Professional Documents
Culture Documents
Abstract-A fiber-optic magnetic sensor is designed based on the far. The optical-phase retardation of this sensor has been
Lorentzian force. Optical phase sensitivities of 0.67 and 0.11 analyzed [8]. Frequency-independent measurements of the
r a d / ( A / m ) for ac and dc magnetic fields are obtained with a linearity
ac magnetic field also have been achieved [9], and dc sen-
of 3 decades using 100 mm of Al-coated sensor fiber. A configuration
for multipoint ac field measurement with frequency-independent sen- sors of this type have been found usable in simple multi-
sitivity is proposed. Electric current is also measurabkwith a 2.6- plexed systems [7] by introducing the FDM [lo] tech-
rad/pA sensitivity. nique. Although the Lorentzian force-type magnetic
sensor appears to be a good alternative to the magneto-
I. INTRODUCTION strictive sensors, only preliminary experiments have been
reported to date.
F IBER-OPTIC magnetic sensors have been reported
utilizing the Faraday effect [ 11 and the magnetostric-
tive effect [2]-[5]. The latter have been found to be the
This paper describes the design methodologies for sen-
sor head, fiber-coating material, and thickness as well as
for operating conditions. Techniques for sensitivity en-
more sensitive. Magnetostrictive materials, such as Ni or hancement and multipoint measurement are presented. An
metallic glass are bonded to or coated around the single- application of the sensor to electric current measurements
mode optical fiber in one arm of a fiber-optic interfero- is also described.
meter. For ac detection, a dc bias magnetic field is applied
to the sensing element to optimize detection sensitivities. 11. PRINCIPLE
Bucholtz et al. [3] detected an ac level as low as 70 Shown in Fig. 1 is the proposed sensor head, in which
ff /& ( = 5.5 X lo-’ A/m/&) at 34.2 kHz per 34 a length of conductive-coated sensing fiber freely spans
m of sensing fiber. A technique for dc magnetic-field de- two supports; the fiber ends are clamped. An electric cur-
tection has also been reported [4] that overcomes environ- rent I ( A ) is injected into the fiber coating in the presence
mentally induced low-frequency fluctuations by applica- of orthogonal magnetic fields B ( T ). The Lorentzian force
tion of an ac dither magnetic field. A compact fiber-optic is then induced which is equally distributed and orthogo-
magnetostrictive sensor [5] with a 5-cm amorphous metal nal to the fiber. Magnitude of the induced Lorentzian force
wire transducer has also been proposed that yields f ( N / r n ) is written as
a sensitivity of 1.3 X lop3A / m / & at 1 Hz. Here, a
practical limitation of these techniques is closed loop con- f = BI. (1)
trol of the dc magnetic field to eliminate the hysteresis of fcauses a deflection in the fiber and the resultant optical-
magnetostriction. phase changes are measured with an interferometer.
As an alternative to magnetostrictive sensors, Lorentz- If a sensing fiber with an equally distributed ac load or
ian force-type fiber-optic magnetic sensors have been re- an ac Lorentzian force f sin u t , plus equally distributed
ported [6], [7], These sensors utilize a conductor-coated static load& both orthogonal to the fiber, behaves as a
sensing fiber in one arm of a Mach-Zehnder interferom- clamped-clamped beam [ 111 and, the damping factor, the
eter. The sensing fiber spans two supports and is exposed secondary effects of shear and rotary inertia are neglected,
to an ac/dc magnetic field with a dc/ac electric current, the differential equation governing the small deflection at
respectively, flowing in the coating. The alternating Lor- any point y ( x , t ) on the fiber is
entzian force then causes elastic vibrations in the fiber
which induce optical-phase changes proportional to the
magnitudes of the measured magnetic fields. Phase sen-
sitivity of 0.11 rad/(A/m) with a minimum detection
level for an ac magnetic field of 7.1 X lop3A/m/& Although the Lorentzian force is applied normal to the
with a 10 cm length of sensing fiber has been obtained so fiber, its deviation in the x direction is neglected. E is
Young’s modulus of the fiber, Z, is the moment of inertia
of area, A is fiber cross section, y is fiber weight per unit
Manuscript received December 4, 1989; revised March 26, 1990. volume, and g is the acceleration of gravity. If the static
The author is with NTT Transmission Systems Laboratories, 1-2356,
Take, Yokosuka-shi, Kanagawa 238-03, Japan. deflection due to& is denoted as y 1( x ) and the deflection
IEEE Log Number 9036662. due tofsin u t as y 2 ( x , t ) , then y ( x , t ) = y , + y 2 . There-
TI
__
.-......
