320
OPTICS LETTERS / Vol. 18, No. 4 / February 15, 1993
Mode-locked fiber laser gyroscope
M. Y. Jeon, H. J. Jeong, and B. Y. Kim
Department of Physics, Korea Advanced Institute of Science and Technology, 373-1, Kusong-Dong,Yusung-Ku, Taejon, Korea
Received September 30, 1992
We describe a new form of an all-fiber-optic gyroscope that utilizes a fiber laser. The Sagnac interferometer
is used as a loop reflector with a modulated reflectivity to produce actively mode-locked optical pulses. The
rotation-induced phase shift is obtained from the timing shift of the pulses. Experimental results compare
well with the theoretical predictions.
We propose and demonstrate a novel fiber-optic gy-
loop reflector. 9
roscope in the form of a mode-locked laser consisting
of a laser cavity formed by a planar mirror at one
reflector is a function of the rotation rate or any nonreciprocal phase shift introduced between the counterpropagating waves in the Sagnac interferometer.
A fiber-optic phase modulator located near one end
of the fiber coil, as shown in Fig. 1(a), can be used to
modulate the optical loss in the cavity by modulating
the phase difference between the counterpropagating
waves. When the frequency of the loss modulation is
the same as the frequency spacing of the longitudinal
modes of the laser [i.e., Af = c/n(L. + 2Le), where
n is the refractive index], mode locking takes place,
and the output of the laser become a series of short
pulses. The timing of the pulses is determined such
that the oscillating pulses in the cavity pass through
the loss modulator at the time of minimum loss. On
the other hand, the depth of optical loss modulation for the system in Fig. 1(a) becomes maximum
when the modulation frequency is fm = c/2Lcn. At
this frequency, the modulation provided by the loop
end, a gain medium, and a Sagnac interferometer
at the other end. The Sagnac interferometer serves
as a reflector as well as a rotation-sensing element.
The output of the gyroscope is a series of short
optical pulses. The separation of two sets of optical
pulses in the time domain changes as a function
of rotation rate, providing a much simplified signal
processing compared with that in conventional fiberoptic gyroscopes.
The Sagnac effect introduces differential phase
shift between the countercirculating optical waves.
In a ring-laser gyroscope,' the differential phase shift
is translated into beat frequency of the laser lines
oscillating in the opposite directions, which could be
measured with high accuracy. For interferometric
fiber-optic gyroscopes, however, the basic intensity
output from the Sagnac interferometer has a
sinusoidal dependence on the differential phase shift.
Relatively complicated electronic/optical signal processing is required in order to recover the differential
phase information that is linearly proportional to
the rotation rate. A number of closed-loop2 - 4 and
open-loop5 - 8 signal-processing techniques have been
developed for this purpose. It is, however, still felt
that a simpler signal processing than the existing
ones is desirable for better performance and/or lower
The reflection coefficient of the loop
Le
Output1
Amplifier
Output2
(a)
cost.
The novel gyroscope configuration described in
this Letter combines the characteristics of the
ring-laser gyroscope and the interferometric fiberoptic gyroscope. The entire optical circuit operates
as a laser without gain competition between the
counterpropagating waves in the sensing loop. The
rotation-induced Sagnac phase shift is translated into
spacing of the mode-locked pulses instead of the beat
frequency.
A schematic of the simplified mode-locked fiber
laser gyroscope (MLFLG) is shown in Fig. 1(a). It
consists of a laser cavity formed by a planar mirror
at one end and a Sagnac interferometer at the other
end, with an optical amplifier in between. When
the optical gain provided by the amplifier is greater
than the round-trip loss, the system operates as a
cw laser since the Sagnac interferometer acts as a
0146-9592/93/040320-03$5.00/0
Pulses
R
f
tf
Ar
f
Vill
1f
ill-I
W-W
t
i
I
AO/
. I'
I
-4
: OR
(b)
Fig. 1. (a) Schematic of the MLFLG. Le, fiber length
outside the sensing loop; L4, fiber length of the sensing
loop; PM, phase modulator. (b) Response of the gyro
output to the phase-difference modulation.
© 1993 Optical Society of America
February 15, 1993 / Vol. 18, No. 4 / OPTICS LETTERS
Length-matchingfiber
(provided by Polaroid) was used as the optical amplifier.
I Oscilloscope
I
I Generator
Fig. 2. Experimental setup of the MLFLG. PC's, polarization controllers; DC, directional coupler; PM, phase
modulator; Pol, polarizer. The black dots indicate splice
points.
reflector is pure amplitude modulation without any
phase modulation, which simplifies the mode-locking
process. In order to operate the gyroscope at f m =
Af, the optical length of the fiber cavity outside of the
sensing loop should be half that of the sensing loop
(Le = L4/2).
