Investigation of Magneto-Inductive Sensors for Low Magnetic Field Measurements
Huan Liu, Changfeng Zhao, Hongpeng Wang, Xiaobin Wang, Haobin Dong, Zheng Liu, Jian Ge, Zhiwen Yuan, Jun Zhu, Xinqun Luan
aa r X i v : . [ phy s i c s . i n s - d e t ] J u l Investigation of Magneto-Inductive Sensors for LowMagnetic Field Measurements
Huan Liu,
Member, IEEE,
Changfeng Zhao, Hongpeng Wang, Xiaobin Wang, Haobin Dong,Zheng Liu,
Senior Member, IEEE,
Jian Ge, Zhiwen Yuan, Jun Zhu, and Xinqun Luan
Abstract —The surface resistance reduction can further sup-press the surface power loss of a superconducting radio-frequencycavity. Since the surface resistance of a superconducting radio-frequency cavity is mainly originated from the magnetic fluxtrapping, and thus the corresponding magnetic field strengthcould be measured to reflect the residual resistance. A fluxgatemagnetometer is always employed to measure the ambient sur-face magnetic field of a superconducting radio-frequency cavity.However, this kind of equipment is relatively larger than thecavity and always need expensive cost. In this paper, we developeda magneto-inductive (MI) magnetic sensor, which is smaller,lighter weight, and lower cost than the fluxgate magnetometer.The specifications such as the noise floor, resolution, etc., aremeasured. In addition, a comparative observation of the magneticfield between the proposed MI sensor and a highly precise isconducted. The experimental results identify the capability ofthe proposed MI sensor in weak magnetic detection.
Index Terms —Magnetic field, magneto-inductive sensor, super-conducting radio-frequency cavity.
I. I
NTRODUCTION T o achieve a High-Q operation of the superconductingradio-frequency cavity, numerous researches have beenimplemented [1], [2]. To be specific, since the cryogenic lossis tightly coupled to the High-Q operation, it is necessary toreduce the surface resistance of the superconducting radio-frequency cavity which is mainly originated from the magneticflux trapping within the period of the cavity cooling downprocedure [3], [4]. As the cavity cools down to a supercon-ducting condition, the partial ambient magnetic field shouldbe trapped into the radio-frequency cavity. In this case, wecan obtain the surface residual resistance variations thoughmeasuring the corresponding magnetic field strength [5], [6]. This work is partly supported by the National Natural Science Foundationof China under Grant Nos. 41904164 and 41874212, the Foundation ofWuhan Science and Technology Bureau under Grant No. 2019010701011411and 2017010201010142, the Foundation of National Key Research andDevelopment Program of China under Grant No. 2018YFC1503702, theFoundation of Science and Technology on Near-Surface Detection Laboratoryunder Grant Nos. 6142414180913, TCGZ2017A001, and 614241409041217,and the Fundamental Research Funds for the Central Universities, ChinaUniversity of Geosciences (Wuhan) under Grant No. CUG190628.H. Liu, C. Zhao, H. Wang, X. Wang, H. Dong, and J. Ge are with Schoolof Automation, China University of Geosciences, Wuhan, China, and HubeiKey Laboratory of Advanced Control and Intelligent Automation for ComplexSystems, Wuhan, China (email: [email protected]).Z. Liu is with School of Engineering, Faculty of Applied Science, Uni-versity of British Columbia Okanagan Campus, Kelowna, Canada (email:[email protected]).Z. Yuan, J. Zhu and X. Luan are with Science and Technology on Near-Surface Detection Laboratory, Wuxi, China.
