Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.2 9/18/84; site bbncc5.UUCP
Path: utzoo!linus!decvax!ucbvax!ucdavis!lll-crg!seismo!harvard!bbnccv!bbncc5!tappan
From: tappan@bbncc5.UUCP (Dan Tappan)
Newsgroups: net.micro.mac
Subject: Re: MacSCSI board query
Message-ID: <728@bbncc5.UUCP>
Date: Thu, 17-Oct-85 07:57:05 EDT
Article-I.D.: bbncc5.728
Posted: Thu Oct 17 07:57:05 1985
Date-Received: Sun, 20-Oct-85 07:23:45 EDT
References: <2469@ut-ngp.UTEXAS> <284@ukecc.UUCP>
Reply-To: tappan@bbncc5.UUCP (Dan Tappan)
Distribution: net
Organization: Bolt Beranek and Newman, Cambridge, MA
Lines: 28

In article <284@ukecc.UUCP> crowe@ukecc.UUCP (Donald "Deke" Crowe) writes:
>Some of us at U of Ky are doing three of these hard disk systems and we have 
>already etched the boards and assembled the disk controllers, etc.  Now we are 
>all worried that the Dobb's article has some typos.  We noticed at first that
>the address decoding doesn`t make a great deal of sense.  But decoding can be
>done a million different ways and still work.  We have talked to John Bass at 
>Fastimes and he says the new ROMs will not work.  This is logical because the 
>new ROMs will be decoded where the SCSI board is being addressed.  There is no
>way to get around this because there are no more address lines available from
>the ROM sockets.>

Actually the current MAC ROM's are 23256's, for which pin 1 is a NC,
however (reading from the "IM underground") pin 1 is connected to A16 -
which should let you simply plug in the higher density ROM's (23512?).

More to the point. The MacSCSI design uses A20 as part of it's address 
select logic. A20 is connected to one of the chip enable lines on the ROM.
In other words, the ROM address space is 2Meg, the MacSCSI sits in the
high Meg of that, the ROM sits in the low, even with the new ROM's the
ROM is only going to take up 128KB of that Meg.

I have been working on a version of Macscsi that uses a 74139 to do
the decoding (1 chip as opposed to the two in the Dr. Dobbs article).
I'm still waiting for my NCR5380 before I know how well it's going to
work.

Dan Tappan
BB&N Labs