IEEE Instrumentation and Measurement

Technology Conference
Budapest, Hungary, May 21–23, 2001

Frequency Response of a Thin Cobalt Film Magnetooptic Sensor

Guillermo Robles and Romano Giannetti
Departamento de Electrónica y Automática
Universidad Pontificia Comillas de Madrid
Phone: +34-91 542 2800, email: [email protected]

Abstract – The magnetooptic effect is due to a change in the polarization 36.9 pF

of the light when it is reflected or passes through a magnetized material.
The rotation of the polarization plane is proportional to the magnetic field.
The great advantage of using a magnetooptic sensor to measure currents or
magnetic fields is its wide bandwidth. It can reach the magnitude of several
GHz. This fact is widely known, however no effective measurements have
been taken. In this paper we present the frequency response of a cobalt thin
film used as magnetooptic material. It was first excited by several sinusoidal 10.9 Ohm 106.65 mH
magnetic fields at different frequencies. The range of frequencies studied in Fig. 1. Equivalent circuit of the Helmholtz coil used in the first part of the
the first experiment reached 200 Hz, which is suitable to measuring power experiment.
lines currents or magnetic fields. Because the coil that creates the magnetic
field has a great impedance at higher frequencies, an alternative method
based on magnetic impulses has been designed to obtain high frequency 5.13 pF
data. With the latest experiments we have been able to measure frequencies
as high as 2 MHz, obtaining a flat response.

Keywords – Sensors, Ferromagnetic materials, Magnetic field measurement

0.358 Ohm 1.03 uH
I. INTRODUCTION Fig. 2. Equivalent circuit of the coil used in the magnetic pulse generator

When a polarized beam passes through a magnetized material

its polarization plane rotates [1][2][3]. This effect was first Initially, the setup included a Helmholtz coil to create the mag-
discovered by Michael Faraday in 1845. The law that regulates netic field. Calibrating measurements of the magnetic field
the rotation is (1). with a fluxometer indicate that the coil is able to give 85 Oe/A.
The equivalent circuit is displayed in Figure 1. It has been pos-
I sible to reach 179 Hz without problems or drawbacks. How-
 =V ~
d~l
H (1)
ever, for frequencies higher than that, the impedance would
grow too high, see table I, and the magnetic field would turn
to be too weak for our purpose. Nevertheless, interesting re-
~ is the magnetic field, V is the sults can be obtained in that frequency range and they will be
Where  is the rotation angle, H reported in the next section. It is necessary, then, to design a
Verdet constant (which depends on the magnetooptic charac- more complex excitation circuit or try to fabricate another coil
teristics of the material) and d~l is the length of the path traveled with a higher magnetic field per ampere ratio. None of these
by the light inside the material. Obviously, for the same thick- alternatives would fit our purposes if signals at very high fre-
ness and the same material, the rotation angle is proportional quencies are going to be measured.
to the magnetic field. The magnetooptic material used in the
sensor is a thin cobalt film [4][5]. The next step considered was to minimize the radius of the
coil so that the magnetic field would increase proportionally.
When designing a sensor, one of the questions that arises is The turned wire should be very thin and it would not support
what type of signals are going to be measured, and conse- high currents during long periods of time. Hence, the average
quently, what range of frequencies are necessary to be detected. value of the intensity should be very small and so the energy
The aim of this paper is to calculate the bandwidth of the sensor delivered in every cycle. The adopted solution was to create
based on a thin cobalt film. Then, the uses of the sensor could magnetic pulses and then, to study the time response of the
be stated. The thin film cobalt sample is 200 Å thick. We also system with the aid of a system identification tool.
intend to test in future experiments, if there is a relationship
between the thickness and the bandwidth of the material. The sample was introduced between a small pair of coils.
0-7803-6646-8/01/$10.00 ©2001 IEEE

