Intellivoice changes over the years.

[JohnPCAE]

Well, for what it's worth, I tried to isolate some traces. Green is metal where there is nothing underneath, white is where there is something underneath (or where I simply didn't do anything). I have no idea which components are which and whether this is useful, but I guess you could add it to your docs.

 

http://www.mediafire.com/view/7t6pf7g045070nh/GI_SP0256_die_shot_8500w_4.jpg

Edited December 11, 2015 by JohnPCAE

[intvnut]

Well, for what it's worth, I tried to isolate some traces. Green is metal where there is nothing underneath, white is where there is something underneath (or where I simply didn't do anything). I have no idea which components are which and whether this is useful, but I guess you could add it to your docs.

 

http://www.mediafire.com/view/7t6pf7g045070nh/GI_SP0256_die_shot_8500w_4.jpg

 

It looks like you traced out most of the V_DD and GND major traces at least. I have somewhere an XCF where I was trying to do this on layers. I don't have GIMP on this machine, so I can't really look at the XCF files right now.

 

I do have some other images where I at least identified some major structures, such as some counters, a decoder PLA I never fully understood, the program counter and single-level return pointer. Also, all of the pads are labeled with their pin numbers and names as I recall.

 

I also did visually verify that the ROM is the 012 ROM. IIRC, it has an 8:1 interleave, although that's all just based on memory at this point. (By 8:1 interleave, I mean every 8th bit in the ROM array is part of the same byte. This is common in RAM and ROM structures.)

 

FWIW, I did this work in cooperation with the Visual6502 folks. If anyone wants to get down and dirty with actually reversing this to a netlist and is willing to put in the time and effort, contact those folks. Maybe someday we'll have a Visual SP0256! :-) They already have a javascript Visual 6502, 6800 and just recently ARM1. Very cool stuff. The ARM1 version is notable because it was generated from the actual design database (CIF files) with ARM's blessing. It also uses a new WebGL based version of the renderer which is much faster.

 

And a little off-topic: If any of you reading are 6502 enthusiasts, this recent podcast interview with Bill Mensch is great listening.

[JohnPCAE]

I did this in Paint Shop Pro v8. Without the magic wand and colorizer tools, I woudn't have even attempted it. I did trace out some more since I uploaded the file, maybe later on I'll do a little more and upload a newer version. This would have been a lot easier if the starting point was a PNG instead of a lousy JPEG

[JohnPCAE]

This is incredibly tedious...

 

http://www.mediafire.com/view/mk99kpo784kiss4/GI_SP0256_die_shot_8500w_5.jpg

 

Someone else will have to interpret what the little rectangles mean as that's beyond my knowledge...

[JohnPCAE]

http://www.mediafire.com/view/lpmk81ycg339uec/GI_SP0256_die_shot_8500w_6.jpg

[JohnPCAE]

Added a few more sections. Is this in any way useful?

 

http://www.mediafire.com/view/ta38qqfc1tkmsdf/GI_SP0256_die_shot_8500w_7.jpg

[JohnPCAE]

http://www.mediafire.com/view/4ev90bg7t60ed3c/GI_SP0256_die_shot_8500w_8.jpg

 

I isolated a lot more of the top layer; all that's left is the ROM area and the surrounding logic. I'm not sure it's worth doing the ROM area since the contents are known, and I don't much feel like slogging through it. I'll probably complete the surrounding logic at some point, though.

[JohnPCAE]

http://www.mediafire.com/view/1vwt6y20042dj5m/GI_SP0256_die_shot_8500w_9.jpg

 

Finished isolating the top layer for everything except the ROM area (though I did do a tiny corner of it as well).