1. Yuji Otake, Akito Araya and Kazuo Hidano
　Development of a Highly-sensitive/ Broad-band Servo-type Seismometers 　using Magnetic Levitation
Yuji Otake, Akito Araya and Kazuo Hidano
ERI, Earthquake Research Institute
University of Tokyo
My name is Yuji Otake belonging to ERI, Earthquake Research Institute, University of Tokyo. Today, I would like to present my study on a highly sensitive broad-band seismometer using an astatic rotational pendulum that uses permanent magnet magnetic levitation.

2. Abstract
The aim of this study is to realize a highly-sensitive/wideband seismometer for new seismological-observational network. We have developed an astatic rotational pendulum using magnetic levitation for removing the parasitic resonances of a spring to support a pendulum weight form a gravity force and the thermal dependence of the spring constant. The principle is as follows. A cylindrical plunger permanent magnet has a weight at one side of its end edge. The plunger magnet is inserted into a uniform magnetic field generated by a window-frame-type permanent magnet and attached to two crossed leaf spring hinges outside the magnet bore. Magnetic forces occurred by the magnetic field counterbalance a gravity force at the weight. A prototype seismometer using the rotational pendulum has a velocity feedback control system of a weight position, a magnetic spring, and, a capacitance position detector. To realize the stable operation of the rotational pendulum without the unnecessary movements of the plunger magnet, the magnetic field uniformity of the window-frame magnet reached 10-4 by devising its pole shape. The thermal dependence of the magnetic field strength was compensated as much as 9x10-5/°C by Ni-Fe metal, having a negative coefficient of permeability. The natural period of the prototype was 10 s at the maximum and was able to be changed from 5 s to 8 s with the magnetic spring. These values of the period are consistent with theoretical calculations. Observations using this seismometer have been recently carried out. The results of the observations showed that the noise level of the seismometer was less than 10-9 m/s in the frequency range over 1 Hz.

3. Purpose of Development
To observe cheaply easily the vertical/Horizontal components of normal free oscillations of the Earth at many points.
Normal Free Oscillations
Seismometer
Atmospheric Pressure
Use for Earth Tomography
Less than m/s
Less than 0.01 Hz
Earth
10-6 m/s
Less than 0.1 Hz
The purpose of developing the seismometer is to observe cheaply easily teleseismic events for obtaining an image of the earth’s interior by tomgraphyic method　using the earthquake waves. For the telseismic events (earthquake), it cold be more than 10-7 and, of course, it’s depend on a magnitude and these frequency of the wave forms reached, from the point where the earthquake generated, to the observation point are less than 1 Hz. Some time it is more than 1 hours. For detecting the vertical components of normal free oscillations of the Earth at many points in the world is also purpose. This oscillation is generated by a change of atmospheric pressure or ocean waves. The signal level of the free oscillations are less than m/s and these frequency are less than 0.01 Hz.
Earthquake
To observe Telseismic Events

4. Traditional Seismometer
How to counterbalance gravity at a weight and a restoring force of a spring.
Natural period
Frequency
Parasitic resonance
Pendulum Response
Atmospheric Pressure
Ground Motion
Immovable Point by Inertia
Weight
This slide shows a principle of an ordinary seismometer that uses a pendulum, a mass and a spring system. To measure a ground motion, we need a reference point that is a immovable point of the pendulum. To makes the immovable point, we use a moment of inertia of the pendulum over a frequency range of its natural period. If we want to a low frequency response for detecting elastic waves generated by the teleseismic events and the normal free oscillation, the seismometer should have a long natural period such as more than 10s. The seismometer also uses a feedback control system for a pendulum mass position to fix the mass positon to the ground(seismometer frame), because, if a large vibration comes at location of the seismometer, then the pendulum makes sacale out without the feedback control. In the case of feedback control, the feedback force is exactly correspond to a ground vibration. Traditionally, seismometers use a suspension spring to counterbalance gravity at pendulum weight and restoring forces of the suspension spring and a hinge spring. The potential function of the spring is almost the same as that of gravity for obtaining an astatic state of the pendulum, a long natural period. The astatic state means the weak spring constant of the pendulum and a weak restoring force of the spring. The main causes of low frequency noise over the natural period of the pendulum in detection noise of a seismometer are a change of buoyancy at the pendulum weight, generated by a change of atmospheric pressure and thermal expansion of a seismometer structure such as a pendulum rod. The main cause of high frequency noise could be parasitic resonances of the suspension spring. Of course, in this case, we omit electrical noise of a displacement sensor and some circuit. The noise also causes high frequency detection noise.
buoyancy
Thermal Expansion