417
 Frequency Helmholtz coil Small coil
180 Hz 22
0.358
Laser
500 Hz 54
0.358

1.000 Hz 107.2
0.358

10.000 Hz 1.067 k
0.364
Power Supply Polarizer

TABLE I
M AGNITUDE OF THE IMPEDANCE OF THE COIL FOR DIFFERENT
FREQUENCIES .
Signal Helmoltz
generator Ampli Coil

These were connected together in the same way as if they were TDA7294
a tiny Helmholtz coil. The length of each one was 6 mm; the Magnetic Field and
outer diameter, 4 mm and the inner diameter, 1,5 mm. This beam direction
one, being so small allowed us to obtain high magnetic fields. Cobalt sample
The equivalent circuit is shown in Figure 2. Polarizer

The measurements taken using these two coils and the interest-
ing results obtained with magnetic pulses will be presented in Sensor
the following sections.
Fig. 3. Setup scheme to create a magnetic field and to measure the
II. MEASUREMENTS magnetooptic Faraday Effect.

Several sets of measurements have been taken using two dif-

ferent magnetic field generators. The input is the intensity averaging method was used for this frequency point. Never-
through the coil or the magnetic field created by the coil. The theless, it can be observed that the response is linear from zero
output is the signal detected by a photodiode amplified with an to 179 Hz.
intensity-to-voltage amplifier. The spectra of these signals an
the relation between them is calculated obtaining the frequency Similar plots have been calculated for other thickness of cobalt
response of the system. If this curve is flat, we can assert that film. The response is also linear but the gain is different as it
the bandwidth of the cobalt is beyond the maximum frequency was expected.
measured.
B. Small coil
A. Helmholtz coil

For the first coil, a function generator has been used to create a It is possible to compute the frequency response of a system
sinusoidal wave of different amplitudes and different frequen- studying the response to a step in the input signal. Following
cies. This signal is amplified with a TDA7294 and the output this argument the use of a magnetic pulses generator has been
is connected to the Helmholtz coil, see Figure 3. considered. The first idea is to discharge a capacitor through a
coil of N spires.
The plot of the amplitudes of the input signal versus the output
of the sensor is shown in Figure 4. As it was explained before, The radius should be small enough to create a high magnetic
the more frequency we try to impose, the less intensity and field. However, the laser beam and the magnetic field must be
magnetic field we get due to the high autoinductance. This is parallel to each other and perpendicular to the cobalt sample.
why the plot has more values in the region of low magnetic Hence, the light must pass through the coil, then through the
field. At low frequencies we have a maximum magnetic field sample and finally, hit the photodiode at the sensor.
in 16 mT, and the minimum field measured is 400 T at any
frequency. The current through the spires would depend on its impedance.
It must be low so that the current can be high. The impedance
Notice that the slope of the line is the ratio between the out- of the coil has been measured and its equivalent circuit is
put and the input to the system considered. The frequency re- shown in Figure 2.
sponse can be plotted calculating the slope of the linear func-
tion for each frequency, Figure 5. The value corresponding These characteristics maximize the magnetic field as is shown
to 110 Hz has a slight difference compared to the rest of the in (2), where a is the radius of the spires, L is the length of the
frequencies. This is probably due to the fact that a different solenoid and N is the number of spires.

418
 300
1K
2 Hz
250 9 Hz + 1000 uF Coil
21 Hz
60 V
-
61 Hz
200
110 Hz
Sensor Voltage (mV)

179 Hz Trigger
150

100

50 Fig. 6. Circuit designed to create a magnetic pulse.

Input and Output Time Plot

0 1
0 2 4 6 8 10 12 14 16 18
Magnetic field (mT)

0.5
Fig. 4. Plot of the output signal versus the current in the coil (proportional to

Output
the magnetic field) for a 200 Å cobalt sample and different frequencies.
0

−0.5
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
5 −3
x 10

60
4 50
Gain (mV/mA in coil)

40

30
Input

3
20

10
2
0

−10
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
1 Time (seconds) −3
x 10

