Chrisa850

If you really do have ROM version 3, turn on the computer and press control-reset. At the prompt, try booting the external SmartPort device by typing PR#5 and pressing return.

I see schematic for version 1.1 is available. Was there a PCB revision published for 1.1? Your OSHW Lab page still includes PCB version 1.0.

You would need a good understanding of the NTSC RS-170A video standard and how the color burst reference is used for subtractive color. I'm not surprised that S-video (Y/C) works correctly as the chroma channel isn't "composited" on the luminance channel, making reference to the color burst signal much less critical.

The NTSC color burst frequency is 315/88 = 3.579545MHz. I have used 3.579545MHz in every Atari I have ever had to change a crystal in, including an 800XL I serviced last month. The error in frequency from what is written in the manual is .00003MHz, which is within the NTSC color burst tolerance. Even Console5 uses them as a replacement for 2600 and 800XL. https://console5.com/store/atari-2600-ntsc-color-crystal-oscillator-c015510-replaces-3-579575-mhz.html Since the NTSC color burst appears to be the issue in question, I would try using the exact NTSC color burst frequency and see what the results are.

It appears from looking at a factory schematic that the oscillator may be a Pierce-gate design optimized to use a series resonant crystal. You would normally *not* use a crystal requiring an 18 or 20pF load capacitor in that design. Load capacitors are commonly used in parallel resonant Colpits oscillator design where a capacitor is placed directly across the crystal to force the crystal to resonate at the proper frequency. I see no such capacitor in the 800XL design. Has anyone tried a series resonant crystal like an ECS 35-S-1X? https://www.mouser.com/ProductDetail/ECS/ECS-35-S-1X?qs=CNVyLAqdhH3RajNnZxGUjg%3D%3D

I have been performing Claus's old "MACE 48K Mod" since it was first published back in 1982. I was a member of the Michigan Atari Computer Enthusiasts (MACE) when the article appeared in the club newsletter. I've probably modified more than 50 machines over the years. I've never experienced any issues with a 48K application not properly functioning. Does anyone have recommendations on which programs may be troublesome so I can do more extensive testing? Back in 2021 I created a PDF document outlining how I perform the mod using (now) less expensive 41256 dynamic RAM for one of the YouTube vintage computer channels. If anyone would like a copy just let me know. I'll have to try Claus's new version with upgraded CAS timing to see how it performs. If anyone here was a MACE member, I used to work for the computer store that stayed open and served pizza following the monthly meetings.