Sensors and Actuators A, 37-38 (1993) 417-421 417

Race-track fluxgate sensors

Pave1 Rlpka Czech Technrcal Unwerszty, Electroteehnrcal Faculty, Department of Measuremeni, 166 27 Prague 6 (Czech Repubhc)

Abstract

A fluxgate sensor smtable for the vector measurement of small d c magnetic fields IS descrlbed The sensor core of oval (race-track) shape IS etched from 35 pm amorphous tape After symmetnzatlon, 1 nT sensor offset and 10 pT rms noise are achieved The tuned mode for sensor excltatlon utihzmg non-hnear resonance 1s discussed The optimum pick-up co11 length 1s derived 9 mm diameter, 3 mm thick sandwch-type rmg-core sensors made by the same technology are sultable for portable magnetometers with a resolution better than 1 nT The methods for parametnc amphficatlon of the sensor output voltage, current output and detectlon of higher harmonics are drscussed

Introduction

Fluxgates are the most suitable sensors for the mea- surement of magnetic fields m the range 0 1 nT to 1 mT Their noise level 1s lower than that of magnetoreslstors, senuconductor, magnetostnctlve, optical and other types of solid-state room-temperature d c magnetlc- field vector sensors [l] After the traditional open-core type, the rmg-core 1s the most widely used sensor geometry In addition, other sensor constructions, such as a thin-film sensor with flat cod [2] or an oval (race-track) tape-wound sensor have appeared [3] Sen- sors made from etched sheets m oval (race-track) shape have been reported m ref 4 Although the sensitivity was increased due to the lower demagnetization, the spurious output signal (which may be suppressed by core rotation m the case of rmg-core sensors) was very high, causing mterfacmg problems, large offset and noise of about 20-40 pT rms (64 mHz- 10 Hz) [s]

Race-track core sensors

The sensor core consists of 12 sheets of oval (race- track) shape etched from 35 pm thck amorphous tape made from low-magnetostnctlon cobalt-based amor- phous alloy Vltrokov 8116 (Co6,Fe4Cr7Si8B14) pro- duced by the Institute of Physics SAV, Bratislava, or Vltrovac 6025 (Vacuumschmelze Hanau) The tapes were used as received, even urlthout relaxation anneal- mg The core length 1s 70 mm, the width 12 mm and the track urldth 2 mm

The present effort 1s to lower the sensor noise and offset The spurious feedthrough signal was shown to be caused mostly by sensor non-symmetry due to the

0924-4247/93/$6 00

vanatlon of the thckness of the amorphous core tape Ths may be lowered by etchmg the cores from the (more homogeneous) central part of the tape and proper matchmg of mdlvldual core sheets and balanced by a small piece of the same matenal The remammg lack of symmetry 1s caused by mhomogenelty of the exatatlon field due to the wmdmg geometry and non- umformlty of the magnetic properties A further sym- metnzation may be performed by adjusting the excltatlon and pick-up cods Although this process may reduce the asymmetry by three orders of magmtude, it 1s not practical for the mass-production process A sensor core geometry that allows the balance to be adjusted snnply by moving the pick-up co11 was pro- posed

Using the described balancmg procedures for the sensor core of amorphous Co-based Vltrokov 8116 matenal, the sensor feedthrough was lowered to a value corresponding to a measured field of units of pT The sensor offset was reduced below 1 nT and the noise level to well below 10 pT rms m the frequency range 50 mHz-IO Hz, which corresponds to about 50 pT peak-peak, as illustrated m Fig 1

The sensors were measured in two basic modes (1) the (traditional) voltage-output mode, (11) the current-output (short-circuited) mode mtro-

duced by Pnmdahl et al [6] The excitation current necessary for the best perfor-

mance was as high as 1 5 A (peak-peak) for 400 turns of the excitation wmdlng For this purpose the parallel resonant arcmt described by Acuna [7] was used Tlus clrcmt allows the p-p excitation current I,, to be increased to more than ten tnnes the amphtude of the smewave or tnangle current I from the current source supplying the parallel combmation of the excltatlon cod

