Does anybody know how long the light will stay in one of the two
polarization states of a PM fiber? Assuming perfect incoupling along
one of the two axis of the fiber, i would imagine that the light does
not stay in that state for infinite long fibers due to fiber
imperfections/scattering/whatever. Does anybody have a typical length
for this, or can this effect be ignored with respect to normal
attenuation? (For my application i would need PM fibers of up to 1 km
@ 633 nm)

On 29 Oct 2003 02:23:49 -0800, bassw@hotmail.com (Bas) wrote:

Does anybody know how long the light will stay in one of the two
polarization states of a PM fiber? Assuming perfect incoupling along
one of the two axis of the fiber, i would imagine that the light does
not stay in that state for infinite long fibers due to fiber
imperfections/scattering/whatever. Does anybody have a typical length
for this, or can this effect be ignored with respect to normal
attenuation? (For my application i would need PM fibers of up to 1 km
@ 633 nm)


Hi

Have tried it just for some 10 meters. So what I can say is: give it a
try

At this length we sent the two channels of an interferometer through
the single-mode fiber and separated them again at the end.
Actually it is classical polarization optics.

One of the beams were frequency shifted with a Bragg-cell before the
fiber input, so coupling to the other polarization would lead to a
crosstalk interference signal not wanted (having the frequency of the
shift, i.e. 40 MHz).

We collected scattered light from particles and since this intensity
is much weaker than the beams, I can say that crosstalk was below a
level of 10 to minus 5 or so.

JK wrote:

On 29 Oct 2003 02:23:49 -0800, bassw@hotmail.com (Bas) wrote:

Does anybody know how long the light will stay in one of the two
polarization states of a PM fiber? Assuming perfect incoupling along
one of the two axis of the fiber, i would imagine that the light does
not stay in that state for infinite long fibers due to fiber
imperfections/scattering/whatever. Does anybody have a typical length
for this, or can this effect be ignored with respect to normal
attenuation? (For my application i would need PM fibers of up to 1 km
@ 633 nm)


Hi

Have tried it just for some 10 meters. So what I can say is: give it a
try

At this length we sent the two channels of an interferometer through
the single-mode fiber and separated them again at the end.
Actually it is classical polarization optics.

One of the beams were frequency shifted with a Bragg-cell before the
fiber input, so coupling to the other polarization would lead to a
crosstalk interference signal not wanted (having the frequency of the
shift, i.e. 40 MHz).

We collected scattered light from particles and since this intensity
is much weaker than the beams, I can say that crosstalk was below a
level of 10 to minus 5 or so.

This is interesting. Normally the polarization purity of a PM fibre is
only ~20 dB or so after a long length, though the *orthogonality* of the
two modes will be much better than this--more like the 10**-5 you
quote. What do your optics look like, exactly, and how are you
adjusting the polarization? How stable is it with time and temperature?

One dirty secret with PM fibre is that although the average polarization
is more stable than in regular SM fibre, the rate of change of
polarization with temperature is actually comparable or even worse. The
coupling goes down, but since the birefringence is so high, it changes
very rapidly with temperature. Thus I normally think of PM fibre as
being good for stabilizing the operating point of an interferometer, so
that another technique such as modulation-generated carrier can pull
decent data out of it.

At this length we sent the two channels of an interferometer through
the single-mode fiber and separated them again at the end.
Actually it is classical polarization optics.

One of the beams were frequency shifted with a Bragg-cell before the
fiber input, so coupling to the other polarization would lead to a
crosstalk interference signal not wanted (having the frequency of the
shift, i.e. 40 MHz).

We collected scattered light from particles and since this intensity
is much weaker than the beams, I can say that crosstalk was below a
level of 10 to minus 5 or so.

This is interesting. Normally the polarization purity of a PM fibre is
only ~20 dB or so after a long length, though the *orthogonality* of the
two modes will be much better than this--more like the 10**-5 you
quote. What do your optics look like, exactly, and how are you
adjusting the polarization? How stable is it with time and temperature?

