Calculation of Fluxgate Sensor Parameters

İlker Yağlıdere1,2 and Ece Olcay Güneş1

1
Electronics and Communication Engineering Department, Istanbul Technical University, Istanbul, 34469, Turkey
[email protected], [email protected]
2
ASELSAN Electronics Inc., 296. Street, 06370, Ankara, Turkey

Abstract Since the most popular geometries suffer from high

demagnetization effect which limits the sensitivity and increases
In this work, we present two ring-core fluxgate sensor the sensor noise, calculating demagnetization factor accurately
designs to investigate the accuracy of parameter calculations becomes more of an issue. Many empirical formulas [13, 14]
such as inductance and resistance of the windings, and an analytical formula [15] have been proposed for the ring-
demagnetization factor and apparent permeability. All these core, but none of them succeeds for a wide range of dimensions.
parameters as well as relative permeability are measured and For example, the proposed formula in [13] was validated for
some simulations are performed through a FEM (Finite only four cores having identical diameter and height values. If a
Element Method) analysis tool. Measurement results are fluxgate is to be designed with a commercial strip wound core,
compared with the calculated theoretical values in addition relatively thick cores in the market are the only alternative. The
to FEM analysis results. In the first design, a crystalline core fluxgate output voltage formula and the effect of
stacked from circular permalloy laminations is used while an demagnetization factor on the output voltage are given in [16].
amorphous core wound from Vitrovac 6025Z strip is The first goal of this study is measuring the demagnetization
preferred for the second design. Demagnetization factors are factors of the two ring-core sensors designed and comparing
calculated by using well-known empirical and analytical them with the values calculated by using three well-known
formulas in the literature. The results point out that a new formulas. The second goal is performing some finite element
demagnetization factor formula is necessary because the analyses to investigate the formulas. Finally, we aim to examine
three formulas used miscalculate for the designed sensors. the consistency of current formulas. Physical design details are
Moreover, calculating the demagnetization factor by using described in Section II. Sections III and IV give the calculation
finite element analysis method is found to be reliable. and measurement results while Section V concludes with the
comparison of the results.
1. Introduction
2. Physical Design and Dimensions
Fluxgate sensors have come a long way since Aschenbrenner
and Goubau started to construct the first fluxgate magnetometer A fluxgate core stacked from four circular permalloy
in 1928 and published their paper in 1936 [1]. Sensitive sensors laminations is used in the first design (sensor-1) while a core
were designed for ship and submarine detection during the wound from 45 turns of 19 µm thick Vitrovac 6025Z strip is
Second World War [2]. After the war, they have been widely used in the second design (sensor-2). The physical dimensions
used as compasses in shipping and aviation as well as in rocket of the sensors are tabulated in Table 1 and some mechanical
and missile systems [3]. The Soviet research satellite Sputnik 3 details are illustrated in Fig. 1.
launched in 1958 was the first space application of fluxgates [4].
Since 1980s, fluxgates have been used for observing the Table 1. Physical dimensions of the sensors
variations in Earth’s magnetic field [5]. Starting with 1990s,
micro-fluxgates were developed, but they suffer from low Sensor-1 Sensor-2
Dimension Symbol
sensitivity and high noise [6]. Sweden's second microsatellite mm mm
Astrid-2 with a digital fluxgate payload was launched in 1998 to lamination thickness tlamination 0.3 -
explore the electric and magnetic fields in the upper ionosphere strip thickness tstrip - 0.019
[7]. In 1999, Denmark's first satellite Ørsted was launched for outer diameter (bare) ODbare 12.17 10.00
mapping the earth’s field [8]. NASA used fluxgates in THEMIS inner diameter (bare) IDbare 8.81 8.00
satellite in 2007 [9]. By 2015, NASA started the height (bare) hbare 1.20 3.80
“Magnetospheric Multiscale Mission” comprising four outer diameter (boxed) ODboxed 13.65 11.60
identically instrumented spacecrafts equipped with ASIC inner diameter (boxed) IDboxed 7.70 6.50
fluxgates, to study the microphysics of magnetic reconnection height (boxed) hboxed 2.80 5.10
by using Earth's magnetosphere as a laboratory [10]. One of the excitation coil wire dia. dexc 0.20 0.20
recent applications is precise electric current sensing. Texas pickup coil wire dia. dpickup 0.40 0.14
Instruments announced the DRV421 and DRV425 CMOS pickup coil length lpickup 15.10 13.38
integrated fluxgate sensors specified for operation over the pickup coil width wpickup 20.00 16.30
extended industrial temperature range of -40°C to +125°C in pickup coil height hpickup 5.0 6.5
2015 and they can replace the classical Hall probes [11, 12].
 Fig. 1. Core dimensions

