On Wed, Apr 15, 2009 at 1:53 AM,  <ruud.baltissen@apg.nl> wrote:
> Hallo allemaal,
> Ethan> ... and restored them.
>
> What did you have to do?

One of them had a drive with a defective track zero sensor (IIRC), but
since the component that required replacing was apparently inside the
sealed part of the HDA, I sent the drive out for an after-market
refurb - this was in the late 1980s when it was still possible to get
drives refurbished.  I got the whole job done, new platters, etc.,
along with the repaired sensor, and it cost (in 1989 dollars) about
$70.  This was less than a like-new TM603S was going for at the time,
so I considered it worth it for a "zero hour" disk.  Had I known that
an ST225 or ST251 would have worked without changes, I probably would
have used one of those rather than refurb an old drive like a TM603S.

The other one was rather simple - a previous owner had probably
replaced the disk mechanism, and it failed the formatting operation.
I found a 4-head drive (TM602S) but the jumper was in place for a
6-head drive (TM603S) - the formatter tried to format the missing
heads and failed.

> Three years (?) ago I received three 9060's and one 9090 from Jaj Janssen (Duisburg, Germany) for repair, whom on its turn got them from Anders. After hearing this http://www.dienstagstreff.de/ gave me another 9060 to repair as well. I managed to get only one 9060 to life again :( Three harddisks were broken but this was not a problem as I found out that I could use Seagate ST-225's as replacement.

Yep.

> Four DOS boards were brooken but I managed to repair two.

Good.  Those are somewhat easy to repair for someone with 6502 repair
experience.  They are not particularly complicated, no more so than
the main board in a 4040 or 1541.  I'd ensure any 2114s were working
and then use something like a code tracer or Fluke 9010A analyzer with
a 6502 pod to see what the DOS board was doing, test memory, test the
bus lines, etc.  After that, it's probably time to poke around with a
scope and/or a logic analyzer.

> The other two are repairable as well IMHO.

Even better.

> But the biggest problem were, what I call, the "middle boards". These bords have been placed between the harddisk and the DOS board. They are equiped with a lot of, partly unknown, logic around a CPU like AMD-IC. They convert the SASI interface from the DOS board to the MFM interface of the harddisks.

Yep.  I don't know that I ever had good documentation for that board
except what came in the C= service manual (which wasn't enough to
repair defective electronics, and it certainly wasn't enough to
understand the low-level formatting process).

> And I still wonder why C= designed things in this way.

SCSI wasn't really around yet, so SASI was the "easy" way of
offloading all the ST-506 peculiarities to a product designed,
debugged and built by someone else.  In that era, it wasn't uncommon
to have a "bridge" between the host and the drive - later than the
D90x0, in the early days of SCSI, the Adaptec ACB4000 was very popular
- it was found with Ataris, BBC Micros, Sun 3s, Amigas, and more.

> I have been trying to figure out how the SASI interface worked but without any good info on SASI and without any good source listings, I gave up: I needed the time for, IMHO, more important projects. But I intend to use the info gathered with my 1541IDE project to replace this middle board and MFM-HD with an IDE-HD.

I only know a little about SASI, but it's essentially proto-SCSI.  I'd
always thought about trying to modify the DOS board firmware to speak
enough SCSI to talk to an embedded drive.  At the time I was
considering this 20MB-50MB embedded SCSI drives were cheap enough to
make the effort worthwhile, even if the filesystem was limited to
16MB.  Electrically, that 50-pin interface on the DOS board should be
5MB/sec-narrow-SCSI compatible (parity might be an issue though).

I think the info about SASI was largely in Shugart docs, and probably
mostly with the S-100 crowd.  By the time disk were hitting the home
market in quantity, SASI was replaced by SCSI.

If you know anything about the earliest SCSI implementations
(especially how drives and controllers handled the lack of an IDENTIFY
packet to query the drive geometry), SASI will look very familiar.
It's probably possible to untangle the code on the DOS board by using
an ACB4000 programmers manual as a guide to deciphering the packets
sent.  Knowing what packets the "middle board" supports would be very
helpful, but one could make educated guesses based on other SASI and
early SCSI "bridge" boards.

-ethan


       Message was sent through the cbm-hackers mailing list