. Check that there's some resistance across C51, just in case something's gone short and is pulling Vref down. Also check the resistance of R69. It looks like that might connect to pin 28, which is Rset and also controls the DAC gain. The datasheet for the ATT20C490 (looks to be similar) recommends 147ohm. If I read the datasheet correctly, if that resistor goes open circuit then the gain will go to zero.

It might also be worthwhile trying to measure the voltage on pin 32 with the card powered up. I think it should be around 1.2V from an internal regulator. There's a bit more risk poking around live boards.

I've never tried baking in an oven. I'd guess most ovens would struggle to get hot quickly enough to reflow the solder without having the board too hot for too long. Plus it might singe the labels. I'd just touch a soldering iron on any suspect joints, with a bit of flux. If it is a cracked joint then it might help just prodding at components round that area whilst watching to see if anything flickers on the screen.

WJG6260 wrote on 2021-03-23, 22:58:

debs3759 wrote on 2021-03-23, 22:55:

The picture was fine, thanks. I have all the data I need from it.

Sounds good! I hope this helps!

By the by, I checked out your site and quite enjoyed it! I can't wait to see it grow!

Thanks. It has over 20K cards listed so far, but only ATI/AMD and NVidia cards at the moment. I'm slowly adding other cards, but waiting for the person who puts my databases onto the site to work on his script for adding the third database.

snufkin wrote on 2021-03-23, 23:15:

Hope the work with the BIOS does fix it. If not, do you have a multimeter? I'm guessing a lot here, but it looks like pin 32 (pin 1 is the one marked on the slikscreen by R63, count anti-clockwise from there) should connect to C51, which goes to ground. Check that there's some resistance across C51, just in case something's gone short and is pulling Vref down. Also check the resistance of R69. It looks like that might connect to pin 28, which is Rset and also controls the DAC gain. The datasheet for the ATT20C490 (looks to be similar) recommends 147ohm. If I read the datasheet correctly, if that resistor goes open circuit then the gain will go to zero.

It might also be worthwhile trying to measure the voltage on pin 32 with the card powered up. I think it should be around 1.2V from an internal regulator. There's a bit more risk poking around live boards.

I've never tried baking in an oven. I'd guess most ovens would struggle to get hot quickly enough to reflow the solder without having the board too hot for too long. Plus it might singe the labels. I'd just touch a soldering iron on any suspect joints, with a bit of flux. If it is a cracked joint then it might help just prodding at components round that area whilst watching to see if anything flickers on the screen.

Thank you! I like this idea too. I do have a multimeter, but my probes are a little on the larger side. Nevertheless, it should still work, and I’lol give it a check and see. I think you might be onto something, and I’ll give it a reflow if need be. I have a pretty solid soldering iron, so I guess that’ll do the trick.

@debs3759 Awesome! I’m excited to see this project keep growing! 20k cards is quite impressive! How large do you foresee it growing with that third database?

Also, @all—interesting developments during my further tinkering.

Full disclosure: the motherboard on which I’m using this is an Alaris Cougar. I tried a Boca SuperX VGA Cirrus GD5426 VLB video card that I have, and I had a similar issue—no image, but VGA signal detected by the monitor. I know that card works (it was new old stock when I bought it, and I still occasionally use it, but it’s mostly kept in bubble wrap in a box), so I pulled it, moved the CPU speed jumper to >33MHz (even though the Cougar is running the BL3 at 25MHz X 3), and the card worked absolutely fine. I know the ARK1000VL is more tolerant of 0WS setups, so I figured maybe there’s something else at play?

I know the Cougar is fine and it works perfectly, but it is a really quirky board. That being said, I am not so sure it’s the motherboard because I moved the Powergraph C33 into my other 486, which runs an MTech R407e. That board works great and has no weird quirks, really, and it likes the SuperX and the ARK1000VL just fine. It, however, does NOT work with the Powergraph C33.

Further, I’m a bit conflicted because the Powergraph WAS working when I first got it, before messing with it and running MTRRLBFE.

That being said, I tried the Powergraph one more time in the Cougar, and I got an image for a SPLIT second, and then it disappeared again. I rebooted, and the image stayed longer, but faded post-boot. The Boca and ARK1000VL work fine, and even the ATI MACH32 from my MTech 407e system worked.

WJG6260 wrote on 2021-03-24, 00:42:

@debs3759 Awesome! I’m excited to see this project keep growing! 20k cards is quite impressive! How large do you foresee it growing with that third database?

The goal is to keep growing it until I run out of places to get new (and old) cards and specs from. My aim is to collect data eventually for as many cards as possibly. I'm still going through old sites on the Wayback Machine, and have started gathering data from eBay listings (at least, those with good photos) and by googling cards to find specs. It'll keep me going for years. Being housebound due to ongoing health issues gives me lot of time to fill 😀