2. 790
TECHNICAL PHYSICS LETTERS Vol. 36 No. 9 2010
A.M. KURBATOV, R.A. KURBATOV
We have made two fibers, with diameters 80 and
90 μm. Based on the first fiber, we obtained a polariz
ing fiber, and based on the second fiber, we obtained a
polarization maintaining fiber with losses amounting
to 0.35 dB/km.
Polarizing 500 m long fiber due to its winding into
60 mm diameter coil showed x mode losses of
3 dB/km and a dichroism of 30 dB/km. Light source
spectrum width in these measurements was approxi
mately 20 nm. It is a good result considering that a
birefringence value of 3.4 × 10–4
is not very high. Since
winding this fiber with a diameter larger than 60 mm
did not led to dichroism, it is clear that we have got a
bend type polarizer.
W fiber is known due to the fact that even its fun
damental mode may exhibit cutoff at finite wave
length. Cutoff means that mode effective RI becomes
equal to quartz cladding RI:
Calculation of the W fiber fundamental mode cut
off wavelength (threshold) is quite simple. However,
the mode cutoff by no means always means its large
losses. That is why it is necessary to clarify real mech
anisms of fundamental mode losses. Primarily we will
consider the following two of these:
(1) Fundamental mode radiation tunneling into
external quartz cladding and touching the absorbing
coating (in straight fiber).
(2) Bending losses.
In our fiber, the fundamental mode cutoff wave
length is approximately 2.2 μm. First of the loss mech
anisms according to calculations leads to the fact that
fundamental mode losses become significant already
neff n3.=
at wavelength of 1.8 μm. Bending losses due to wind
ing into 40 mm diameter coil shift the beginning of
losses approximately to 1.55 μm. Below we will only
consider the bending losses.
Most important question is the dependence of
bending losses on the parameter χ = τ/ρ, where τ is the
RC radius and ρ is the core radius. According to calcu
lations, RC in our fiber may be arbitrary thin because
the operation wavelength (1.55 μm) is far from funda
mental mode cutoff threshold (2.2 μm).
Figure 3 shows the behavior of wavelength λ0 at
which bending losses are 1 dB/km as a function of
parameter χ (this could be named the fundamental
mode bending cutoff threshold). The same figure pre
sents a plot of the fundamental mode spot diameter
(MFD) evolution as dependent on χ. In order to cal
culate the bending influence, we generalized the
approach that was developed in [6] for conventional
two layer fibers. To confirm the results, we also mod
eled the problem by the method of supermodes [7, 8],
which were calculated by finite difference method
(one of supermodes always significantly resembles the
fundamental mode of straight fiber, so its attenuation
determines bending losses).
From Fig. 3, one can see the following. First of all,
at all χ the bending cutoff shift does not decrease
below 1.55 μm. Second, this curve initially goes down,
then (in the region 1.3 < χ < 1.42) it has approximately
constant level (1.55 μm), and eventually it grows
slowly. This behavior of λ0(χ) curve can be explained
by the model of fundamental mode coupling to higher
order attenuated modes. The coupling coefficients of
these modes monotonically decrease when χ grows.
Synchronism of these modes first increases sharply
and prevails over decrease in the coupling coefficients
(losses grow), then it increases smoothly (losses do not
change), and eventually it stabilizes (losses decrease).
Third, the MFD at χ < 1.6 sharply grows and this
gives the limitation to RC width from below, so one has
to use RC with χ > 1.6.
Calculations using the methods described above
show that, at a birefringence larger than 8 × 10–4
, it is
possible to obtain fiber with large dichroism in a wide
0.005
−0.011
Fig. 1. Initial preform RI profile for W fiber PANDA. Fig. 2. W fiber PANDA cross section photograph.
RI values in various elements of PANDA fiber
Parameter RI Value
Core RI, n1 1.4655
Reflecting cladding RI, n2 1.451
Outer cladding RI, n3 1.46
Stress applying rods RI 1.4515
3. TECHNICAL PHYSICS LETTERS Vol. 36 No. 9 2010
NEW OPTICAL W FIBER PANDA FOR FIBER OPTIC GYROSCOPE SENSITIVE COIL 791
spectral range (100 nm and more) and with low losses.
