The Modeling and Simulation of Improved

First Edition 2008
© NIK DIN MUHAMAD 2008
Hak cipta terpelihara. Tiada dibenarkan mengeluar ulang mana-mana bahagian artikel,
ilustrasi, dan isi kandungan buku ini dalam apa juga bentuk dan cara apa jua sama ada
dengan cara elektronik, fotokopi, mekanik, atau cara lain sebelum mendapat izin bertulis
daripada Timbalan Naib Canselor (Penyelidikan dan Inovasi), Universiti Teknologi
Malaysia, 81310 Skudai, Johor Darul Ta’zim, Malaysia. Perundingan tertakluk kepada
perkiraan royalti atau honorarium.
All rights reserved. No part of this publication may be reproduced or transmitted in any
form or by any means, electronic or mechanical including photocopy, recording, or any
information storage and retrieval system, without permission in writing from Universiti
Teknologi Malaysia, 81310 Skudai, Johor Darul Ta’zim, Malaysia.
Perpustakaan Negara Malaysia Cataloguing-in-Publication Data
Modelling and control of power converters and drives / editor Nik Din Muhamad.
Includes index
ISBN 978-983-52-0648-1
1. Electric current converters. I. Nik Din Muhamad.
621.313
Editor: Nik Din Muhamad

Pereka Kulit: Mohd Nazir Md. Basri & Mohd Asmawidin Bidin
Diatur huruf oleh / Typeset by

Fakulti Kejuruteraan Elektrik
Diterbitkan di Malaysia oleh / Published in Malaysia by

PENERBIT
UNIVERSITI TEKNOLOGI MALAYSIA
34 – 38, Jln. Kebudayaan 1, Taman Universiti,
81300 Skudai,
Johor Darul Ta’zim, MALAYSIA.
(PENERBIT UTM anggota PERSATUAN PENERBIT BUKU MALAYSIA/
MALAYSIAN BOOK PUBLISHERS ASSOCIATION dengan no. keahlian 9101)
Dicetak di Malaysia oleh / Printed in Malaysia by

UNIVISION PRESS SDN. BHD.
Lot. 47 & 48, Jalan SR 1/9, Seksyen 9,
Jalan Serdang Raya, Taman Serdang Raya,
43300 Seri Kembangan,
Selangor Darul Ehsan, MALAYSIA.
v Contents
CONTENTS
Preface vii
Chapter 1 Control Loop Design of DC-DC Converter 1

Systems Using PSpice
Chapter 2 Simulation of Power Converters with 18

Sliding Mode Control Using PSpice
Chapter 3 Dynamic Evolution Control for Step-down 37

DC-DC Converter
Chapter 4 Digital PI Controller Design for Power 50

inverter Using MATLAB-SIMULINK
Chapter 5 Overview of Direct Torque Control of 67

Induction Machines
Chapter 6 Study on Stability and Performances of 99
DTC Due to Stator Resistance Variation
Chapter 7 A Quick Dynamic Torque Control for 118

Direct Torque Control Hysteresis-based
Induction Machines
vi Contents
Chapter 8 Improved Torque Capability through 133

Overmodulation for Direct Torque
Control Hysteresis-based Induction
Machines
Chapter 9 The Modeling and Simulation of Improved 148

DTC of PMSM Drives Using
Matlab/Simulink
Chapter Modeling and Simulation of a Network of 162

