|
Attendance (no order)
- Ernie Bergmann
- Mike Dudek
- Greg Lecheminant
- Piers Dawe
- Joe Gwinn
- Vivek Telang
- Lew Aronson
- Tom Lindsay
- John Jaeger
- Petar Pepeljugoski
- Andre Van Schyndel
- Paul Wachtel
- Pavel Zivny
- Ali Ghiasi
Agreed agenda
-
Attendance
-
Agenda
-
Review previous minutes
-
Discussions
-
TWDP, waveforms, etc.
-
Other
- Jitter waveform (see attached)
- Next call
Discussions
- D2.0 is out for comment.
- It can be found on the private area of the LRM web site.
- Comments due 4/28.
- Those who are not members should contact David Cunningham on commenting
procedures.
- TWDP, waveforms, etc.
- Piers made a point that pre-processing is part of the overall
process of going from measurement to result. We should see if folks can get
the same result starting from the same raw waveform.
- Tom will send all raw waveforms to Piers for uploading next to the
pre-processed waveforms.
- There appears to be some variation in penalties depending on sample
frequency and interpolation methods. Instrumentation companies commonly
use sin(x)/x method. Pavel to look into this more. ClariPhy uses cubic
spline (in MATLAB). Companies should try different methods.
- Numerous pre-processed TP2 waveforms are now on the web in the private
area.
- A separate file contains OMA and Zero values. Tom will update this to
include links to eye patterns (if they exist) in other presentations and
add other descriptive material, if available.
- Broadcom (Vivek) has corroborated ClariPhy penalties from previous week
(see http://grouper.ieee.org/groups/802/3/aq/public/upload/TP2measuredwaveformpenalties_finiteEQsTP3D1.1stressors.pdf.). ~0.1
dB worst case differences. Using same pre-processed waveforms, but
independently developed EDC engine, so this helps validate TWDP code.
- Tom to forward Vivek's results to Piers for upload.
- Non-monotonic behavior in results is explained by BER-based MMSE EDC
method. Taps are determined based on Gaussian assumption, but penalty is
based on BER with ISI pdf that is not Gaussian. See description in Annex A
of D2.0 for more info.
- Waveform analysis
- Main concern is that, in some cases, lab waveforms exhibit higher
penalties with finite equalizers than the Gaussian waveform, even when
their PIE-D values are lower. Close to 0.5 dB has been seen. Not clear how
this might impact budget or coverage.
- Need more waveforms to analyze that represent the
expected/desired range of performance over process, environment, aging,
etc. No need to disclose details of how waveforms were generated, but
committee should be sure that penalty limits allow what is appropriate for
cost-effective designs. Some impairment variables were suggested:
bias, temperature, and electrical coupling & matching.
- Not discussed on call - be sure captured waveforms are not
inverted!!
- (Uncorrelated) jitter
- Reviewed jitter waveform Tom had included in meeting announcement.
Agreed it is better than present square wave and should be submitted unless
upcoming work dictates otherwise.
- Lots of discussion about which edge - rising vs. falling (or both),
single vs. many (or all), worst vs. a rss method, etc. No conclusion, but
suggested that folks get in lab and determine what is required. A goal of
the spec is reasonable accurate and simple.
Next call - 2 weeks. This will give TP3 some time to generate some new
stressors.
-
Date: Thurs, Apr 14, 05
-
Time: 9:00 AM Pacific
-
Duration: 1:30 max
-
Number: 401-694-1515
-
Access code: 421721#
Please send comments or corrections.
Thanks,
Tom Lindsay ClariPhy Communications tom.lindsay@clariphy.comphone:
(425) 608-0209 or (949) 480-9210 cell: (206) 790-3240 fax: (425)
608-0232
----- Original Message -----
Sent: Wednesday, March 30, 2005 9:53
AM
Subject: [10GMMF] Reminder for TP2 call
3/31/05
Dial-in details
-
Date: Thurs, Mar 31, 05
-
Time: 9:00 AM Pacific
-
Duration: 1:30 max
-
Number: 401-694-1515
-
Access code: 421721#
Proposed agenda
Tom
|