2006-01-0939

Development of a Contactless Hall effect torque sensor for

Electric Power Steering
Didier ANGLEVIEL
Moving Magnet Technologies S.A.

Didier FRACHON, Grald MASSON

Moving Magnet Technologies S.A.

Copyright 2005 SAE International

ABSTRACT
In this paper, we will present an innovative torque
sensor design which could be preferably used in Electric
Power Assisted Steering (EPAS) application. This torque
sensor is a non-contact Hall effect design. The
specificity of this structure is its ability to measure the
shift angle between two rotating shafts linked by a
torsion bar. This measurement is done with stationary
electronic components. This unique structure generates
enough magnetic flux variation to measure angular shifts
from +/-1 to +/-8 with low-cost standard Hall ASICs
available from various suppliers. This torque sensor has
convinced automotive industry due to its good
performances, its compact dimension and a price
compatible with the market expectation for this
application.
This paper will explain the basic magnetic principle of
this torque sensor, the improvements which have lead to
a structure compatible with automotive industry
requirements, some measurement results and possible
sensor integration in the application.

the driver to the steering wheel. This torque signal

usually comes from the shift measurement between two
shafts linked by a linear torsion bar. The main concern in
this measurement is due to the fact that these shafts are
also rotating with the steering wheel and the integration
of electronic on rotating parts is definitely a blocking
point. An example of BLDC motor and torque sensor
typical integration on the steering column is shown on
the following drawing. The BLDC exerts its torque to the
steering column through a worm gear mechanism.

BLDC motor

Torque sensor
housing

INTRODUCTION
Nowadays, for cost and pollution reasons, the
development of new vehicles is strongly driven by a
research of fuel consumption reduction. One interesting
way for such fuel economies is the replacement of
current hydraulic power steering by EPAS using high
power Brushless DC (BLDC) motors. The typical fuel
economy realized by shifting from a hydraulic power
steering to an EPAS is about 5 percent. With the
improvement of the BLDC motor structures and the
increase of the electric power available with batteries,
the market shares of EPAS should grow quickly in the
next few years [1].
In order to drive efficiently the BLDC motor, the ECU
requires accurate information on the torque applied by

Figure 1: Electric Power Steering

Moving Magnet Technologies (MMT) has developed an
innovative low cost structure answering all the
requirements of torque sensing for EPAS. This torque
sensor is based on a non-contact Hall effect principle
and uses standard programmable Hall ASICs available
on the market. MMT has applied for two patents
regarding first a basic structure of this torque sensor and
then an optimized design [2] [3]. These torque sensor
designs are presented hereafter.

BASIC DESIGN AND SENSOR PRINCIPLE

The MMT torque sensor basic structure is shown
hereafter.
Hall probe + Electronic (fixed
stationary on the housing)

On these two first views of the torque sensor, the rotor is

realized with magnets of the same polarity (red parts)
alternated with ferro-magnetic areas from the yoke. This
magnet configuration could be replaced by a more
industrial ring magnet with alternated north and south
poles on its surface as shown below.

Rotor (ferro-magnetic
yoke+magnets) fixed on a
rotary shaft

Figure 4: Rotor made with multipolar ring magnet

One couple of north and south poles of the magnet is
associated to one tooth on each stator to create an
elementary sensor as shown below.

Stator teeth (one of each

stator)

2 stators (ferromagnetic parts) fixed

on a rotary shaft

Figure 2: Exploded view of the MMT torque sensor

This sensor is made up of the following sub-assemblies:
The rotor (grey and red parts), fixed on one
rotary shaft of the steering column, is made by a
plurality of permanent magnets fixed on a soft
ferro-magnetic yoke.
The stators (blue and yellow parts) are fixed on
the second rotary shaft of the steering column,
linked to the first one by a torsion bar. These
stators are fixed one toward the other. They are
made of soft ferro-magnetic material.
One or two (for redundancy) Hall ASICs fixed
stationary on the housing. The Hall probe is
inserted in a measurement slot created between
the two stators. With the MMT design, we
generate enough flux variation to use the Hall
ASICs available from Melexis (MLX 90251 or
redundant MLX 90277) or Micronas (HAL 8XX).
The following figure shows the sensor in a working
configuration.

