604 IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL 38, NO 2 . APRIL 1989

Optical Fiber Feedback SQUID Magnetometer SEIICHI NAITO, MEMBER, IEEE, YOSHIHIRO SAMPEI, AND TAKAHIRO TAKAHASHI

Abstruct-This paper describes an optical fiber feedback supercon- ducting quantum interference device (SQUID) magnetometer which was developed to improve electromagnetic interference characteristics. The SQUID consists of an RF SQUID probe, an RF amplifier, two mul- timode fibers, and a SQUID control unit. Phase-locked pulse width modulation (PWM) was used to construct a flux locked loop (FLL) cir- cuit in the SQUID control unit.

The operation of the optical fiber feedback SQUID is stable when a common mode voltage of ac 100 V/50 Hz is applied. It has an energy resolution of 1 x lo-’’ J/Hz.

This paper also describes the measurement of an auditory evoked field from the human brain in a magnetically shielded room using the fiber feedback SQUID with a gradiometer type pickup coil.

I. INTRODUCTION HE SQUID magnetometer has an extremely high sen- T sitivity for detecting very weak magnetic fields. In

biomagnetic measurements, a SQUID magnetometer can make significant contributions to studies of activity in the human body [ 11.

During the past 15 years, the sensitivities of SQUID magnetometers have been improved and many biomag- netic signals such as those for magnetocardiograms [2] and magnetoencephalograms [3], have been measured in the human body with the SQUID magnetometer. Measur- ing systems employing the SQUID magnetometer have been developed to detect biomagnetic signals with and without a magnetically shielded room. In these systems, a SQUID probe in the cryostat and a head amplifier are connected to the SQUID control unit and to a readout sys- tem which includes the computer with electrical cables. Usually, the readout system is placed outside the mag- netically shielded room. The SQUID may detect noise signals from electrical equipment and cables due to its extreme sensitivity.

The SQUID magnetometer is especially sensitive to electromagnetic interference between the SQUID probe and the SQUID control unit. Recently, multichannel SQUID magnetometers have been developed [4], [5] which hold promise [6] for detecting the magnetic field produced by the human brain simultaneously at many points. For these complicated systems, rejection of the electromagnetic interference would prove to be impor- tant.

Our focus is on improving the electromagnetic interfer- ence characteristic of the SQUID magnetometer. This pa-

Manuscript received June 10, 1988. The authors are with the Yokogawa Electric Corporation, 2-9-32, Naka-

IEEE Log Number 8825810. cho, Musashino-shi, Tokyo 180, Japan.

Lock i n AMP +-

FeedBack

r-l Cryostat

T ? +Es -Es

U

Fig. 1. Block diagram of optical fiber feedback SQUID magnetometer.

per describes a new method for improving the SQUID magnetometer using optical fiber feedback with common mode rejection characteristics. It also reports an auditory evoked field measurement using an optical fiber feedback SQUID magnetometer.

11. CONFIGURATION OF THE FIBER FEEDBACK SQUID

Fig. 1 shows a block diagram of the optical fiber feed- back SQUID magnetometer. The SQUID magnetometer consists of an RF SQUID probe, an RF amplifier unit, multimode fibers, and a SQUID control unit. The RF am- plifier unit operates on battery power. The SQUID probe, which operates at 4.2 K, consists of a thin-film SQUID device and an input coil. The SQUID device, with a sub- micrometer-width niobium thin-film microbridge on a sapphire substrate, is used as an RF SQUID operating at 22 MHz [7]. The RF amplifier unit includes a 22-MHz oscillator and demodulator which produces feedback and a reference AF signal (20 kHz). E/O, O/E converters are also included. The SQUID control unit consists of a flux locked loop (FLL) circuit and O/E, E/O converters. A phase-locked pulsewidth modulation (PWM) circuit is used to combine the feedback and the reference signals.

The demodulation circuit in the RF amplifier unit con- sists of a low-pass filter and a monostable multivibrator. The low-pass filter produce a feedback signal from the PWM signal that is applied to the SQUID . The mono- stable multivibrator produce a lock-in reference signal which is also applied to the SQUID.

