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

Re: Finally Fun

OK, Roger, I am willing to discourse with you for the good of the
enterprise, but only if you will answer my queries. One that I missed an
answer on is the following:

(From roger)
>Cisco's proposal has the following A, B sections.
>A- Use the Signal pair 1,2 and 3,6 to deliver power on a new switch.
>B- Use unused pair (4,5 and 7,8 ) to deliver power for mid-span.

(My question again)
There are at least two of us here that remain confused as to what you mean
in by "A, B".

Does this mean that Cisco is planning on using "A" and "B", or
Cisco is using "A" and others should do "B"
or... what?

I confess that I don't watch your web site, but our marketeers do, as you
might guess, and they noticed the conflicts between a couple of your press
releases. Please clarify.

Now it is my turn to answer some queries for you: (delimited by *****)

At 05:16 PM 5/10/00 -0700, R karam wrote:
>I want to thank Larry for all his worries, he is beginning to De-bore me.
>and would appreciate any inputs or concern that we can address
>to make the next meeting worthwhile for everyone and tackle the issues.
>I got some of Larry's feedback down here to keep the email resonable.
>between the "--"
>Larry's input on the MLT3 with/without power:
>Well, actually, here is an area where I think that just using a normal
>signal is inadequate. I don't seem to be alone in this, beause IEEE 802.3d
>(MAU Conformance Test Suite, on your CD) was deemed necessary to test for
>proper implementation, and it was for this reason that I put in in my List
>Of Requirements (Have You Looked).
>To put it in a nutshell, these tests look at the ability of a an Ethernet
>link to deal with bursts of out-of-band interference, minimum-level signals
>and the like. Having magnetics operating near saturation may affect the
>ability of the system to cope with such things. 
>>> Larry, why  would  we run the magnetic near saturation to start with ?
>>> please specify where in your list this concern has risen so it won't be
>>>     guess work on my part- thank's.

(Larry's response)
You know, in my previous message I said that you had probably been in on
the early deliberations in which the magnetics vendors stated what the
capabilities were for maintaining DC cancellation in magnetics, and, based
upon which presentations, the Committee leaned toward using the (4,5 and
7,8) wires. This was before your product announcement in January.

However, thinking back, I really don't remember seeing you or Bob Bell.
Bill Q and Carl N were definitely in the deliberations in Kauai, as was
Kevin B. Also, you say below you have only been to two meetings, and
January and March would fill that out, all right.

Anyway, that and at least one earlier meeting were where the questions were
raised (see the presentations) and the concern level heightened.

The "bottom line" performance estimates that we got from Pulse and Bel were
that they could control the DC balance or flux cancellation  within 10% of
the DC power delivered over variations in process, time, temperature and
phase of the moon (+/-35 mA or so of bias). We were therefore rather
surprised (to put it mildly) to see the bar raised so much in January.

I hope that answers your question so far.

>Larry's input on the Magnetic department
>When you are running 100TX you have 20 mA signals, but up to 8 mA of
>allowed DC unbalance. 10BASE-T is DC balanced, but you have 50 mA signals.
>So the two schemes sort of help each other out in the magnetics department,
>but is that still true with a big DC bias present? (I don't know-- but this
>is the kind of thing that will have to be sorted out, as well as the prior
>Ethernet clauses that deal with such things.)
> >well Larry I can measure this, but given that the current is balanced to
>>a decent  level, we expect the unbalance to chew away at the 8ma margin,
>>how was the 8ma number chosen?, and if the OCL (open circuit inductance),
is way above the 350uh
>>to start, you are not on the edge of saturation, and your phy is doing
BWL (base line
>>wander correction) would this still be a problem.  Will try to address
this issue.  (you gonna tip me

(Larry's response)
Well, the "decent level" is the question. There was never any doubt that
big enough magnetics would solve it. Whether the magnetics proposed do it
is a good question for the magnetics vendors!

The 8 mA number is a direct quote from Paragraph 9.1.7 of TP-PMD, which is
the basis of 100TX. 

They don't directly say where it comes from, but I'm going to risk a wild
guess that it comes from the Baseline Wander discussion in Appendix 2 of
TP-PMD, where they present the infamous "killer packet" that is stated to
produce 0.75 volt of DC offset. This, divided by the characteristic
impedance of the transmission system of 100 ohms, gives 0.75V / 100 Ohms
which handily comes out to be 7.5 mA or about 8 mA with a little tolerance
and windage thrown in. I might point out (and PHY vendors should chime in
if I am wrong) that as I understand it all of the baseline wander
compensation is at the receiver end of the link. There is no pre-distortion
of the data stream at the transmit end.


