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

RE: [EFM] 10/16/02 FEC Call Minutes

I agree with your test setup.
Since it's expected that the highest MPN penalty will be 
at the max and min wavelength, it is important to force the 
min and max wavelength.

I understand that you have preliminary results.
I think it can be helpfull to have the following data:
1. Optical power at the receiver.
2. Error probability for the first few bits after locking and compare it
to coutinuous case. 
3. Probability of unlock as a function of optical power (or BER)

Best regards,

-----Original Message-----
[] On Behalf Of larry
Sent: Thursday, October 17, 2002 3:29 AM
Subject: [EFM] 10/16/02 FEC Call Minutes

EFM FEC Conference Call Minutes
Hosted by EFMA
October 16,2002
10-11:20 PST


Eric Lynskey, UNH IOL
Piers Dawe, Agilent
Meir Bartur, Zonu
Lior Khermosh, Passave
Mike Peppler
Masoud Khansari, Centillium
John Limb, Broadcom

1. Spent all the meeting discussing the proposed MPN test methodology
and block diagram from Meir and the proposed CDR test methodology and
block diagram from Eric. Copies attached. The FEC group wants buy in
from EFM fiber and optical experts on the proposed test setups and
methodology BEFORE spending the effort.

2. MPN Testing

a) All attendees agreed that this test set-up and proposed methodology
would produce valuable insight into the distance improvement possible
with FEC in MPN limited links.

b) Meir's plan  is to do a family of plots of MPN penalty vs Normalized
dispersion, Dn (Dn = BDLs) with each plot at a different line BER. Meir
proposes using a fixed L of fiber (say 25Km) and vary the laser Temp to
change s (s = spectral width), resulting in a change to Dn. From the
family of plots can then be used to determine the Dn difference between
two different BER's.  The reason for fixing the fiber L is to avoid a
connector attenuation variability issue resulting when you plug and
unplug connectors to insert different L's of fiber.  Also, one fixed
fiber keeps the number of variables in the set up to a minimum. MPN
penalty is determined by measuring the receive power with and without
the 25Km fiber loop (at the same BER). Meir also wants to plot BER vs
received power.

c) John and Piers feel that doing the testing proposed by Meir using at
least two different L's would provide more meaningful results.  The
Agrawal  theory says that it is the Dn and k parameters alone that
determines MPN penalty for a given BER, so if the theory is correct, the
family of plots for the two different fiber lengths will overlap.
Connector attenuation variability may not be an issue since we are
concerned at what happens in the MPN dominant region  of the MPN vs Dn
curve where the slope is very steep.

d) John said he might be able to do some MPN testing in his lab too.  He
will use different lengths of fiber.

e) Meir's MPN test plan looks good so far but we would like more
feedback from other experts within EFM.  Again, will Meirs test set up
and methodology answer the question, what distance improvement can you
achieve using FEC with MPN limited links?

3. CDR Testing

a) Eric of UNH IOL is doing this testing and has some very preliminary
results.  Test block diagram and methodology attached.

b) With all due respect to Eric, the time spent discussing CDR testing
was constrained due to the lateness of the hour.

c) Eric is measuring delay time from light in to SERDES preamble out.
While Eric's measurements are of great value to EPON, the specific FEC
issue we would like to answer is how does a low BER affect the CDR lock
time.  This can be determined by looking at the delta time it takes
electrical data pulses into the SERDES to lock the CDR at different

Larry again stressed the importance of buy in on the test methodologies

We are making good progress on the testing front thanks to Meir, Eric
and to those who have commented on their proposed methodologies.
Granted, up until this point only a few have seen the proposed test
methodologies, however, with the wide distribution of these minutes I
expect (and hope) to get more valuable input.



larry rennie wrote:

> Date: Wednesday, October 16, 2002
> Time: 1:00 p.m. (EST) = 10:00a.m (PST)
> Duration: 1 hour
> Chair: Larry Rennie, National Semiconductor
> Dial in: 1-800-331-8016 (toll free), 1-816-650-0666 (direct) 
> Participant code: 744004
> Hosted by EFMA.
> Proposed Agenda/Discussion:
> 1. CDR low BER Testting.  Status.
> 2. MPN low BER testing.  Status.
> 3. Presentations for November EFM meeting..
> Larry