The phase-locked PWM circuit and waveform are shown in Fig. 2(a) and (b), respectively. Voltage E,, which is proportional to the output of the SQUID mag- netometer, is an input voltage for the PWM circuit. Volt- ages +E, and -E, are standard voltages. An operational

0018-9456/89/0400-0604$01 .OO 0 1989 IEEE

NAITO er al.: OPTICAL FIBER FEEDBACK SQUID MAGNETOMETER

0 e3 - -

605

t--t

Ei

amplifier A1 is used with R1, R2, and C as an integrator. CLOCK is a 20 kHz sawtooth wave. Operation of the PWM circuit was described in [8]. Ei is expressed as

where we assume that the switch S1 is connected to +E, for time t l and to -E, for time t2 . The phase of the PWM output signal ( e3) is locked to the sawtooth CLOCK sig- nal as shown in Fig.. 2(b). This is different from [8]. Equation (1) shows that the pulsewidth is proportional to the output voltage of the SQUID magnetometer. Thus the lock-in reference signal AF and the feedback signal can be combined and transmitted together to the RF amplifier unit in a multimode optical fiber. A digital signal is avail- able for counting pulsewidth.

The SQUID signal is amplified by an RF amplifier with a cascade JFET input, with a 2-dB noise figure, and then converted into an optical signal by a light-emitting diode (LED) with a wavelength of 820 nm. The optical signal is transmitted to the SQUID control unit by the 30-m mul- timode step index fiber (quartz with 100 pm/140-pm core/cladding diameter) and can be detected by a p-i-n photodiode.

111. SIGNAL DISTORTION When the SQUID signal is transmitted on the optical

fiber, signal distortion must be minimized. A p-i-n pho- todiode has a linear response but an LED does not. Dis- tortion in the transmitted signal degrades the performance of the SQUID magnetometer, especially the magnetic flux resolution. We also examined signal distortion in the op- tical transmission line. The experimental result is shown in Fig. 3 . The signal distortion depends on the bias cur- rent and the signal amplitude of the LED. The minimum

0 20 40 60

AMPLITUDE (mAp-p)

Fig. 3 . Distortion in optical transmission line.

k - c r y o s t a t -1 I I k- v cmn 4 SQUID CONTROL UNIT

Fig. 4. Experiment of common mode voltage in optical fiber feedback SQUID.

- I rf osc I ELECTRICAL CABLE

OUTPUT ,I ___________-------

I ., j \Lock in AMP +

K SQUID CONTROL UNIT

Fig. 5. Experiment of common mode voltage in conventional SQUID.

distortion was found to be 0.7 percent in the optical trans- mission line with a bias current of 20 mA and an ampli- tude of 20 mA,-,.

IV. COMMON MODE CHARACTERISTICS

Higher common mode rejection will certainly be needed when very weak magnetic fields such as those from the human brain are measured with a SQUID magnetometer. The experimental setup for evaluating common mode characteristics for optical fiber feedback SQUID’S and conventional SQUID’S are shown in Figs. 4 and 5 , re- spectively. An ac common mode voltage was applied be- tween the RF amplifier unit and the SQUID control unit in each experiment. Waveforms of the fiber feedback SQUID output and applied magnetic field (input) are shown in Fig. 6 for the case where common mode volt- ages of (a) 0 V and (b) ac 100 V were applied. It can be , shown that the operation of the SQUID is stable even

606 IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 38. NO. 2. APRIL 1989

(a) F r e q u e n c y ( H z )

Fig. 8. Noise spectrum of optical fiber feedback SQUID.

OUT P U T ( S O m V / d i v . )

OPT1 FIB

I N P U T ( 1 V / d i v . )

HEAD PHONE Fig. 6. Output of optical fiber feedback SQUID with common mode volt- age of (a) 0 V and (b) ac 100 V/50 Hz. MAGNET1 CALLY

SHIELDED ROOM

Fig. 9. Block diagram of AEF measurement.

R F AMP OUT P U T

( 5 0 m V / d i v

OUT P U T ( O . S V / d i v .

R F AMP OUT P U T

( S O m V / d i v .

OUT P U T ( 0 . 5 V / d i v

when a common mode voltage of ac 100 V/50 Hz is ap- plied.

Waveforms of the RF amplifier output (SQUID signal) and SQUID magnetometer output for a conventional SQUID with no applied magnetic field are shown in Fig. 7 for the case where when common mode voltages of (a) 0 V and (b) ac 30 mV/50 Hz were applied. The RF am- plifier output and SQUID output are modulated by the common mode voltage. It can be shown that the conven- tional SQUID produces a false output when a low-level common mode voltage is applied, even where no mag- netic field is present.

