1

Abstract - A sensor less space vector pulse width modulated (SV-PWM) direct torque controlled (DTC) induction motor drive, having reduced torque ripple even at low operating speeds and maintaining constant switching frequency has been studied and implemented. It utilizes MRAS based closed loop speed estimation and symmetrical switching scheme for the semiconductor devices. In the scheme presented in this paper a combination of high- pass and low-pass filter is used to avoid DC drift, problem of saturation and to enhance the response of speed estimation. Detailed simulation studies are carried out to verify the efficacy of the system. The viability of the scheme is confirmed by performing detailed experimental studies.

Index Terms—Induction motor drive, direct torque control (DTC), space vector pulse width modulation (SV-PWM), fixed switching frequency, model reference adaptive scheme (MRAS).

I. INTRODUCTION

VER the years direct torque controlled (DTC) [1], [2] induction motor drive has become significantly popular

so far as its application in industry is concerned. This is due to its simplicity in control algorithm compared to that of vector controlled induction motor drive while having almost similar torque response characteristics as that of vector controlled drive. However, performance of DTC based drive at low speeds is not satisfactory as flux estimation during low speeds cannot be achieved accurately. Moreover, modern high performance drives are expected to operate without sensing speed of the system by means of a speed encoder. In view of this extensive research is being carried out in the field of sensor less DTC based induction motor drive. Several sensor less speed estimation schemes have been proposed and among them Model Reference Adaptive Scheme (MRAS) [4], [7], [8] has become popular because of its simplicity. However, in this scheme rotor speed estimation is inaccurate due to the presence of low frequency component at the output of the low pass filter which is used to estimate rotor flux in reference and adaptive model of MRAS. In this paper a modified flux estimation scheme is utilized, wherein a combination of high pass filer and a low pass filter is used in both reference and adaptive model of MRAS, which improves the performance to a considerable extent. In order to reduce torque ripple and

Abhijit Choudhury is pursuing his M.Tech from Department of Electrical Engineering, Indian institute of Technology, Bombay, India (e-mail: abhijic@ee.iitb.ac.in).

Prof. Kishore Chatterjee is with the Department of Electrical Engineering, Indian Institute of Technology, Bombay, India as a professor (e-mail: kishore@ee.iitb.ac.in).

device switching loss, symmetrical switching strategy of space vector pulse width modulation [3], [5] is utilized in this scheme. Detailed simulation studies have been carried out to verify the performance of the scheme. A laboratory prototype utilizing digital signal processor, DSP TMS320F28335 has been developed. Extensive experimental validations are carried out to establish the viability of the scheme.

II. PRINCIPLES OF OPERATION

A. Machine Modeling

The machine model equations in general reference frame are given below, where ‘g’ denotes the general reference frame.

gs gV i jw pg g g

s s s sR (1)

g r0 i j(w w ) pg g gr r r rR (2)

g gs s m rL i L ig

s (3)

g gr r m sL i L ig

r (4)

e L r m rT T Jp B (5)

e

3PT ( i i )

4g g g gsd sq sq sd (6)

Where Vs is stator voltage, Rs,Rr are stator and rotor resistances respectively, is, ir are stator and rotor current, λsd,λsq are stator flux linkages, Ls,Lr are stator and rotor self inductances, Lm is the mutual inductance, Te,TL are the electromagnetic and load torque, Wr,Wg are rotor and reference field speed. For stationery reference frame the machine model equations will be modified to the following equations, where ‘α’ axis is aligned with the magnetic axis of the stator coil, R.

sV i ps s sR (7)

r0 i -jw pr r r rR . (8)

s s m rL i L is (9)

r r m sL i L ir (10)

me *

r s

L3PT ( * )

4 L L s r (11)

Speed Sensor Less Direct Torque Controlled Induction Motor Drive With Constant Switching

Frequency Operation Abhijit Choudhury and Kishore Chatterjee

O

978-1-4577-0776-6/11/$26.00 ©2011 IEEE

2

Where (λsα

,λrα) are the stator and rotor flux linkages

respectively. From Eq. (9) rotor current can be derived as

s sr

m

( L i )i

Ls

(12)

Combining Eq. (10) and (12)

*ms s

r

L( ) L i

Ls r (13)

Where, 2

* s r ms

r

L L LL

L

(14)

And stator current can be expressed as

ms * *

s r s

Li ( )

L L Ls

r

(15)

sI

sV ts s

m

r

L( )

L r

*s sL i

s

s r

Fig. 1. Phasor diagram of stator and rotor flux linkage space vector.

