News on the remake of the Mill 6809

[Alex Freed <alex_news@alexfreed.com>]

OK, a slow boat from China brought a few PCBs and I have assembled one card.

http://alexfreed.com/FPGApple/Mill6809/redesign_mill6809.jpg

Smaller and simpler than the original due to a CPLD replacing the memory 
remapping ROM and some glue logic.

The good news is that it works, but the bad news is that it takes an 
instrument to see that.

Here is a snapshot:

http://alexfreed.com/FPGApple/Mill6809/mill6809_analyzer.PNG

I'm trying to use the good old demo1.bas from the disk image.

 480  REM FFFE -> 40FA=JMP $1000
 490 X =  PEEK (ROM) * 256 +  PEEK (ROM + 1)
 500  REM
 510  POKE (X),126: REM  7E
 520  POKE (X + 1),CODE / 256
 530  POKE (X + 2),0
 540  POKE SLOT + 2,128: REM  RUN

This code wants to write a jump instruction to the address, presumably 
$40FA, pointed to by the last 2 bytes in standard Apple ][ ROM. 
Unfortunately my enhanced 2e has $FAC3 at FFFE-FFFF. So as one can see 
on the analyzer screen the 6809 fetches $FAC3 and jumps to that address. 
Which of course lives in ROM and can't be patched. Oops...

Sure I could load a standard monitor into the "language card", switch to 
it and then try it again, but this is a bit ugly. Also I'm pretty 
confident the basic counting test will work - after all both CPUs are 
running. Where should I take a better test? My main concern is that 
knowing that the LS14 was replaced with LS04 while adding the memory 
mapping for OS09 I suspect that timing may need to be tuned. Of course 
my address mapping CPLD is an order of magnitude faster than a ROM used 
in the original version, but "too fast" can also be an issue. So I need 
some tests before declaring the card operational. I want something that 
runs with all 3 versions of the memory map - virgin, flipped A15 and 
OS09 remapping.

-ALex.