(snip)
There are more than 2 ways of getting a linear polarisation out.
Remember that the PM fiber is stress induced and so may act as
waveplate. So normally the polarisation state changes along the length
just as inside a polarization retarder.

The 2 stable ways are to send the input linear polarisation in aligned
accurately with one of the principal axes of the fiber.
Therefore I claimed to use a lambda half waveplate for adjustment of
this.

JK wrote:

At this length we sent the two channels of an interferometer through
the single-mode fiber and separated them again at the end.
Actually it is classical polarization optics.

One of the beams were frequency shifted with a Bragg-cell before the
fiber input, so coupling to the other polarization would lead to a
crosstalk interference signal not wanted (having the frequency of the
shift, i.e. 40 MHz).

We collected scattered light from particles and since this intensity
is much weaker than the beams, I can say that crosstalk was below a
level of 10 to minus 5 or so.

This is interesting. Normally the polarization purity of a PM fibre is
only ~20 dB or so after a long length, though the *orthogonality* of the
two modes will be much better than this--more like the 10**-5 you
quote. What do your optics look like, exactly, and how are you
adjusting the polarization? How stable is it with time and temperature?

(snip)
There are more than 2 ways of getting a linear polarisation out.
Remember that the PM fiber is stress induced and so may act as
waveplate. So normally the polarisation state changes along the length
just as inside a polarization retarder.

The 2 stable ways are to send the input linear polarisation in aligned
accurately with one of the principal axes of the fiber.
Therefore I claimed to use a lambda half waveplate for adjustment of
this.


So, the reason you're not answering the question is that the design is
proprietary, is that right? Is it part of a product? Being mysterious
about it doesn't lend credibility, but if this is for real, I very much
want to learn how to do it.

You're claiming a very very pure polarization coming out of your PM
fibre, which is rather surprising. Fibres in general are really lousy
for instrument building, unless it's an embedded or sacrificial
application. It would be a significant advance in the field to obtain
10**-5 polarization purity over time and temperature in an unattended
application--even a Faraday-rotator mirror (FRM) has trouble doing
10**-5.

Hi

I realized yesterday that the interferometer is completely
differential and a crosstalk level up to a few percent will work fine.
The interferometric signal is collected in backscatter and collects
equal amount of light from each beam.
Sorry, 10**-2.

But anyway, remember that the function of the fiber is classical
polarization optics.

The use of a lambda half plate for adjustment is not feasible with a
laser diode, but is per example very feasible to HeNe and argon laser
beams.

Anyway we have found that for the question rised originally: Be
carefull by using the fiber right. If not, a kilometer of length will
lead to terrible fluctiations in output polarization state.

JK wrote:
I realized yesterday that the interferometer is completely
differential and a crosstalk level up to a few percent will work fine.
The interferometric signal is collected in backscatter and collects
equal amount of light from each beam.
Sorry, 10**-2.

But anyway, remember that the function of the fiber is classical
polarization optics.

The use of a lambda half plate for adjustment is not feasible with a
laser diode, but is per example very feasible to HeNe and argon laser
beams.

Anyway we have found that for the question rised originally: Be
carefull by using the fiber right. If not, a kilometer of length will
lead to terrible fluctiations in output polarization state.

That makes a good deal of sense. If you get round to it, I'd be
interested in a more complete description of the system. Even getting
the polarization crosstalk to be a common-mode effect is a very
worthwhile achievement.

Actually this use of the PM fiber leads to a very accurate
interferometric measurement of Doppler shift in a flow. The noise
contribution from the fiber is very low, probably about a few Hz.
Our competitors tried by using 2 fibers instead, since they were
blocked by our patent. Actually this pick up noise in the range of a
few kHz. Surprisingly this was acceptable to many measurements, - I
would never have dreamed of such a noisy solution.