V. NOISE SPECTRUM AND SENSITIVITY Fig. 8 shows the noise spectrum measured by a spec-

trum analyzer when no magnetic field was applied to the SQUID magnetometer. The noise spectrum was flat from 0.1 to 100 Hz. 50-Hz line noise and its harmonics were not observed. The sensitivity was measured by observing the jump of the flux quantum 9o in the SQUID magnetom- eter output. It was found to be about 1 V/a0 , and the flux resolution was 3.2 x lop4 4j0/&.

VI. AUDITORY EVOKED FIELD MEASUREMENT An auditory evoked field (AEF) was measured in a

(b)

Fig. 7. RF amplifier output and output of conventional SQUID with com- mon mode voltage of (a) 0 V and (b) ac 30 mV/50 Hz.

magnetically shielded room using an optical fiber feed- back diagram Of the AEF measurement is shown in Fig. 9. The Japanese vowel /a/ was used as

A

NAITO et al.: OPTICAL FIBER FEEDBACK SQUID MAGNETOMETER 607

3.2 x +o/& and can detect the magnetic evoked field produced by the human brain.

The method described here may be applicable for the multichannel SQUID system to reduce the electromag- netic interference between the channels.

0.5 PT

ACKNOWLEDGMENT - 100 mSec

Fig. 10. AEF wave from auditory cortex.

sound stimuli [9]. The duration of the stimulus was about 70 ms and the period was 1 s. The sound stimuli were recorded on a tape recorder and transmitted to the ear through an air tube and the type of earphone used on air- lines. There were no electric cables feeding into the mag- netically shielded room. A first derivative gradiometer with a diameter of 30 mm and a baseline of 70 mm was used as a pickup coil. Fig. 10 shows the AEF wave from the auditory cortex, the component of the magnetic field perpendicular to the skull. The signal-to-noise ratio was improved by an average of 64.

VII. CONCLUSION An optical fiber feedback SQUID magnetometer can

improve electromagnetic interference characteristics. Two multimode fibers 30-m long and phase-locked PWM are used to construct an FLL circuit. It can be constructed with only one optical fiber using a multiwavelength opti- cal transmission system. The optical fiber feedback SQUID magnetometer was stable with a common mode voltage of ac 100 V/50 Hz. It has an flux resolution of

The authors wish to acknowledge the encouragement of Dr. M. Ibuka and wish to thank G. Uehara and Y. Kikuti for their helpful discussions.

REFERENCES [l] G. L. Romani, S . J. Williamson, and L. Kaufman, “Biomagnetic in-

strumentation,” Rev. Sci. Instr., vol. 53, no. 12, pp. 1815-1845, 1982. [2] D. Cohen and D. McCaughan, “Magnetocardiograms and their vari-

ation over the chest in normal subjects,” Amer. J . Cardiol., vol. 29, no. 5, pp. 678-685, 1972.

[3] D. Cohen, “Magnetoencephalography: Detection of the brain’s elec- trical activity with a superconducting magnetometer, ” Science, vol. 175, pp. 664-666, Feb. 1972.

[4] S . J . Williamson et a l . , “Magnetoencephalography with an array of SQUID sensors,” in Proc. 10th Int. Cryogenic Engineering Conf., pp.

[5] M. Kajola et a l . , “Low nolse seven-channel DC-SQUID magnetom- eter for brain research,” in Proc. 6th Int. Con5 Biomagnetism, pp.

[6] K. Shirae, H. Furukawa, M. Katayama, and T. Katayama, “Proposal of the multichannel SQUID amplifiers for biomagnetic measurement,” in Proc. 6th Int. Con$ Biomagnerism, pp. 132-133, 1987.

[7] M. Ibuka, H. Hosomatsu, and S. Naito, “A SQUID magnetometer using a niobium thin-film microbridge,” IEEE Trans. Instrum. Meas.,

[8] M. Tomota, T. Sugiyama, and K. Yamaguchi, “An electronic multi- plier for accurate power measurement,” IEEE Trans. Instrum. Meas.,

[9] Y. Kikuchi, T . Tsunoda, S. Naito, T. Takahasi, and G. Uehara, “A sound feature extraction mechanism in human brain,” in Proc. 6th Int. Con$ Biomagnetism, pp. 40-41, 1987.

339-348, 1984.

110-111, 1987.

vol. IM-30, pp. 251-254, 1981.

vol. IM-17, pp. 245-251, 1968.