Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

RE: Signal vs. Idle debate

(I sent this out last week but it didn't seem to make it into the
reflector. Try #2:)

Paul Yew and others,

There are a number of Committee members that are concerned about (and,
indeed, have serious reservations about) the probable side effects of
superimposing comparatively high current DC on the signaling pairs. To date
we have had no systematic demonstration by Cisco or anyone else that this
arrangement is benign, and one of the charter items for the Committee is
that the DC powering scheme chosen "shall not degrade the transmission

And, as it happens, Cisco apparently did not choose to adopt some of the
isolation requirements that the Committee reaffirmed from existing 802.3

So it isn't quite such an opened-and-closed book situation as the marketing
hype would have you believe. And the onus is on Cisco to provide enough
information to make their case, the "trust us" statements appearing here

At the March meeting Cisco asked the question "what do we have to do to
convince the Committee?". I have posted an answer to this at the IEEE
802.3af public web site:

It is titled "Recommended quantitative measurements of the parameters" and
is comprised of extracts from IEEE 802.3 and ANSI TP-PMD specifications.
These are the "numbers" that we all test our products against (hopefully)
and constitute the criteria for which ANY scheme (not just Cisco's) will
have to pass muster, in addition to the special requirements set forth by
the Committee.

I am sure that most of the Committee members are familiar with these
criteria, particularly in view of the fact that  many of the members
participated in the standards committees that drew up these requirements in
the first place.

A solid presentation of measurements of these parameters in an actual
operating system (with margins, not just PASS or FAIL results) would do a
lot to give the Committee sufficient information to evaluate the viability
of the Cisco design or any other that is proposed.

(Comments on the requirements list are welcome)


Larry Miller