Re: [10GMMF] Notes from Aug 24 meeting on TP3 definition
Hi Michael,
Thank for providing fast resume. Please add my name to the list since I
was attending.
I have some comments about the marked questions.
1. Linearity.
Why should we argue about ISI linearity? The laser linearly
converts amplitude into intensity and what we can measure on DCA in
every case is just optical intensity. Out target should be to reproduce
a synthesized optical signal whose detected intensity fits ell with
worst case fiber samples. Why should we discuss about electric field
propagation and how modal components overlap? That will include non
linear overlapping but we will never manage the electrical field from
the source point of view.
As a second point I do not see how to relate intramodal
dispersion with non linear mode overlapping. Those are two different
things. We will have non linear pulse contribution overlap from
different fiber modes due to square law detection of overlapping
electric fields. Each modal electric field goes under intramodal
dispersion (chromatic dispersion) by himself, without any kind of
interaction with spatially adjacent propagating modes. I agree
intramodal dispersion is negligible for such distances.
In addition, according to Petre proposal (to which I agree) we
remain in the electrical domain at all sections (page 4-5 of Petre
presentation) and both LPF act on electrically down-converted signals.
The MSE algorithm works in the electrical domain.
I have some questions for Petre:
1. The metric you use for defining MSE convergence is PSR
on page 3. Since the error is at the denominator of the in integrand
function you should reach convergence for increasing PSR, isn't?
2. Once you have computed PSR, how do you proceed in
changing A0, A1, A2, T1 and T2? Do you use a multidimensional
optimization tool?
Please let me know your feedback,
Best regards
Stefano
-----Original Message-----
From: owner-stds-802-3-10gmmf@LISTSERV.IEEE.ORG
[mailto:owner-stds-802-3-10gmmf@LISTSERV.IEEE.ORG] On Behalf Of Michael
Lawton
Sent: Dienstag, 24. August 2004 20:43
To: STDS-802-3-10GMMF@LISTSERV.IEEE.ORG
Subject: [10GMMF] Notes from Aug 24 meeting on TP3 definition
1. Attendees
Aronson, Lew
Babla, Chet
Dawe, Piers
Ewen, John
Fiedler, Jens
George, John
Ghiasi, Ali
Hallemeier, Pete
Hanberg, Jesper
King, Jonathan
Latchman, Ryan
Lawton, Mike
Lindsay, Tom
Peters Weem, Jan
Popescu, Petre
Shanbhag, Abhijit
Sun, Yi
Witt, Kevin
Zona, Bob
1. Questions on Stressed Rx Test E-O and E-O-E linearity
Lew presentd some slides addressing the points raised by Abhijit on this
topic. In these slides 2 questions are addressed:-
i) With the ISI block before the E to O converteer will there be
intramodal dispersion and non linerar ISI after the the PIN (due to the
square law detector)
- the response was that the intramodal dispersion will be
negligble due the the very short lengths of fiber
- for the non linearities arising from the square law detector
Lew said he had a colleague who had shown him some references which
indicated that the E-O-E process in fact turns out to be linear (cross
products average out).
ii) The impact of non-idealities in the E to O conversion
- Lew stated that this could be an issue with a slow or
non-linear source, and a requirement should be included in the
characterisation portion of the test, namely
o The test signal should be defined optically and
measured with a linear reference receiver
o mismatch between electrical ISI and received ISI
should be corrected with either a better source or compensation.
We are awaiting experimental results before we close this agenda item.
2. Designing the ISI generator for receiver testing
We reviewed the latest set of Petre's slides, dated 24/8/04.
There was some discussion regarding the need for LPF2 on p4. It was felt
that this probably was needed in order for the fitting algorithm to be
truly comparing like with like.
There were some questions on the pulse shaping on pages 4 and 5. Petre
will address these next week.
There was some discussion regarding fitting with a pulse shape with
normalised area vs normalised height. Not clear which is best. The
approach used currently fits to a normalised height.
There was a discussion on whether a very high speed source could present
the test receiver with any problems and hence it should be avoided.
No-one presented any concerns on this subject.
Please let me know if you have any comments or corrections.
Next meeting Tuesday August 31st.
Best Regards
Mike