Figure 3: MMT torque sensor basic structure

Magnet poles of the rotor

Figure 5: View of an elementary sensor in a

beginning of the stroke position with flux lines
A magnetic flux variation is generated in the stator teeth
when the magnet transitions are moving in front of them.
This flux variation is a linear function of the magnet
angular position and we have demonstrated the
following analytical equation:

Bm =

4 B r Da L
1 .F .Da + 4. .E .H

Where:
Bm is the flux measured in the slot between two
stator teeth
Br is the magnet remanence
Da is the average diameter of the air gap
L is the magnet thickness
is the angular position of the magnet transition
1 is the angular length of one stator teeth
F is the thickness of the slot between the two
stator teeth
is the flux leakage coefficient
E is the radial air gap length
H is the teeth radial length

As shown on the previous figures, in this sensor, we

multiply the number of elementary sensors in order to
increase the flux variation in the measurement slot. The
flux variations generated in each elementary sensor are
integrated in an external disc on each stator. These two
discs create a radial measurement slot in between the
stators. By design, the induction on a given radius in the
measurement slot is constant as shown on the 3D
simulation results hereafter:

Flux concentration areas

(redundant design)

Flux collection area

Figure 7: Flux collector/concentrator

Figure 6: 3D simulation result showing the induction

in the measurement slot
MMT has demonstrated that an analytical equation of
the flux variation in the measurement slot is:

The use of these flux collectors/concentrators allows a

simplification of the stator design to L-shape stator. The
axial collection of the magnetic flux on the stator could
be done on the outside or the inside of the disc flux
integrating area of the stator. Even if an outer collection
of the flux leads to a little increase of the axial length of
the sensor, MMT strongly advises this solution. Indeed,
this outer collection of the flux decreases the flux
leakage between the flux collector (18% increase on the
flux variation) and the sensitivity to magnet transition
rotating in front of the Hall ASIC is also decreased.
These stator modifications lead to the design in the
following figure.
External flux collection

4 NZB r Da L
Bm =
NDa ZF 1 + 4ES m
Where:
N is the number of elementary sensors
Z is the axial dimension of the permanent
magnet
Sm is the measurement surface.
MMT has also developed some rules in order to size the
angular length of the stator teeth according to the stroke
to measure and the diameter of the steering column on
which we will mount the sensor.
This first structure has convinced some automotive
partners of MMT with its performances and we have
started a new development phase in order to improve
the sensor performances and use some lower-cost
parts.

DESIGN IMPROVEMENT TO AN OPTIMIZED

AUTOMOTIVE SENSOR DESIGN
In order to increase the flux in the measurement slot, a
good solution is to collect the flux from the stator with
stationary ferro-magnetic parts, and then concentrate
this flux on the Hall ASIC with a reduced surface
measurement slot. These parts are shown in the figure
below.

L-shape stators

Figure 8: MMT design with simplified stators and

external flux collector
The second modification on the stator shape has been
done on the shape of the teeth. Indeed, in our first
design, the stator teeth were overlapping the magnet on
their axial direction. In order to optimize the axial
dimension of the sensor, the collection area of one stator
(disc area) was axially close to the end of the teeth of
the other stator. In this proximity area between both
stators, flux leakage was occurring. Some 3D
simulations and prototype measurements have shown
that shortening the teeth of the stator involves an
increase of the flux variation in the measurement slot
without any deterioration of other sensor performances.
3D simulation has also shown that the closer we are
from the stator flux integration disc area, the higher is
the induction and so the leakage from one tooth to the
next one on the other stator. In order to increase the
cross section of the teeth when the flux density
increases (decrease of the induction in the teeth), we
have designed some trapezoidal shape teeth. We have
also validated this shape modification of the teeth with
3D magnetic simulations and prototype measurements.

