Re: SCSI CD-ROM Drive

[dempson@actrix.gen.nz (David Empson)]

<ell.ee.ess.aich.ee.are.jay.tee> wrote:

> Nathan Mates (nathan@visi.com) wrote:
> : Matt Ackeret <mattack@area.com> wrote:
> : >Mike Hackett wrote:
> : >>From what I've read, the RamFast (what is the proper capitalization
> : >>on that anyway? :-) ) does not support drives that are =strictly=
> : >>SCSI-2. 
> 
> : >I though that SCSI-2 was by definition a superset of SCSI-1.

Actually, I would classify it as a _subset_ in some ways - it is much
stricter about the command set supported by drives.  It also defines
(and standardises) a lot more device types than SCSI-1.

> :    I believe it is, but there are some drives that appear to *require*
> : the additional features of SCSI-2 not present in SCSI-1, which is what
> : Mike Hackett was trying to get across originally. 
> 
> :    I believe the story on why certain Quantum and other SCSI drives
> : don't work with the RamFast is that in SCSI-1, the host adaptor card
> : was not required to have a SCSI ID of its own. In SCSI-2, there is the
> : possibility that devices can request (and require) that.
> 
> From what I've read in the literature, SCSI-1 also required the host
> card to have a SCSI ID;

The key question is whether the host implements the "Arbitration" phase
of the SCSI communication sequence.

If the host doesn't implement arbitration, and is not able to act as a
target (in response to another host), then it should be able to get away
with not having a SCSI ID - there is no time at which it needs to
identify itself to the drive.

This is the situation with the RamFast.


The SCSI-2 standard allows a drive to require arbitration support by the
host.  Some recent Quantum drives (notably the Fireball series) do
require this, so they don't work on a RamFast.

All versions of the Apple SCSI cards support arbitration, so they don't
have problems.

-- 
David Empson
dempson@actrix.gen.nz
Snail mail: P.O. Box 27-103, Wellington, New Zealand