Fig. 7. Time response of the input signal and the output of the sensor. Notice
0
0 20 40 60 80 100 120 140 160 180 200
that the units at the input are volts. This voltage must be converted into
Frequency (Hz) current dividing by the impedance of the coil over a wide range of
frequencies.
Fig. 5. Frequency response for the 200 Åcobalt sample.

sured. In Figure 7, it is shown the signal measured at the sensor

N0 I and the time response of a voltage step created by the discharge
B~ = q

2 ~ux (2)
a2 + L2
of the capacitor. The measurements have been taken with an
osciloscope Tektronix TDS 544A. The sampling frequency was
5
107 samples per second. The maximum frequency visible
The circuit is quite simple. When the MOSFET is off, a 1000 F would be 25 MHz, theoretically. The total time sampled was 1
is charged up to 60 V through a resistor of 1 K
. The time con- ms, so the minimum frequency we can appreciate is 1 KHz.
stant should be less than the trigger cycle. In the experiment
the gate is in low level during 5 seconds which is enough to In order to obtain the time response of the intensity in the
finish the charge of the capacitor. When the MOSFET is on, the spires, it is necessary to perform the following calculations:
current is abruptly conducted through the coil and the R DSON
of the MOSFET. This resistence is the same order of magni- The frequency response of the voltage step is obtained calcu-
lating the Fast Fourier Transform (FFT ) of the time response of
the input, let it be called V~ (f ). Then it is computed Z (f ), the
tude as the impedance of the coil. In order to reduce it, up to
4 MOSFET were connected in parallel, with the same trigger
frequency response of the equivalent circuit of the spire in the
same points as V~ (f ). Dividing these two functions we get the
input. The driver was an IR2110 and the MOSFETs were some
IRF530.
frequency response of the intensity discharged through the coil.
The current in the spire must be calculated, it cannot be mea- Then it is calculated the inverse FFT to get the time response of

419
 Intensity Time Response −8 Spectrum estimate
180 x 10
5

160
4

140
3

Output
120
2

100
1
Intensity (A)

80
0
5 6 7 8
10 10 10 10
60

−4
40
x 10
1

20 0.8

0
0.6

Input
−20
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0.4
Time (seconds) −3
x 10
0.2
Fig. 8. Time response of the intensity in the spires. This intensity is
proportional to the magnetic field. 0
5 6 7 8
10 10 10 10
Frequency (Hz)

Fig. 9. Spectrum estimate of the signals. Above 2 MHz the energy of the
the current, see Figure 8. signal is practically negligible.

If we look back to (2), and once we have obtained the current Amplitude of the Frequency Response
−25
in the spire, we could deduce the value of the magnetic field
−30
theoretically. However, this equation is valid for the value of
−35
the magnetic field inside an ideal solenoid and infinite long. In
Gain (dB)

−40
the experiment, the solenoid was far from being ideal. It was
divided into two coils, very likely to the shape of a Helmholt −45

coil, and the sample was placed between them. The magnetic −50

field should not follow an easy equation as (2), but a much −55
10
0
10
1
10
2
10
3
10
4 5
10 10
6

more complicated one or none at all. It is preferable to fol-

Phase of the Frequency Response
low an indirect way of measuring the magnetic field. As it has 0

been shown before, in Figure 4, the relationship between the −20

magnetic field and the response of the magnetooptic sensor is
Degrees

−40
linear, at least up to 200 Hz. We can use these data to calibrate
the response and convert the signal of the sensor into a mag- −60

netic field. Moreover, we can compare it with the theoretical −80

one. −100
0 1 2 3 4 5 6
10 10 10 10 10 10 10
Frequency (Hz)
C. Frequency Response up to 2 Mhz
Fig. 10. Frequency response of the system. Output voltage at the sensor
versus input current in the coil.
The input and output to the measuring system in study were
analysed with the assistance of MATLAB and the System Iden-
tification toolbox. The energy of the signal is shown in Figure the frequency response of the coil, the magnetic field or the
9. It can be seen that the signal is negligible for frequencies current through the coil disminish at high frequencies.
larger than 2 MHz, so in spite of been able to observe signals
containing frequencies of 25 MHz, practically those signals are However, the frequency response representing output voltage
hidden by the noise in the output. For the time being, we will versus input voltage —voltage drop in the coil— is absolutely
only consider frequencies up to 2 MHz; nevertheless, we are flat, see Figure 11. These results are unquestionably encourag-
studying different approaches to this drawback and we hope to ing and we are very confident in reaching responses at higher
overcome it introducing a white noise signal as excitation. frequencies in a short period of time.

