TF536 + 68000 relocator and ROM board project.

[derkom]

Aha, I never knew the tolerances were that tight. Thanks for the tip. I'll bear that in mind if I ever make a new revision.

[stephen_usher]

I know I've asked this before, but I got no replies, does any of the Atari ST hardware other than the CPU monitor the /HALT line?

Is it safe to treat it as unidirectional signal from the system to the CPU and not the other way around?

i.e.

MMU/GLUE etc. ----> CPU /HALT

rather than

MMU/GLUE etc. <---> CPU /HALT

I guess that if the CPU is halted then it doesn't matter as the rest of the system isn't actually doing anything.

[stephen_usher]

I tried to simplify the grounding wires this evening... and it backfired a bit on the 68000. I think that I'm also starting to wear out the socket on the motherboard as I have to press the card in a certain position to get it to boot and not give RAM and ROM spurious ROM CRC errors.

Anyway...

Here's a quick video demonstrating switching between CPUs and TOS versions. There are a few bad pixels when running on the 68000 due to the ground noise issues. Oh, and I forgot to show the "Turbo LED" lit when running on the TF536. D-oh!

https://youtu.be/UWGA0sDTgrw

Eventually I'll put switches on the outside of the case to be able to change the CPU and TOS versions.

[stephen_usher]

I reverted my ground simplification as it didn't work.

Anyway, as a test I ran Gembench with the machine switched to the 68000 and TOS 1.04, which is basically game compatibility mode and got this interesting result...

IMG_2093.JPG

Now, the clock hasn't been changed but could the 74HCT157D chips switching /BG, /BGACK and /E be adding wait states? I saw the 512 colour images PP's hard disk converted games use before loading the games have strange colour effects as if the timing was off too, mostly at the bottom of the screen.

[exxos]

Was you getting 100% as a stock machine ?

[stephen_usher]

Yep.

[exxos]

Very odd.. I mean BGACK etc are only used really for the blitter in those tests... E is for the ACIA which is disabled anyway... I can tell you that if you skew the 8MHz from the GLUE to CPU by even a few ns you get odd stuff like that start to happen.. Though normally the video starts shifting 16 pixels to the right as well. But overall it is possible that the logic chipd could be delaying something too much causing the slowdown..But by what you have mentioned, I don't really see how that is possible.

[stephen_usher]

I know. Weird.

Anyway... onwards!

Here's a video with a full demonstration of the board capabilities...

https://youtu.be/MBtW-hqjgg8

[derkom]

Now, the clock hasn't been changed but could the 74HCT157D chips switching /BG, /BGACK and /E be adding wait states? I saw the 512 colour images PP's hard disk converted games use before loading the games have strange colour effects as if the timing was off too, mostly at the bottom of the screen.

FWIW, on my Stacy, 68HC000 directly plugged in on the left, switched via muxes on the right...

Screenshot 2020-10-29 at 23.46.59.png

For whatever reason, blitting was a hair different, but everything else is identical.

[stephen_usher]

Fair enough. Sometimes things are strange. :-)