from the example photo you offer, I'd say that some datalines between the memory and imagine128 chip have a bad/failed contact, which could also be a bad chip...

Did you do any chip squeezing to see if you can make the lines change?

Thanks for your replies! The squeeze test was brilliant 😁

I narrowed the issue down to one corner of this DRAM chip here.

If I put pressure on it, the artifacting changes and will sometimes disappear. I took a closer look with a microscope and poked a few of the legs at that corner. At least five of them have bad solder joints. I can freely move the legs side to side. Will need to do a reflow of that area and see if the issue is fixed.

Thanks again! Will report back on the results.

Wow ! Great work guys ! Hope you can fix it easily 😀

Cl gd5424 doesn’t have a pci interface. How does this work?

Cl gd5424 doesn’t have a pci interface. How does this work?

Anonymous Coward wrote on 2022-07-06, 01:52:

Cl gd5424 doesn’t have a pci interface. How does this work?

My uneducated guess would be that the Imagine chip contains some kind of bridging logic that allows the Cirrus chip to work as if it was connected through an ISA (or VLB) to PCI bridge .

Success! There were actually seven pins on that memory chip with bad solder joints. I just reflowed them with a bit of fresh solder and the card now works perfectly.

I did a YouTube video on the repair for anyone interested in seeing it done under a microscope.

https://youtu.be/NDT8kTUil0k

Thanks again for the tips!

btw Afaik IBM didnt manufacture any ram at that point in time, so those chips are most likely rebadged NEC anyway 😉

lemonlime wrote on 2022-07-08, 18:54:

Success! There were actually seven pins on that memory chip with bad solder joints. I just reflowed them with a bit of fresh sol […]
Show full quote

Success! There were actually seven pins on that memory chip with bad solder joints. I just reflowed them with a bit of fresh solder and the card now works perfectly.

I did a YouTube video on the repair for anyone interested in seeing it done under a microscope.

https://youtu.be/NDT8kTUil0k

Thanks again for the tips!

Good work! I've got one of the late 128 Series 2s without the Cirrus Logic chip so if the newer VBIOS isn't easy to find, I should be able to dump mine

pancakepuppy wrote on 2022-07-08, 21:27:

Good work! I've got one of the late 128 Series 2s without the Cirrus Logic chip so if the newer VBIOS isn't easy to find, I should be able to dump mine

I've been searching for the newer VBIOS without success. If you don't mind taking a dump of your card's BIOS when you get a chance it would be very much appreciated!

Thanks again.

Great job and excellent video!
I'm also interested in the newer VBIOS, as it might offer a chance to put my #9 card to good use (it's a Mac-only card, so no Cirrus Logic chip onboard).

Alright, I popped the VBIOS out of my "legacy-free" Imagine 128 Series II and read it out with a universal programmer. Let me know if you try this out on an older card and how it goes! 😀

Edit: For reference, my card's VBIOS part is a National Semiconductor NM27C256V. I think, in this package, it's one-time programmable. If the older ones are also OTP parts you may need to get a new EE/EPROM to program.

Thank you! I'll try it and report back - my card also has a OTP chip (NM27C512) so it might take more than a few days if I don't find a suitable chip somewhere around me.

pancakepuppy wrote on 2022-07-21, 00:37:

Alright, I popped the VBIOS out of my "legacy-free" Imagine 128 Series II and read it out with a universal programmer. Let me know if you try this out on an older card and how it goes! 😀

Edit: For reference, my card's VBIOS part is a National Semiconductor NM27C256V. I think, in this package, it's one-time programmable. If the older ones are also OTP parts you may need to get a new EE/EPROM to program.

Thanks very much for doing that! Will see if I can find a programmable chip for it.

I ordered some chips (AT27C512R and W27C512P) and I'll have to wait their arrival for further testing (512kbit because that's why my card originally uses, I'll just concatenate the file 2x).

Tried what I had at hand, closest I could find was a W49F002 so I had to concatenate the file 8 times in order to fill it 😁
I expected it not to work (and of course it didn't!) since the chip enable and output enable pins are 22 and 24 compared to 23/25 on the original chip.
Another chip I had available was a SST39VF512 - closer in size but more incompatible - above pins still shifted, and Low Voltage version. Arrrgh!!!

W27C512P chips arrived today, but no luck in getting my card to work.
I don't know what the correct route is in this particular case (original chip is 512kbit, new BIOS is 256kbit) so I tried all 3 possible combinations (image concatenated to fill the entire chip, or using just the first and then the second half of the chip), all result in "no videocard" beep code 🙁

pancakepuppy wrote on 2022-07-21, 00:37:

Alright, I popped the VBIOS out of my "legacy-free" Imagine 128 Series II and read it out with a universal programmer...

Could you please share some photos of your board? I can't seem to find the variant without the CL chip on the net, and I want to compare it with mine.

Images attached