Now developed electromagnetic torque can be written as

me *

r s

L3PT ( * )

4 L L s r (16)

me *

r s

L3PT (| || | sin )

4 L L s r (17)

Where, s r (18)

Where | |s and | |r

are the magnitude of the stator and

rotor space vector which rotates at synchronous speed. ‘ ’ denotes the angle between them while rotating in synchronous speed.

B. Basic DTC scheme

In the basic DTC scheme for induction motor drive a simple switching table is used, where depending upon the states of change in flux and torque, particular state vector is chosen. The table used for this purpose is shown TABLE 1.

The input to the table is change in flux and torque which comes from two separate hysteresis block. Hysteresis block used for flux is of two state and for torque is three state. From the machine terminal voltage and line current, stator flux, torque and state of the voltage vector are estimated and all these information are fed back to the closed loop controller to generate the switching signals. In fig. 2 six active switching vector and two null vector position are shown.

TABLE 1 SWITCHING VECTOR TABLE FOR BASIC DTC

1V(100)

2V (110)3V (010)

4V(011)

5V(001)6V(101)

8V (111)

7V (000)

1

2

3

45

6

eVs

Fig. 2. Voltage vector position.

Because of the presence of this hysteresis loop torque

produce by the induction machine becomes very much fluctuating as we can observe it from the simulation results. But the main advantage of this scheme is the simplicity in implementation as it requires no reference frame transformation which was the basic need for FOC drive. The stator flux is estimated as follows:

s s s(V i R )s dt (19)

Magnitude of stator flux can be estimated as

* 2 2( ) ( )s s sd sq (20)

Position of the vector can be estimated as

-1tan ( )sqe

sd

(21)

Torque of the machine can be estimated as

*e

3PT ( i i )

4 sd sq sq sd (22)

*e eT T Te (23)

*e e s (24)

Where Te, λe are the reference electromagnetic torque and

stator flux linkage. The speed of the machine can be sensed by

3

using a encoder or can be estimated by using a sensor less speed estimation scheme.

C. Space vector modulation technique (SV-PWM) with symmetrical switching scheme

To reduce the torque and flux ripple of the basic DTC scheme, a switching strategy utilizing space vector pulse width modulation (SV-PWM) has been used. The main advantage of this technique is its simplicity and at every instant of time the actual position of the voltage space vector can be located precisely. This leads to better control and fast transient response in machine operation. The basic concept of this scheme is as follows:

If the reference voltage vector is lying in the first sector as

shown in the Fig. 2, it can be constructed by its two associate active vector V1(100) , V2(110) and two null vector V7(000) or V8(111) in the same sequence for a specified time. The total time for which the active voltage vectors are applied is called ‘effective time’. Power transfer from inverter to motor can take place during this time only. These time intervals can be estimated by comparing volt-second of reference voltage vector with the applied voltage vectors and it can be expressed as

1 z

sin( )3T T

sin3

e

a

(25)

2 z

sinT T

sin3

ea

(26)

0 z 1 2T T (T T ) (27)

Where s

dc

V

2( )*V

3

a

, zT is the sampling time, 1T , 2T , 0T

are time intervals for which 1V , 2V and 7V or 8V is applied,

e is the angle between the voltage space vector sV and the

one of the active vectors( 1V to 6V ) in the direction of

rotation the reference vector, and dcV is the DC link voltage of

the inverter. To generate the required inverter control signal, SVM

requires only two reference voltage vector ( Vsd , Vsq

) in stator

flux oriented reference frame. In this scheme [6] these two vectors are generated by the closed loop controller as shown in fig. 3. The switching pulses are generated depending upon the magnitude and phase angle of the required reference voltage vector as expressed in the following equations:

2 2|V | ( ) ( )s sd sqV V (28)

-1tan ( )sqe

sd

V

V

(29)

Timing of the voltage vector applied by the inverter is done by symmetrical switching pattern. If Tz is the total switching time interval, then T1,T2 will be time intervals during which two active voltage vectors are applied, depending upon the sector in which it is lying. T0 is the time interval during which the null vector is applied.

Fig. 3. Block diagram of Control strategy.