6
Fig. 1. Clamped fiber span with an ac Lorentzian force applied normal to
the fiber.
polarization I
maintainingfiber
with Ai coating
ac
source
acancor
dc SourCB lock-in
amp.
fore, the time-dependent fiber-length change A Lacis given Fig. 2. Basic experimental setup.
by line integrating y ( x , t ) and subtracting the dc term as
11 I
1560 JOURNAL OF LIGHTWAVE TECHNOLOGY. VOL. 8. NO. 10. OCTOBER 1990
Q
U
103
-
- 0.8 ' di=l .O
t
Eex9
/!&
.l-
.-
c) 0.2
0.6
0.4.
0
. di=l.O
Ni, di=l.O
Au,
: measured values
- : calculated values
10'
5 10
N+ 112 (N: order of resonance)
100 100
ac magnetic field: 2
> 7 Aim ms.. 230 Hz fl
--
0
E 80 - dc magnetic field: -80
0.3Wm U
a
temperature $:
fluctuation -60
3
2- 5
-40 .-
Fig. 6. Cross section of the AI-coated sensor fiber. 4
A
4
-20
-e
A’ room temperature:
21.8 degree(C) f
L
lo2
>
--
n
”
E 10’ ature raise in the coating.
-;
a
loo 102
>
E E
E -” **
r,
.-
J
applied dc: 500 mA
I -
44
a
n
10’
**
10.. a loo
3 10.2 10’’
ac:137 mA rms.
.-
10’‘ I 1o 2 1 o3
discrepancies. The 100-mm sensor span was thus found
to be the optimum. applied current frequency(f), Hz
Fig. 9. Interferometer output due to the joule heating effect.
C. Clamp
If the fiber supports give the fiber the freedom of rota-
tion, the span-center deflection can be five times larger E. Sensitivity Enhancement
than with clamped supports [ 111 thus theoretically provid- As shown in the analysis, the ac Lorentzian forcefrep-
ing higher sensitivity. This, however, could not be real- resents an ac or dc magnetic field strength, which when
ized so far without damaging the fiber, so all experiments multiplied by fo determines the interferometer output. This
used clamped supports. fo can be increased by increasing the interaction between
the dc current and dc magnetic field, thus enhancing the
D. Applicable Electric Current sensitivity. The sensitivity dependence of ac field mea-
Shown in Fig. 8 are the sensitivity dependence of ac surements on the dc magnetic bias field was measured as
field measurements on dc and the corresponding temper- shown in Fig. 10. The interferometer output linearly in-
ature raise in the coating measured with a thermistor sen- creased with dc bias field in the region lower than -20
sor. The sensitivity linearly increased with dc current up kA/m. However, the output peaked at dc -28 kA/m.
to 500 mA. However, at dc currents higher than 500 mA, Here, with zero dc bias field, fo is almost equal to 5.7 X’
though no additional noise was observed with a spectrum N/mm ( = gravitational force). With a dc field of
analyzer in the frequency range higher than 100 Hz, - 28 kA/m and dc current of 500 mA, fo 1.76 x lop5 =
low-frequency noise emerged due to convection-induced N/mm ( = Lorentzian force) 5.7 X +
N/mm.
fluctuations in the fiber-coating temperature. Therefore, Therefore, sensitivity should ideally be enhanced by a
the maximum usable dc with this setup was determined to factor of 32 between the two cases. This, however, was
be - 500 mA. For dc magnetic field measurements, ac is not fully realized in the experiment. The discrepancy may
used. AC with frequency w does not yield an optical-phase be attributed to initial bending at clamps, torsion, and an-
shift at w due to the joule heating effect. The shift occurs isotropy of the coating. With dc bias fields higher than 20
at 2w as shown in Fig. 9. However, application of a dc kA/m, the Lorentzian force is more than 20 times bigger
bias current to the ac generates additional optical-phase than the gravitational force. This suggests that the high
shifts at w . Therefore, care must be taken when applying tensile force generated in the fiber by the large Lorentzian
a dc bias current so that the joule heating effect is kept force has suppressed the magnetically induced fiber vi-
negligibly small. bration amplitudes thus decreasing thc sensitivity.
