Re: [10GMMF] task 4 Oct. 4th call -- Polarization
Vipul,
thank you for your pertinent comments. Let me try to explain why
polarization works in multimode fiber when offset is taken into account.
I completely agree with you on the fundamental mode solution for the
axial symmetric fiber.
Let us continue to assume the fiber has rotational symmetry and the
vector wave equation reduces to the well known Helmoltz scalar equation.
According to weakly guiding approximation WGA, all modes belonging to
the same mode group exhibit the same propagation constant and therefore
have the same group velocity (this is strictly true within the WGA, but
it is not relevant for the discussion).
One important statement is that the whole mode solution (geometrical
mode structure) refers to the assumed polarization of the electric
field. It is customary to assume the x-axis oriented along the linear
polarization of the electric field. At this point, if mode excitation
would be still axial symmetric (OFL), any polarization effect would be
observed. "Everything" would be the same independently from any
azimuthal position.
What is important now is the result of the overlapping integral at the
launching section to calculate the amount of intensity transferred to
each mode group. If the excitation is eccentric, defined through some
offset coordinate, the coupling coefficients will depend on the relative
orientation of the linear polarization (reference x-axis) and the
(r,phi) coordinate of the offset center.
This is the important point. For a given mode group assembled by several
modes with the same propagation constant, the amount of intensity
transferred between two offset fibers depends on the relative
orientation of the offset respect to the light polarization. In
addition, at the same radial coordinate region belong different mode
groups with different group velocity.
Assuming the profile is not optimized we experience a polarization
induced DMD which leads to pulse distortion after few hundreds of meters
at the 10GbE scale.
I hope to have been clearer in this picture description. This is not in
contradiction with well known mode theory. It just consider one
different situation where polarization, offset and profile optimization
play the role in getting pulse distortion.
Please, I apologize if something was missing or uncorrected. Looking
forward in receiving your comments,
Best regards
Stefano
-----Original Message-----
From: owner-stds-802-3-10gmmf@IEEE.ORG
[mailto:owner-stds-802-3-10gmmf@IEEE.ORG] On Behalf Of Vipul Bhatt
Sent: Donnerstag, 7. Oktober 2004 20:34
To: STDS-802-3-10GMMF@listserv.ieee.org
Subject: Re: [10GMMF] task 4 Oct. 4th call -- Polarization
I appreciate the work done by Joerg and his team, and the polarization
experiment report seems well organized, but...polarization penalty for
multimode links? I hope someone can educate me, because the way I see
it, polarization shouldn't make any difference to DMD.
I am starting with the assumption that only the input polarization was
changed, and everything else was unchanged -- i.e., the excitation
region didn't shift, the input field intensity pattern remained the
same, etc.
DMD changes only if the power distribution among mode groups changes. A
mode group consists of one or more LP modes having nearly identical
propagation constant. My contention is that a change in input
polarization not only leaves power in each mode group intact, but it
also leaves power in each mode intact.
We normally characterize a mode as LP(m,n), where m is the azimuthal
number and n is the radial number. But actually, each LP(m,n) mode
consists of two (if m=0) or four (if m>0) distinguishable modes having
the same propagation constant -- distinguishable because of two possible
polarizations and two possible azimuthal orientations, sine and cosine.
For example, mode group 6 consists of LP(5,1), LP(3,2) and LP(1,3), and
has a total of 12 distinguishable modes.
Power in each LP mode is the sum of the powers carried by its
distinguishable modes. These distinguishable modes share the power, the
proportion being decided by how the input polarization aligns with the
polarization axes of the distinguishable modes. When the input
polarization direction is changed, the proportion of power among these
distinguishable modes changes, but the total power in an LP mode remains
the same. Therefore, DMD should not be affected by changes in input
polarization.
Please tell me what I am missing. Thank you.
Vipul
Vipul_Bhatt@ieee.org
+1-650-941-6290
-----Original Message-----
From: owner-stds-802-3-10gmmf@IEEE.ORG
[mailto:owner-stds-802-3-10gmmf@IEEE.ORG]On Behalf Of yusun@IEEE.ORG
Sent: Wednesday, October 06, 2004 9:04 AM
To: STDS-802-3-10GMMF@listserv.ieee.org
Subject: [10GMMF] the minitues of task 4 Oct. 4th call
Dear colleagues,
I attached the minutes of tele-conference of task 4 group on Oct. 4th. I
would appreciate your comments. The proposed next task 4 meeting is Oct
11th, 2004 (Monday), 11am east coast, 8 am west coast and 4 pm UK.
Best regards,
Yu