Generally, a fluxgate sensor is employed to measure theambient surface magnetic field of a superconducting radio-frequency cavity with high precision. However, this approachis not available to achieve a higher spatial resolution since thefluxgate sensor’s size is relatively large [7], [8]. For instance,a commonly used and accepted fluxgate sensor, dubbed Mag-03, which is developed by with dimensions of 32 mm x 32mm x 225 mm [9], [10]. In recent years, a kind of magneticsensor using magneto-inductive (MI) technique has graduallyentered the field of weak magnetic detection [11], [12]. Themeasurement principle of the MI sensor is conspicuouslydifferent from that of the fluxgate sensor. Because of itsminiature size, the MI sensor has been commonly employedin many applications, such as compassing, inertial navigation,security system, etc [13]–[15].In this paper, we develop a MI sensor to measure thesurface magnetic field of the superconducting radio-frequencycavity. The electrical properties including the noise floor,resolution, magnetic tracking performance, etc., are evaluatedin the Laboratory of Advanced Magnetic Sensor and IntelligentImaging Detection, China University of Geosciences, Wuhan,China. Moreover, a measurement system using the MI sensoris developed to conduct an outdoor contrast experiment witha commercial Overhauser magnetometer and its performancefor magnetic anomaly detection is verified. R MI Sensor Inductor H Schmitt Trigger Oscillator Output I Fig. 1: Diagram of the electrical components involved in themagneto-inductive technique.II. M
AGNETO -I NDUCTIVE S ENSOR
The MI sensor mainly consists of a resistor-inductor cir-cuit, in which an electronic component is sensitive to theexternal magnetic field. To be specific, as shown in Fig. 1, the period during the charging and discharging procedures ofthe MI sensor inductor between two thresholds (lower andupper) through the Schmitt trigger oscillator, is proportionalto the external magnetic field strength H . Generally, the totalmagnetic field H t is composed of the external magnetic field H and the generated field kI of the circuit as: H t = H + kI (1)where k stands for the conversion factor of the sensor coil, and I stands for the generated current. In this case, the Schmitttrigger would operate the I through the resistor R to oscillatesince the generated voltage of R is over than the setting triggervalue. More details about the operation principle of the MIsensor can be referred to literature [16].A MI sensor module is constructed as shown in Fig. 2,which consists of one sensor coil (PN 13101) for z axis,two sensor coils (PN 13104) for x and y axis, respectively.Further, an application-specific integrated circuit controlleris embedded in this module, which can transfer the analogdata to a digital format. In this case, we can obtain theexternal magnetic field value through deriving the digital datato a micro-controller directly, which can avoid the additionalsignal conditioning module and acquisition module as that ina fluxgate sensor.Fig. 2: Three-axis MI hybrid sensor.III. E XPERIMENTAL RESULTS AND DISCUSSION
In this section, the specifications of the proposed MI sensorincluding the resolution, the noise floor, the stability, andthe magnetic tracking performance are evaluated by using amagnetic shielding cylinder with three layers, which can fur-ther suppress the external magnetic field and electro magneticinterference. In this case, the measurement range of the MIsensor is set as -800 µ T to 800 µ T, the digital resolution is13.33 nT per least significant bit, and the sampling rate is 10Hz.
A. Platform setup
We set up a magnetic field measurement system usingthe proposed MI sensor as illustrated in Fig. 3, this systemmainly consists of four modules, i.e., a MI sensor, a micro-controller, a USB serial controller, and an upper computer.Hence, the magnetic field data collected by the MI sensorcan be transferred to the computer through the serial port, and ultimately achieve real-time monitoring. Further, a leastsquares based data smoothing method is employed to suppressthe fluctuations of the anomalous data, further improving thesignal to noise ratio (SNR) of the measured data.Fig. 3: Testing platform of the proposed MI magnetic fieldsensor system.
B. Resolution
Theoretically, the resolution stands for the minimum detec-tion value of the MI sensor, and the standard deviation can beadopted to evaluate the resolution [17]. The standard deviationcan be written as: δ = vuut n − n X i =1 ( y i − y i ) (2)where δ stands for the standard deviation, y i stands for theaverage value, and n stands for the number of the collecteddata. Fig. 4 shows the collected continuous data within 10minutes. Through Eq. 2, we can calculate the resolution asabout 8.28 nT. This value is lower than the digital resolutionwe set as 13.33 nT, because we implement an averagingoperation for every ten data to further improve the resolution.Ideally, the larger the number of averaging data, the betterthe resolution. However, this will influence the dynamicalresponse characteristic of the MI sensor. Sampling Point -40-30-20-10010203040 M agne t i c f i e l d ( n T ) Fig. 4: Zero applied field measurement results.
C. Noise floor
Since the frequency characteristic of the noise features a1/ f dependence, the power spectral density at 1 Hz is alwaysemployed to evaluate the noise floor of a sensor [18], [19].Hence, we collect the magnetic field data consecutively within1 hour. Fig. 5 shows the evaluating results of the powerspectral density. We can see that the noise floor of the 1/ f region is relatively high, and it decreases slowly overall withup and down fluctuations as the frequency increases. As thefrequency increases around 1 Hz, the total noise tends to bestable and the noise floor of the developed MI sensor is about0.235 nT/Hz / at 1 Hz. -3 -2 -1 Frequency (Hz) -3 -2 -1 n T / H z / Fig. 5: Power spectral density of the proposed MI sensormodule.
D. Stability
The stability of the sensor represents how constant thecollected data is when there is no external magnetic field [20].To evaluate the stability of the proposed MI sensor, the sensoris also placed in the magnetic shielding cylinder without anexternal magnetic field but might be accompanied by a littlebit residual of the geomagnetic field. The system is operatingfor about 1 hour with a sampling rate of 10 Hz. Fig. 6 showsa histogram with the distribution of the collected data. It canbe seen that the distribution is normal with an approximatesymmetrical shape, and the value of the symmetrical centeris about 0 nT. In addition, the randomness of the observedvariations is corresponding to the intrinsic noise of the sensor.