D. Scheme for modified stator flux estimation

Estimation of stator flux s by utilizing (19) requires the

involvement of an integrator and its associated initial value problem. In the controller implementation, integration is generally realized by utilizing a low pass filter. The accuracy of the flux estimation depends on the cut off frequency of the low pass filter used. Generally a low cut off frequency below 5 Hz is chosen for the low pass filter [4], [7]. As a result, low order harmonics including dc component get added to the sensed signals. This dc component in the sensed voltage results from the dc offset that is being added to it to get it processed by the DSP chip. In actual practice the dc offset that is added to the signal before getting processed by the DSP and the dc offset that is subtracted from the signal after getting processed by the DSP cannot be accurately maintained to be same due to component, noise and temperature variation. This leads to variation in the magnitude of the estimated stator flux. In order to overcome this problem a modified flux estimation scheme has been proposed in this paper wherein a cascade combination of high pass and low pass filter is used. The high pass filter is employed to filter out all the low order harmonics that are present in the sensed signal as shown in fig. 4 and the low pass filter is used to estimate the flux. The same strategy is used for estimating rotor flux in adaptive and reference model used for sensor less speed estimation scheme as expressed in (30) and (31). The sensor less scheme used for speed estimation is a Model Reference Adaptive scheme (MRAS) shown in Fig. 5. The schematic control block diagram of the sensor less scheme is depicted in Fig. 6.

4

*rs s

m

L( ) - L iLr s

(30)

M r r

R

1(L j )

Ta a ar si dt (31)

p I

1(K K ) *rw

s (32)

Fig 4. Stator flux estimation block.

Fig. 5 MRAS based speed estimator.

Fig. 6. Sensorless SV-PWM scheme.

E. Simulation studies

Detailed simulation studies are carried out on MATLAB/Simulink platform to verify the operating behavior of the proposed scheme. Machine parameters chosen for carrying out the simulation study are provided in TABLE 2. Switching frequency of SV-PWM is kept at 10 kHz and simulation step size is kept at 20 µsec. Performance of the basic DTC scheme is shown in Fig. 7 and 8. Fig. 7(a-c) shows the stator current and torque response due to step change in speed command from 62.8 red/sec to 125.66 red/sec. Performance of the drive due to a step change in load torque command from 0 Nm to 7.0 N.m is depicted in Fig. 8(a-c). It can be observed from this figure that torque ripple is considerable, this is due to the fairly large width of the hysteresis band provided to keep the switching frequency of the switching devices within their rating. Simulated performances of the modified scheme utilizing space vector

pulse width modulation are depicted in Fig. 9 and Fig. 10. The simulated behavior of the system when a step change in load torque from 0 Nm to 7.0 N.m is initiated at 0.8 sec, is shown in Fig. 9 (a) to (c). From these simulated performances, it can be seen that even though there is a step change in the load torque, speed of the drive has almost remained constant. Fig. 10 (a-c) shows the response of the modified scheme for a step change in speed command from 62.8 red/sec to 125.66 red/sec. The step change in the speed command is initiated at 0.5 sec. From the aforementioned simulated performances it can be inferred that the system is having a fast toque response while having minimum torque and speed ripple.

Fig. 7. Simulation results of the basic DTC during change in speed. a) rotor

speed; (b) torque; (c) motor current.

Fig. 8. Simulation results of the basic DTC during change in load torque.

a) rotor speed; (b) torque; (c) motor current.

5

Fig. 9. Simulation results of the modified DTC during change in load torque.

a) rotor speed; (b) torque; (c) motor current.

Fig.10. Simulation results of the modified DTC during change in speed.

a) rotor speed; (b) torque; (c) motor current.

F. Experimental Validation

A laboratory prototype of the scheme utilizing DSP TMS320F28335 as the core controller has been developed to validate the performance of the proposed scheme. The flux estimated by employing a low pass filter in the conventional scheme is shown in Fig. 11(a) while the flux estimated by employing a high pass filter followed by a low pass filter is show in Fig. 11(b). It can be inferred that the low order frequency components which are present in the estimated flux in case of the conventional scheme have been eliminated by employing a combined high pass and low pass filter in the modified scheme. Measured starting performance of the scheme is shown in Fig. 12 (a) to (c). Measured transient response of the drive when the load torque is step changed from 0 Nm to 7.0 N.m while the speed command is maintained at 83.7 red/sec is shown in Fig. 12(d) to 12(f).

Fig. 11. Experimental results depicting differences in estimated flux in the conventional scheme vis-a-vis modified scheme when the speed of operation is 63 red/sec (a) estimated stator flux using conventional low pass filter; (b) estimated stator flux by using modified method.

Fig. 12. Experimental results for modified DTC during 83.7 red/sec speed of operation. (a) machine speed with no load; (b) machine torque; (c) stator current; (d) speed during load change; (e) change in load torque; (f) stator current variation.

6

Transient performance of the drive when the speed command is step changed from 62.8 red/sec to 125.6 red/sec is shown in Figures 13(a) to 13(c). Measured steady state performance of the drive while it is running at a speed of 125.6 red/sec is depicted in Figure 14(a) to (c).

