Swinburne University of Technology | CRICOS Provider 00111D | swinburne.edu.au

Swinburne Research Bank http://researchbank.swinburne.edu.au

Tashakori, A., & Ektesabi, M. (2013). A simple fault tolerant control system for Hall

Effect sensors failure of BLDC motor.

Originally published in Proceedings of the 8th IEEE Conference on Industrial Electronics and Applications (ICIEA 2013), Melbourne, 19–21 June 2013

(pp. 1011–1016). Piscataway, NJ: IEEE.

Available from: http://dx.doi.org/10.1109/ICIEA.2013.6566515.

Copyright © 2013 IEEE. This is the author’s version of the work, posted here with the permission of the publisher for your personal use. No further distribution is permitted. You may also be able to access the published version from your library. The definitive version is available at http://ieeexplore.ieee.org/.

A Simple Fault Tolerant Control System for HallEffect Sensors Failure of BLDC Motor

A. Tashakori, IEEE Student Member and M. Ektesabi, IEEE MemberSwinburne University of Technology, Melbourne, Australia

atashakoriabkenar@swin.edu.au

Abstract—This paper presents a novel fault tolerant controlsystem for Hall Effect position sensors failure of permanentmagnet brushless DC (BLDC) motor in a closed loop controlscheme. The proposed system is capable to detect and identifythe Hall Effect sensors breakdown based on sensors signals andDiscrete Fourier Transform (DFT) analysis of the measured linevoltages of motor. In this paper behavior of BLDC motor isstudied for Hall Effect sensors breakdown through simulationmodel. BLDC motor simulation model were validated firstby experimental data under no fault condition. Analyzing thesimulation results of the various sensor breakdowns leads todevelop an expert system for Hall Effect sensors failure diagnosisin BLDC motor. A simple method is presented to generate thecommutation signal of faulty Hall Effect sensor to maintain theproper operation of BLDC motor after fault occurrence. Theproposed fault tolerant control system does not need massivecomputational efforts and can be implemented as a subroutinein the main control program of the BLDC motor. The simulationresults show correct performance of the proposed fault tolerantcontrol system.

I. INTRODUCTION

BLDC Motors have been used widely in different industrialand commercial applications since 1970’s. It is a novel DCmotor that is commutated electronically due to absence ofbrushes. High efficiency, high speed ranges, high dynamicresponse (due to permanent magnet rotor) and less mainte-nance requirements (due to absence of brushes) are immediateadvantages of BLDC over brushed DC motor. Complex controlalgorithm because of electronic commutation and higher man-ufacturing price due to high cost of ferromagnetic materialsare disadvantages of BLDC compare to the brushed DC motor[1]. However BLDC motors are suitable and of interests forthe applications such as drive train of electric vehicle thatefficiency and precise controllability of electric motor havecritical effect on the performance of overall system.

Electronic commutation of BLDC motor depends on accu-rate detection of permanent magnet rotor position. There aretwo main control strategies of BLDC motor based on rotorposition detection methods. In the first method sensors areused to detect the rotor position and in the second methodthe position of rotor is detected through sensorless techniques.Optical encoders are used for applications with high resolutionrequirements and Hall Effect sensors for applications with lowresolution requirements [2].

Hall Effect sensors are mounted inside the BLDC motorwith 120 electrical degree phase difference to detect rotorposition. Output of each sensor is high (logic ’1’) for 180

electrical degree and is low (logic ’0’) for the next 180 degreewith respect to rotor position. Decoding Hall Effect signalsresult to choose the proper voltage space vector for switchingof the three phase Voltage Source Inverter (VSI) drive ofBLDC motor. In this paper position sensor failure of a threephase (star connection) permanent magnet BLDC motor withthe inbuilt Hall Effect sensors is discussed. BLDC motor andthree phase VSI drive are shown in Fig. 1.