5. A principle of an astatic rotational pendulum using a plunger permanent magnet in a uniform magnetic field environment.
The positive rotational moment of a plunger permanent magnet generated by a magnetic field B cancels a negative rotational moment acted at a weight by gravity.
To reduce the noise of the seismometer, to remove the parasitic resonances and the thermal dependence (expansion), and to obtain long natural period in a seismometer pendulum, we devised a new principle that employs magnetic levitation to support the pendulum weight. We call it a vertical astatic rotational pendulum. The pendulum has a cylindrical type permanent magnet that a weight is attached at one side of its end edge. The plunger magnet is inserted into a uniform magnetic field. Magnetic forces applied to the plunger magnet moving as a compass counterbalance to the gravitational force at the weight. For realizing a vertical astatic rotational pendulum with a long natural period and a wide dynamic range, it is essential to have the opposite amplitudes and the same shapes of functions that are a gravity moment and a magnetic moment at the pendulum weight

6. The vertical component seismometer using an astatic rotational pendulum
Capacitance Position Sensor
Window-frame Magnet
Magnetic Spring
Pendulum Weight
Plunger Magnet
Feedback Coil
　To realize the new principle that I already mentioned in the previous slide, we made this seismometer on the slide. This slide shows a structure of a vertical component seismometer using the astatic rotational pendulum. The left side photo is a front view, and the right side photo shows the back. The rotational pendulum seismometer comprises a weight, a plunger permanent magnet, a window-frame-type permanent magnet, a magnetic spring, a feedback coil, and a capacitance position sensor. The feedback coil and the magnetic spring are similar to a voice coil for a speaker which a plunger magnet is into a solenoid coil. The window frame magnet has a rectangular parallel pipe bore that is enclosed with a iron pole pieces and rectangular parallel pipe permanent magnets. The plunger magnet, which is inserted into the rectangular parallelepiped bore of the window-frame magnet, is attached to the positioning mechanism of the weight. The mechanism can adjust the rotational axis position of the plunger magnet to the axis of the hinges. The movie shows a free oscillation of the pendulum. The natural period of the oscillation is about 5s.
Free oscillation
Natural Period : about 5s

7. The magnetic field distribution of the window-frame-type permanent magnet
The uniformity of a magnetic field around the center plane, where the plunger magnet is placed, is 10-4.
Plunger Magnet
This slide shows measured magnetic field distribution at the center plane of the window frame magnet bore, and also shows its FEM calculation. For obtaining a stable operation of the rotational pendulum seismometer and for eliminating its unnecessary movement. The field tilt ( not good uniformity) conducts distorted functions of the gravity and magnetic moment at the plunger magnet that mentioned at the pervious slide. The distored functions derive an unnecessary movement of the rotational pendulum such as an upward force on the plunger magnet pendulum inclined to the horizontal center plane of the bore above. We designed the window frame type magnet with which the rectangular parallel pipe bore was enclosed with a iron pole pieces and rectangular parallel pipe permanent magnets. At first, we tried the flat shape of the pole pieces. However, in this case uniformity of a magnetic field in the bore was not so good. The field uniformity was about 10^-3 that was not sufficient for our purpose. There is a tilted magnetic field near the bore opening. Therefore the distribution is compensated for obtaining a more flatter magnetic field by the tilted magnetic pole surface near to the opening, than a magnetic field generated by a flat pole. The tilted pole is shown in the FEM calculation’s figure. This calculation graph shows one fourth of a quadrant of the window frame magnet. The field flatness on the measured result is more than 10^-4 adjacent to the end edge of the plunger magnet. The magnetic flux density is 0.23 T around the center. The Gauss meter was used for the measurement.
Measured data
Central Plane
FEM calculation

