Revista Mexicana de Física

ISSN: 0035-001X
rmf@[Link]
Sociedad Mexicana de Física A.C.
México

Rice, J.
Vector Analysis of Magnetic Fields for Hall Sensor Design
Revista Mexicana de Física, vol. 58, núm. 2, diciembre, 2012, pp. 241-244
Sociedad Mexicana de Física A.C.
Distrito Federal, México

Available in: [Link]

How to cite
Complete issue
Scientific Information System
More information about this article Network of Scientific Journals from Latin America, the Caribbean, Spain and Portugal
Journal's homepage in [Link] Non-profit academic project, developed under the open access initiative
 Revista Mexicana de Fı́sica S 58 (2) 241–244 DICIEMBRE 2012

Vector Analysis of Magnetic Fields for Hall Sensor Design

J. Rice
Allegro MicroSystems Inc. Lafayette, Colorado, USA,
jrice@[Link]
Recibido el 25 de junio de 2010; aceptado el 9 de mayo de 2011

Hall sensors are commonly employed in conjunction with permanent magnets to determine vector components of motion; i.e. position,
direction and speed. In this paper, a vector analysis method of relating scalar magnetic flux reported by a Hall sensor to vector components
of magnetic field is offered as a tool to validate, compliment, or as a substitute to field mapping and Finite Element Methods.

Keywords: Vector analysis; magnetic field expansion; potential theory; Hall sensor

Sensores de efecto Hall se emplean en conjunto con imanes permanentes para determinar los componentes de vector de movimiento, es decir,
posición, dirección y velocidad. En este artı́culo, un método de análisis de vectores para describir el flujo escalar magnético que reporta
un sensor efecto Hall relacionado con las componentes de vector del campo magnético se introduce como una herramienta de diseño para
validar, complementar o para sustituir la elaboración de un mapa de campo y los métodos de elemento finito.

Descriptores: Análisis vectorial; expansión de campo magnético; teorı́a de potencial; sensor Hall.

PACS: [Link]; [Link]; [Link]

1. Decomposition of dipole magnetic field into 2. Construction of bar magnet vector field
cartesian coordinate vector coefficients
The vector components representing a bar magnet with
The magnetic field for a dipole is expressed by the inductance dimensions2xb , 2yb , 2zb , and residual inductanceBr are de-
vector Bdipole in Eq. (1). [1] termined by convolution of the unit dipole vector components
m over the bar magnet volume bounded by ±xb , ±yb , ±zb ;
Bdipole = Cm 3 [3 (m̂ · r̂) r̂ − m̂] (1)
r Eq. (6), for each of the vector coefficients of Eqs. (3), (4)
Wherer is a vector representing the location of a point rel- and (5).
ative to the center of the dipole, and m̂ is a unit vector
representing the direction between the poles. The constant Zzb Zyb Zxb
Cm = µ0 /4π. bar magnet
B∈(x̂,ŷ,ẑ) (x, y, z) = Br B̂∈(x̂,ŷ,ẑ)
The vector Bdipole is decomposed into the orthogonal co-
ordinate system in Eq. (2), where the coefficientsBx̂ , Bŷ and −zb −yb −xb

Bẑ represent the components directed along the Cartesian co- (x − ξ, y − ψ, z − ζ)

dipole
dξdψdζ (6)
ordinate unit vectors.
Bdipole = Bx̂ x̂ + Bŷ ŷ + Bẑ ẑ (2) The general proposition of Eq. (6) is a remarkable out-
come of Potential Theory, and the next few pages demon-
Expressions for the coefficientsBx̂ , Bŷ and Bẑ are
strate the value of Eq. (6) in Hall sensor applications.
determined by introducing into equation (1); location
r = (x − x0 ) x̂ + (y − y0 ) ŷ + (z − z0 ) ẑ , with the simplify-
ing assumption that the center of the coordinate system is the
center of the dipole, x0 = y0 = z0 = 0 . The choice of mag- 3. Validation of magnetic field mapping
netization; m = 1/µ0 ŷ allows the expressions in Eq. (3), (4),
One direct application of the vector magnetic field compo-
and (5) for the inductance vector coefficients of a unit dipole.
nents solved by Eq. (6) is to validate or supplement magnetic
dipole 1 3·x·y field data generated by field mapping. To illustrate this appli-
B̂x (x, y, z) = · (3)
4π (x2 + y 2 + z 2 ) 52 cation mapped datasets, generated by scanning a calibrated
à sensor in several planes parallel to a magnet, are compared to
dipole 1 3 · y2 calculated datasets (see Table I).
B̂y (x, y, z) = · 5
4π (x2 + y 2 + z 2 ) 2 The magnetic poles are along the y-axis. The measured
! and calculated field profiles in the plane y = 9.75 mm are
1 shown in the last two columns of row 1. The measured and
− 3 (4)
(x2 + y 2 + z 2 ) 2 calculated datasets have a correlation of > .99. In row 2 the
1 3·z·y field profile in the plane at x = 9.75 mm from the same mag-
dipole
B̂z (x, y, z) = · (5) net is investigated. The mapped and calculated data sets in
4π (x2 + y 2 + z 2 ) 52
this plane also have a high correlation of > .98.
242 J. RICE