Fig. 1. BLDC motor drive system

Failure of Hall Effect sensors effect directly on the appliedvoltages to the BLDC motor and degrade the performanceof overall motor drive. Development of fault tolerant controlsystems increases reliability of electric motor drive. The faulttolerant control system mainly should accomplish the follow-ing tasks,

1) fault detection;2) fault identification;3) remedial strategies.Various possible faults may happen in a Hall Effect sensor

such as flaws in the core (corrosion, cracks, residual magneticfields and core breakage), changes in the bias current, changein the magnetic properties of the ferrite core due to temperaturevariations, changes in the orientation of the induced magneticfield in the sensor (due to mechanical shocks or other reasons)[3]. Any of these faults may result to breakdown of the HallEffect sensor in BLDC motor. Unbalanced positioning of Hall

Effect sensors is another common fault in BLDC motor thatincreases the low-frequency harmonics in torque ripple anddegrades the overall drive performance [4], however it doesnot result in sensor breakdown.

There are a few research work on Hall Effect sensor failureof BLDC motor specifically for electric vehicle applications.Jeong et al. have presented a control strategy that providesfault tolerance to the major sensor faults which may occurin an interior-permanent-magnet-motor (IPMM)-based electricvehicle propulsion drive system [5]. Position sensors fault isdetected through difference between the estimated rotor angleand the actual measured one through a sensorless algorithmbased on extended EMF in rotating reference frame. In thisapproach reconfiguration to sensorless control scheme is intro-duced to rectify the fault and maintain the proper operation ofthe motor after fault occurrence is detected. Complexities ofsensorless control scheme and transition algorithm to snesor-less control are the main drawbacks of the proposed method.Simulation results of BLDC motor have been presented duringHall Effect sensors failure for a lunar rover wheel application[6]. Decoded switching signals of VSI and current waveformsof all phases during Hall Effect sensor breakdown are shown;however there is not a comprehensive discussion on thepresented simulation results, the fault diagnosis technique orremedial strategies in the paper.

Signal analysis, model based, and expert systems (knowl-edge based methods) are three basic classifications for faultdetection and diagnosis algorithms for BLDC motor [7]. Inthis paper, an expert fault diagnosis system of Hall Effectsensor failure in BLDC motor is discussed. In the section twobehavior of BLDC motor is studied during Hall Effect sensorfailure. Fault diagnosis technique and remedial strategies arediscussed in sections three and four respectively.

II. HALL EFFECT POSITION SENSOR FAILURE

A three phase low voltage BLDC motor is used as a prac-tical test motor to validate simulation results. Experimentalsetup of BLDC motor is shown in Fig. 2.

Fig. 2. Experimental setup

BLDC motor is simulated in Simulink using SimPowerSys-tems library. The experimental test motor specification datawhich is given in Table I have been used in the simulationmodel. The three phase variable source inverter drive of BLDCmotor is simulated using MOSFET switches. A Proportional

Integral (PI) controller is designed to adjust the duty cycle ofhigh frequency Pulse Width Modulation (PWM) signal basedon speed error to control the speed of motor in a closed loopscheme. An embedded code is used in Simulink model toproduce high frequency PWM signal. A duty cycle controlledPWM signal is applied to all switches of VSI.

TABLE IEXPRIMENTAL BLDC MOTOR SPECIFICATION

Description Value Unit

DC voltage 24 V

Rated speed 3000 RPM

Rated Torque 0.28 N.m

Phase resistance 2 ohm

Phase inductance 4.60 mH

Inertia 4.43e-6 kg.m2

Torque constant 0.069 N.m/A

Poles 10 -

Experimental BLDC motor and simulation model are testedunder the same operating conditions for 2000 RPM referencespeed and 0.1 N.m load torque. The line voltage and corre-sponding Hall Effect signal of phase A for experimental setupand simulation model of BLDC motor are shown in Fig. 3.

Fig. 3. Line voltage and Hall Effect signal of phase A of BLDC motor

It can be seen in figure that the simulation results are in

a good agreement with the test data of experimental BLDCmotor. Subsequently Hall Effect sensor faults are applied to thevalidated simulation model of BLDC motor. Characteristics ofBLDC motor such as speed, torque, line voltages and currentof all phases are studied for failure of Hall Effect positionsensors.

Breakdown of the position sensor cause the output signalof sensor to be whether constant high (logic ’1’) or constantlow (logic’0’) and it does not change according to the rotorposition. Therefore behaviour of BLDC motor is studied forthe both conditions (HA = 0 and HA = 1) separately duringfailure of Hall Effect position sensor of phase A.