E. Magnetic tracking performance
Further, we adopt the magnetic field shielding cylinder togenerate a changeable magnetic field while the step valuesincluding 50 nT and 100 nT, to identify its magnetic trackingperformance. The measurement results are shown in Fig. 7.We observe that the proposed MI-based magnetometer cantrack the variation of 100 nT effectively, while for the 50 nTthe magnetic tracking performance is less satisfactory (with -50 -40 -30 -20 -10 0 10 20 30 40 50
Magnetic field (nT) N u m be r o f c o rr e s pond i ng v a l ue Fig. 6: Distribution of collected data within one hour.major fluctuations) than that of 100 nT. In addition, there isone thing that needs to be emphasized. During the experiment,we observe that when the step value is lower than 50 nT, themagnetic tracking performance is not obvious due to the ownnoise properties of the MI sensor. Hence, we only depict thetesting results of 100 nT and 50 nT in this paper.
Sampling Point M agne t i c f i e l d ( n T ) Step value: 50 nTStep value: 100 nT
Fig. 7: The measurement results of the magnetic trackingperformance.
F. Geomagnetic comparison test
To further identify the performance of the developed MIsensor system, a comparison testing between this sensor anda commercial Overhauser sensor system was implemented toobserve the magnetic field during the same period. Moreover,we place a iron (the shape is cylinder and the dimension isabout 80 mm x 80 mm x 200 mm) near the aforementionedtwo sensors with different distances (30 cm and 100 cm), toverify the capability for magnetic anomaly detection. The localmagnetic field strength is about 49660 nT and the magneticfield measurement results are shown in Fig. 8.
Sampling point M agne t i c f i e l d ( n T ) Overhauser sensorMI sensor
Remove the iron Remove the ironPlace the iron with a distance of 100 cmPlace the iron with a distance of 30 cm
Fig. 8: Contrast test between the proposed MI sensor and the commercial Overhauser sensor.Firstly, the trends of the geomagnetic field intensity mea-sured by these two instruments are basically the same, whileduring the anomaly detection stage, the two recording curvesare opposite and symmetric. The main reason is that in order toavoid magnetic interference, the two magnetometers are placedfar away from each other (5 m in this paper), and the differentsensor positions will generate a magnetic field gradient. Be-sides, since the iron is placed between the two sensors, therewill generate two opposite magnetic anomalies at the head andthe tail of the iron according to its magnetic field distributioncharacteristics [19]. Further, the fixed magnetic anomalies ofbuildings, vehicles, and cables in the test environment alsogenerate magnetic field gradients, but they do not affect thecontrast between the two sensors. Under the influence ofmoving pedestrians and vehicles in the test environment, theexternal magnetic interference cannot be completely avoided.The consistent variation trend of the commercial Overhausersensor and the MI sensor indicates that our proposed MI sensorhas good magnetic field tracking performance and is sensitiveto the magnetic anomalies in the measured area.IV. C
ONCLUSIONS
In this study, we developed a miniature MI sensor modulewith dimensions of 1 cm x 1cm x 0.2 cm to measure thesurface magnetic field of the superconducting radio-frequencycavity. The laboratory testing results demonstrate that theproposed MI sensor can detect a low magnetic field strengthas low as 8.28 nT when the sampling rate is 10 Hz, and itstypical noise floor is about 0.235 nT/Hz / at 1 Hz. For thefield testing results, the proposed MI sensor shows good mag-netic field tracking performance, and has the ability to catchmagnetic anomalies. Consequently, the proposed MI sensorshows a great scope for magnetic anomaly detection, and theminiature property makes it overcome the shortcomings of therelated commercial devices (especially the fluxgate sensor) formeasuring the surface magnetic field of the superconductingradio-frequency cavity, i.e., bulky size, heavy, high power, etc. Further, this study also lays the foundation for future worksto improve the accuracy of the MI sensor for magnetic fielddetection. R EFERENCES[1] R. Ueki, T. Okada, M. Masuzawa, K. Tsuchiya, T. Kawamoto,K. Umemori, E. Kako, T. Konomi, and H. Sakai, “Study on magneto-resistance sensors for low magnetic field measurements,”