Support material used in sensor-1 is shown in Fig. 2. In this

configuration, the toroidal core close fits to the support material
and the structure is mechanically stable, which makes it ideal for
commercial sensors.
Sensor-2 is produced with a different support geometry,
which allows inserting and removing the core easily through the
support. When using such a configuration, it is easy to measure
the pickup coil inductance with and without the core; therefore,
it is appropriate for scientific studies. Sensor-2 is shown in Fig.
2 and some parameters of the designed sensors are tabulated in Fig. 2. Support material for (a) sensor-1 and (b) sensor-2
Table 2.

3. Relative Permeability Measurements π (ODbare − IDbare )

l path = , (2)
ODbare
ln( )
Relative permeability measurements are performed by IDbare
measuring the inductance of the excitation coils (Lexc) at near
DC (100Hz) with a Keysight U1733C handheld LCR meter. The where bare dimensions and Nexc represent the dimensions of the
permeability values are then calculated from the measured core without the plastic case and excitation coil turn number,
inductance values. By assuming that a non-significant respectively. The effective path length (lpath) is not calculated by
demagnetizing field occurs for the circular excitation field along directly averaging the inner and outer circumferences. Because
the toroidal core and it can be neglected, the calculated effective the flux lines inside the toroidal core concentrate in the shorter
permeability (µeff) values are accepted as relative permeability paths, the effective path length is somewhat smaller than the
(µr) values. The following general equations are used for average circumference. Although it has a minor effect on the
excitation coil inductance and path length calculations: path length for thin ring-cores, it is not neglected in calculations.
Measured relative permeability values are given in Table 2.
μ 0 μ r N exc 2 Acore
Lexc = , (1)
l path

Table 2. Some important parameters of the designed sensors

Parameter Symbol Unit Sensor-1 Sensor-2

material - - permalloy 6025Z
path lengtha lpath mm 32.67 28.16
cross sectional area Acore mm2 2.016 3.20
resistance of unit length σexc mΩ/m 538.3 538.3
winding number Nexc turns 102 86
coil inductanceb Lexc mH 15.82 26.71
relative permeabilityc µr - 19610 25288
cross sectional area Apickup mm2 100.00 105.95
resistance of unit length σpickup mΩ/m 168.9 1079
winding number Npickup turns 36 78
acalculated by using (2).
bmeasured.
ccalculated from the measured coil inductance.
 4. Calculations and Measurement Results in which knon-uniformity is a magnetic field non-uniformity
correction factor (also known as Nagaoka's coefficient).
Coil resistances are calculated by multiplying the wire length Nagaoka's coefficient is generally between 0.7 and 0.95 for
(lwire, in mm) with the wire resistivity per unit length (σ). For pickup coils and always smaller than 1. The constant can be
excitation coils, wire length is calculated as calculated precisely by using Lundin's handbook formula [17]
for coils with a circular cross section. Some approximations can
lwire = N excl1−turn + 200 be used for coils with a rectangular cross section. In this study,
knon-uniformity values are calculated by using the measured Lair
= N exc ( ODboxed − IDboxed ) + 2hboxed + 4d exc  + 200. (3) values. Inductance of the pickup coil with core is calculated as
follows:
For pickup coils, wire length is calculated as
μ0 N 2
lwire = N pickupl1−turn + 200 L pickup = k non − uniformity
l pickup
( Apickup + 2(μa − 1) Acore ). (8)

 ( )
= N pickup  2 w pickup + h pickup + 4d pickup  + 200.

