On Tue, Apr 28, 2009 at 5:21 AM, Anders Carlsson
<anders.carlsson@sfks.se> wrote:
> Hello,
>
> ED> I like the idea of hacking a real D9090 .. It's not like my efforts
> ED> will be easy to share since so few people have these drives.
> RB> You, Hervé and me makes three already :)
>
> How difficult would it be to make a D90X0 replica? Mostly I'm thinking about
> the DOS and SASI boards.

I would recommend a replica as far as the DOS board, but _not_ the SASI board.

> I realize there are a lot of chips, many
> undocumented as mentioned before. I'm not entirely sure what the point would
> be, in particular if MFM hard drives are getting rare to find and few still
> work.

Exactly.  Designing something new with an MFM interface is probably
not going to reap many rewards.

> Do there exist any interfaces from ST-412/506 to e.g. CF/SD/whatever?
> Since those drives were used in many PCs and so on, there should exist a
> market.

I know from the ClassicCmp mailing list, there's a re-occurring topic
about such an interface.  The "problem" is that to design an
ST506/ST412 device emulator that just drops in where a real drive goes
is more complicated than it sounds.  The data rate to the heads is
5Mbit/sec, and the format is whatever the controller says it is.  In
the case of the D90x0, it's 32 sectors of 256 bytes.  PCs used 17
sectors of 512 bytes (less header information, so more room for data).
  The controller feeds the data via analog cables to the drive, which
mostly does amplification and pulse shaping.  The emulator would have
to be able to emit or accept waveforms at that rate to be a proper
drive substitute.

It's not an impossible problem to solve, but it is a non-trivial problem.

> Well, I'm just trying to come up with a way to have more people join in on
> the fun, if you base your hacking on the 9000 series HDDs rather than making
> a completely new design. If everyone are sitting with the same kind of DOS
> board we have something in common.

The DOS board from a D90x0 drive is like most C= boards - a CPU or
two, some ROM, some RAM, and a bus interface (IEEE-488 in this case).
With modern microcontrollers, it's trivial to replace 10 chips from
the 1982 era with 1 chip, so you'd just need to pick a storage medium
(SD or CF are cheap, common, and easy to talk to) and work out how to
physically interface to the IEEE side of things (40-pin
microcontrollers have lots of I/O pins, but are easily available in
multiple varieties for under $10 each).  It might be fun and easy to
hack together a 6502-based DOS board replacement, but it would be
several times larger than and cost many times more than an AVR-based
IEEE drive replacement.  Obviously for this application, any AVR-based
design would have freely-available schematics and layout files so
anyone could build their own boards and have a starting point for
additional projects.  Once the bus routines are written and debugged,
there's a lot one could do with a PET given a generic peripheral
platform to start from (serial, printer, data collection, etc).

I'm not saying that nobody should hack a real DOS board, but they are
limited in number, and a fresh design could be done very cheaply and
shared with everyone much more easily than making folks try to find a
dead D90x0 drive.

> Regarding the question how common those hard drives were, I can chime in
> this was the total inventory from the storage of a former Commodore/PET
> dealer and serviceman:
>
> * About 40 mixed IEEE dual floppy drives (3040, 4040, 8050, 8250, 8250LP)
.
.
.
> * Four D90X0 hard drives (two working, of which one with a ST-225)
>
> Notice I didn't find a single 1541 or better IEC drive, so clearly the focus
> once had been on PET customers. I do realize this distribution is probably
> not representative for the real world, that you would find one 9000 series
> hard drive for every ten floppy drives you come across, or even less likely
> one HDD for every fifth PET.

My experience in the US is much less than one hard disk per five PETs
I've seen - perhaps one in twenty.

-ethan


       Message was sent through the cbm-hackers mailing list