[802.3ae_Serial] From serial PMD call 22 Jan: test fest and stressed eye specification
Present
-------
Piers Dawe Agilent SPG
Dubravko Babic Alvesta
Raj Savara Network Elements
Juergen Rahn Lucent
Jan Peeters Weem Intel, Optical Platform Division
Benny Christensen Intel (Giga)
Greg LeCheminant Agilent T&M
Stretch Camnitz Agilent T&M
Adam Healey Agere
Tom Lindsay Stratos Lightwave
Peter Ohlen Optillion
Petar Pepeljugoski IBM
February Test Fest
------------------
Questions about proposed lab fest in Santa Rosa, e.g. what kit is available,
what tests?
Request for bad examples as input for the test development. Bad 10G
transmitters would be perfect! A 1G transmitter tested at 2G, for example,
might be a surrogate.
Discussion of confidentiality around the test fest. Should the host (Greg,
Agilent) schedule lab time per company and collaborative time? NDA? The
host will try to write some kind of proposal, would like more input from
transceiver vendors on what to measure.
New stressed eye generator
--------------------------
What proportion of vertical eye closure from interfering sine wave, from
filter?
What range of frequencies for sine wave?
How to calibrate?
With or without pattern dependent jitter?
Lone bit pattern or what?
Think will be using scope (DCA) to calibrate, so can't see low
probability events, and as pattern, SJ and interferer are not synchronised,
won't get a simple clean eye diagram. Can we agree what we will see, both
the trace and a mask that would bound it? Would be beneficial to get an
actual scope screen shot into the standard, to avoid misunderstandings.
Do we have a mask for the stressed eye?
Note predictable relation between SJ and (T)J on one hand, and
interferer amplitude and VECP on the other.
How to specify? Maybe 4 dimensions out of 6 or so candidate parameters?
This will lead to a hard spec on pulse shrinkage, at high probability, which
the bathtub didn't provide.
What is the metric for the proportion of jitter in which both edges of a bit
move in the same direction, (like clock-induced jitter?) and in which they
move in opposite directions (e.g. pulse shrinkage)? Some kind of
statistical correlation coefficient? And what score does an actual (direct
modulated) transmitter have?
Two key areas to find out:
What do transmitters actually do?
What do receivers react to?
A deliberate high pass filter in a receiver would help its performance in
the stressed eye, except that we should use an adequately long test pattern
such as 2^31-1 so that any such approach does not avoid its baseline wander
penalty. And we won't be able to calibrate the stressed eye with such a
long pattern without next-generation test equipment. So we are resigned to
calibrating on a shorter pattern, maybe 2^10 or 2^15, and testing receivers
with a longer one. Need to try out really short simple patterns, and eye
line capture.
Petar said that he could look at the proposed stressed eye generator by
simulation; please send him suggestions on exactly what to simulate.
TDP (Transmitter and Dispersion Penalty)
----------------------------------------
Try out TDP method at 850 and 1310 nm. Prove that the transversal filter,
for 850 nm virtual TP3, works. Can this be proved at a lower bit rate e.g.
Gigabit?
The test fest hosts would like examples of real transmitter and receiver, to
answer questions such as: how does a product receiver compare with a BERT?
Eye mask
--------
Juergen pointed out that the eye mask changes I had mentioned related
primarily to 1310 nm and he did not want the 1550 nm mask lowered.
Not test related
----------------
There are items to be decided (such as comments) which do not rely on test
procedure development. Address them when? by email? Suggested Piers
and/or David Kabal list out or group the issues in the comment database.
Next phone meeting
------------------
The next PMD teleconference will be tomorrow, Tuesday, at the usual time and
coordinates:
4:15 pm GMT = 17:15 CET = 11:15 am EST = 8:15 am PST, Tuesday 29
January 2002
+1(816)650-0631 Access code 39209
Piers