8. Compensation of temperature dependence of magnetic field strength by NiFe metal (for Window-frame Magnet)
Very old technique: A watt-hour meter uses it.
The vertical scale is 5 mm/div and the horizontal scale is 50s/div.
The movement of the pendulum weight position dependent on temperature is about 10 mm/degree.
The thermal dependence of magnetic field strength is less than 9 x 10-5/K.
This slide shows the compensation method used to eliminate the temperature dependence of the magnetic field in the window frame magnet bore by NiFe metal. The method is employed in an ordinary watt hour meter. The arrangement of the window frame magnet and the compensation metal is described. The equivalent magnetic circuit of the window frame magnet with NiFe compensation metal is also shown. The optimum area of the metal was determined by the equation on the slide which calculates the ratio between the whole area (Amet) of the metal and the whole area (Aperm) of the permanent magnets. Bmet is the magnetic flux density in the metal and Bperm is the density in the permanent magnets. After installation of the metal to the window-frame magnet, an experiment was carried out. The result of compensation of the magnetic field strength by the compensation metal NiFe is described the left side below. The figuer shows the equilibrium point of the pendulum weight, which moves proportional to the temperature. The point in the figure moved about 3mm during a measurement with an environmental temperature change of 0.2 K. The strength change of the magnetic field was compensated to be less than 9x10-5/K. This change corresponds to an acceleration change of 2.3 x 10-5 m/s2/ K. (The vertical scale is 5 mm/div and the horizontal scale is 50s/div. This graph shows that the movement of the pendulum weight position, which is dependent on the temperature, is about 15 mm/K.)

9. The change of the natural period of the rotational pendulum, when changing the current of the magnetic spring.
　The solid circle are experimental values and the pink solid squares are calculated values. This natural periods has been measured by the capacitance sensor 10 times. The error bars are standard deviations for 10 measurements. The black line is a curve fitted by a 2nd order polynomial.
The change of the natural period of the rotational pendulum, when changing the current of the magnetic spring.
The solid circle are experimental values and the pink solid squares are calculated values. This natural periods has been measured by the capacitance sensor 10 times. The error bars are standard deviations for 10 measurements. The black line is a curve fitted by a 2nd order polynomial.
　 I set the differential equation up to subtract the negative force of the magnetic spring from the positive restoring force of the leaf spring, and obtained the natural period of the system by using the equation on the slide. Where E (98 GPa) is Young’s Modulus, Is (the leaf spring thickness is 0.12 x 10-3 m and the length is m.) the geometrical moment of the inertia of the leaf spring, l (0.2m) the length of the pendulum arm, n (10 turns) the winding number of the magnetic spring coil, i the coil current, mp the pole strength (1.9 x 10-4 wb) of the plunger magnet, r (0.015m) the coil radius and I (the weight size is X, Y, Z = 0.023, 0.002, 0.05 m and the material is aluminum.) moment of inertia of the weight. From this result solved with the equation, I confirm the rotational pendulum works in accordance with the principle. The combination between the rotational pendulum and the magnetic spring works well without any problems. The magnetic spring makes a natural period be longer and can changes the detection characteristic of the rotational pendulum.

10. The PID feedback system for the rotational pendulum
P component
D component
I component
This slide describes a block diagram of a servo-type seismometer using the astatic rotational pendulum and the PID (Proportional Integral Differential) feedback control circuit to make a flat frequency response (from about 13 Hz to about Hz) of velocity detection for vibration. This circuit comprises a pre-amplifier connected to a capacitance position detector, a high-pass filter for phase compensation at the point of the open loop gain 1 (that determines an upper corner frequency of about 13 Hz) in a Bode diagram, an integrator (low-pass filter) for the integral feedback control to obtain low-frequency stability of the seismometer, and a high-pass filter (that determines a lower corner frequency of about Hz) to control by a differential feedback method that allows velocity detection.
Total gain 5800

11. Observational set up for measuring the earth’s background vibration noise
The noise was measured with the astatic rotational pendulum seismometer at Matushiro Observatory of Japan Meteorological Agency in March, 2003 and 2004.
This slide shows the observational set up for measuring the earth’s back ground vibration noise measured with the rotational pendulum seismometer, terrestrial magnetism measured with the flux gate magnetometer, and atmospheric pressure observed with the barometer. The system was settled in the vault of Matushiro Observatory belonging to JMA (Japan Meteorological Agency). To reduce the effect of the terrestrial magnetism (usually several hundreds nT a hour), the seismometer was installed into a magnetic shield enclosure. The enclosure has 3 layers with 2 mm permalloy plates. The magnetic field strength in the enclosure was measured and reduced by as much as one several hundredth of an external magnetic field. Output signals from the seismometer, the barometer, and the flux gate magnetometer were recorded with a digital oscilloscope having an ADC with 16 bit, 1 kHz sampling.