A. Hall Effect signal is constant zero

Hall Effect sensor fault of phase A (HA = 0) is applied tothe validated BLDC motor model at t = 0.5s while the motorwas running under 0.1 N.m torque load at 2000 RPM speed.Speed and torque characteristics of BLDC motor for HA = 0fault condition are shown in Fig. 4.

Fig. 4. Speed and torque characteristics of BLDC motor for HA = 0

It can be seen in Fig. 4 after fault occurrence that the speedof motor is oscillating and electrical torque ripple amplitudeis increased remarkably. The line voltages of BLDC motorfor HA = 0 fault condition are shown in Fig. 5. Voltageof neutral point of BLDC motor is not stable during highfrequency PWM switching, therefore line voltages of BLDCmotor are measured with respect to negative terminal of DClink of inverter.

TABLE IISENSORS FAULT EFFECT ON THE SWITCHING SIGNALS

Fault type Open swiches of VSI

HA = 0 S1 , S6

HA = 1 S5 , S2

HB = 0 S3 , S2

HB = 1 S1 , S4

HC = 0 S5 , S4

HC = 1 S3 , S6

Hall Effect sensor failure causes change of the switchingsignals of VSI and effect directly on the applied line voltage of

Fig. 5. Line voltage of BLDC motor for HA = 0

the BLDC motor. Effect of various Hall Effect sensors fault onthe switching signals of VSI are given in Table II. Switchingsignals S1 and S6 are constant zero (switches S1 and S6

remain open circuit) for HA = 0 fault condition. Change ofapplied voltages cause variation of the stator phase currentsof the BLDC motor that effects directly on electrical torqueof motor and increases torque ripple. Phase currents of BLDCmotor for HA = 0 fault condition are shown in Fig. 6.

Fig. 6. Phase currents of BLDC motor for HA = 0

B. Hall Effect signal is constant one

Hall Effect sensor failure of phase A (HA = 1) has beenapplied to the validated BLDC motor model at t = 0.5s whilethe motor was running under 0.1 N.m torque load at 2000RPM speed. Speed and torque characteristics of BLDC motorfor HA = 1 fault condition are shown in fig. 7.

It can be seen that speed and torque responses of BLDC mo-tor for HA = 1 are quite similar to the ones for HA = 0 faultcondition. However by checking switching signals of VSI, itcan be observed that switches S2 and S5 are constant zero(switches S2 and S5 remain open circuit) after fault occurrencewhich are not the same as previous fault condition. Thereforeapplied voltages to the motor are completely different forHA = 1 fault condition. The line voltages of BLDC motorfor HA = 1 fault condition are shown in Fig. 8.

Fig. 7. Speed and torque characteristics of BLDC motor for HA = 1

Fig. 8. Line voltage of BLDC motor for HA = 1

Phase currents of BLDC motor for HA = 1 fault conditionare shown in Fig. 9. It is depicted from figure that currentof all phases are deteriorated after fault occurrence. Phasecurrents for HA = 1 are approximately symmetric projectionsof the phase current waveforms of HA = 1 fault conditionwith respect to the time axis.

Fig. 9. Phase currents of BLDC motor for HA = 1

III. FAULT DIAGNOSIS

Each Hall Effect signal of BLDC motor has specific valueat each instant of time with respect to permanent magnet

rotor position. Electronic commutation is done by decoding theposition sensor signals. Decoding rules of Hall Effect signalsto choose a proper switching vector of VSI are shown in TableIII. As it can be seen in table there is no condition that allthree hall signals being one or zero at a same time.

TABLE IIIDECODING RULES OF HALL EFFECT SIGNALS

Rotor angle Hall A Hall B Hall C Conducting(Electrical degree) switches

30-90 1 0 1 S1 , S4

90-150 1 0 0 S1 , S6

150-210 1 1 0 S3 , S6

210-270 0 1 0 S3 , S2

270-330 0 1 1 S5 , S2

330-30 0 0 1 S5 , S4

The addition of Hall signals introduced by (1) is a faultsignature for Hall Effect sensors breakdown. Maximum possi-ble value of Hf is 2, where the minimum possible value is 1(1 6 Hf 6 2) for each specific electrical angle section. If Hf

value goes over of these limits Hall Effect sensor failure isdetected. Hall Effect sensors Fault Flag (HFF) is introducedfor sensor fault detection. HFF is set to ’1’ if Hf value ismore than 2 (it means that one of the position sensor signalsis constant one), HFF is set to ’-1’ if Hf value is less than 1 (itmeans that one of the position sensor signals is constant zero)and HFF is ’0’ in case of no fault. Maximum fault detectiontime is the time of on electrical rotation of rotor which is quitefast.

