You might face some surprises such as shorted tantalum capacitors though 😀

OK... that CGA card does look like the new-style due to the resistors on it, but looking at pictures of new CGA cards I can't find one with the Motorola chip instead of the other one. Does this mean anything?

keenmaster486 wrote:

OK... that CGA card does look like the new-style due to the resistors on it, but looking at pictures of new CGA cards I can't find one with the Motorola chip instead of the other one. Does this mean anything?

AFAIK only the resistors impacted on 8088MPH. Scali does mention the Hitachi CRTC's Hsync width difference on his blog, but I don't know if it impacts on the demo.

SquallStrife wrote:

AFAIK only the resistors impacted on 8088MPH. Scali does mention the Hitachi CRTC's Hsync width difference on his blog, but I don't know if it impacts on the demo.

Yes, the difference in the resistors (it's the DAC ladder that converts RGBI colours to analog) results in slightly different colours between old and new CGA. That's why you have to select the correct mode in the final version of the demo (party version was designed only for old style CGA, and some parts looked incorrect on new CGA).

The 6845 CRTC chip does have some effect on the demo, but you won't normally notice it with the final version. Namely, Hitachi and various other clones have slightly different behavior in some cases. For our demo, the hsync width matters. In the party version we chose a value of '0', which corresponds to 16 on a real Motorola 6845. This gave us the most stable image with the best possible colours in the high colour modes. However, various chips interpret a value of '0' as 0 rather than 16. As a result, you do not have ANY horizontal sync at all, and the image cannot be stabilized, because the monitor/TV can never lock onto any horizontal sync pulses.
In the final version, you can modify the hsync width in the calibration screen. Since we don't use a default of '0' in the final version, you won't really be affected. You just tweak the values until you get something that works on your card.

Ah, thanks! That makes sense now.

Also, the reason for the AMD vs. Intel 8088, with an Intel FPU...

The earliest steppings of the Intel 8088 had some major interrupt handling bugs. IBM wanted to get those CPUs out of the affected systems, so when you bought the FPU upgrade kit, it came with an AMD 8088 (all AMD 8088s not having the bug), and instructions saying to replace the CPU.

If your Intel CPU was copyrighted 78,81, you could leave it in place, but if it was only copyrighted 78, then you actually did need to replace it.

Interesting. Does this say something about the actual vintage of my system board?

Not really, and AFAIK by the time the XT came out, all of the Intel 8088s were fixed. (But, because IBM was dual-sourcing 8088s, Intel and AMD chips were about equally common.)

You have the datecodes on the chips - i.e. 8445 which would mean week 45 in 1984. Latest I saw on your board was 8522.

bhtooefr wrote:

Not really, and AFAIK by the time the XT came out, all of the Intel 8088s were fixed. (But, because IBM was dual-sourcing 8088s, Intel and AMD chips were about equally common.)

Interesting! Out of curiosity, where other clone chips, like that exotic K1810WM86, affected, as well ?

Just asking. This one was "developed" since '82, so it perhaps was based of an older design.
I know very little about this chip, though. I just accidently stumbled on this one,
while I was searching for the rare 286 clones MME U80601 and КР1847ВМ286..

The claims I've heard are that no clone chips were affected but, well... maybe that one is?

Only way to find out would be to get one or a machine with one, and try it. (Do note that Soviet chips tended to use metric pin spacing of 2.50 mm, instead of the rest of the industry's standard of 0.1" (2.54 mm).)

Ah, okay. Thanks for the reply. I was just curious, as this one likely wasn't, uhm.., created the usual way..
Would be interesting if someone who has such a machine could perform some sort of compatibility test.