Sie sind auf Seite 1von 2

Mo.4.B.

ECOC 2008, 21-25 September 2008, Brussels, Belgium

Accurate measurement of fourth order dispersion coefficient in short


highly nonlinear fibers
J.M. Chavez Boggio, S. Moro, N. Alic, and S. Radic
University of California, San Diego, 9500 Gilman Drive, La Jolla, CA 92093
Abstract
An improved four-wave-mixing method capable of measuring ultra-low values of the fourth-order dispersion
coefficient in short highly nonlinear fibers is introduced and validated experimentally.
Introduction
The accurate measurement of the fourth order
dispersion coefficient (4) is relevant in several
applications in nonlinear fiber optics, and especially
important for devices such as fiber optical parametric
amplifiers (FOPAs). For example, the magnitude of 4
greatly impacts the available bandwidth and gain
flatness in FOPAs [1].
Recently, a few methods based on parametric
amplification (PA) [2-3] and four-wave mixing (FWM)
[4] were proposed for 4 measurement. The PA
based-methods require a high pump power in order to
provide exponential gain. In [2], as much as 60W was
used to pump a fiber of only 2m in length.
Furthermore, these methods often entail inaccuracies
in determining the exact peak gain frequency due to
gain sidelobes width. On the other hand, the method
reported in [4] relies in conventional FWM, requiring
much lower power levels for equal fiber length, and
does not require gain. This renders the FWM method
potentially more suitable for accurate measurements
of 4 in very short (few meters) fibers.
In this paper we propose a simple way to improve the
accuracy of the measurement of the ratio between the
third and the fourth order dispersion coefficients
(3/ 4) reported in [4].

(X,Y), by varying l and measuring pm . From these


fitting parameters we obtain 3/4 and 0. As
illustrated in Fig. 1, phase matching (maximization of
the generated FWM product) is obtained by tuning the
noise source around 0.
A

FWM product
pm

ASE noise

Fig. 1. Schematic of the measurement of pm. The


noise is tuned at three locations: the continuous line
corresponds to the phase matched case (the FWM
product was maximized), while the dotted and dashed
lines correspond to the no phase matching cases.
Errors in measuring 3/ 4 originate from inaccuracies
in measuring the frequency pm . In very short fibers
the generated FWM product can be very broad, due
to its width scaling as 1/3L(l pm )2 [4]. This fact
will enhance the experimental uncertainty in
determining pm . Furthermore, the FWM power
2
scaling with L , requires an elevated sensitivity of the
optical spectrum analyzer (OSA) used to visualize the
FWM product. With the objective of analyze the error
in short fiber measurements, consider the noise to be
slightly detuned from the phase matching condition by
c. It can be shown that the generated power (to the
first order in c and normalized relative to the power
that would be generated in the phase matched case)
is given by
1
( l )

(2)

= sinc 2 ( pm l )2 c L 3 + 4 pm
12

Theoretical background and results


The measurement of 3/4 is based on the FWM
interaction between a noise source (centered at a
frequency A and positioned in close vicinity to 0)
and a laser at frequency l (which is far from 0). The
noise should have a bandwidth of A < 1 nm (thus
A << | - A|). The FWM process satisfy the
condition = 1 + 1 l, where 1 and 1 are two
Fourier components within the noise spectrum, and
is the frequency of the generated wave (see Fig. 1).
Therefore, the resulting FWM field at an arbitrary
frequency is the sum of the contributions from a
continuum of pairs of noise components at 1 and 1.
The FWM power is [4]

The ratio indicates the rate of reduction of the FWM


power as the noise is detuned from the phase
matching condition. Eq. 2 was obtained under the c
> 2A assumption. Figure 2 shows dependency on
the following parameters: 4 (values 1 and 810-5
4
ps /km), L (4, 10, and 50 m), and the separation
between the laser and the generated product (25 and
3
35 THz). The 3 was fixed to 0.04 ps /km. Note in
Figs. 2(a-c) that as the fiber length is reduced the
generated power is less sensitive to the detuning from
the phase matching condition. Therefore, errors in
determining pm will be important in very short fibers.
On the other hand, the importance of laser location is
characterized by comparing Fig.2(b) (25 THz
separation between the laser and the generated

PFWM () = 2 2 Pl L2 sinc 2 ( L / 2) d1 S (1 ) S (1 ) (1)


0

where is the nonlinear coefficient, Pl is the laser


power, L is the fiber length, and S(1) and S(1) are
the noise power spectral densities. It can be shown
that when A << |A l|, the FWM contributions
from all the pairs of noise frequencies have identical
wave-vector mismatch: = 3(c 0)(c l)2 +
4(c l)4/12, where c = (1 + 1)/2. The phase
mathing condition = 0 can be expressed in the
linear form Y = A + BX, where Y = (pm l)2, X =
(pm + l), B = 483/ 4, and A = B0.
Consequently, the coefficients A and B can be
obtained by simple linear fitting to the measurements