TABLE I. Magnetic Field over Plane Surfaces TABLE II. Translation

magnet; dimensions = 15 × 3.5 × 3.5 mm3 , Br = 13050G magnet; dimensions = 6 × 3 × 6 mm3 , Br = 12030G
Illustration Plane (mm) Measured Calculated Transform
Illustration Vector components
PS0 = T PS

y = 9.75 (τx , τy , τz )
= (0, y, 0)

x = 9.75
(τx , τy , τz )
= (6, y, −6)

The absolute accuracy of the measured to calculated data

is subject to inaccuracies of up to +/- 10%. These inaccura-
cies come from; i) variations in magnetization and ii) mea- (τx , τy , τz )
surement error which could be from sources such as scan = (4, −5, z)
set-up, centering, step size, or calibration of sensor. Still,
the high correlation demonstrated between the data calcu-
lated from Eq (6) and the measured data allows the use of
the calculated magnetic vector field of the bar magnet to sup- 6. Orbital applications
plement the measured data, where measurement of the field
Many common Hall sensor applications impose an orbital or
is unavailable or expensive.
circular constraint upon the magnet trajectory. A magnet may
orbit about an arbitrary axis relative to the sensor, but math-
4. Sensor normal unit vector ematical convenience allows a coordinate system to be cho-
sen which aligns the axis of the magnet orbit with one of the
A Hall effect sensor is only sensitive to the field normal to its Cartesian axis. While the magnet is in motion, the sensor is
surface area. The unit vector, n̂sensor = nx x̂ + ny ŷ + nz ẑ, at a relative stationary observation point.
represents orientation of the sensor in Cartesian coordinate The mathematical tool of affine transformation [2] is em-
system. The flux sensed is then the dot product of the sen- ployed, to track relative position of sensor to magnet during
sor normal unit vector n̂sensor , with the magnetic inductance motion. Scripted forms of the position transforms for orbital
vector Bmagnet , as shown in Eq. (7) motion are shown in Table III, Column 2. The matrix forms
of the affine transforms used in this analysis are provided in
B sensor = n̂sensor · Bmagnet (7) the appendix.
Column 5 of Table III illustrates the flux selected from the
Graphs that follow in this paper illustrate the variation of inductance vector by the sensor orientation. The sensed flux
all three orthogonal components i.e. Bx , By ,Bz , with mag- can be converted into a digital signal by integrating a com-
net position. It should be understood from equation (7) that parator into the Hall device. For this illustration, two switch
the sensor orientation selects the scalar flux from the induc- points set at 50 G and 10 G create the hysteretic property of
tance vector components. switch state versus angle.
The initial conditions of the magnet orientation relative to
its motion can be used to classify and illustrate common or-
5. Sensing motion within a plane bital applications. Three initial magnet orientations amenable
to case studies are tangential, radial and axial.
Table II illustrates the use of the volume integral from equa-
tion (6) to calculate the sensed flux for position of a mag- 6.1. Orbit with tangential aligned poles
net in plane relative to a Hall sensor. The illustration rep-
resents common applications where the sensor face, magnet Rotation about the z-axis corresponds to the magnetic poles
poles and magnet motion, all are oriented along Cartesian aligned tangential to the center of orbit. This magnet orien-
axis. When so aligned, the flux measured by the Hall sen- tation with the sensor placed at approximately the radius of
sor is simply one of the unit vector coefficients Bx , By or the orbit is one of the most common applications of Hall sen-
Bz . In Table II the relative position of the magnet to sensor sor technology; a lid sensor, found in flip phones, notebook
is parameterized by affine translation. computers and office equipment.

Rev. Mex. Fis. S 58 (2) (2012) 241–244

 VECTOR ANALYSIS OF MAGNETIC FIELDS FOR HALL SENSOR DESIGN 243

TABLE III. Orbits

magnet; radius 3mm, thickness 2.5mm, Br = 12300G
Illustration Position Transform Vector Components Sensor Normal Flux & Switch State

Ps0 −Pm0
= Rz −(T Ps −Rz T Pm ) n̂ = ŷ
Ps = (0, −10, 0), Pm = (0, 0, 0) 0
Rz− n̂
n̂ = ¯ − ¯
(τx , τy , τz ) = (20, 0, 0) ¯Rz n̂¯

Ps0 −Pm0
= Rx −(T Ps −Rx T Pm ) n̂ = ŷ
Ps = (0, 8, 0), Pm = (0, 0, 0) 0
Rx− n̂
n̂ = ¯ − ¯
(τx , τy , τz ) = (0, 20, 0) ¯Rx n̂¯