Fig. 13. Experimental results for modified DTC during speed change from 62.8 red/sec to 125.6 red/sec. (a) change in speed; (b) torque response; (c) stator current.

Fig. 14. Experimental results for modified DTC during steady state operation at 125.6 red/sec. (a) speed; (b) torque; (c) stator current.

TABLE 2 MACHINE PARAMETER USED FOR SIMULATION OF SPEED SENSOR LESS DIRECT TORQUE CONTROL (DTC) SCHEME

Induction motor kW rating and No of pole

2.1 kW, 4

Stator and Rotor Resistance (Rs, Rr)

7.83 Ohm, 7.55 Ohm

Stator and Rotor Self Inductance (Lss, Lrr)

21.55 mH, 21.6 mH

Mutual Inductance (Lm) 0.4535 H Motor Inertia (J) 0.06 kg-m2

III. CONCLUSION

A sensor less space vector pulse width modulated direct torque controlled induction motor drive utilizing a modified stator flux estimation logic has been proposed in this paper. It comprises of a combination of a high and a low pass filter to reduce the low order harmonics and the dc offset associated with the process involved in estimating the stator flux by

integrating the relevant stator voltages. The proposed modified scheme demonstrates a good transient and steady state performance, which is confirmed by performing detailed simulation and experimental studies.

IV. REFERENCES

[1] I.Takahashi and T.Noguchi, “A new quick response and high efficiency strategy of an induction motor,” in Conf.Rec.IEEE-IAS Annu.Meeting,1985,pp.495-502.

[2] M.Depenbrock, “Direct self control for high dynamics performance of inverter feed AC machines,” ETZ Arch..,vol.7, no.7, pp.211-218, 1985.

[3] Thomas G.Habetler, Michele Pastorelli and Leon M.Tolbert, “Direct Torque Control of Induction Machine Using Space Vector Modulation,” IEEE trans.Industry Application, vol. 28, no.5, September 1992.

[4] Cristian Lascu and Ion.Boldea , “A Modified Direct Torque Control for Induction Motor Sensorless Drive,” IEEE trans. Industry Application, vol. 36, no.1, Feb. 2000.

[5] Yen-Shin Lai and Jian-Ho Chen, “A New Approach of Direct Torque Control of Induction Motor Drive for Constant Inverter Switching Frequency and Torque Ripple Reduction,” IEEE trans. Energy Conversion, vol.16, no.3, Sept 2001.

[6] Yuttana Kumsuwan, Suttichai Premrudeepreechacharn and Hamid A.Toliyat “Modified Direct Torque Control Methode For Induction Motor Drive Based on Amplitude and Angle Control of Stator Flux,” in Elsevier, Journal, 27 Feb. 2008.

[7] M.Messoudi, H.Kraiem, M.Benhamed, L.Sbita and M.N.Abdelkrim, “A Robust Sensorless Direct Torque Control of Induction Motor Based on MRAS and Extended Kalman Filter,” in Leonardo Journal of science, 12 january-June 2008.

[8] Pereda.J, Dixon.J and Rotella.M, “Direct Torque Control for Sensorless Induction Motor Drives Using an Improved H-Bridge Multilevel Inverter,” presented at IEEE Industrial Electronics conf., pp. 1110-1115, Nov.2009.

V. BIOGRAPHIES

Abhijit Choudhury was born in Tripura, India, in 1985. He received the B.E degree from Tripura Engineering College (NIT Agartala) in 2007 and presently perusing his M.Tech from Indian Institute of Technology, Bombay. He was with Asea Brown Boveri (ABB Ltd) from 2007 to 2008 as a management Trainee. During this period he was responsible for designing of high voltage Induction machine used for different industrial application and for machine drives

application. After that in 2008 he had joined as a Research Assistant M.Tech in Indian Institute of Technology, Bombay. His fields of interests includes Power Electronics and Electrical Machine Drives.

Kishore Chatterjee was born in Calcutta, India, in 1967. He received the B.E. and M.E. degrees in power electronics from M.A.C.T., Bhopal, India, and Bengal Engineering College in 1990 and 1992, respectively, and the Ph.D. degree in power electronics from the Indian Institute of Technology, Kanpur, in 1998.

From 1997 to 1998, he was a Senior Research Associate at the Indian Institute of Technology, Kanpur, where he was involved with a project on

power factor correction and active power filtering, which was being sponsored by the Central Board of Irrigation and Power, India. He has been with the Department of Electrical Engineering, Indian Institute of Technology, Bombay, since 1998 where is currently a Professor. His current research interests are modern var compensators, active power filters, utility-friendly converter topologies, and induction motor drives.