978-1-4244-2228-9/08/$25.00 (c) 2008 IEEE

Vol. 1 - 77

Mo.4.B.3

ECOC 2008, 21-25 September 2008, Brussels, Belgium

-1

3/4 = (1630 330) ps as quoted in Table I. The


error in the measurement is only 20% even though
the laser was tuned over a rather small range
(between 1445 and 1475 nm). For the fiber with L =
18m the measurement was first performed with 4
followed by 12 laser positions. As explained above,
increasing the number of points and increasing the
separation between the laser and the FWM product
result in error reduction. For the fiber with L = 9m the
conventional way (reported in [4]) resulted in large
errors. In sharp difference, when 3/4 was measured
using the proposed approach, the obtained result was
significantly closer to those obtained with the longer
segments even when only 3 laser positions where
used. Finally, the 5m fiber was tested using the new
approach but non-negligible errors are evident, which
we attribute to the limited sensitivity of the OSA due
to the low FWM power. In this HNLF we have
3
measured that 3 = 0.038 ps /km, consequently in the
case of L = 9m we can calculate 4 = (2.5 0.5)10-5
ps4/km, while for L = 50m we have 4 = (2.35
-5
4
0.5)10 ps /km.

product) and Fig. 2(c) (35 THz separation) where the


4 was kept at 110-5 ps4/km. Note that when this
separation is increased the FWM power is less
sensitive to detuning from the phase matching
condition. Finally, the effect of changing 4 can be
observed by comparing Figs. 2a (4 = 10-5 ps4/km)
and 2b (810-5 ps4/km), while keeping (l pm)/2 =
25 THz. The increment by a factor of 8 in 4 does not
introduce a considerable change to the spectrum.
It can be concluded from the results in Fig. 2 that in
the case of the 4m fiber, it is likely that a noise
detuning c = 0.12 THz (0.1 nm) will result in a
negligible change of the FWM power. This implies an
error in measurement of pm that will be comparable
to c. As shown in [4], in order to have less than 10%
error in measurement of 3/4, the error in
measurement of pm should be less than 0.05 nm. It
is important to note that a further improvement of
accuracy of 3/ 4 measurements is possible by taking
advantage of the symmetrical behavior of relative to
c. The main idea consists of detuning the noise at
two different frequencies in order to obtain identical
FWM power. The respective FWM peak frequencies
are measured and the actual pm is simply the
average of these two frequencies. The advantage
comes from measuring the FWM power in a region
that has a more rapid change in power as c is
changed.
1.0

50 m

a)

10 m
4m

1.0
0.8

0.6

0.6

0.4

0.4

0.2

0.2

0.0

0.0

-0.8 -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8

50

-1630 330

1445-1475

18

-1627 83

1413-1487

12

18

-1574 284

1428-1504

-1528 330

1445-1475

-2242 410

1445-1475

l (nm)

# points

Table I: Results for several fiber lengths.

Figure 3 shows a typical example of the way we


measured pm to increase the accuracy.

-0.8 -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8

c (THz)

1.0

10 m
4m

c (THz)

0.8

50 m

b)

3/4 (ps-1)

-55.5

50 m

c)

10 m

Power (dBm)

0.8

L(m)

4m

0.6
0.4
0.2
0.0
-0.8 -0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8

c (THz)

-56.5
-57.0
-57.5

pm c

pm

pm + c

-58.0

Fig.2. as a function of detuning from phase


matching condition. The details are in the text.

1693

1694

1695

1696

Wavelength (nm)

The experimental setup was identical to the one


used in [4]. An amplified spontaneous emission (ASE)
noise source is filtered using a tunable band-pass
filter (BPF) and then amplified with a booster EDFA.
The ASE source together with a tunable external
cavity laser (ECL) was coupled into the fiber under
test (FUT) using a 3 dB coupler. The coupled powers
into the FUT were <2 mW at l and 0.15 0.6 W from
the ASE. Spectra were measured with an OSA with
0.01 nm resolution.
A highly nonlinear fiber (HNLF) was cut down to four
different lengths of L = 50, 18, 9, and 5 m and the
ratio 3/4 was measured in each fiber. Our results
are summarized in Table I. First, in order to have a
reference value we tested several times the fiber with
L = 50 m and obtained similar values. A typical one is

978-1-4244-2228-9/08/$25.00 (c) 2008 IEEE

-56.0

Figure 3. Example of measuring pm from


detuning the noise.
Conclusions
We demonstrated a fast and simple measurement
method the fourth order dispersion in short fibers (<
10m) with an error of 20 %.
References
1. M. Hirano et al. ECOC PD, Glasgow, 2005.
2. B. Auguie et al. IEEE Photon. Tech. Lett., 18
(2006) p. 1825.
3. G.K.L. Wong et al. Opt. Exp., 13 (2005) p. 8662.
4. J.M. Chavez Boggio et al. J. Opt. Soc. Am. B, 15
(2007) p. 2046.

Vol. 1 - 78

Das könnte Ihnen auch gefallen