Re: WTD: tandon 603s or 602s

From: Ethan Dicks (
Date: 2006-07-02 10:25:58

On 7/2/06, "André Fachat" <> wrote:
> Besides, has anyone 9090/9060 schematics of a higher quality than
> those that are on Those basically are block blobs
> with lines.... :-(

I have a factory-original D9060/D9090 service manual at home, but that's
10,000 miles away and I won't be near it for six months.  Sorry.

> On the other hand: If really an SASI board is used,
> then the DOS does not need to know about drive schematics - the
> SASI-to-ST506 needs to know that. Is that a 6502-based board, or
> is it programmed by the main DOS CPU?

There are two boards in a D90x0 drive - the "DOS board", which would
look familiar to anyone with experience inside a Commodore disk drive,
and a Tandon? Xebec? SASI-to-ST506 bridge card.  The bridge card has
some micro controller and is superficially like an Adaptec
ACB4000-type SCSI bridge card.  Same era, same basic functionality,
just a little older.  It was not created specifically for the D90x0;
it's an off-the-shelf bridge card. The DOS card has, IIRC, two
6502-class processors (one might be a 6504, can't remember for sure) -
just like C='s 5.25" floppy drives - one speaks IEEE-488 and knows
about C= DOS commands, etc., and the other processor handles the SASI
bus.  The DOS board only knows about geometry, not about deep-down
drive details.  The IEEE-aware processor gives general sorts of
instructions to the SASI-aware processor on the DOS board, and the
SASI-aware processor formats SASI packets and wiggles lines to
implement the SASI protocol.  My plan was to understand the
SASI-processor well enough to tweak the packet formation and any
required handshaking changes to allow a modern embedded SCSI drive to
go right on the DOS board, bypassing the SASI-to-ST506 bridge card.
The DOS processor still would only be able to address about 16MB, but
perhaps it would be easy enough to give the drive multiple units (0:,
1:, 2:...) and at least use up a good slug off of an old 200MB drive.

I have grubbed around in the code a bit and belive it's possible, but
I can't estimate how difficult it would be to accomplish any of this.
I'm certainly available for questions if anyone else cares to start a
disassembly attempt, but I won't be starting/resuming one anytime in
the next few months.


       Message was sent through the cbm-hackers mailing list

Archive generated by hypermail pre-2.1.8.