Date   : Wed, 08 Mar 2006 21:36:29 +0000
From   : Fragula <fragula@...>
Subject: Re: 6502s and SASI/SCSI.

Howdy Phil!

Phil Blundell wrote:

> Oh right, yeah, probably.  The ones I have lying around are Xebec and
> Emulex, but I'm not sure exactly where they came from.
The card driving the RD-51 on the MicroPDP is an Emulex.

> I think so.  Like you say, the basic "read block" and "write block"
> commands are pretty much the same: it's just the more esoteric stuff
> that tends to vary.
Well, we probably don't need a heck of a lot of esoteric stuff, except
where that stuff is needed (i.e. to get a drive to spin up, reset etc.
on those that need it.)

> It probably doesn't make a whole lot of difference, to be honest.  Since
> the transports are different you're always going to need some level of
> translating intelligence, and once you reach that point it doesn't
> matter much whether you are doing SCSI->SCSI bridging or SCSI->ATA
> command set translation.
Ahh.. I've been labouring under the assuption that the ADFS was talking
more-or-less SCSI, i.e. pidgeon SCSI with none of the more advanced
stuff that is now neccessary even to get the thing to start up, that
translation per-se was not needed for more than just a few things (i.e.
drive initialisation, formatting, unexpected message handling and the
like.) that could maybe be handled by a few vector traps to a utility ROM.

> I thought a bit more about the idea of a S(A|C)SI->ATA bridge board
> during an idle moment today.  So long as you could live with
> asynchronous transfers only (which I think is all the Acorn thing can do
> anyway)
PMSL! Wot To UDMA-133?? :'D If its as fast as a floppy, its fast enough,
though faster is almost always good.

> you could probably do the whole thing with just a single
> microcontroller, ATmega165 or some such. 
Carefull with the "you" if you don't mind.. I'm getting a bit senile
now, and I wasn't that good to start with.

> It'd be pretty easy to have
> different partitions on the drive show up as individual LUNs, which I
> guess is what ADFS would want.
Ahhh.. Yes I think thats right. There not being any device IDs used with
the ACB rings a bell, just LUNs. It really is that old. <sigh> There are
probably only two valid LUNs, which probably gives a max drive useage of
1 gig, at least from within ADFS.

> No, it needed a little hardware mod; that was the PAL thing I mentioned
> earlier on.
Not too bad. Real Parroty would be better though, if it was a 1Mhz<>SCSI
board-from-scratch affair.

Cheers!

M.