>Larry's input on Detection Pulses:
>I was talking about the detection pulses, not what you get after the power
>is on. I do not think you have shown anything of that, at least not yet, or
>I missed it. My reason for asking is that there seems to be a certain
>amount of filter de-Q'ing in your differential scheme, similar to what we
>found in the common mode scheme because of the low load impedance on the
>filter output.
>And I think that a digital filter would be a damned clever solution to this!
>For the detection pulses, we control them, their amplitude, frequency- even
>polarity, they are differential and much lower in frequency, the filter is
a low pass
>filter I remind everyone, so ok, it shaves off 100mv that I designed
for.... why
>is any filter mismatch a worry at this point?.  Also as promised 90% the
>filter will be gone- you gotta start somewhere....

(Larry's response)
I hope this is something you can show us. At this point the only way we can
see data is to replicate your system, which is pretty non-productive.


Finally, here I am again on the playing fields, could you tell us a little
>about your approach so we can share our worries about it with you, for we
have been
>to two meetings so far, and your concerns toward our approach never made it 
>to me so they can be addressed.

(Larry's response)
This is clearly a legitimate request, and one that will be addressed. 

As I have said, up until recently, we have had little bandwidth for this
project and frankly we have had some difficulty getting our management
behind it. However, we do owe Cisco a debt of gratitude in that your
blow-the-doors-off transient approach has been a wake-up call that has
greatly facilitated our efforts in raising internal interest and awareness!

Therefore, you may expect increased activity in the form of presentation of
engineering tests and results in the next couple of meetings.


(Now Larry's Sermon)

Now I wanna preach a bit. I think that you believe that there is no
technical merit to our resistance to using the signal pairs for powering,
and that we are just playing politics. Well, it just ain't so. To
illustrate I am going to quote at length from my upcoming,
not-yet-published posthumous autobiology:

" scene: ... last summer (1999)...

Larry sat down to thoroughly consider the matter. The PHY detection scheme
that had been presented was clever, really clever, and he had said so. It
seemed to offer a robustness against false detection that none of the other
proposals had. And yet...

Almost without realizing it he picked up the well-worn copy of IEC 11801.
As he carelessly riffled through the pages he chanced upon the crosstalk
requirements of Section 6.2. Suddenly the figures on the page appeared to
glow and take on an eerie brightness.

As the seconds passed the glow increased into a blinding light.

It was a monstrous, grisly light.

A hideous light.

A boiling, pestilential light.

A light that would have disfigured hell.

In burning figures the table (6.5) showed the MINIMUM amounts of isolation
between pairs guaranteed by the cable vendors at various frequencies, all
neatly adjusted for the attenuation of the link at these frequencies.

It was phenomenal. Unbelievable! Yet there it was, a Compact with Reality:

At 100 Mhz it was 4 dB. Not too interesting.....

But by the time you got down to 62.5 MHz it was 13 dB! A factor of nearly

And as you went down in frequency, the numbers just got better and better:

23 dB (14:1) at 31.25 MHz!
28 dB (25:1) at 20 MHz!
20 dB (31.6:1) at 16 MHz, the Category 3 limit!
35 dB (56:1) at 10 MHz!

Why, by the time you got down to 4 MHz you had a guaranteed minimum
isolation of 40 dB, a hundred to one!!!

Larry choked at the realization. Why, at lower frequencies the unused cable
pairs could be anything from AC open to AC shorted with hardly any effect
at all on the signal-carrying pairs! And if the cable were to be terminated
in its correct characteristic differential and common-mode impedances....

He could hardly believe it. Could this FREE isolation be real? There was
only one way to find out-- hook up some cables on the trusty network
analyzer and see if it panned out.

The figures did not lie. In fact, the cable sample used was even BETTER
than the advertised numbers. 

Even with the crummy NEXT of RJ-45 connectors, the isolation below 30 MHz
was at least 15 dB (5.623:1) under the worst case conditions he could
produce by wiggling cables. And these were REALLY bad cables.

Awe-struck by this Truth, Larry thought, "Clearly, this is not an advantage
to be lightly disposed of by attaching power supply parts directly to the
signal wires! No indeed! Not when you can reduce the risk of interaction by
nearly an order of magnitude by simply using those wires that hadn't been
used previously!"

It seemed almost a Meeting with Destiny, seeing a terrific use for these
previously neglected, so-called "unused" pairs of wires.

Larry shuffled off, rather shaken by this experience. Clearly, this was
going to take some consideration and quite possibly about three fingers of
single-malt Scotch to fully absorb it all...."

(End of Quote, with due apologies to Douglas Adams)

No, Roger, this is not about politics. Ya picked the HARD set of wires to
use, ole buddy!

Regards & no offense,