Subject: Re: Archive Python DAT problem with ahc 2940W
To: None <email@example.com, firstname.lastname@example.org>
From: Matthew Jacob <email@example.com>
Date: 09/29/1997 21:13:44
>>The drive functioned fine when I configured the BIOS not to negotiate sync.
>>Would the termination problem not exist independently of sys/async config?
>If you don't do sync, you don't load the bus as much and termination
>is less of an issue.
Actually, I believe it's more of an electrical thing. When in async mode,
there is an explicit wait for REQ* to be asserted, then you assert ACK*,
then you wait for REQ* to be deasserted and then you deassert ACK*. If your
termination is really fouled up, you just get a really long transition
time for the signals- eventually they get driven to the desired state.
In sync mode, for all but data phases, the same thing applies. In data
phases, there is an agreement between the initiator and the target that
REQ* (and associated data + parity) will be arriving so many nanoseconds
apart for a burst of <offset> bytes. If your termination is fouled up,
this won't be correctly clocked, and and the fifo latches usually get
screwed up about how many pulses (and bytes of data) actually moved.