IEEE Trans-actions on Applied Superconductivity , vol. 30, no. 4, pp. 1–4, 2020.[2] K. Umemori, T. Dohmae, E. Kako, T. Konomi, T. Kubo, M. Masuzawa,T. Okada, G.-T. Park, A. Terashima, K. Tsuchiya et al. , “Improvement ofmagnetic condition for kek-stf vertical test facility toward high-q study,”in
Proc. 18th Int. Conf. on RF Superconductivity (SRF’17) , 2017, pp.444–447.[3] A. Romanenko, A. Grassellino, A. Crawford, D. Sergatskov, and O. Mel-nychuk, “Ultra-high quality factors in superconducting niobium cavitiesin ambient magnetic fields up to 190 mg,”
Applied Physics Letters , vol.105, no. 23, p. 234103, 2014.[4] M. Masuzawa, K. Tsuchiya, and A. Terashima, “Study of magneticshielding materials and fabrication of magnetic shield for superconduct-ing cavities,”
IEEE Transactions on Applied Superconductivity , vol. 24,no. 3, pp. 1–4, 2013.[5] K. Umemori, H. Hara, H. Inoue, E. Kako, T. Konomi, T. Kubo, H. Sakai,K. Sennyu, H. Shimizu, M. Yamanaka et al. , “Vertical test results ofnitrogen doped srf cavities at kek,” in . JACOW, Geneva,Switzerland, 2016, pp. 2154–2157.[6] H. Liu, H. Dong, J. Ge, Z. Liu, Z. Yuan, J. Zhu, and H. Zhang,“Apparatus and method for efficient sampling of critical parametersdemonstrated by monitoring an Overhauser geomagnetic sensor,”
Reviewof Scientific Instruments , vol. 89, no. 12, p. 125109, 2018.[7] B. Bai, H. Liu, J. Ge, and H. Dong, “Research on an improved resonantcavity for overhauser geomagnetic sensor,”
IEEE Sensors Journal ,vol. 18, no. 7, pp. 2713–2721, 2018.[8] H. Liu, H. Dong, J. Ge, Z. Liu, Z. Yuan, J. Zhu, and H. Zhang, “A fusionof principal component analysis and singular value decomposition basedmultivariate denoising algorithm for FID transversal data,”
Review ofScientific Instruments , vol. 90, no. 3, p. 035116, 2019.[9] B. Instruments, “Mag-03 three axis magnetic field sensors,” 2014.[10] H. Liu, H. Dong, J. Ge, and Z. Zhao, “Magnetic field gradient detectorbased on the nuclear overhauser effect,”
Chinese Journal of ScientificInstrument , vol. 36, no. 3, pp. 592–600, 2015.[11] H. Liu, X. Wang, H. Wang, J. Bin, H. Dong, J. Ge, Z. Liu, Z. Yuan,J. Zhu, and X. Luan, “Magneto-inductive magnetic gradient tensorsystem for detection of ferromagnetic objects,”
IEEE Magnetics Letters ,vol. 11, no. 1, p. 8101205, 2020. [12] M. Moldwin, B. Bronner, L. Regoli, J. Thoma, A. Shen, G. Jenkins, andJ. Cutler, “New magneto-inductive dc magnetometer for space missions,”in
AGU Fall Meeting Abstracts , 2017.[13] H. Liu, H. Dong, Z. Liu, J. Ge, B. Bai, and C. Zhang, “Construction ofan overhauser magnetic gradiometer and the applications in geomagneticobservation and ferromagnetic target localization,”
Journal of Instrumen-tation , vol. 12, no. 10, p. T10008, 2017.[14] Y. Huang, J. Ge, H. Dong, and H. Liu, “An automatic wideband90 ◦ phase shifter for optically pumped cesium magnetometers,” IEEESensors Journal , vol. 17, no. 23, pp. 7928–7934, 2017.[15] H. Liu, X. Wang, J. Bin, H. Dong, J. Ge, Z. Liu, Z. Yuan, J. Zhu, andX. Luan, “Magnetic gradient full-tensor fingerprints for metallic objectsdetection of a security system based on anisotropic magnetoresistancesensor arrays,”
AIP Advances , vol. 10, no. 1, p. 015329, 2020.[16] A. Leuzinger and A. Taylor, “Magneto-inductive technology overview,”
PNI white paper , 2010.[17] H. Dong, H. Liu, J. Ge, Z. Yuan, and Z. Zhao, “A high-precision fre-quency measurement algorithm for fid signal of proton magnetometer,”
IEEE Transactions on Instrumentation and Measurement , vol. 65, no. 4,pp. 898–904, 2016.[18] H. Liu, H. Dong, J. Ge, B. Bai, Z. Yuan, and Z. Zhao, “Researchon a secondary tuning algorithm based on svd & stft for fid signal,”
Measurement Science and Technology , vol. 27, no. 10, p. 105006, 2016.[19] H. Liu, H. Dong, Z. Liu, J. Ge, B. Bai, and C. Zhang, “Noisecharacterization for the fid signal from proton precession magnetometer,”
Journal of Instrumentation , vol. 12, no. 7, p. P07019, 2017.[20] H. Liu, H. Dong, Z. Liu, J. Ge, W. Luo, C. Zhang, Z. Yuan, J. Zhu,and H. Zhang, “A comprehensive study on the weak magnetic sensorcharacter of different geometries for proton precession magnetometer,”