[10GMMF] Influence of mechanical stress to polarization effect
David and all,
I am posting this to the group because I think that discussion is
interesting for the general understanding of the polarization behaviour.
David, I agree that your observation of the limited transmission length
where a polarization can measured with a polarizer is very interesting.
That gives an indication about the length scale on which this kind of
effects change in the fiber.
But this measures just the linear polarization. We performed (nearly all
of our)experiments also using circular polarized launch for the
transmission over MMF and found out that the effects are absolut similar
to the linear polarized case. You will see the same pulse variations and
the same size of the effects by vibrations of the fiber.
I did not stated this seperately because my picture of the situation is
that even if you launch linear polarized light you will have a large
part of circular polarized light after a certain length of fiber. That
is easy to understand if you think of the large amount of birefringence
which is present (especially in the MMF). Therefore you will measure no
linear polarization after a certain fiber length.
About 15 years ago I developed a measurement equipment to measure the
mechanical stress in the fibers and performed several measurements of
the axial mechanical stress in drawn fibers (SMF and also MMF). Some
description and measurements of stress variations near a fiber splice
you will find in the following paper:
"Fast photoelastic stress determination: application to monomode fibres
and splices", Th. Rose, D. Spriegel, J.-R. Kropp, Meas.Sci.Technol. 4,
Maybe that can be uploaded to the server. In the references you will
find some more literature and there is out even more.
I am sorry but I have no example of stress profiles in MMF anymore but I
will give brief explainations.
Generally you have to destinguish between stresses caused by the
material composition of core and cladding and the stress induced by the
drawing procedure. Looking on Fig.2 of the reference these influences
can be extracted for the single mode case from the presented splice
measurement. The top measurement shows the stress distribution across
some standard fiber. The observed stress profile is mainly generated by
the drawing procedure (this profile is individual to that fiber, you can
find a large variation of stress distributions which are strongly
connected with the drawing parameters of the production). The bottom
stress profile from the center of the splice shows the contribution from
the material compositon because the fiber was melted in this area and
cooled down without any additional external stress.
In multimode fibers the material composition part of the stress is
following within the core roughly the radial parabolic composition
trend. This stress is much larger than in the single mode case because
of the higher Ge content. Superimposed to that you have also in the MMF
the additional stress induced by the drawing process. As far as I
remember the stress in the MMF is in general about around 5 times larger
than in SMF.
These are old findings but I want to share that info because I think
that these stresses and the resulting birefringence is the basic reason
of our observations. In my understanding the geometrical imperfections
like non circularity of the core will have a minor impact on the
Maybe John Abbot can provide further infos about that subject.
From: email@example.com [mailto:firstname.lastname@example.org]
Sent: Mittwoch, 3. November 2004 18:18
To: Kropp Joerg (IFFO MOD CE); Ysun@optiumcorp.com;
Cc: Stefano.Bottacchi@libero.it; Fiedler Jens (IFFO MOD CE)
Subject: RE: First Draft summary of experimental observations
Thanks for putting this together - we'll review this and get you
feedback by Friday.
Simon is preparing our slides and we should have them ready for Monday.
Just to clear to everyone, I intend to split of some of our results for
the task 2 view of the work.
Jeorg on Monday's call you asked if it was the connector rather than the
long fibre that was suppressing the polarisation effect. We have
repeated the measurements and you were right - the connector was the
problem. But the interesting thing is that for the long patchcord the
degree of polarisation is zero (for linear polarisation).
Also, in response to John Abbott feedback, you will all get an updated
version of my analytical calculations on the polarisation effect via the
reflector. I have converted the calculations from HMG modes to LGG modes
per Johns request. The result is that I have ended up back where I
started - axisymmetric index profiles (per our current 10GBASE-LRM
models) don't predict polarisation dependent impulse responses.
I'm of the view that we have gathered enough experimental data on this
effect to deal with it at an engineering level: a theory would be nice
but is, in my opinion, optional.