12. The　wave form and the frequency spectrum of the earth’s background vibration noise　at Matushiro, March 03”
Rotational Pendulum
(m/s)/√Hz)
Marine Microseism
10-6 m/s
10**-2
10**-4
10**-6
10**-8
buoyancy
10**-10
This slide shows observation results measured with the rotational pendulum seismometer at Matushiro. The right above side drawing shows wave forms of marine microseism, with which the amplitude is about 10^-6 m/s and the frequency is about 10 Hz. The below right side graph describes a velocity frequency spectrum of the Earth’s background noise. The amplitude of the Earth’s background noise is about 10^-10 m/(s√Hz) in the frequency region over 1 Hz. This value is close to the specifications of the most sensitive seismometer, such as STS-I. However, the rotational pendulum seismometer dose not have sufficient sensitivity to measure the Earth’s background vibration noise under 0.1 Hz. One of the origins of this problem is a change in the buoyancy at the weight caused by a variation of the atmospheric pressure. The value of the buoyancy change by a rough estimation is 5.5 x 10^-8m/s2/kPa. (calculated under the following conditions: a weight of 0.8 kg, a weight volume of 1.56 x 10^-4 m3, and a temperature of 20 °C. This value is slightly under the measurement values in the frequency band from 0.1 Hz to 0.01 Hz in the figure. )
10-10 (m/s)/√Hz
10**-12
10**-14
10**-2
10**-3
10**-1
10**-0
10**1
10**2
Frequency [Hz]

13. The　wave form and the frequency spectrum of the earth’s background vibration noise at Matushiro, March 04”
Marine Microseism
Rotational Pendulum
(m/s)/√Hz
10-6 m/s
10**-3
10**-4
10**-5
10**-6
10**-7
buoyancy
March 12, 04” AM 2h 2min
This slide shows observation results measured with the rotational pendulum seismometer at Matushiro. The right above side drawing shows wave forms of marine microseism, with which the amplitude is about 10^-6 m/s and the frequency is about 10 Hz. The below right side graph describes a velocity frequency spectrum of the Earth’s background noise. The amplitude of the Earth’s background noise is about 10^-10 m/(s√Hz) in the frequency region over 1 Hz. This value is close to the specifications of the most sensitive seismometer, such as STS-I. However, the rotational pendulum seismometer dose not have sufficient sensitivity to measure the Earth’s background vibration noise under 0.1 Hz. One of the origins of this problem is a change in the buoyancy at the weight caused by a variation of the atmospheric pressure. The value of the buoyancy change by a rough estimation is 5.5 x 10^-8m/s2/kPa. (calculated under the following conditions: a weight of 0.8 kg, a weight volume of 1.56 x 10^-4 m3, and a temperature of 20 °C. This value is slightly under the measurement values in the frequency band from 0.1 Hz to 0.01 Hz in the figure. )
10**-8
10-８ (m/s)/√Hz
10**-9
10**-10
10**-2
10**-3
10**-1
10**-0
10**1
10**2
Frequency [Hz]

