A Simulation Model of Hall Sensor Misalignment in BLDC Motors

Chung-Wen Hung
Department of Electrical Engineering, National Yunlin University of Science & Technology,
123, Sec. 3, University Road, Douliou, Yunlin, 64002, Taiwan

Chu-Lin Hsu
Department of Electrical Engineering, National Yunlin University of Science & Technology,
123, Sec. 3, University Road, Douliou, Yunlin, 64002, Taiwan
E-mail: [email protected], [email protected]
www.yuntech.edu.tw

Abstract
A simulation model of Hall sensor misalignment in BLDC motors is proposed in this paper. The Hall sensor is
popularly used in brushless DC motor to decide commutate and rotational speed. However, the accuracy of Hall
sensor position is limited in most of BLDC motors, and it is lack of suitable simulation model for Hall Sensor
installation misalignment. The details of simulation model are discussed in this paper, and simulation results are
also provided to show the model isworkable.
Keywords: simulation model, BLDC motor, Hall sensor, misalignment.

reference for BLDC motor control integrated circuit.

1. Introduction However, those two studies didn’t discuss any
BLDC motors are popularly used nowadays due to its simulation. [3] focused on commutation behavior
high power density, low maintenance requirement, and situation of BLDC motor, and proposed a simulation
easy control. Most BLDC motors have Hall sensors model in MATLAB. [4] presented an efficient
which are used to decide commutation timing and simulation model for variable sampling system, which is
estimate rotational speed. However, it is impossible to caused by low resolution effect of Hall sensors.
install Hall sensors in accurate position without error. However, it is a lack of simulation model for the Hall
This misalignment effect will cause torque ripple due to sensor position misalignment.
wrong commutation timing and incorrect speed
2. The Simulation Method
feedback.
This phenomenon is highlighted in some studies and To simulate the Hall sensors in BLDC motors, the
these papers tried to propose solution. [1] proposed a rotation information, distance, is necessary. It is used to
Hall-effect-sensor-based position observer to mitigate decide the hall sensor signals. The most of BLDC motor
torque ripple in permanent-magnet synchronous simulation models only provide the rotational speed
machines, and the hardware and simulation validation output, so an integrator will be used to perform the
showed the proposed algorithm is workable. [2] rotation distance. As shown in Fig. 1, to simplify
proposed an auxiliary circuit for BLDC motors, to reject explanation, the pole pair number is set to 1; Hx
the Hall sensor signal noise and capture time interval. indicates Hall sensor whether x is a, b or c. When the
The circuit is built in field programmable gate array, BLDC motor rotational angle arrives 0 or π radians,
FPGA, and the experimental results show the circuit means the point O reaches origin or Ha, the output of
works as expecting. This work could be a good Hall sensor A will change due to the change of magnetic

© The 2016 International Conference on Artificial Life and Robotics (ICAROB 2016), Jan. 29-31, Okinawa Convention Center, Okinawa, Japan

52
Chung-Wen Hung, Chu-Lin Hsu

field direction. The detail is presented in Fig. 2. If and described as (1)-(3). Here, % indicates remainder
without reverse, the output of the integrator is monotony operator and θpresents the integrated rotational angle.
increasing. However, only the angular information
within 0 to 2π is used, so the remainder divided by 2π is
useful for the decision of Hall sensor signal. Similarly,
when point rotes 5π/3 or 2π/3 radians, the output of the
Hall sensor B will change. Again, Hall sensor C will
change at π/3 and 4π/3 radians.

Fig. 3. Simulation model of Hall-sensor

Fig. 1. Diagram of Hall-sensor and relative rotor positions

Fig. 4. Simulation model of Hall-sensor B and C

HallA(θ) = 0, if (θ % 2π) < π (1)

Fig. 2. The relationship between Rotation angle and Hall = 1, if (θ % 2π) ≧ π
sensor output HallB(θ) = 1, if (θ % 2π) ≦ 2π/3, or (θ % 2π) > 5π/3 (2)
= 0, o. w.
3. The Proposed Simulation Model HallC(θ) = 1, if (θ % 2π) ≧ π/3, and (θ % 2π) < 4π/3 (3)
As discussed in the previous section, an integrator is = 0, o. w.
used to integrate rotational speed to rotational distance,
which is equivalent to rotational angle; a remainder
block is built to calculate rotational angle within 0 to 2π. Fig. 5. Equivalent simulation model
This block comprises a divider, a rounder, and an adder.
A relational operator is applied to decide Hall sensor 4. The Proposed Hall Sensor Misalignment
Effect Simulation
signal. The simulation model is shown in Fig. 3. Note,
to simplify the calculation, the output of the remainder Before discussion of Hall Sensor Misalignment Effect, a
block is ratio of 2π, and the second input of the modified simulation model is proposed which is
relational operator is changed to 1/2 from π. The other equivalent to the model discussed in previous section.
relational operator could be skipped, because remainder The model shown in Fig. 4, every hall sensor channel
function will change the output to zero when divisible. includes two relational operators and a logic operator.
However, the output of Hall Sensor B is positive, when However, an equivalent operation could be performed
rotational angle is larger than 5π/3 or less than 2π/3. with adding an angle shifter to the integrator output. For
And again, Hall Sensor C output high when lager than the Hall sensor B, the output is high till the point O runs
π/3 and less than 4π/6. The details are shown in Fig. 4, to 2π/3, and then keep in low before it runs to 5π/3. If
4π/3 is added into to the integrator output, the
© The 2016 International Conference on Artificial Life and Robotics (ICAROB 2016), Jan. 29-31, Okinawa Convention Center, Okinawa, Japan