Hf = HA +HB +HC (1)

However identification of faulty sensor is impossiblethrough Hall Effect sensor Fault Flag. As it is discussedin previous section, the line voltages of BLDC motor aredeteriorated due to position sensor failure. Therefore DFTanalysis is used for pattern recognition of the line voltages.DFT of line voltages are calculated by (2) for specific intervalsof time. The minimum time interval for proper fault detectionis one electrical rotation of motor. Spectral Energy Density(SED) of computed frequency spectrum is determined by (3).SED difference of successive time intervals are calculated andanalyzed to identify the faulty position sensor.

V (f) =

N−1∑n=0

Vne−j2πk n

N k = 0, 1, ..., N (2)

Em(f) = |V (f)|2 (3)

εm = Em(f)− Em−1(f) (4)

Calculated SED errors of the BLDC motor line voltagesfor Hall Effect sensor failure of phase A for HA = 0 andHA = 1 faults are shown respectively in Table IV and TableV. As it is shown in the tables, it is possible to distinguish two

failure conditions of Hall Effect sensor through SED errors.The simulation model is also run for position sensor faultsof phases B and C. The line voltages of BLDC motor forHall Effect sensor faults of all phases are studied through thesimulation model.

TABLE IVSED VALUES FOR HA = 0 FAULT CONDITION

Description Phase A Phase B Phase C

SED befor fault [Em−1(f)] 957 942 938

SED after fault [Em(f)] 807 1044 1083

SED error [εm] -150 102 145

TABLE VSED VALUES FOR HA = 1 FAULT CONDITION

Description Phase A Phase B Phase C

SED befor fault [Em−1(f)] 957 942 938

SED after fault [Em(f)] 1109 925 802

SED error [εm] 152 -17 -135

Hall Effect Identification Flag (HIF) of each phase isintroduced for faulty sensor identification. Numeric values aregiven to HIF of each phase according to SED errors of allthree line voltages of BLDC motor as below,

• HIF is ‘-1’ if SED error is negative;• HIF is ‘1’ if SED error is positive.A multidimensional knowledge based table is developed for

position sensors fault diagnosis according to Fault detectionand identification flags (HFF and HIF) by analyzing thesimulation results of BLDC motor for Hall Effect positionsensor faults of all phases. One of the advantages of thistechnique is that it is not necessary to know the exact linevoltages of BLDC motor for different speed and loads inadvance. Multidimensional knowledge based table for positionsensor fault diagnosis of BLDC motor is shown in Table VI.

TABLE VIRULE BASED POSITION SENSOR FAULT IDENTIFICATION TABLE

Fault type HIF HIF HIF HFFphase A phase B phase C

No fault X X X 0

HA = 0 -1 1 1 -1

HA = 1 1 -1 -1 1

HB = 0 1 -1 1 -1

HB = 1 -1 1 -1 1

HC = 0 1 1 -1 -1

HC = 1 -1 -1 1 1

IV. REMEDIAL STRATEGY

After identification of the faulty Hall Effect sensor, cor-responding sensor signal is disconnected and is substitutedwith a generated commutation signal by microcontroller. Com-mutation signal is generated based on 120 electrical degree

phase difference from the other two Hall Effect signals. A newand simple method has been introduced to calculate electricaldegree delays with respect to the time [8]. If motor speed isknown and it does not change during commutation intervals(controller keeps the motor speed constant), the time of oneelectrical degree rotation of BLDC motor can be determinedin seconds by (5), where P is the pole numbers and ωref isthe reference speed of the controller in RPM [8].

t =1

P/2(6 ∗ ωref )(5)

The proposed technique is implemented as embedded Mat-lab function in Simulink file of BLDC motor model. Faulttolerant control system of BLDC motor is tested under 0.1N.m load torque and 2000 RPM reference speed for HA = 0fault occurrence at t = 0.5 s. Speed characteristics of BLDCmotor with the implemented fault tolerant control system isshown in Fig. 10.