10 Adjustable Speed Drives
Index 175
vii Preface
PREFACE
Power electronics technology is used for efficient control of

motors in electric cars, industrial and consumer motor drives, and
the development of more reliable, lightweight switching power
supplies for sophisticated computer and communication
equipment. Recent advances in microprocessor technology are
leading to the integration of microelectronics and power
electronics technologies for smart and efficient control of motor
drives and power supplies.
Switching power supply research has been spurred by needs
for higher performance, smaller size, and lighter weight power
sources to be compatible with the shrinking size of all forms of
communication and data handling systems. Everyday usage of
portable battery operated equipment in everything, from home
appliances and hand-tools to mobile communication devices, has
also prompted increased switching power supply research. Static
DC-DC converters and DC-AC inverters provide natural interfaces
with new direct energy sources such as solar cells, fuel cells, and
thermoelectric generators, and form the integral part in most
uninterruptible power sources.
Energy Conversion Department has emerged as a leading
department for power electronics and drive research and education
in Malaysia. Research programs undertaken by the department are
diversified and include the following disciplines: modeling,
simulation, and control of power electronics; grid-connected
inverters, inverters for photo-voltaic (PV) applications; and electric
drives.
viii Preface
A vast amount of technology in switching power supply and

electric drives has been generated at ENCON in the past three
years. The research results have been published in numerous
conference proceedings and technical journals. These papers serve
as invaluable teaching tools and reference guides for power
electronics engineers. To facilitate future referencing, ENCON has
organized in three books. Each book is grouped according to these
topical areas:
Part I : Modeling and Control of Power Converters and Drives

Part II : Recent Advances in Power Inverters
Part III : Power Electronics Applications in Renewable Energy
ENCON will strive to maintain, update and expand the series.

Future plans for Energy Conversion Publication Series are
additional books with new topics of interest to industry and power
electronics community.
Nik Din Muhamad

Faculty of Electrical Engineering
Universiti Teknologi Malaysia
2008
Control Loop Design of DC-DC Converter Systems Using PSpice 1
1
CONTROL LOOP DESIGN OF DC-DC
CONVERTER SYSTEMS USING PSPICE
Nik Din Muhamad

Abdul Jaafar Shafie
1.1 INTRODUCTION
In recent years, there are many SPICE-based models developed
for simulating dc-dc converters. In general, three types of model
can be identified: the detailed model, the large-signal averaged
model, and the small-signal model. All three models are valuable
in analysis and design of dc-dc converter systems. The small-
signal model is required to design the control system of a dc-dc
converter. There are a number of well-documented techniques and
guidelines for designing the control system of a dc-dc converter [1-
6]. Nevertheless, the design of a feedback compensator is still not a
simple task, especially for a new designer. The task involves the
tedious, mechanical and human error prone computation of the
transfer functions, and repetitive fine-tuning of compensation
network component values. Moreover, the designer’s judgment
and experience are often required in the design process.
In this paper we demonstrate that the design guidelines can be
programmed in compact form in PSpice. For this purpose, an
option available in PSpice called Analog Behavioral Modeling
(ABM) is used. We choose to program the control loop design
procedure [4] due to its generic, systematic and a widely adopted
procedure [7-10]. The chosen design procedure is made as a
subcircuit model and stored in PSpice’s library. In this manner,
2 Modeling and Control of Power Converters and Drives
the design procedure is treated as a library component, which

makes it easy to use. The methodology of development is
presented in detail. The presence of the design guideline’s
subcircuit model along with the existing models makes a new
approach to using PSpice simulator in the design process. The
proposed approach includes both design as well as simulation tools
thus making extensive use of the SPICE in dc-dc converter design
cycle. Moreover, both the flexibility and capability of SPICE as a
stand-alone program can be enhanced. Using SPICE-aided design
approach in dc-dc converter development can reduce the time and
cost. Throughout this paper, a buck converter with voltage mode
control is used as an example to develop and verify the control
loop design procedure. The general procedure, however, can still
be readily extended to include other converters with different
control schemes provided that their transfer functions and design
guidelines are available.
1.2 DESIGN BASED ON SMALL SIGNAL MODEL
1.2.1 Buck Converter Power Stage

