Tokamac build and programming XC1736 chips

[DrGonzo]

Mind if I ask where you picked up the legit xc3030A? I'm still not convinced mine are counterfeit and still suspect issue(s) with my xc1736 and/or burning process, but I'd buy a few and swap them in to give it a try.

Thx!

 

[zigzagjoe]

@JC8080 sent me a Presto card that is supposed to work in a Color Classic. I've dumped the PROM.... drumroll, it's the same one! So whatever is responsible for working/not working in a color classic seems to be either a board revision (perhaps different color classic boards?) or component selection on the card.

Here are some pictures of the board in question.




Related, I have validated you can just dump these PROMs via a logic analyzer or even with a microcontroller (as I did). Given default OE/Reset polarity...

  SetVCC(1);

  digitalWrite(PIN_CE, 1); // cs = deassert
  digitalWrite(PIN_RESET, 1); // reset = assert / OE = deassert
  digitalWrite(PIN_SEREN, 1); // disable serial interface
  digitalWrite(PIN_CLK, 0);
  BitDelay();

  digitalWrite(PIN_CE,0);
  digitalWrite(PIN_RESET,0);
  BitDelay();

  int i = 0;
  while (digitalRead(PIN_CEO)) {

    Serial.print(digitalRead(PIN_DATA) ? '1' : '0');

    digitalWrite(PIN_CLK,1);
    BitDelay();
    digitalWrite(PIN_CLK,0);
    BitDelay();
   
    i++;
  }
  Serial.println();
  Serial.println(i);

This should result in the same number of bits as the PROM is rated to contain. To transform this to a binary dump, for each 8 bits, reverse the bits, then convert to a byte.

Given a bitstream from such a dump containing a series of bits like so: 0110001010101111
Split it into groups of 8 bits. 0123456789012345 -> 01234567 | 89012345
Now reverse the bits in each byte. 01234567 | 89012345 -> 76543210 | 54321098
Convert grouping of 8 bits into bytes, done.

 

[JC8080]

The CC logic board that is confirmed to work with the Sonnet Presto that @zigzagjoe has is PN 820-0367-C, photos attached. The Presto came in this machine when I bought it, so unfortunately I don't know any history of the board beyond that.

 

[zigzagjoe]

The CC logic board that is confirmed to work with the Sonnet Presto that @zigzagjoe has is PN 820-0367-C, photos attached. The Presto came in this machine when I bought it, so unfortunately I don't know any history of the board beyond that.

This spurred a good idea, thank you. Bolle and I have been brainstorming and pouring through schematics; my working theory is perhaps something's going terribly awry at handover to the Presto/Tokamac 040, perhaps there might be some issue with bus mastering on the Color classic. So few cards actually used bus mastering on the LC slot an color classic specific errata could have went unnoticed until a chipset or board revision fixed it. I am only aware of one card (apple 820-0532) that does bus mastering in the conventional sense (not like an accelerator).

So to that end I'd like to see some more pictures of boards known to work or not work so we might check board revisions.

 

[Phipli]

This spurred a good idea, thank you. Bolle and I have been brainstorming and pouring through schematics; my working theory is perhaps something's going terribly awry at handover to the Presto/Tokamac 040, perhaps there might be some issue with bus mastering on the Color classic. So few cards actually used bus mastering on the LC slot an color classic specific errata could have went unnoticed until a chipset or board revision fixed it. I am only aware of one card (apple 820-0532) that does bus mastering in the conventional sense (not like an accelerator).

So to that end I'd like to see some more pictures of boards known to work or not work so we might check board revisions.

Keep in mind I have made a Presto and it wouldn't work in my early LC (with the big fat Cap on the SCSI chip). I sent it to... Max? And It worked fine for him.

The issue might not be limited to CCs.

 

[zigzagjoe]

Keep in mind I have made a Presto and it wouldn't work in my early LC (with the big fat Cap on the SCSI chip). I sent it to... Max? And It worked fine for him.

The issue might not be limited to CCs.

Right. That is my thinking behind checking ASIC revisions, as I recall some of the LCs had issues with accelerators.