The spectra analysis was performed with MATLAB. The re- III. NEXT STEPS
sults are shown in Figure 10. The data in this figure are the
frequency response of the whole system, output voltage versus The experiment described above was done with a single trig-
input intensity. The cut-off frequency is 40 KHz. Then, the ger at the input gate of the MOSFET. Therefore, the information
gain decreases with a slope of 20 dB/dec. This fact is due to in terms of energy contained in the data is somewhat limited.

420
 Frequency response
2
10 The complete experiment should be repeated for a set of sam-
ples with different thickness. It is exceedingly interesting to
0
10
test the bandwidth dependece with the thickness of cobalt. It
Amplitude

will be applied to maximize the bandwidth according to the

−2
10
thickness in the case there is an appropiate relationship.

IV. CONCLUSIONS
−4
10
5 6 7 8
10 10 10 10
In this paper we present a characterization study of thin cobalt
200
film magnetooptic sensors. An exhaustive analysis for low fre-
quency was performed presenting results up to 200 Hz. More-
100 over, an alternative method to increase the range of frequencies
Phase (deg)

0
studied has been developed. The current limit of the experi-
ment has reached a frequency as high as 2 MHz. It is being
−100 studied another technique to widen the maximum frequency
studied at present. The final response of the system has turned
−200
10
5
10
6 7
10 10
8
to be linear and extemely flat as it was desired.
Frequency (Hz)

Fig. 11. Frequency response of the system. The input and output are the References
voltage drop at the coil and the voltage mesaured at the sensor respectively.
The data above 2 MHz is not significative and should be ignored. [1] G.W. Day and A.H. Rose. Faraday effect sensors: The state of the art.
In Fiber Optic and Laser Sensors VI, volume 985, pages 138–150. SPIE,
1988.
[2] G.W.Day, M.N.Deeter, A.H.Rose, and K.B.Rochford. Faraday effect sen-
The next step in developing new measurements is to introduce sors for magnetic field and electric current. In Interferometric Fiber Sens-
a different signal through the MOSFET gate. This new signal ing, volume 2341. SPIE, 1994.
must contain more information about high frequency harmon- [3] Marvin J.Freiser. A survey of magnetooptic effects. IEEE Transactions
on magnetics, 4(2):152–161, June 1968.
ics. A white noise will accomplish the requirements. It can be [4] R.Carey, B.W.J.Thomas, and D.M.Newman. The magneto-optical prop-
simulated with a PRBS (Pulsed Random Binary Signal) at the erties of thin cobalt films. Thin Solid Films, (67):35–40, 1980.
gate. However, in order to observe high frequencies the charge [5] G.Robles and R.Giannetti. Magneto-optic faraday effect current sensor
based on thin cobalt films. In IEEE IMTC’2000 Proceedings, pages 1548–
of the capacitor should be quick enough to be discharged at a 1551, 2000.
high rate. If a single capacitor is not sufficient, an array can be [6] Jerald Graeme. Photodiode Amplifiers. Mc Graw-Hill, 1995.
[7] Henry W. Ott. Noise Reduction Techniques in Electronic Systems. Wiley-
disposed in an array and sequentially discharged through the Interscience, 1988.
same coil.

The high frequency noise, which is present in Figure 7, should

be reduced to improve the signal-to-noise ratio. This noise is
thought to be thermal noise due to the resistors and noise from
the photodiode. We will explore different techniques to dis-
minish its effect in the final signal [6][7].

421