IEEE TRANSACTIONS ON MAGNETICS, VOL. 50, NO. 5, MAY 2014 6500103

A Low-Noise Fundamental-Mode Orthogonal FluxgateMagnetometer

Robert Bazinet1, Alfredo Jacas2, Giovanni A. Badini Confalonieri2, and Manuel Vazquez2

1Phoenix Geophysics Ltd., Toronto, ON M1W 3K5, Canada2Instituto de Ciencia de Materiales de Madrid, Consejo Superior de Investigaciones Científicas, Madrid 28049, Spain

We introduce a low-noise fundamental-mode orthogonal fluxgate magnetometer for use in magnetotelluric surveys. The fluxgatemakes use of rapidly quenched amorphous wire having vanishing value of saturation magnetostriction constant (λ ≈ 10−7) anddisplaying ultrasoft magnetic behavior. The design of the fluxgate consists of multiple U-shaped sensor heads where pairs of wirepieces are inserted as core material. The novelty of this system resides in the use of a quadruple sensor head. Digital signal processingwith effective electronic noise suppression allows this magnetometer to achieve a noise floor of 0.8 pT/Hz1/2 for frequencies above10 Hz. The possibilities of in situ application are discussed and guidelines on noise suppression strategies are given.

Index Terms— Amorphous microwire, low noise, magnetometer, orthogonal fluxgate.

I. INTRODUCTION

MAGNETOTELLURIC surveys require very sensitive acmagnetometers. Systems based on induction coils are

the common sensor of choice for such surveys as they provideappropriately low-noise level, typically 0.1 pT/Hz1/2 at 1 Hz(Phoenix Geophysics Ltd MTC50 sensor, for example) [1].This comes at a high logistic price, as these sensors arelarge and heavy (1.5 m long, 10 kg). Fluxgate magnetome-ters are small but unfortunately do not provide the perfor-mance level required. The best commercially available unitis specified at 6 pT/Hz1/2 [2], which value is also typicalof various results reported in the literature. More recently,giant magnetoimpedance (GMI) magnetometers have beenshown to be competitive substitutes for fluxgate sensors,generally showing higher bandpass and better high-frequencyperformance. However, they present higher 1/f noise in thelow-frequency region, with relatively strong perming effectand low-frequency temperature drifts [3]–[5]. Alternatively,superconducting quantum interference device magnetometersdo provide the required performance but their use in fieldconditions is mostly impractical.

Fundamental-mode orthogonal fluxgate (FMOF), makinguse of magnetic microwires and originally described in[6], offers a series of advantages over more conventionalsecond harmonic fluxgate magnetometers. The main differ-ence between FMOF and the conventional second harmoniccounterpart resides in the physical principles involved in theexcitation of the magnetic core. In FMOF, this is achievedby passing throughout the wire a dc current to which anac sine wave, of lower amplitude, is superimposed. As aconsequence, the magnetization vector of the core, instead ofreversing, rotates from its radial position toward the axis of thewire. The amplitude of this rotation is detected by a sensing

Manuscript received May 27, 2013; revised September 6, 2013; acceptedNovember 16, 2013. Date of publication November 26, 2013; date of currentversion May 1, 2014. Corresponding author: G. A. Badini Confalonieri(e-mail: gabadini@icmm.csic.es).

Color versions of one or more of the figures in this paper are availableonline at http://ieeexplore.ieee.org.

Digital Object Identifier 10.1109/TMAG.2013.2292834

coil in the form of induced voltage, which is sinusoidal at afundamental frequency [6]. The use of magnetic microwires ascore material, where the circumferential close magnetic fluxof the wire can be excited by a circumferential field obtainedby passing a current along the wire axis, allows for simplersensor design together with the easiness to miniaturize it.

Traditionally, the functionality of FMOF was limited by therelatively high output noise [7], understood to arise mostlyfrom Barkhausen noise. Recently, however, Butta and Sasada[8] made an important contribution toward understandingthe source of noise in FMOF, and reducing it down to thecompetitive value of 1.8 pT/Hz1/2, by careful control overthe amorphous microwire magnetic anisotropy. Other aspectsto consider are the contribution to the noise arising from themagnetically harder ends of the wire [9] and the excitationparameters [10].

With noise values of a few pT/Hz1/2, further ther-mal and electronic contributions become relevant, alongsideBarkhausen noise, to the system noise level. In the followingsection, we present the development of an FMOF sensor with anoise level of better than 1 pT/Hz1/2, which, while not beingas good as the best large induction coils, it is neverthelessgood enough for a wide range of geophysical applicationsas detailed in the patents applied on several aspects of thismagnetometer [11].

II. EXPERIMENTAL DETAIL

In this section, we describe the elements involved in thefluxgate magnetometer.