(4)
Demagnetization factor is calculated using the following
well-known formulas in which t, d and A denote core thickness,
The term 200 in (3) and (4) is the length of wire outside the coil.
average core diameter and cross sectional core area, respectively
Apparent permeability is calculated by using the measured
[13-15]:
inductance values:
t
L pickup − Lair Apickup Dglobal = 0.223 , (9)
μa = + 1. (5) d
Lair 2 Acore
A
When µr and µa are known, then global demagnetization factor Dlocal = 1.83 , (10)
Dglobal can be found as follows [13]: d2

1   4(d + t )   A
μr − μ a 2
Dglobal = . (6) Dlocal = ln   − 1 . (11)
μ a ( μr − 1) π   h    (d + t )2

Pickup coil inductance without the core is calculated as
Measured values in addition to the calculated values by using
the given formulas and FEM analysis software are tabulated in
μ0 N 2 Apickup Table 3. Demagnetization factor, apparent permeability and the
Lair = knon −uniformity , (7)
l pickup pickup coil inductance are the parameters that are difficult to be
calculated as seen in Table 3.

Table 3. Comparison of the measured and calculated values for the sensors designed

Sensor-1
Sensor-1 Error Sensor-2 Sensor-2 Error
Parameter Unit
Measured
Calculated % Measured Calculated %
Rexc Ω 0.82 0.79 -3.66 0.86 0.85 -1.16
Rpickup Ω 0.27 0.26 -3.70 4.00 4.10 +2.50
Lair mH 9.1 9.1 0.00 56.0 56.0 0.00
knon-uniformity - 0.844 - - 0.925 - -
19.00a +5.55 214.5a +117
14.20b -21.1 76.00b -23.2
Lpickup µH 18.00 99.0
15.34c -14.8 96.33c -2.70
d
17.07 -5.18 94.62d -4.43
0.0357a -9.69 0.0209a -71.4
0.0671b +69.7 0.1446b +98.4
Dglobal - 0.0395 0.0729
0.0556c +40.5 0.0773c +6.11
d 0.0805d
0.0440 +11.3 +10.4
27.96a +10.7 47.84a +249
14.89b -41.0 6.91b -49.6
µa - 25.26 13.71
17.98c -28.8 12.92c -5.76
d
22.70 -10.1 12.42d -9.44
aCalculated by using the empirical formula D
global=0.223(t/d).
bCalculated by using the empirical formula D 2
local=1.83(A/d ) and (12).
cCalculated by using the analytical formula given in (11) and (12).
dFEM analysis results.
 The following relation can be used for transition between 7. References
local and global demagnetization factors in a ring-core fluxgate
[14]: [1] H. Aschenbrenner and G. Goubau, “Eine anordnung zur
registrierung rascher magnetischer störungen”, Hochfreq.
1 Electroakustik, vol. 47, no. 6, pp. 177-181, Jun., 1936.
Dglobal = 2 Dlocal + . (12)
μr − 1 [2] H. How and C. Vittoria, “Generation of high-order
harmonics in insulator magnetic fluxgate sensor cores”,
IEEE Trans. Magn., vol. 37, no. 4, pp. 2448-2450, Jul.,
For high values of µr, the second term can be neglected.
2001.
Pickup and excitation coil resistances as well as pickup coil
[3] A. Grosz, M. Janosek, Haji-Sheikh and S. C.
inductances without the core are calculated precisely. However,
Mukhopadhyay, eds., “High Sensitivity Magnetometers”,
calculated demagnetization factor and related parameter (e.g.
Springer, Switzerland, 2017.
pickup coil inductance and apparent permeability) values are
[4] K. Shirae, “Noise in amorphous magnetic materials”, IEEE
quite different from the measured values. These differences are
Trans. Magn., vol. 20, no. 5, pp. 1299-1301, Sep., 1984.
all caused by the errors in demagnetization factor calculations.
[5] R. Langel, et al., “The MAGSAT mission”, Geophys. Res.
For both sensors designed, (9) underestimates the
Lett., vol. 9, no. 4, pp. 243-245, Apr., 1982.
demagnetization factor, whereas (10) and (11) overestimate the
[6] Y. Liu, et al., “Improved performance of the micro planar
factor. In contrast to the three formulas, FEM analyses