14. The　wave form and the frequency spectrum of the earth’s background vibration noise at Matushiro, March 04”
Marine Microseism
CMG 3T
(m/s)/√Hz
10-6 m/s
10**-3
10**-4
10**-5
March 12, 04” AM 2h 2min
10**-6
10**-7
Filter between 0.1 Hz to 0.01 Hz
This slide shows observation results measured with the rotational pendulum seismometer at Matushiro. The right above side drawing shows wave forms of marine microseism, with which the amplitude is about 10^-6 m/s and the frequency is about 10 Hz. The below right side graph describes a velocity frequency spectrum of the Earth’s background noise. The amplitude of the Earth’s background noise is about 10^-10 m/(s√Hz) in the frequency region over 1 Hz. This value is close to the specifications of the most sensitive seismometer, such as STS-I. However, the rotational pendulum seismometer dose not have sufficient sensitivity to measure the Earth’s background vibration noise under 0.1 Hz. One of the origins of this problem is a change in the buoyancy at the weight caused by a variation of the atmospheric pressure. The value of the buoyancy change by a rough estimation is 5.5 x 10^-8m/s2/kPa. (calculated under the following conditions: a weight of 0.8 kg, a weight volume of 1.56 x 10^-4 m3, and a temperature of 20 °C. This value is slightly under the measurement values in the frequency band from 0.1 Hz to 0.01 Hz in the figure. )
10**-8
10**-9
10-８ (m/s)/√Hz
10**-10
10**-2
10**-3
10**-1
10**-0
10**1
10**2
Frequency [Hz]

15. We need further refinement to obtain the best data.
Summary
　We need further refinement to obtain the best data.
We proofed the principle of the vertical astatic rotational pendulum.
We successfully accomplished a long natural period of more than10 s.
The magnetic field uniformity was about 10-4 in the bore of the window-frame magnet.
The thermal dependence of magnetic field strength was less than 9 x 10-5/K.
The noise level of the rotational pendulum seismometer in a frequency range over 1 Hz was less than 10-9 (m/s)/√Hz.
The rotational pendulum seismometer has an effect from a change in buoyancy caused by a variation of atmospheric pressure.
Make a chamber.

16. Another Developments Abstract
Horizontal Component Seismometer with Magnetic Levitation
Development of new instruments with new ideas is important to open up a new window of earth science. In this study, we have developed a seismometer using a new principle to obtain further sensitivity than present one. This seismometer uses a horizontal pendulum having weak leaf-spring crossed hinges and using magnetic levitation for a pendulum weight. The levitation employs permanent magnets magnetized by N, S strip, alternatively. This pendulum has been built and evaluated by experiments. The results of the experiments are as follows. 1. The prototype of the pendulum achieved a natural period more than 11 s. 2. The natural period of the prototype was changed with a magnetic spring. These values of the period were consistent with theoretical calculations. 3. The prototype with a feedback control circuit, a feedback control coil and a capacitance displacement sensor worked as a servo-type seismometer. 4. The observations using the magnetic levitation seismometer have been carried out and showed capacity to measure microseisms.

17. 　A horizontal component seismometer with magnetic levitation using plane permanent magnets with NS alternative magnetization strips.
Levitation
　By using the NS alternative magnetization, we can obtain a uniform potential energy just above the magnet surface. A perpendicular direction referred to the movement direction of a pendulum must be mechanically constrained with a hinge, because of Earnshaw’s low.
TP. 32 A horizontal component seismometer with magnetic levitation using plane permanent magnets with NS alternative magnetization strips.
This system uses plane permanent magnets with NS alternative magnetization strips. The magnets are opposite placed respectively. By using the NS alternative magnetization, we can obtain a uniform potential energy just above the magnet surface. It can stably levitate more than 2 kg. A perpendicular direction referred to the movement direction (a strip direction) of a pendulum must be mechanically constrained with a hinge, because of Earnshaw’s low. The perpendicular direction is unstable.

18. The picture of the horizontal component pendulum with magnetic levitation of the weight using the plane permanent magnets
TP.33 The picture of the horizontal component pendulum with magnetic levitation of the weight using the plane permanent magnets.
The horizontal pendulum comprises a magnetic levitation system with NS strips, a weight, a hinge using hight-elinvar metal, a magnetic spring, feedback coil and capacitance position detector. The hinge position can be changed with a X-Y-Z stage.

19. A free oscillation of the pendulum and an output signal of the horizontal servo-type seismometer with magnetic levitation for the weight
Marine Microseism
The wave form is measured by the capacitance position meter (0.2V/mm). The natural period of the pendulum is about 8 s.
TP.36 A free oscillation of the pendulum with magnetic levitation for the weight.
This wave form of a free oscillation of the horizontal pendulum is measured by the capacitance position meter (0.2V/mm). The natural period of the pendulum in this case is about 8 s.
It is so difficult to obtain this wave form without position feedback.

20. Laser Scale Servo-type Seismometer