To increase the dichroism, it is also possible to apply
absorbing/scattering materials located in quartz clad
ding [9, 10].
Generally, the proposed W fiber combines advan
tages of two convenient fibers. First of these is the fiber
with the RI difference between core and quartz clad
ding equal to Δn13 = n1 – n3 (see table); the second
fiber has the RI difference Δn12 = n1 – n2. In the first
fiber, when winding it, one can ensure wide single
polarization spectral window, because birefringence
against Δn13 is significant. But in this case one will not
obtain the desired MFD. In second fiber, there is no
problem with MFD, but there is no chance to obtain
wide single polarization spectral window, because
birefringence against large Δn12 is small. The proposed
W fiber has simultaneously a wide spectral window (as
that in the first fiber) and the opportunity to obtain
desired MFD (as that in the second fiber).
Based on the same W structure, Panda fiber with
diameter 90 μm operates as polarization maintaining
(PM fiber). In this case, dichroism is removed to
longer wavelengths, but due to this x mode losses are
simultaneously sharply reduced. At present, based on
the structure described above, we obtained PM fiber
samples with losses amounting to 0.35 dB/km, which
is not far from the 0.2 dB/km limit.
Small losses in FOG coil could be used in different
ways. For example, in FOG there is a minimal level of
optical signal power reaching the photodetector, for
which photodetector and preamplifier electronic
noise is suppressed. Power reserve which is obtained
due to low loss fiber application could be used to
employ other methods of additional signal phase mod
ulation. This reserve can also be used for coil winding
with several kilometers length, which will improve
FOG sensitivity.
As for h parameter, it appeared to be 2 × 10–5 m–1.
This is good result for birefringence B = 3.4 × 10–4
,
considering that h parameter depends on B approxi
mately as B–2
. Further birefringence growth is purely
technological problem and it is associated with stress
applying rods doping.
Finally, material losses in stress applying rods are
not larger than several hundredths of dB/km (due to
fundamental mode tight confinement in the core). For
the same reason, sensitivity to twisting is absent, and
this also gives to these fibers certain advantages when
using in fiber optic gyroscopes.
Acknowledgments. Authors are grateful to FIRE
RAS 226 Laboratory Head G.A. Ivanov for his help in
fiber fabrication.
REFERENCES
1. A. L. Tomashuk, K. M. Golant, and M. O. Zabezhailov,
Fiber Opt. Technol. Mater. Dev., No. 4, 52 (2001).
2. V. Dangui, H. K. Kim, M. J. F. Digonnet, and
G. S. Kino, Opt. Express 13, 6669 (2005).
3. S. O. Konorov, L. A. Mel’nikov, A. A. Ivanov,
M. V. Alfimov A. V. Shcherbakov, and A. M. Zheltikov,
Laser Phys. Lett. 2 (7), 366 (2005).
4. S. Kawakami and S. Nishida, IEEE J. Quant. Electron.
QE 10 (12) (1974).
5. A. M. Kurbatov and R. A. Kurbatov, RF Patent
no. 2250482 (Priority from 16.09.03; Register
20.04.05).
6. C. Vassallo, J. Lightwave Technol. LT 3, 416 (1985).
7. P. L. Francois and C. Vassallo, Appl. Opt. 22, 3109
(1983).
8. J. A. Besley and J. D. Love, IEEE Proc. Optoelectron.
144, 411 (1997).
9. A. M. Kurbatov and R. A. Kurbatov, RF Patent
no. 2250481 (Priority from 19.05.03; Register
20.04.05).
10. A. M. Kurbatov and R. A. Kurbatov, RF Patent
no. 2269147 (Priority from 26.05.04; Register
27.01.06).
1.0
λ0, μm
χ
1.80
1.8
1.77
1.74
1.71
1.68
1.65
1.62
1.59
1.56
1.53
1.50
10.0
9.8
9.6
9.4
9.2
9.0
8.8
8.6
8.4
8.2
8.0
MFD, μm
1.2 1.4 1.6 2.0 2.2 2.4 2.6 2.8 3.0
Fig. 3. Behavior of wavelength λ0 at which bending losses
are 1 dB/km (solid line) and fundamental mode MFD
(dashed line) depending on the RC width.