53
 A Simulation Model of

summation will be fed into the remainder block and shifter which is used to add position error is needed, as
relational operator block, which are the same as Hall shown in Fig. 8. It is more intuitive and can avoid
channel A. The outputs sill are equivalent to the missing the other relational operator in Fig. 7. Note that,
previous model. Similarly, a 4π/3 shifter performs the an extra shifter could be added into channel A if there is
same effect for channel C. The calculation requirement any position error for the Hall sensor A.
may not less than previous one. However, it is simpler
and easier to understand. Moreover, it implies an
advance what will be introduced in next section.
In most BLDC motor, the accuracy of hall sensor
installation is limited, as presented in Fig. 6. The hall
sensor may not be installed in correct position, it will
cause wrong interval of HALL sensor signal. A novel
simulation model is proposed to simulate this
misalignment effect, and the position error is added to
the two relational operators in Fig. 4. As shown in Fig. 7,
the Hall sensor B channel is modified, and it can work
well to reflect the misalignment effect.
Fig. 8. Equivalent misalignment simulation model

(a) Matlab Simulink demonstration (b) proposed models

Fig. 9. Hall sensor signal comparison without misalignment

Fig. 6. Diagram of Hall-sensor with misalignment

Fig. 10. Simulation result of Hall sensor B with misalignment

5. Simulation Results
To verify the proposed simulation model work well, the
proposed model is add to the BLEDM DEMO model in
Matlab Simulink. The input of the proposed model is
Fig.7. Misalignment simulation model of Hall-sensor B the BLDC motor speed, and the output is Hall sensor
output. Fig. 9 presents the Hall sensor signal
However, misalignment effect could be simulated easier, comparison between the Matlab Simulink
if the previous equivalent simulation model is selected. demonstrational and proposed models without
Because rotor and Hall sensor are symmetric, only one misalignment. It shows that the two outputs of Hall
© The 2016 International Conference on Artificial Life and Robotics (ICAROB 2016), Jan. 29-31, Okinawa Convention Center, Okinawa, Japan

54
Chung-Wen Hung, Chu-Lin Hsu

sensors are the same. Next, to show the misalignment

effect, the proposed misalignment model is used to
replaced, and the channel B result is shown in Fig. 10.
Obviously, the proposed simulation model could
simulate the misalignment effect well.

6. Conclusion
A Simulation Model of Hall Sensor Misalignment in
BLDC Motors is proposed in this paper. For most
BLDC motors, it is very difficult and expensive to
install Hall sensors in correct position accurately. And it
will cause torque ripple and wrong speed feedback. The
proposed model could simulate exactly this
misalignment effect, and it is can provide a good
simulation reference to engineers for designs of BLDC
motor controllers.

Acknowledgements
This work is partially supported by the Ministry of
Science and Technology, ROC, under contract No. NSC
103-2622-E-224-056 and 104-2221-E-224-012.

References
1. P. B. Beccue, S. D. Pekarek, B. J. Deken, and A. C.
Koenig, Compensation for Asymmetries and Mis-
alignment in a Hall-Effect Position Observer Used in
PMSM Torque-Ripple Control, IEEE Transactions On
Industry Applications, Vol. 43, NO. 2, MARCH/APRIL
2007, pp. 560-570.
2. C.-W. Hung and J.-H. Chen, A Hall Sensor Auxiliary
Circuit with Noise Rejection and Time Interval Capture
Functions for BLDC Motors, Sensor Letters, Vol. 10,
2012, pp. 1178–1184.
3. B. Tibor, V. Fedák, and F. Ďurovský, Modeling and
Simulation of the BLDC Motor in MATLAB GUI, in
Proc. 2011 IEEE International Symposium on Industrial
Electronics (ISIE), 27-30 June 2011 pp.1403-1407
4. C.-W. Hung, C.-T. Lin and C.-W. Liu, An Efficient
Simulation Technique for the Variable Sampling Effect
of BLDG Motor Applications, in Proc.33rd Annual
Conference of the IEEE Industrial Electronics Society,
Nov. 5-8, 2007 (Taipei, Taiwan, 2007), .pp. 1175 - 1179

© The 2016 International Conference on Artificial Life and Robotics (ICAROB 2016), Jan. 29-31, Okinawa Convention Center, Okinawa, Japan

55