1562 JOURNAL OF LIGHTWAVE TECHNOLOGY. VOL. 8. NO. IO. OCTOBER 1990
;pan 100mm
ac measurement
frequency 245 Hz
dc bias field
If the sensor fiber is oriented such that the gravity and the multimoded He-Ne laser is calculated as 3 x lo-' -
Lorentzian force are not collinear, sensor vibrations can- rad, which was obtained by using an interferometer path
not be precisely expressed by (2). For sensitivity en- imbalance of -40 mm and a spectral line width of -25
hancement, however, a strong dc Lorentzian force is ap- kHz [16]. Several combined modes would then yield a
plied and the portion of the gravitational force shrinks to -
phase noise of low4rad which roughly coincides with
one tenth or less of the entire applied dc load on the sensor the minimum detected phase level of the sensor. The lin-
fiber thus providing freedom of sensor orientation. earity trend in Fig. 11 and the above noise consideration
suggests that the minimum detectable magnetic field can
V. MAGNETIC FIELDMEASUREMENT be improved by eliminating the phase noise, namely by
By using the fabricated sensor head and optimum op- nulling the interferometer-path imbalance and/or by intro-
erating conditions, magnetic field measurements were ducing a single-frequency light source with a narrow
conducted and attained good linearity over three decades spectral line width. If a phase sensitivity of lop6rad [15]
as shown in Fig. 1 1 . A phase sensitivity of 0.67 is attained with 1-Hz bandwidth, the minimum detection
rad/(A/m) of ac magnetic field was obtained as deter- of 1.5 X A/m/& and 9 x A/m/& for
mined by the J 1(max) method [15]. The previously re- ac and dc magnetic fields is indicated by extrapolation.
ported ac measurement sensitivity [7] obtained without the
sensitivity-enhancing technique can be interpreted as a VI. APPLICATIONS
sensitivity of 0.17 rad/(A/m) with dc 500 mA. Thus, A. Frequency-Independent Multipoint Measurements
the sensitivity in the present measurement is higher by a Using One Current Source
factor of 3.8. The minimum detected ac field was 1.27 X To date, multipoint measurement with the Lorentzian-
lop3A/m/& at 245 Hz, which corresponds to a phase force magnetic sensor has been demonstrated only for dc
sensitivity of -0.85 m rad/&. In the case of dc mea- field measurements with a current source connected to
surement, the inteferometer yielded a nonzero value even each sensor head one by one [7]. This section proposes a
with zero measurand field due to the interaction between multipoint ac magnetic sensor using only one frequency-
the ac and geomagnetic field. This value was then sub- multiplexed current source and yet attains frequency-in-
tracted from each output thus obtaining the result propor- dependent sensitivities. Shown in Fig. 12 is the response
tional to the dc magnetic-field strength. However, sensi- of the sensor to a dc magnetic field using a frequency-
tivity enhancement was not introduced in the dc field multiplexed current fi and f2. fi was adjusted to the fun-
measurements because, even if a strong dc is superim- damental resonance of the sensor and& was scanned. If
posed on ac, the resultant dc Lorentzian force changes fi andf2 were not too close each other, the sensor could
with measurand dc field and accordingly changes the sen- consistently be vibrated at fi . As the frequency difference
sitivity. The dc field measurement attained a phase sen- decreased, however, the sensor output was found to fluc-
siti ' y of 0.11 rad/( A/m) and a minimum dectection of tuate at the beat frequency. The minimum frequency dif-
7.8 a lop3 A/m/& was obtained with ac 300 mA ference required for realizing stable measurement was
rms . about 8 Hz.
The interferometer output noise at 100 Hz - 1 kHz Based on this, the proposed multipoint sensor is out-
without any magnetic-field signal was found to be at the lined in Fig. 13. The measurand magnetic-field frequency
rad level with a spectrum analyzer. Electrical is fB, the resonant frequency for each tandem sensor head
noise measured with no optical input to the detectors (Si- is 5 ( i = 1 , 2, -), and the current source provides
p-i-n photo-diode ) and the light source intensity noise currents composed of frequencies f i i (i = 1 , 2, * * ). fii -
were both -20 dB lower than the -
lop4 rad level. is adjusted equal to 5 + fB or 15 - fB 1 [9] to yield a
Therefore, the laser phase noise is assumed to have lim- Lorentzian force with frequency to cause a resonance
ited the sensor sensitivity. Phase noise for each mode of for each sensor head. Another Lorentzian force is simul-
1 -
OKAMURA: FIBER-OPTIC MAGNETIC SENSOR UTILIZING THE LORENTZIAN FORCE 1563
10
>
--
E
s
a
L
3
8
c)
10
E fiuctuation
2 ,d
f.- ,
, I
I
dc magnetic field: 240 Alm
applied ac: 10 mA rms.
