NetNews Usenet Archive 1993 #3/NN_1993_3.iso/spool/comp/periphs/scsi/6646

Internet Message Format | 1993-01-23 | 1.9 KB

Path: sparky!uunet!usc!elroy.jpl.nasa.gov!ames!agate!apple!goofy!applelink.apple.com
From: [e-mail] (Scott Smyers)
Newsgroups: comp.periphs.scsi
Subject: Re: You will be terminated!
Message-ID: <[e-mail]>
Date: 22 Jan 93 20:26:12 GMT
References: <[e-mail]>
Sender: [e-mail]
Organization: Apple Computer, Inc.
Lines: 30


The AMD glitch eater circuit helps the chip ignore glitches which occur
when signals are de-asserted on the SCSI bus.  Because the SCSI bus is
open collector, all the current needed for de-assertion of signals comes
from the terminators, which are at the ends of the bus of course.  As a
result, the when a driver releases a signal there is often a glitch as
the in-rushing current from either end of the bus works to bring the
voltage back up to the de-asserted level.

I believe that the glitch eater is only on the REQ signal (I could be
wrong on this).  The data lines are not edge sensitive, but the REQ and
ACK signals respond to each other as the result of a transition.  If the
REQ signal is dirty, the other device may believe that it has received 2
REQ's when there has only been one.

Anyway, the glitch eater helps the chip work better in environments which
are noisy, which is the case when you have many devices and/or long
cables and/or poor cables and/or you got up on the wrong side of the bed.
 It does this by "eating" glitches which occur on the REQ signal.  The
glitch eater circuit also provides better immunity in clean environments.

The bottom line is that the glitch eater has absolutely nothing to do
with termination.  You still need termination.  I know of no chip that
has termination built in.  In fact, unless you could disable it, on chip
termination is not very useful (it limits the device to being at the end
of the bus).

------------------------------------------------
Scott Smyers
Apple Computer