Figure 1(b) shows the reflectivity of the Sagnac interferometer (R) with respect to the phase difference
(AO
k),demonstrating the timing of the optical pulses
in the presence of a sinusoidal phase-difference modulation. Optical pulses will be produced at the time
when the net phase difference between the counterpropagating waves in the sensing loop is zero
and the reflectivity from the Sagnac interferometer
is unity. Therefore, for every cycle of the phasedifference modulation, two optical pulses are produced if the rotation-induced nonreciprocal phase
shift (AOR) is less than the amplitude of the phasedifference modulation (0im). Without any rotation
input, the optical pulses are equally spaced. With
rotation input, however, the two sets of optical pulses
will be shifted in time by the same amount but
in opposite directions. The amount of timing shift
represents the rotation rate with the relationship
At = T sin-'( AR)
321
A dichroic mirror (99% reflection at 1.06 /.tm
and >80% transmission at 0.8 ,um)was glued to one
polished end of the Nd-doped fiber. The other end
of the Nd-doped fiber was spliced to one port of the
directional coupler that forms the gyroscope sensing
coil. A high-power laser-diode array (500 mW at
808.4 nm) was used as the pumping source. The
fiber sensing coil consisted of a 600-m-long singlemode fiber with the cutoff wavelength of 790 nm
wound around a spool with a radius of 8 cm. An
extra length of fiber of 182 m was spliced between
the Nd-doped fiber and the directional coupler to
increase the amplitude modulation. Although it was
not long enough to satisfy the condition described
above (L4 = '/2L4), it was sufficient for demonstrating the operating principle of the gyroscope. With
150 mW of absorbed pump power, 3.6 mW of cw laser
output power at 1.06 1tum was obtained. A phase
modulator was placed at one end of the fiber coil,
and polarization controllers were installed inside and
outside of the sensing coil. The coupling ratio of
the fused directional coupler was 85% at 1.06 p-m.
A polarizer may be necessary outside the loop for a
stable reciprocal operation of the MLFLG but was not
used for our experiment. The polarization controller
inside the sensing coil was used to provide reciprocity
for the interfering optical waves. A sinusoidal modulation signal was applied to the phase modulator at
fm = 195.4 kHz, which corresponded to the longitudinal mode spacing of the laser. The phase modulation
(a)
(1)
where T = 1/fm . If the sense of the rotation is reversed, so is the direction of the timing shift that
can be measured with reference to the electrical
signal applied to the phase modulator. By measuring the time separation of the optical pulses, the
rotation rate can easily be obtained. More than two
optical pulses per phase modulation cycle can be
produced depending on the amplitude of the phasedifference modulation and the rotation-induced phase
shift. Note that enough phase modulation amplitude should be provided to cover the desired dynamic
range of the gyroscope. Another possibility is to use
a triangular phase-difference modulation waveform
that produces a linear scale factor instead of Eq. (1).
If the splitting ratio of the directional coupler is exactly 50%, no light output is expected from output 2.
A light signal from output 1 with less than a 100%
reflecting mirror could be used as the output of the
gyroscope.
The experimental setup of the MLFLG is shown
in Fig. 2. A 29-m-long double-clad Nd-doped fiber
(b)
2 ps/div
Fig. 3. Signal output from the MLFLG: applied electric signal to the phase modulator (upper traces) and
the mode-locked optical pulses from the MLFLG (lower
traces) with (a) no rotation input, and (b) a rotation rate
of 15 deg/s (2 As/division).
OPTICS LETTERS / Vol. 18, No. 4 / February
322
Sagnac Phase Shift A9,(rad)
0.5
1.0
1.5
2.0
15, 1993
2.5
U,
-2
a)
Rotation Rate (deg/s)
pure phase modulation is used for the mode locking
by placing the phase modulator outside of the sensing
coil, no timing shift for the pulses was observed. The
polarization controller outside of the sensing coil did
not influence a portion of the pulses. Although a
fiber amplifier was used for our experiment, there
are also possibilities with the use of semiconductor
amplifiers.
In conclusion, we have proposed and demonstrated
a novel mode-locked fiber laser gyroscope whose output is the timing shift of mode-locked pulses.
This research was supported by the Agency for
Defense Development under contract number 90-1-1.
References
Fig. 4. Response of the timing shift of the pulses as a
function of the rotation rate.
1. F. Aronowitz, in Laser Applications, M. Ross, ed.
amplitude was 3.12 rad. Figure 3(a) gives the gyroscope signal without rotation input showing the
equally spaced pulses with 50-ns pulse width.
Figure 3(b) shows the output pulse train when the
gyroscope was rotating at the rate of 15 deg/s. It
can be clearly seen that the timing of the two sets
of pulse trains is shifted in opposite directions as
predicted. Figure 4 shows the shift of the timing of
the pulse At as a function of rotation rate for three
different values of phase modulation amplitudes
(0kin= 2.8, 4.5, and 5.4 rad). The dotted curves correspond to theoretical curves obtained with Eq. (1)
and agree well with experimental results. When a
3. B. Y. Kim and H. J. Shaw, Opt. Lett. 9, 375 (1984).
4. H. C. Lefevre, in Optical Fiber Sensors II, B. Culshaw
and J. Dakin, eds. (Artech, Dedham, Mass., 1989),
(Academic, New York, 1971), Vol. I, p. 133.
2. J. L. Davis and S. Ezekiel, Opt. Lett. 6, 505 (1981).
Chap. 11.
5. K. Bohm, P. Martin, E. Weidel, and K. Petermann,
Electron. Lett. 19, 997 (1983).
6. A. D. Kersey and R. P. Moeller, Electron. Lett. 26, 1251
(1990).
7. B. Y. Kim and H. J. Shaw, Opt. Lett. 9, 378 (1984).
8. K. Toyama, K. A. Fesler, B. Y. Kim, and H. J. Shaw,
Opt. Lett. 16, 1207 (1991).
9. D. B. Mortimore, IEEE J. Lightwave Technol. 6, 1217
(1988).