@ 1993 - Elsevler Sequoia All nghts reserved

418

50

PT /diV

I I I I I I I I I I I

start 0 s stop 31 969 s

Rg 1 Output of the RTVK2 race-track fluxgate sensor The cahbratlon step IS 200 pT

linear mode - tuned mods

fact tranriant

s - tuned mode N 0

fast transient

(4 Rg 2 Charactenstics of the excltatlon cod tuned by C,, = 0 5 pF and driven by the current source (a) Basic diagram of the tuned sensor wth voltage output The exatation and pick-up (sensmg) cods have I$ and N2 turns, respectively C, tunes the sensor output (b) Exc&&on current ID( p-p vs excitation voltage V (c) Second harmomc component (with re- spect to the first harmomc)

and the tuning capacitor Figure 2 shows the non-lmear behavlour of the arcmt on mcreasmg the current- source amphtude from zero to the cntlcal current IA, the clrcmt fips from the lmear (non-saturated) mode to the tuned mode, m which the energy IS exchanged between the exutation co11 and the tunmg capacitor m each period The cod IS deeply saturated by large cur- rent peaks and the cn-cmt stdl works m the non-linear

Rg 3 Excitatton current parameters for the workmg pomt A from Fig 2 (border of the stable tuned mode) as a function of the tuning capacitor C,, (a) peak-peak amplitude, (b) second har- monic chstortlon

resonant mode even when the supply current Z IS de- creased, until the point B, from which the circmt flips back mto the lmear mode The values m Fig 2 were measured for the tuning capacitor C,, = 0 5 PF Figure 3 shows the excltatlon parameters (current amplitude and spectral punty) for the cntlcal point A as a func- tion of C,, The part of the workmg curve between points A and B IS quasistable as It is not autostartmg, I e , it 1s not automatically reached after swltchmg on the power (without special starting clrcmts) Such an operatmg pomt IS advantageous, as the second har- momc dlstortlon of the excltatlon current IS lowered due to the filtenng effect of the resonant clrcmt

The current sensltlvlty was 0 2-O 5 pA/nT, depend- mg on the sensor cross se&on The pick-up cod geome- try optilzatlon was studled Figure 4 shows the voltage sensltivlty for Merent parts of the sensor A short pick-up co11 having 100 turns was displaced from the core centre and the sensltlvlty was measured as a

419

Fig 4 Voltage sensltlvlty as a function of the posltlon of the short p&up co11 The lon@udmal displacement IS measured from the of the 70 mm long sensor core to the axis of the 10 tnm long sensmg cool wtth 100 turns

axis

function of the co11 position As the drop m sensltlvtty m the vlcmrty of the core ends IS progressive, the optimum pick-up coil length may be estnnated as about 40 mm (the core length being 70 mm) Such a short sensing coil may be used only in the case when the functions of the feedback and sensing cods are sepa- rated

Mmmture sandwrch-type sensors Beside the described low-noise race-track sensors,

numature sandwich-type low-cost fluxgates were also developed Magnetic sheets from the amorphous mate- nal of the rmg shape were glued between two plastic covers of the same shape and directly wound by the exntatlon co11 Two sensor sues were tested

(I) a mmtature sensor with 9 mm core diameter and 3 mm thickness,

(u) a low-field sensor of 21 mm diameter and 7 mm thick

The sensors were used m both the current-output and voltage-output modes

T’he senslhvlty m the voltage mode was increased by tuning the sensing (pick-up) co11 by the parallel capacl- tor The (non-linear) mechanism of parametnc amphfi- cation has already been described analytically m ref 8 and the stab&y was mvestlgated usmg numerical mod- els [9] Figure 5 shows the tunmg curve for the large sensor the maximum sensihvlty for fourth harmomcs IS reached for C = 11 nF (amplification factor 13 5), for the same capacitor value a moderate local sensltlvlty maxmmm appears at the second harmomc The clrcmt 1s unstable for C > 40 nF Figure 5 illustrates the large

dependence of the sensitlvlty on the arcwt parameters m the regon of high amphfication Even m the case of the feedback-type magnetometer, when the sensor sense- tlvlty vanatlons are suppressed by the large d c amphfi- cation m the loop, substantial problems with the offset changes caused by the phase shifts remam Degradation of the sensor noise caused by parametnc ampltication was documented m ref 10 The small tuning capacttor can remove the ‘rmgmg’ at the sensor output, which IS caused by the parametnc amptication of the hgher harmomcs by the parasitic capacitances of the pick-up cod, cables and input amphfier, the waveforms are shown m Fig 6 The output of the sensor tuned to the fourth harmoruc IS shown m Fig 7 for merent field values