The combination of these two modifications of the teeth

shape leads to a 12% output signal increase in the
measurement slot. Figure below shows the evolution of
the stator shape.

Figure 9: Teeth shape modification

These stator shape modifications lead to a part which
could be stamped and formed easily and which allows a
large saving in weight as compared with previous design
(typically about 50%).

Figure 11: Output signal programmed on one probe

(0.5 to 4.5 V)

Today, MMT optimized torque sensor design for steering

column application is shown on the figure below:

Figure 12: Non-linearity and hysteresis

measurement over the stroke
With the available ASICs, we are also able to program
the gain to optimize the non-linearity at the no torque
position as shown hereafter.

Figure 10: Optimized MMT torque sensor magnetic

structure

PROTOTYPES MEASUREMENT RESULTS

Based on this optimized torque sensor, MMT has
developed a family of designs with the same overall
dimensions (magnet: OD=37.5 mm, ID=34.75mm,
L=8mm) for strokes of +/-8, +/-4, +/-1. In order to
adapt the prototypes to the stroke, we have played on
the stator teeth shape and number.

Figure 13: Non-linearity optimized on the middle of

the stroke

In this paper we will focus on the measurement results

for the +/-8 stroke prototype. A table in appendix A
gives a summary of these typical measurement results
for the complete sensor family.
MEASUREMENT RESULTS ON +/-8
The figures hereafter will show the output signal of this
prototype,
the
non-linearity
and
hysteresis
measurements, the output signal stability on one turn on
mid-stroke position (no torque) and on end-stroke
position (maximum torque).

Figure 14: Stability on one turn at no torque position

We have also made a study on the variation of the

hysteresis as a function of the stroke measured.
14 stroke sensor: hysteresis relative to the stroke variation
0,35
Hyst (% of FS)

0,3
0,25
0,2
0,15
0,1
0,05
0
0,01

Figure 15: Stability on one turn at maximum torque

position

0,1

10

100

Stroke cycle (deg)

Figure 18: Peak to peak value of the hysteresis as a

function of the stroke measured

INFLUENCE OF MECHANICAL AND THERMAL

PARAMETERS
The results presented hereafter are similar for the whole
torque sensor family. The view hereafter shows the
direction for the off-centering on the sensor.

We have also measured the influence of rotor

displacement during the life of the sensor (in use) for a
given programming of the Hall ASIC.
Axial shift (along Z) in USE

NL (% of full stroke)

0,5

0,3

0,1

-0,3

-0,2

-0,1

0
-0,1
Axial shift [mm]

0,1

0,2

0,3

Figure 19: Rotor axial shift influence

Figure 16: Directions for off-centering

Radial shift (along X & Y) in USE

V a ri a ti o n i n % o f th e fu l l
stro k e

Temperature influence on sensitivity probe A

2
1.5
1
0.5
0
-0.5
-1
-1.5
-2

1
0,9
0,8
0,7
0,6
0,5
0,4
0,3
0,2
0,1
0

X shift
Y shift

NL(% full stroke)

With the standard Hall ASICs used in MMT, we are able

to compensate the non-linearity of the output signal over
an automotive temperature range (-40C to 150C) as
shown on the figure below.

-0,15

-0,1

-0,05

0
0,05
X & Y shift [mm]

0,1

0,15

Figure 20: Rotor radial shift influence

On the measured prototype, the radial air gap between
magnet and stator teeth is 0.5 mm. It is possible to
increase this mechanical gap in order to decrease the
influence of the radial shift if required.
-8

150C

-6
25C

-4
-40C

-2

Stroke position (deg)

The performances of these prototypes are suitable with

the requirements of automotive industry for EPAS.