Fig. 14. Demonstration of the multipoint sensor with sensor resonant fre-
ac frequency: t i =PZOHz(fixed)
1, :sweep quencies at 364 and 500 Hz.(a) Multiplexed ac current frequencies (164
lo
Hz,300 Hz) and magnetic-field frequency (200Hz).(b) Interferometer
' loo 10' 102 3 output spectrum, which are dominant at 364 and 500 Hz but weak at 36
and 100 Hz.
ac frequency difference,lf ,- f4, Hz
Fig. 12. Sensor response to a frequency-multiplexed current.
dc magnetic field:
measurand magnetic field > 160kAfm
/tB ....
resonant
frequen?
fi
.I..,
+ C4sinh
4% fg
-x --
Ayw2
and b = JEl,g/Ay.The boundary conditions: y = 0 and
[SI K . P. Koo, F. Bucholtz, D. M. Dagenais, and A. Dandridge, “A
compact fiber-optic magnetometer employing an amorphous metal
wire transducer,” IEEE Photon. Technol. Lett., vol. 1 , no. 12, pp.
464-466, 1989.
[6] K . A. Arunkumar, “Ultrasensitive fiber-optic magnetic field sensor,”
d y / & = 0 at x = 0 and x = L determine the constants in Proc. Ann. Meeting OSA, 1985, THK4.
[7] H. Okamura, “Fiber-optic magnetic sensor utilizing metal-coated
as fibre,” Elecrron. Lerr., vol. 23, pp. 834-835, 1987.
cl = K ( ( B - B ’ ) + (AA’- B B ’ ) + 1}/2(1 - BB’)
[8] H . Okamura, “Optical phase retardation of the Lorentzian force-tyoe
fiber optic magnetic sensor,” Electron. Lett., vol. 10, pp. 622-623,
1989.
C, = K ( ( A + A ’ ) - ( A B ’ + B A ’ ) } / 2 (1 - B B ’ ) [9] H. Okamura and Y. Fujii, “Lorentzian force-type fiber-optic ac mag-
netic sensor realizing frequency-independent sensitivity,” Electron.
C3 = K - C1 Lett., vol. 24, pp. 1090-1091, 1988.
[lo] F. Bucholtz, A. D. Kersey, and A. Dandridge, “Multiplexed non-
linear interferometer fiber sensors,” in 4th OFS Tech. D i g . , 1986,
c, = -c, (A2) pp. 63-66.
where [ l l ] L. S . Jacobsen and R. S. Ayre, Engineering Vibrarions wirh Appli-
7 cation to Structures and Machinery. New York: McGraw-Hill, 1958.
[12] C. D. Butter, and G. B. Hocker, “Fiber optic strain gauge,” Appl.
Opt., vol. 17, pp. 2867-2869, 1987.
[I31 Y. Sasaki, K. Okamoto, and T . Hosaka, “Polarization-maintaining
and absorption-reducing fibers,” in 5th OFC., Phoenix, 1982, THcc6.
[14] D. A. Jackson, A. Priest, and A. Dandridge, “Elimination of drift in
a single-mode optical fiber interferometer using a piezoelectrically
stretched coiled fiber,” Appl. Opt., vol. 19, pp. 2926-2929, 1980.
[I51 D. A. Jackson, A. Dandridge, and S. K. Sheem, “Measurement of
small phase shifts using a single-mode optical-fiber interferometer,”
(A3) Opt. Lett., vol. 5, pp. 139-141, 1980.
A L is obtained as [16] A. E. Siegman, B. Daino, and K. R. Manes, “Preliminary measure-
ments of laser short-term frequency fluctuations,” IEEE J. Quantum
Electron., QE-3, p. 180, 1967.
[17] G. L. Tangonan, D. I. Persechini, R. J . Morrison, and J . A. Wy-
socki, “Current sensing with metal-coated multimode optical fibres,”
Electron. Letr., vol. 16, no. 25, pp. 958-959, 1980.
The time-dependent component A La, in A L is approxi-
mated as *
AL,, = 2 j L
0 [2 akax ay, + (z)2] ak. (A5) Haruo Okamura received the B.S. and M.S. de-
grees in mechanical engineenng from the Tokyo
Institute of Technology, Japan, in 1971 and 1973,
Therefore, by using y , ( x ) = (fo/24EIz) (L2x2 - 2Lr3 respectively.
+ x 4 ) and y 2 ( x , t ) thus obtaining
He joined the NTT Laboratories in 1973. He
was engaged in research and development on fi-
AL, = K i p
sin wt + K2 ( 2f’ sin
l 7 ) 2wt
ber-optic submarine cable transmission systems.
He is presently engaged in research on interfero-
metric optical-fiber sensor technologies and un-
A~CO~EI, Ayw dersea acoustic imaging systems.
Mr. Okamura is a member of the Japan Society
(A6) of Mechanical Engineers and the Japan institute of Electronics, Informa-
where K 1 and K2 are constants. tion and Communication Engineers.
11- 1 -