The debate on the screen is not the theory or validity of equalization
technique, which has been proved on the copper of known media application.
The debate here is the fiber frequency response characteristics with
complicated DMD factor. The DMD characteristics of the fiber is not
As ordinary approaches, we should define the problem (fiber bandwidth
related to DMD) first, then we can chose the tools. There are many variety
of tools available.
The eye pattern ISI, or amplitude reduction is not frequency response at
all. For a linear BW media, copper, the frequency response is defined, and
given by vendor. You can design an equalizer without any test all. Fiber
is quite different.
Edward S. Chang
NetWorth Technologies, Inc.
[mailto:owner-stds-802-3-hssg@xxxxxxxx]On Behalf Of Kishore Kota
Sent: Friday, July 21, 2000 12:10 AM
To: Jonathan Thatcher; vipul.bhatt@xxxxxxxxxxx; HSSG
Cc: Vivek Telang
Subject: RE: Equalization
If you are interested in the potential for signal processing techniques
to mitigate the interference effects in optical links, it will be very
useful for this group to collect some data. Specifically, if this group
would collect receiver sample data at atleast 2X (preferably 4X)
the baud rate, it would facilitate a lot of analysis. A number of signal
processing people who have been monitoring this group (including yours
truly) would be willing to calculate the potential benefit of any
equalization techniques and present the data on this reflector and at
the next meeting.
Since I have seen numerous eye diagrams presented to this
group, it should be possible to use the same lab setup and
a digitizing oscilloscope to collect this data. For this data to be useful,
the transmitted bits should be random and at least a few thousand baud
intervals of samples to allow extraction of the necessary parameters. The
sampling could be completely asynchronous as long as the sample rate is at
least twice the baudrate. Given this data, it is relatively straightforward
for those in the signal processing field to calculate the effectiveness of
equalization techniques. Numerous techniques are available to combat linear
and certain types of non-linear deterministic effects. The random effects,
if any, would take more work (and more data).
Cicada Semiconductor Inc.
811 Barton Springs Road, Suite 550
Austin, Texas 78704
512-327-3500 x104 voice
> -----Original Message-----
> From: owner-stds-802-3-hssg@xxxxxxxx
> [mailto:owner-stds-802-3-hssg@xxxxxxxx]On Behalf Of Jonathan Thatcher
> Sent: Thursday, July 20, 2000 7:51 PM
> To: 'vipul.bhatt@xxxxxxxxxxx'; HSSG
> Subject: RE: Equalization
> Not a punishment, just something I've been noodling over and
> wasn't sure if
> we were ready to ask.
> Specifically, I have been curious about the potential for the DMD response
> (assumes no launch control) to be non-deterministic. In short,
> what happens
> to that thing we used to call "modal noise" in a DMD-sensitive system?
> It seems likely that a simple experiment could be created if one
> had a BERT,
> Scope, LX transceiver, and a spool of the "bad DMD fiber" to see if the
> general shape of the pulse into the Rx would change under various types of
> "abuse." I'm not talking about dynamic changes like wrapping it around a
> mandrel or anything that would not be "typical" in an installed
> infrastructure. Just fiber blowing in the wind, the temperature of the Tx
> changing... stuff like that.
> Now, you could argue that these types of things would be "random"
> in nature.
> Even so, if the response of the system to these changes is significantly
> non-linear, then the effect could end up unmanageable. I hope otherwise.