Classical fluxgate magnetometers evaluate the output slgnal at the second harmomc of the excitation fre- quency The relative senntivlty on the m&mdual even harmonics depends on the core-matenal charactens- tics and excltatlon parameters The described sensor shows an eqmvalent untuned sensitlvlty of 14mV/mT for both second and fourth harmomcs The mforma- tion contamed at the higher harrnomcs may be ex- ploited usmg the pulse-detection method introduced by Pnmdahl [6]

The current mode sunphfies the interface electromc cmzmts A decrease of the pick-up coil area increases the current sensihvlty, as derived m ref 5 The short- cmzmted mode allows only 75 turns of the pick-up co11 (mstead of the usual 4000) to be used

Because of the small amount of active magnetic matenal, the power consumpfion for the whole one-us

420

2501

*H**( 2nd harmonics u - - 4th harmomcs

0 , I I I I I I I #r-i- 0 5 10 15 20 25 30 35 40

Tuning capacitor [nF]

Rg 5 Tunmg curve of the rmg-core sensor voltage sensltwty on second and fourth harmomcs vs tumng capacitor value

Fig 6 Waveforms of the nng-core sensor top trace, excttation current, vetical scale 400 mA/dw, bottom trace, untuned sensor output voltage for H = 0, vertd de 500 mV/dw

magnetometer may be below 100 mW A portable mstru- ment usmg the described sensor is under development The expected accuracy IS 0 5% FS, the resolution 0 1 nT, the noise level 100 pT p-p for a 21 mm diameter sensor and 500 pT p-p for a mmiature 9 mm dla,meter sensor

Conclusious

The highly sensitive race-track sensors described m the first part of the present paper are suitable for laboratory apphcatlons, where the large sensor size (70 mm length) causes no problems A smgle sensor can

Fig 7 Output voltage of the sensor tuned to the fourth harmomc for H = 0 (top trace) and H = 10 mT Vertical scale 500 mV/dlv

measure the magnetic field m one directIon, a feedback system using a large temperature-compensated or ther- mostatted co11 system IS necessary to u&e the sensor 10 pT rms noise fully A typical apphcatlon might be stations momtormg vanatlons of the Earth’s magnetic field The optimum pick-up cod length IS much lower than the sensor core length The described amorphous-core sensors have parameters comparable to those of the best permalloy-core ones

Small-sized sandwch sensors made by the same tech- nology of etchmg from thm sheets were developed for low-power magnetometers The achieved resolution IS better than 1 nT, which IS sticlent for most of the portable apphcatlons

421

Acknowledgements

Part of the expernnental work (Figs 1 and 2) was performed durmg the author’s stay at the Department of Electrophyslcs, Danish Techmcal Umverslty The author thanks 0 V Nielsen and J R Petersen of this Department for suggestlons and support

References

1 P Rlpka, Review of fluxgate sensors, Sensors and Actuators A, 33 (1992) 129-141

2 T Satz, Fluxgate sensor m planar nucroteclmology, Sensors and Actuators, A21 -A23 (1990) 799-802

3 D I Gordon and R E Brown, Recent advances m flux-

gate magnetometry, IEEE Tram Magn , MAG-8 (1972) 76- 82

4 P Ripka, Improved fluxgate for compasses and posttlon sen- sors, J Magn Magn , 83 (1990) 543-544

5 F Pnmdahl, P kpka, J R Petersen and 0 V Nielsen, The shortclrcmted fluxgate sensltw~ty parameters, Meas Scz Technol, 2 (1991) 1039-1045

6 F Pnmdahl, J R Petersen, C Olm and K H Andersen, The short-nrcmted fluxgate output current, J Phys Set E Set Instrum, 22 (1989) 349-354

7 M Acuna, Fluxgate sensors for outer planet exploratmn, IEEE Tram Magn , MAG-IO (1974) 519-523

8 M A Player, Parametnc amphficatlon m fluxgate sensors, J Phys D Appl Phys , 21 (1988) 1473-1480

9 R D Russel, B B Narod and F Kolar, Charactenstlcs of the capacltwely loaded fluxgate sensor, IEEE Trms Magn , MAG-19(1983) 126-130

10 F Pnmdahl and P A Jensen, NOW m the tuned fluxgate, J Phys E Scr Instrum, 20 (1987) 637-642