The buck converter power stage with PWM modulator of this
study is depicted in Fig. 1. We assume that the converter is
operated in continuous conduction mode. Using the averaging and
linearization techniques, the control-to-output transfer function of
the buck converter including PWM modulator, can be obtained as:
⎛ s ⎞
⎜⎜1 + ⎟
⎝ ω zESR ⎟⎠
= g co
vô
(1)
1+ +
s2
Qω o ω o2
vˆc s
where
g co = ωo = , ω zESR = , and Q =
Vg
ω oL
1 1 R
,
Vp LC rC
+
Vg + C R
- vo
Driver
-
PWM Modulator
d
vc
+
- Vp
Sawtooth
Comparator Generator
Figure 1 Buck converter with PWM modulator
It can be seen from (1) that the control-to-output transfer function

is dependent on the operating point. As the operating point of the
converter is wide, the conventional way of designing the controller
involves selecting the worst case operating point that is under the
minimum line voltage (minimum Vg) and maximum load current
(minimum R) conditions
1.2.2 K-factor Approach

The loop shaping approach is simple and effective for dealing
with the plants having complex dynamic behavior. Among popular
loop shaping method in power electronic applications is K-factor
approach, introduced by [4]. The main feature of K-factor
approach is that the pole-zero placement and resultant circuit
component values can be obtained without trial-and-error. That is
one of the reasons why the k-factor approach widely accepted by
many researchers [6-10]. The type-3 K-factor compensator is

frequently used for compensation of buck, boost and buck-boost
circuits due to its ability to provide the phase boost, Øboost:
0 ≤ φ boost ≤ 180 o (2)
The transfer function of the type-3 K-factor compensator is given
by
ω co ⎛ K ⎞
⎜ s + 1⎟⎟
2
⎜
Aco K ⎝ ω co
Gc ( s ) = ⎠
⎛ s ⎞
(3)
s⎜⎜ + 1⎟
2
⎟
⎝ K ω co ⎠
where ωco is the desired crossover frequency, K is the pole
frequency and zero frequency control factor. The value of K can be
adjusted depending on the phase boost (Øboost) required to make
the phase compensation.
frequency, ωco, the phase boost that the compensator should be

Given the desired phase margin, PM, and the crossover
φ boost = PM − 90 o − φ co
provided:
(4)
and the K-factor is calculated from equations 2 and 3 as:
⎛φ ⎞
K = tan 2 ⎜ boost + 45 o ⎟
⎝ 4 ⎠
(5)
It is important to point out here that this design procedure is

general in the sense that it can be used to any suitable application.
For specific application, this design procedure must be suited to
the requirements of those applications. In the case of buck
converter, an important constraint is that the crossover frequency
must be less than one-fourth of the switching frequency. This
constraint is required to avoid the large signal instability [1].
1.2.3 Type-3 K-factor Error Amplifier

Fig. 2 shows the circuit diagram of the type-3 K-factor error
amplifier or compensation network. The compensation network
consists of the circuit elements as follows: C1, C2, C3, R1, R2, R3,
and Rbias. This network has three poles and two zeros.
C3
C2 R2 C1 R1
R3
+ vo
vc
-
Rbias
Vref
Figure 2 Type-3 K-factor compensation network
The small-signal transfer function for the compensation network is

given by
= (−1) ×
vo 1
vc s (C 2 + C 3 ) R3
(1 + s( R1 + R3 )C1 )(1 + sR2 C 2 )
⎛ ⎛ C C ⎞⎞
(5a)
(1 + sR1C1 )⎜⎜1 + sR2 ⎜⎜ 2 3 ⎟⎟ ⎟⎟

⎝ ⎝ C 2 + C3 ⎠ ⎠
or
ω (1 + s ω z1 )(1 + s ω z 2 )
= (−1) i
s (1 + s ω p1 )(1 + s ω p 2 )
vo
(5b)
vc
By matching (5a) and (5b) we can obtain
ωi =
1
R3 (C 2 + C 3 )
(7)
ω z1 =
1
( R1 + R3 )C1
(8)
ω z2 =
1
(9)
( R2 C 2 )
ω p1 =
1
(10)
R1C1
C 2 + C3
ω p2 = (11)
R2 C 2 C 3
1.3 PROPOSED APPROACH

Figure 2 shows the flowchart of the overall procedure to
determine component values of the error amplifier. The flowchart
consists of three computational blocks: power stage, K-factor, and
compensator. The computation is, therefore, performed in three
steps:
I. The first step is to determine the magnitude, Aco, and phase,
Øco, of the power stage Gp(s) at the crossover frequency, fco,
which are executed in the block of power stage. The inputs to
the block are: the input voltage, vg, the value of L, the value
of C, the load equivalent resistance R, the ESR capacitor, r.
Besides, the user must also supply to the block the value of
crossover frequency, fco.
II. By knowing Aco and Øco, and specifying the desired phase
margin, PM, the block of K-factor can be employed to
compensation networks. The outputs of this block are: ωi,

systematically find the location poles and zeros of the
ωp12 and ωz12.