Fig. 10. Speed response of fault tolerant controlled BLDC motor drive forHA = 0 fault

As it can be seen fault detection time is remarkably fast.Fault identification takes more time due to DFT analysis of theline voltages. Total time of fault diagnosis is 113 millisecondsfor the simulation model that is fast enough to avoid furtherfaults in BLDC motor drive. Amplitude of speed oscillation isreduced as soon as the signal of faulty Hall Effect sensor issubstituted by microcontroller with the generated commutationsignal. Speed of the BLDC motor was following the referencespeed after almost 0.2 seconds.

V. CONCLUSION

Fault tolerant control system for Hall Effect position sensorsfailure of BLDC motor is discussed in this paper. Behaviorof BLDC motor is studied for position sensor failure situa-tions. Discrete Fourier transform analysis is used for patternrecognition of the line voltages of BLDC motor. Hall Effectsensor failure of BLDC motor is implemented on the verifiedsimulation model. A knowledge based table is developed toidentify the faulty sensor by analyzing the simulation results.Hence faulty position sensor is identified through spectraldensity error of the line voltages; exact knowledge of BLDCmotor voltages for various speed and torque loads is not

needed. Commutation signal of faulty sensor is generated bymicrocontroller through correlation between Hall signals. Theproposed fault tolerant system is capable to detect, identify andrectify the Hall Effect sensor break down in BLDC motor.Simulation results show correct detection and identificationof position sensor fault by the proposed fault tolerant controlsystem. Effectiveness of the remedial strategy is also proven bycorrect performance of BLDC motor under faulty condition.proposed system has a simple algorithm and can be imple-mented with a closed loop control scheme of BLDC motoron a single chip microcontroller. Consequently reliability ofBLDC motor drive is improved.

REFERENCES

[1] A. Tashakori, M. Ektesabi, and N. Hosseinzadeh, “Characteristics ofsuitable drive train for electric vehicle,” in International Conference onInstrumentation, Measurement, Circuits and Systems (ICIMCS 2011), Vol.2, pp. 51–57, ASME, 2011.

[2] A. Tashakori and M. Ektesabi, “Comparison of different pwm switchingmodes of bldc motor as drive train of electric vehicles,” World Academyof Science, Engineering and Technology, vol. 67, pp. 719–725, 2012.

[3] E. Balaban, A. Saxena, P. Bansal, K. Goebel, and S. Curran, “Modeling,detection, and disambiguation of sensor faults for aerospace applications,”IEEE Sensors Journal, vol. 9, no. 12, pp. 1907–1917, 2009.

[4] N. B. Samoylenko, Q. C. Han, and J. Jatskevich, “Dynamic performanceof brushless dc motors with unbalanced hall sensors,” IEEE Transactionson Energy Conversion, vol. 23, no. 3, pp. 752–763, 2008.

[5] Y.-S. Jeong, S.-K. Sul, S. Schulz, and N. Patel, “Fault detection andfault-tolerant control of interior permanent-magnet motor drive systemfor electric vehicle,” IEEE Transactions on Industry Applications, vol. 41,no. 1, pp. 46–51, 2005.

[6] L. Wang, J. Liu, and X. Wu, “Fault analysis on driving motors of lunarrover wheels,” (Beijing), 20-23 Aug 2011. International Conference onElectrical Machines and Systems, ICEMS 2011.

[7] X.-Q. Liu, H.-Y. Zhang, J. Liu, and J. Yang, “Fault detection and diagnosisof permanent-magnet dc motor based on parameter estimation and neuralnetwork,” IEEE Transactions on Industrial Electronics, vol. 47, no. 5,pp. 1021–1030, 2000.

[8] A. Tashakori and M. Ektesabi, “Stability analysis of sensorless bldc motordrive using digital pwm technique for electric vehicles,” in Proceeding of38th Annual Conference on IEEE Industrial Electronics Society, IECON2012, pp. 4898–4903, October 2012.