A. Sensing Element

The core material is an ultrasoft amorphous cobalt alloywire obtained with a rotating-water-bath-casting unit, sim-ilar to the one described in [12], having composition(Co0.94Fe0.06)72.5Si12.5B15 and diameter 120 μm, customdeveloped by ICMM for Quantec Geoscience. CoFe-basedamorphous alloys are among the softest magnetic materialsand, in wire form, are of particular technological interest fortheir magnetic and GMI properties [13]. At the origin of the

0018-9464 © 2013 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.

6500103 IEEE TRANSACTIONS ON MAGNETICS, VOL. 50, NO. 5, MAY 2014

Fig. 1. Schematic view of the sensor head construction.

magnetic properties is the unique intrinsic magnetic anisotropyof this class of materials, which is the result the balancedcontributions from the magnetoelastic anisotropy, the stressesquenched in during the fabrication process, and the shapeanisotropy from the cylindrical geometry of the wire. Wiresexhibiting vanishing values of saturation magnetostriction, λs ,can be obtained with a Co:Fe ratio of approximately 94:6. Atthis proportion, the magnetization reversal mechanism alongthe magnetic easy axis, parallel to the wire axis, occurs by therapid displacement of magnetic domain walls at external fieldvalues below 10 A/m [13].

B. Sensor Head

The basic design of our sensor head is very similar to theone presented in [14], using a U-shaped magnetic core, butthe core is inserted in a single 1 mm diameter and 25 mmlong sensing coil, as shown in Fig. 1.

An additional solenoid, over the basic sensor assembly, nullsthe earth magnetic field.

C. Circuit Electronics

The circuit electronics operates as digital, as much as pos-sible. Nevertheless, some very good low noise analog circuitsare needed for proper operation. The system is actually builtto support three magnetic sensor heads for triaxial operation.Fig. 2 shows a block diagram of the circuitry for a singlechannel. Other channels are identical. The components areeither repeated or shared, as appropriate, for the two otherchannels.

The amorphous wire is driven, as any other fundamental-mode orthogonal sensor, by a sinusoidal ac waveform super-posed on a dc bias current. We use a 96 kHz drive frequencywith a peak amplitude of 40 mA over a 50 mA dc bias.Both the ac waveform and the dc bias are synthesized byprogrammable logic driving a fast 12 bit digital-to-analogconverter.

The sensing coil is parallel tuned at 96 kHz and followedby a very low noise preamplifier and a 96 kHz low-pass filter,which then feeds a 24 bit analog-to-digital converter samplingat 192 kHz and synchronized to the same clock as the driverdigital-to-analog converter.

A floating point digital signal processor chip takes thedata from the analog-to-digital converter, synchronously

Fig. 2. Block diagram of the magnetometer electronics.

demodulates it, applies a low-pass filter and decimates to a1 kHz output sample rate, with 400 Hz effective bandwidth.

A second high-accuracy digital-to-analog converter drivesthe earth field canceling coil. A very well-designed output filterwas necessary to reduce the output noise from this converterto under the sensor intrinsic noise.

Data from the DSP are recorded on a Compact Flash card;1 GB provides for over 12 h of recording. The system iscompleted by a serial link to a laptop computer used as theoperator’s console and by a GPS receiver to which the internalclocks are synchronized. This allows real-time comparison ofdata from several independent units.

III. RESULTS AND DISCUSSION

A. Performance

Noise measurements were performed in a shielded canhaving three layers of μ metal. All the measurements wereperformed in a closed loop and the calibration line and aknown amplitude of 50 Hz field was applied by a solenoidpositioned around the sensor. Acquired data were processedusing Quantec proprietary magnetotelluric processing tools.For the purpose of these experiments, we were only extractingthe power spectrum of the digitized signal.

As currently configured, the sensitivity of the magnetometeris approximately 6900 units/nT. With digital processing, thiscan be changed at will. The dynamic range, without nulling isapproximately 1000 nT. It is limited by the saturation of theanalog electronics.

The noise floor is more significant. Individual sensor headsconsistently provide better than 2 pT/Hz1/2 noise performanceand typically 1.5 pT/Hz1/2 at 1 Hz and higher frequencies.

B. Multiple Sensor Heads

Individual sensor heads may be stacked to improve thenoise. Assuming that the noise from each head is not coherent,the noise from multiple units is the square root of the sumof the noise from each individual unit, while the signal addslinearly. This provides a signal-to-noise improvement of 21/2

for each doubling of the numbers of units.A quadruple unit, summing the output of four individual

sensors, each with its own preamplifier works quite well.

BAZINET et al.: LOW-NOISE FMOF MAGNETOMETER 6500103

Fig. 3. Noise spectra of the quadruple sensor head.

Fig. 4. Noise spectra of the single sensor head.

