On 2/16/10, e.stiebler <emu at e-bbes.com> wrote:
On 2/15/10 9:22 PM, Ethan Dicks wrote:
That may well be true, but my own interest is
primarily in the CBM
implementation of IEEE-488, and that, I think, has a reasonably good
chance of going somewhere.
But the CBM attachment is simply the DIN connector (with serial 488
signals), or where there other onces ?
That's for VIC-20 and C-64 and newer machines. PETs (often badged as
"CBM" machines after the earliest models) have a full-on parallel
IEEE-488 interface, albeit implemented by the 6502 bit-banging VIAs
and PIAs, not with TMS9914s or MC68488 or other GPIB-on-a-chip
implementations.
Look up CBM 2040/3040/4040/8050/8250/9060/9090 disk drives. There
were also a number of printer models 4022 and such, plus a modem, plus
third-party periperhals (TNW serial, etc). PET models include 2001-8,
2001-32N, 2001-32B, 3008, 3016, 3032, 4016, 4032, 8032, SuperPET, and
more (the later "Business" line that I don't know as much about).
The serialized IEEE-488 stuff came after 1982 or so. Parallel
IEEE-488 started with the very first PET (even if the initial ROMs had
problems with diskette drives and only worked reliably with printers),
and continued for quite some time.
Part of my interest is because I have a couple of D9060 and D9090 hard
disks, but the original Tandon TM602S and TM603S drives are somewhat
rare (I have more boards and enclosures than working original drives),
but you can replace them with certain Seagate models - even those are
getting somewhat rare. It would be "better" for the long term to come
up with a newer parallel IEEE-488 drive than depend on finding then
hacking a somewhat uncommon product from 20+ years ago. I've recently
learned more about HP Amigo/CS80 drives, but I do know that
implementing a modern microcontroller-based drive emulator for
Commodore machines shouldn't have any bizarre pitfalls - just listen
on the channel, speak when you are spoken to, and understand a dozen
or two "CBM DOS" commands and do what's asked. The original
implementation fits in a few K of 6502 machine code. A 16K AVR should
be able to handle it, and if not, a 32K part, as long as there's
enough RAM to sufficiently emulate the sub-file-level-access commands.
For simply loading and saving PRG files and getting directories, that
subset of CBM DOS shouldn't be hard at all to do. REL files are
another matter. Those could easily take as long as the entire rest of
the project and more.
-ethan