Posted By
TLC
on 2022-10-30
19:36:30
|  Re: New Project: Hannes-compatible 256k RAM Expansion
I have been also playing around with cleaning up Hannes' / Solder's design in the last couple of years (time to time coming back to the thing). I didn't make it solderless, I merely wanted to clean up the mess imposed by the original design (i.e. piggybacking). So, I kept the classic RAM bank replacement method, and put the control logic on a single daughterboard that plugged into a precision socket in the place of the $FD10 6529B. A soldered-in wire harness connected to the board, and collected the required extra signals. (This was all obviously for the Plus/4.) Of course I have experimented with it a lot. The last prototype also added Csory expansion logic compatibility. (I have decided to drop that, however, from the next prototype... as I found out, the schema is just marginally supported. In practice I have also found the combined Hannes / Solder / Csory thing somewhat less than optimal as per complexity vs. functionality... all in all, finally I became less than enthusiastic about it so it will be dropped.) I have also planned to move the thing to GALs in the next prototype, to reduce size and complexity (not done yet).
Theoretically, at least 4 bits need to be implemented from the $FD16 page register (--> two function select bits and two page select bits). For full compatibility, however, I guess implementing a full 8-bit page register is required. (The extra 4 bits are functionally redundant. Yet, running Hannes/Solder compatible code, given the original design's 6529B, I think rightfully expects a full 8-bit read-writeable register at $FD16. The absence of the "unused" 4 bits may, or may not cause problems in practice, given arbitrary code.) BTW speaking of why "reading" the page register is needed in the first place (with or without implementing every bits), I seem to remember some Hannes/Solder expansion detection logic to rely on finding out whether the $FD16 register existed (at all) first, and if it did, finding out whether it was identical (or, separate) to the $FD10 register.
To first test the thing, I think Synergy's old 256k Support specific demos are perfect. Future World and Dream World come to mind first. Taurin can be also a good candidate. Yep, these old stuffs typically use specific loaders (unlikely to cooperate with an sd2iec).
With regards to a C16 compatible, 41256 based ram expansion design, I suspect, current consumption could be a factor.
|
