Path: utzoo!utgpu!news-server.csri.toronto.edu!rutgers!cs.utexas.edu!sdd.hp.com!decwrl!sgi!rpw3@rigden.wpd.sgi.com
From: rpw3@rigden.wpd.sgi.com (Rob Warnock)
Newsgroups: comp.dcom.lans
Subject: Re: cheap ethernet transceivers
Keywords: cheap Ethernet transceivers
Message-ID: <66117@sgi.sgi.com>
Date: 4 Aug 90 09:51:32 GMT
References: <6Xc1m12w162w@remote.halcyon.wa.com> <1990Jul31.031438.19090@blilly.UUCP>
Sender: rpw3@rigden.wpd.sgi.com
Reply-To: rpw3@sgi.com (Rob Warnock)
Distribution: usa
Organization: Silicon Graphics, Inc., Mountain View, CA
Lines: 44

In article <1990Jul31.031438.19090@blilly.UUCP> bruce@balilly.uucp
(Bruce Lilly) writes:
+---------------
| IC's are described in ``Handbook of LAN Technology'' by Paul J. Fortier,
| published by McGraw-Hill ($64.95).  The pinout of the 15-pin cable is
| shown in ``Keeping the Link'' by Martin Nemzow, also published by
| McGraw-Hill ($39.95). (No, I don't work for McGraw-Hill) There are other
| books; check your local library or technical bookstore, as well as
| application notes from the IC manufacturers.
+---------------

Better still would be to read the Ethernet Specification itself (*not* 802.3!).
It's quite readable, has all the actual requirements for voltages, impedances,
drive currents, rise times, etc., and even has sample circuits for driving and
receiving the 15-pin interface.

WARNING! The sample circuit for a transceiver-cable receiver has a bug, which
is in both Version 1.0 and Version 2.0 of the spec, and I have also seen it
in the schematics for at least one of the DEC-manufactured transceivers. The
"data valid" output of the circuit is mistakenly connected to the *center-tap*
of the terminators for the incoming signal. [Boo! Hiss!] Thus common-mode
signals are sent backwards up the cable. [When copying this circuit, don't
do it that way. ;-} ] A friend of mine at Xerox once confided that they
knew the bug was there, but couldn't change it in time because of the long
review/approval cycle through the DEC/Intel/Xerox loop. When Version 2.0 was
being edited, they'd forgotten just what the bug was, and removed the wrong
wire, leaving only the *bad* connection! It's pretty obvious what "data valid"
*should* be hooked to -- just look for a gate with no output connections.

Otherwise, and on the whole, the Ethernet Spec is a *very* good guide.

You also should look at the data sheets for some of the transceiver chips,
such as the ones made by AMD or National. I believe these have sample circuits
for complete transceivers, including hints/warnings about circuit layout to
meet the stringent stray capacitance and isolation rules.


-Rob

-----
Rob Warnock, MS-9U/510		rpw3@sgi.com		rpw3@pei.com
Silicon Graphics, Inc.		(415)335-1673		Protocol Engines, Inc.
2011 N. Shoreline Blvd.
Mountain View, CA  94039-7311