SENSOR INTEGRATION ON THE APPLICATION

Figure 17: Thermal compensation of the output
signal with standard Hall ASIC

A first solution for the integration of the sensor is an

integration of the sensor parts directly on the steering
column. This first integration will be preferred by

automotive suppliers who deliver the complete steering

column assembly.
Flux collectors +
electronic subassembly

Sensor housing

Rotor subassembly

First steering
shaft

Second steering
shaft

Figure 24: Picture of a cassette torque sensor

Torsion bar

CONCLUSION
Stator subassembly

Figure 21: MMT torque sensor mounted directly on

the steering column

Figure 22: Picture of a prototype torque sensor

integration
The other possible integration of this sensor is the
realization of a stand-alone sensor in a cassette design.
This second integration will be preferred by automotive
suppliers who supply the sensor to steering column
manufacturer.

For the torque sensor technology presented in this

paper, we have identified the following advantages and
benefits:

Hall effect patented non-contact design.

Angular shift measurement between two rotary
parts with a stationary electronic.
Measurement of small angular displacement
(down to +/-1) with field density variation
compatible with current ASICs
Low parts number Easy to industrialize
Parts with design easy to manufacture Low
cost design.
Possible redundant design
Possibility to integrate the sensor parts directly
in the column housing or to deliver it separatly
as a stand-alone unit (cassette design).
Parts adjustable to the stroke to measure.
These advantages have lead to successful collaboration
with automotive suppliers involved in EPAS
development.
MMT is still improving its design, for example with
coupling this torque information with position
information.

ACKNOWLEDGMENTS
Mutlipolar ring magnet
Stuck on steel yoke

Hereafter is a non exhaustive list of people who have

been involved in this development: Claude OUDET,
Pierre GANDEL, Daniel PRUDHAM, Jean Daniel
ALZINGRE, Antoine FOUCAUT, Thierry DORGE,
Yannick RONNAT

REFERENCES
Collector supported by flanges

MLX 90277: two hall probes in one SMD

chip located in the measurement air gap

Figure 23: Exploded view of an MMT torque sensor

in a cassette design

[1]
Passenger Car Steering Systems Preparation
for the Technology of Tomorrow (26-27/04/2005 in
Essen) Introduction presentation by Dr.-Ing. Alois
SEEWALD
[2]
Position sensor, design in particular for
detecting a steering column torsion, US patent
publication number #2004-0011138, Claude OUDET,

Daniel PRUDHAM, Pierre GANDEL, Didier FRACHON,

Didier ANGLEVIEL
[3]
Capteur de position, notamment destine la
mesure de torsion dune colonne de direction, PCT
application number #PCT/FR2005/050571, Daniel
PRUDHAM, Grald MASSON, Antoine FOUCAUT.

CONTACT
Didier ANGLEVIEL, USA Marketing Manager

Moving Magnet Technologies S.A.

1 rue Christiaan HUYGENS
25000 BESANCON
FRANCE
www.movingmagnet.com
Tel. +33 381 418 563
[email protected]

APPENDIX A
STROKE +/-1

STROKE +/-4

STROKE +/-8

8 pole pairs

4 pole pairs

90 Gauss/

72 Gauss/

+/- 300 Gauss

+/- 360 Gauss

+/- 575 Gauss

+/-0.2% of Full Scale

+/-0.28% of Full Scale

+/- 0.25% of Full scale

+/- 0.3% of Full Scale

+/- 0.2% of Full scale

+/-0.05 % of Full Scale

+/- 0.03% of Full Scale

+/- 0.025% of Full scale

+/-0.05 % of Full Scale

+/- 0.3% of Full Scale

+/- 0.38% of Full scale

+/-0.16 % of Full Scale

PICTURE OF THE
PROTOTYPE

SENSITIVITY
INDUCTION OVER THE
STROKE
NON-LINEARITY
HYSTERESIS FOR A
FULL STROKE
MEASUREMENT
STABILITY ON 360 @
NO TORQUE POSITION
STABILITY ON 360 @
MAXIMUM TORQUE
POSITION

12 pole pairs
300 Gauss/