ATI GS3 card with garbled text

[gordonf4vcf]

I managed to make some headway with the 5150 I picked up from Sierra Vista. Managed to use some 2364 ROM adapters to fix my Cassette BASIC problem and also try out Ruud's diagnostics.

My ATI card came in last week though, and I immediately ran into a problem.



Before I try replacing RAM on the card, are there any other suggestions? I found a manual for this card but the smallest jumper pins weren't referenced in the manual. I couldn't get it working at all with the MDA monitor, either in Hercules mode or CGA emulation mode, but I got it working in CGA mode on a GBS-8200 with a CGA signal converter I cooked up based on a TheRasteri hack using XNOR gates and a resistor ladder. The GBS and hack circuit I made seems to work on a Commodore 128 VDC so I think that isn't the problem.

This is the card.

 

[Svenska]

Look at the corruption in detail: 'u' (75h) is rendered as 'q' (71h); 'e' (65h) is rendered as 'a' (61h); 'd' (64h) is rendered as '`' (60h); 'V' (56h) is rendered as 'R' (52h).

In all characters, bit 3 is forced to zero. You probably have a faulty video memory chip.

 

[gordonf4vcf]

I figured it was RAM at first glance, because I ran into this exact problem on a TRS-80 Model 1 and replacing all of the RAM addressed it. I figured I'd see if there was something else I missed first.

The RAM appears to be MB81416 parts, compatible with 4416s. This would suggest the RAM is 256 KB, not 64 KB though. If these are 4416 compatible, I can test these on a Commodore 16 with the DIAG264 ROM.

 

[Svenska]

I think I read that most EGA clones came with a full 256 KB video memory, since that amount is required to use all capabilities.

There is no guarantee that a memory chip is the problem. Could also be a bad contact for one of the data bits, or a broken driver IC. If you have an oscilloscope, measure to find the issue. There is no need to replace all the RAM immediately.

 

[gordonf4vcf]

If you have an oscilloscope, measure to find the issue.

This is Plan A. I have a scope ready for this. I don't want to desolder eight 18-pin ICs if I can help it.

It's odd that this card would be considered an EGA clone with 256 KB RAM. I don't remember ATI claiming this card could emulate EGA on a CGA monitor like it could with the EGA Wonder. It would explain why it would have so much RAM though.

 

[Svenska]

I don't actually know this card. At first glance, I thought it was an EGA clone: The 16 MHz crystal means it's neither CGA (14 MHz) nor VGA (25 MHz), and 256 KB of video memory means its neither MDA (4 KB) or HGC (64 KB). It also has the DIP switches found on EGA cards. But the 6845 CRTC doesn't belong on an EGA, which should have made me pause.

Anyway... I did a little searching and found a better picture of the same card on Wikipedia. The memory chips are MB81416, which are 16Kx4. A pair is 16 KB, so the card has 64 KB total. This is not an EGA card at all, but compatible with both MDA and CGA monitors. Sorry for being misleading.

Doesn't change the fact that bit 3 in text mode is faulty.