double-axis fluxgate sensors with different magnetic core
overestimate with an error about only 10% as shown in Table 3.
materials and structures”, Microsyst. Technol., vol. 22, no.
9, pp. 2341-2347, Sep., 2016.
5. Discussion and Conclusions [7] E. B. Pedersen, et al., “Digital fluxgate magnetometer for
the Astrid-2 satellite”, Meas. Sci. Technol., vol. 10, no. 11,
Two ring-core fluxgate sensors made of amorphous and
pp. N124, Nov., 1999.
permalloy materials are presented. Demagnetization factors are
[8] O. V. Nielsen, et al., “Development, construction and
measured and compared with the values calculated by using the
analysis of the 'OErsted' fluxgate magnetometer”, Meas. Sci.
three popular formulas and quite different demagnetization
Technol., vol. 6, no. 8, pp. 1099, Aug., 1995.
factor values are observed. The measurement results are
[9] H. U. Auster, et al., “The THEMIS fluxgate magnetometer”
validated by FEM analyses. It is concluded that the two most
in “The THEMIS Mission”, Springer, New York, 2009.
widely used demagnetization factor formulas proposed by
[10] C. T. Russell, et al., “The magnetospheric multiscale
Primdahl et al. in 1989 and 2002 as well as the analytical
magnetometers”, Space Sci. Rev., vol. 199, no. 1-4, pp.
formula proposed by M. De Graef et al. in 2006 are not valid for
189-256, Mar., 2016.
the tested dimensions. Great care should be taken while using
[11] M. F. Snoeij, et al., “An integrated fluxgate magnetometer
these empirical and analytical formulas by keeping in mind that
for use in closed-loop/open-loop isolated current sensing”
they can lead to error levels above 90% and 40%, respectively.
in European Solid-State Circuits Conference, Graz, Austria,
The results show that finite element analysis tools offer the most
2015, pp. 263-266.
reliable solutions.
[12] M. F. Snoeij, et al., “Integrated Fluxgate Magnetometer for
Fluxgate demagnetization factor should be calculated
Use in Isolated Current Sensing”, IEEE J. Solid-State
accurately before production because it is the most important
Circuits, vol. 51, no. 7, pp. 1684-1694, Jul., 2016.
design parameter, which determines both sensor noise and
[13] F. Primdahl, et al., “Demagnetising factor and noise in the
sensitivity. Despite being over 80 years since Aschenbrenner
fluxgate ring-core sensor”, J. Phys. E, vol. 22, no. 12, pp.
and Goubau have presented the very first fluxgate
1004, Dec., 1989.
magnetometer, a reliable demagnetization factor formula for
[14] F. Primdahl, et al., “The fluxgate ring-core internal field”,
ring-core fluxgate is still absent. We strongly suggest using
Meas. Sci. Technol., vol. 13, no. 8, pp. 1248, Jul., 2002.
finite element analysis tools for calculating the demagnetization
[15] M. De Graef and M. Beleggia, “The fluxgate ring-core
factor and related parameters instead of using the current
demagnetizing field”, J. Magn. Magn. Mater., vol. 305, no.
demagnetization factor formulas. A second study is planned to
2, pp. 403-409, Oct., 2006.
further analyze the demagnetization phenomenon and propose a
[16] İ. Yağlıdere, “A Sensitive Fluxgate Magnetic Sensor
novel method for calculating the demagnetization factor.
Design”, M.S. thesis, Elec. and Com. Eng. Dept., Istanbul
Tech. Univ., Istanbul, TR, 2010.
6. Acknowledgment [17] R. Lundin, “A handbook formula for the inductance of a
single-layer circular coil”, Proc. of the IEEE, vol. 73, no. 9,
We wish to thank Aktif Neser Elektronik Ltd. Sti. and CST - pp. 1428-1429, Sep., 1985.
Computer Simulation Technology AG for supporting this
research by supplying a demo license for CST EM STUDIO.