<< Previous Message Main Index Next Message >>
<< Previous Message in Thread This Month Next Message in Thread >>
Date   : Sat, 17 Jan 2004 17:21:59 GMT
From   : Pete Turnbull <pete@...>
Subject: Re: Host adapter Issue D

On Jan 17, 15:14, Andrew Benham wrote:
> On Fri, 16 Jan 2004, Pete Turnbull wrote:

> > The Adaptec ACB4000 series
> > store drive geometry on the drive, and read it back as part of
their
> > initialisation process; the Xebec S1410 has to be given the
parameters
> > by the host.  When I built my S1410 system for the Beeb, I wrote my
own
> > formatter, which stored the disk info in a short file on track 0,
and I
> > wrote a little boot program which extracted that and set up the
S1410
> > properly.
>
> My solution to this problem with the Xebec controller was to
disassemble
> their firmware and discover where the default drive geometry was
stored
> in the EPROM. From memory, the controller defaulted to the parameters
> for a standard (at the time) 5MB drive. [ Yes, that's 5MB and not
5GB! ].
> I blew a new EPROM with my drive's geometry in it instead, and that
was
> my operational system for a while.

That's a good enough way.  I sold several systems with Xebecs, so I had
to be sure it would work with a variety of drives.

> My motivation to switch to a ACB4000 controller came when I upgraded
> my Master 128/512 to Acorn's "Alternative Master 128 System ROM".
> This ROM provided ADFS v2.03: this includes a "*VERIFY" command which
> verifies floppies =and= hard disks. However, for verifying hard disks
> it uses SCSI command 2FH - this SCSI command is not supported by
> the Xebec card.

I'd forgotten that.  Yes, you're right.  I vaguely remember writing a
verify command which used the Xebec 09H Read Verify command.  It's just
a read command (08H) that doesn't return data; almost the same as 2FH
except it doesn't use the extended addresses.  AFAIR the Adaptecs don't
support 09H.

When I was looking up the command code for that just now, I discovered
that there was an alternative "Enhanced Option" firmware version that
stored the drive parameters on the drive.  Apparently it used track 0,
automatically adding one to any track address supplied for normal
operations.  The Adaptecs didn't do that; they hid their data in the
intersector gaps.

-- 
Pete                                           Peter Turnbull
                                               Network Manager
                                               University of York
<< Previous Message Main Index Next Message >>
<< Previous Message in Thread This Month Next Message in Thread >>