Date   : Mon, 22 Dec 1997 13:55:00 GMT+2
From   : "Mnr WE Meyer (NW5-44) x2637 (Walter)" <wmeyer@...>
Subject: Re: Hard Drives

Mark
> 
> Would it be possible to change the SASI board that is in seemingly wide use.
> The details of it are at http://www.nvg.unit.no/bbc/doc/HardDrive-v1.0.zip
> In fact it could even be the same board. I'd be very interested to know if
> it is.
> 
The board I have is not the same - it seems some of the interfacing 
chips have been replaced by a 8255.  It could be that the boards are 
functionally similar (the 8255 does not really do a lot on the GSL 
board, and could be replaced by a couple of 74LS devices).  Does 
anybody have a circuit diagram of the SASI board mentioned above so 
that I can check if a similar patch could work on that board?

Anyway, if somebody wants to modify an existing SASI host adaptor,
the problems I had to overcome to get my SCSI drive to work were:

1) On the hard drive, disable parity by setting a jumper, and set 
   its SCSI address to 0.  This might be enough to get some drives 
   working.

2) In my setup, the selection phase did not work correctly.
   (The selection phase is the first step in communications 
   between the host adaptor (AKA the initiator) and the hard disk 
   drive (AKA the target).  According to the standard the 
   initiator is supposed to output the SCSI ID bit of the target 
   (HDD) on the data bus.  This did not happen.)

   To fix it I had to:
   a) Modify the board to output the data bus during the selection    
      phase.  (The GSL board was designed specifically *not* to 
      output any data during the selection phase.  On the GSL board 
      the modification involved cutting the track to pin #11 on the 
      8255, and connecting the disconnected pin #11 to pin #24 on the 
      8255.)

   b) Change the software to write the SCSI ID bits to the data bus.
      (This involved changing a LDA#0:STA &FCC0 to LDA#1:STA &FCC0.
      This pattern occurred once only in the EPROM and each 
      utility program, so changing it was easy.

3) After this everything else worked fine, except that some of the 
   "start up" status messages sent by the drive were interpreted as 
   errors, so after pressing break and typing *. <RETURN>, the 
   computer would respond with Winchester error 00.  The second *. 
   was usually successful.  Hereafter, everything worked without 
   any further problems, until the next reset.

The above solution worked with a GSL host adaptor, connected to a BBC 
Micro using Amcom's E-NET 4.94 fileserver EPROM & formatting 
software.  It is certainly less work to modify an existing adaptor, 
but if you are building something from scratch, I'd rather go for 
system mentioned by Crispin in an earlier posting.

For more information on the SCSI standard, the SCSI-2 specifications 
are available on the web at various locations.  (I found it under 
http://scitexdv.com:8080/SCSI2/Frames/SCSI2.html.) Chapter 6 (Logical 
characteristics) discusses the data transfer process.  Note that for 
SCSI-1 the arbitration phase is not required.

I don't want to clog up the mail list with technical info, so 
if anybody wants more, please contact me directly.

Regards

Walter

--
Walter Meyer (wmeyer@...              )
Department of Physics, University of Pretoria, South Africa