Ps0 −Pm0
= Ry −(T Ps −Ry T Pm ) n̂ = ŷ
Ps = (0, −10, 0), Pm = (0, 0, 0) Ry− n̂
n̂0 = ¯ − ¯ = ŷ
(τx , τy , τz ) = (20, 0, 0) ¯Ry n̂¯

6.2. Orbit with radial aligned poles

[Link].
B∈(x̂,ŷ,ẑ) (x, y, z)
Rotation about the x-axis corresponds to the magnetic poles
aligned radial to the center of orbit. The common configura- Zzb Z2πZr
dipole
tion is the sensor placed external to the magnets radial loca- =Br B̂∈(x̂,ŷ,ẑ) (x−r cos(θ), y−r sin(θ), z−ζ)
tion. Applications include thumbwheel counter or end stop −zb 0 0
locator. This configuration lends itself to multiple magnet
r dr dθ dζ (8)
poles located in a ring around the radius of motion, which
can increase the resolution. A field of a cylindrical magnet has symmetry that is
preferable for applications where the motion is pure rotation.

8. Rotation of a diametric magnet

6.3. Orbit with axial aligned poles
A notable cylindrical magnet application is rotation with
Rotation about the y-axis corresponds to the magnetic poles poles aligned diametrically. Table IV illustrates this applica-
aligned axial to the center of orbit. The sensor is usually tion for sensors aligned along the cylindrical coordinate unit
placed in a plane above or below the plane of the magnet mo- vectors, such that the measured flux is a selection of one of
tion. This configuration is used as a rotation counter, mag- the vector components described in equation (9).
netic interrupt, magnetic door sensor, and lends itself to mul- Bmagnet = Bρ ρ + Bθ θ̂ + Bz ẑ (9)
tiple pole configurations as well.
The sinusoidal flux obtained from rotating a diametric
magnet about its axial center can be exploited by placing
two sensors at locations that produce sinusoids of different
phases. [3] Sinusoidal signals that are 90 degree out of phase
7. Construction of cylindrical magnet vector are said to be in quadrature, and can be interpolated to deter-
field mine sub-rotation angles with precision limited only by har-
monic impurities in the motion signal and the noise in the
In a similar way as the bar magnet the expansion of the electronics [4].
field due to the material presence of a cylindrical magnet
is determined by convolution of the normalized unit dipole 9. Summary
vector field coefficients over the volume bounded by r, θ ∈
(0, 2π), ±z for each of the coefficients in equation (3), (4) This paper offers a vector technique for designing with per-
and (5). manent magnets and Hall sensors. Expansion of the bar and

Rev. Mex. Fis. S 58 (2) (2012) 241–244

244 J. RICE

cylindrical magnet field components is determined by a con- Appendix

volution integral, which has broad application for calculating
the projected field of a magnetized volume. Measured flux Affine Transformation Matrices
data is correlated with calculated data. Affine transforms are
employed to calculate sensor to magnet distance for common Translation
magnet trajectories. The relationship between the vector field  
and scalar flux is shown for popular applications. All calcula- 0 0 0 τx
tions and illustrations in this paper were produced using the  0 0 0 τy 
T = 
 0 0 0 τz 
relative inexpensive and commonly available mathematical
spreadsheet MathCAD. 0 0 0 1

TABLE IV. Rotation Rotate about x-axis

magnet; radius 3mm, thickness 2.5mm, Br = 12300G  
0 0 0 0
Transform  0 cos(θ) − sin(θ) 0 
Illustration
& Normal
Flux Rx = 
 0 sin(θ) cos(θ) 0 


0 0 0 1
Ps0 = Rz− (Ps )
Ps = (0, 7, 0)
Pm = (0, 0, 0) Rotate about y-axis
n̂ = n̂θ  
cos(θ) 0 − sin(θ) 0
 0 0 0 0 
Ps0 = Rz− (Ps ) Ry = 
 sin(θ)

Ps = (0, 7, 0) 0 cos(θ) 0 
Pm = (0, 0, 0) 0 0 0 1
n̂ = n̂ρ
Rotate about z-axis
Ps0 = Rz− (Ps )  
Ps1 = (−5, 3, 0) cos(θ) − sin(θ) 0 0
Ps2 = (−5, 0, 3)  sin(θ) cos(θ) 0 0 
Rz = 


Pm = (0, 0, 0) 0 0 0 0 
n̂s1 = n̂s1 = n̂z 0 0 0 1

1. R. Blakely, Potential Theory in Gravity & Magnetic Applica- 3. J. Viola, 90-degree Rotary Report (Allegro MicroSystems Inc.,
tions (Cambridge University Press, 1996). 2005) available upon request.
2. J. Foley, A. van Dam., Computer Graphics Principles and 4. D.E. Denk, Optoelectronics, Instrumentation and Data Pro-
Practice (2000). cessing 44 (2008) 105.

Rev. Mex. Fis. S 58 (2) (2012) 241–244