Signals from the four preamplifiers are summed before thelow-pass filter shown in Fig. 2.

Fig. 3 shows the noise spectra of the quadruple unit.The spike at 50 Hz is our 1570 pT calibration line. Thehigh-frequency noise floor is approximately 0.8 pT/Hz1/2.

C. Noise Origin

Fig. 4 shows that the noise floor does not change when theexcitation current through the amorphous wire is shut down.This shows that the observed noise is mainly the thermal noiseof the sensing coil—amorphous wire core assembly, with mostof it contributed by the eddy currents in the amorphous wires.The FMOF mechanism is apparently noise free, at least at thislevel.

Electronic noise can be kept lower than the contribution ofthe sensor itself but this needs very careful design of the analogparts of the system. The preamplifier input noise was foundto be of one order of magnitude less than the sensor noise,but we did have problems, in particular, with noise injectionfrom the earth field nulling circuit. The output noise of theoperational amplifier used to buffer out the digital-to-analog

converter was found to be the source of the problem. Both alower output noise operational amplifier and an output filterwere necessary to bring down that noise.

As the signal levels are extremely low, power supplycoupling may also be an issue. Massive supply filtering wasneeded to prevent 1 Hz pulses from the GPS to corrupt thesignal.

IV. CONCLUSION

With careful design, it was found possible to improvethe noise of a FMOF gate magnetometer to values below1 pT/Hz1/2. It looks like the noise is mainly limited bythe eddy-current losses in the amorphous core material. Thisraises the possibility that the noise can be reduced furtherby the use of a smaller diameter amorphous wire activeelement.

ACKNOWLEDGMENT

This work was supported by Quantec Geoscience Ltd. Theteam at the Instituto de Ciencia de Materiales de Madrid, undercontract to Quantec, developed and produced the amorphouswire material.

REFERENCES

[1] (2013, Feb. 21). Field Sensors [Online]. Available: http://www.phoenix-geophysics.com/products/sensors/

[2] (2013, Feb. 21). Three-Axis Fluxgate Magnetometers andMagnetic Field Sensors [Online]. Available: http://www.bartington.com/three-axis-fluxgate-magnetometers.html

[3] B. Dufay, S. Saez, C. Dolabdjian, A. Yelon, and D. Ménard, “Devel-opment of a high sensitivity giant magneto-impedance magnetometer:Comparison with a commercial flux-gate,” IEEE Trans. Magn., vol. 49,no. 1, pp. 85–88, Jan. 2013.

[4] L. Ding, S. Nabily, S. Saez, J. Gieraltowski, and C. Dolabdjian,“Investigation of giant magnetoimpedance magnetic noise comparison,”Sensor Lett., vol. 5, no. 1, pp. 171–175, 2007.

[5] M. Malátek, P. Ripka, and L. Kraus, “Temperature offset drift ofGMI sensors,” Sens. Actuators A, Phys., vol. 147, no. 2, pp. 415–418,Oct. 2008.

[6] I. Sasada, “Orthogonal fluxgate mechanism operated with dc biasedexcitation,” J. Appl. Phys., vol. 91, no. 10, pp. 7789–7791,May 2002.

[7] E. Paperno, “Suppression of magnetic noise in the fundamental-mode orthogonal fluxgate,” Sens. Actuators A, Phys., vol. 116, no. 3,pp. 405–409, Oct. 2004.

[8] M. Butta and I. Sasada, “Orthogonal fluxgate with annealed wire core,”IEEE Trans. Magn., vol. 49, no. 1, pp. 62–65, Jan. 2013.

[9] M. Butta and I. Sasada, “Effect of terminations in magnetic wire onthe noise of orthogonal fluxgate operated in fundamental mode,” IEEETrans. Magn., vol. 48, no. 4, pp. 1477–1480, Apr. 2012.

[10] M. Butta, S. Yamashita, and I. Sasada, “Reduction of noise in funda-mental mode orthogonal fluxgates by optimization of excitation current,”IEEE Trans. Magn., vol. 47, no. 10, pp. 3748–3751, Oct. 2011.

[11] R. Bazinet, “Very low noise magnetometer,” U.S. Patent 8 237 438,Aug. 7, 2012.

[12] I. Ohnaka, “Melt-spinning into liquid cooling medium,” Int. J. RapidSolidificat., vol. 85, no. l, pp. 219–236, 1985.

[13] M. Vazquez, “Soft magnetic wires,” Phys. B, Condensed Matter,vol. 299, nos. 3–4, pp. 302–313, Jun. 2001.

[14] I. Sasada and H. Kashima, “Simple design for orthogonal fluxgatemagnetometer in fundamental mode,” J. Magn. Soc. Jpn., vol. 33, no. 2pp. 43–45, 2009.