III. The block of compensator converts the location of poles and
zeros of compensator to component values of the error
amplifier.
Vg L C R Vp r
Power Stage
Gp(s) fco
Aco Øco
K-factor
PM
ωi ωp12 ωz12
Vo Compensator
Vref
R3
C1 C2 C3 R1 R2
Figure 2 Flowchart for the proposed approach
1.4 PSPICE IMPLEMENTATION

The PSpice simulator is provided with an extension called
Analog Behavioral Modeling (ABM). With ABM the simulator
can be used like a programming language and to solve general
mathematical problems by translating them to an electrical circuit.

ABM in PSpice is able to evaluate expressions that are functions
of circuit variables (voltages, currents, and simulation time) using
the controlled current and voltage sources (G and E devices).
ABM can also be used to solve system of linear and nonlinear
algebraic equations as well as systems of complex, transcendent
and ordinary, differential equations in their implicit or explicit
form. In each case, the equations are converted into electrical
circuits and solved by PSpice with a DC analysis for only
algebraic equations or a transient analysis for systems of algebraic
and differential equations. Editing the input file of PSpice is
relatively more comfortable than programming in MATLAB, C or
other program languages.
1.4.1 Power Stage Block

The function of this block is to compute the magnitude and
phase of the control-to-output transfer function, Gp(s). The control-
to-output transfer function for the buck converter has one zero and
previously given in (1). With s = jω, we see that transfer function

one complex pole-pair. The equation of the transfer function was
is a complex number containing a real part and an imaginary part.

The magnitude of a complex number is the square root of the sum
of the squares of the real and imaginary parts. The phase is the
inverse tangent (arctan) of the ratio of the imaginary part to the
real part. Therefore, the magnitude of the control-to-output transfer
function can be calculated as
⎛ ω ⎞
1 + ⎜⎜ ⎟⎟
2
⎝ ω zesr ⎠
Gco
= A=
vo
⎛ ⎛ ω ⎞2 ⎞ ⎛ ω ⎞2
(12)
⎜1 − ⎜ ⎟ ⎟ + ⎜ ⎟
vc
⎜ ⎜⎝ ω o ⎟⎠ ⎟ ⎜⎝ Qω o ⎟⎠
⎝ ⎠
The phase of the control-to-output transfer function can be

calculated as
ω
ω Qω o
φ = tan −1 − tan −1
ω zesr ⎛ω ⎞
(13)
1 − ⎜⎜ ⎟⎟
2
⎝ ωo ⎠
To implement (12) and (13) in PSpice, the frequency is assigned as
a variable, and other parameters are assigned as constants. The
.PARAM statement is used for this purpose. By using the
.PARAM statement we can create parameters and assign algebraic
mathematical expressions to it.
1.4.2 K-factor Block

This block is used to implement K-factor approach in
designing feedback compensator in PSpice. The inputs to this
function at the crossover frequency, fco (or ωco). By knowing Aco

block are gain, Aco, and phase, Øco, of the control-to-output transfer
and Øco, and speciying the phase margin, PM, the phase boost
required can be calculated as
φ boost = PM − 90 o − φ co (14)
Then, K-factor is calculated as

⎛φ ⎞
K = tan 2 ⎜ boost + 45 o ⎟
⎝ 4 ⎠
(14)
The two poles of the compensator are located at

ω co
ω p12 = (15)
K
The two zeros of the compensator are located at
ω z12 = K ω co (16)
The integrator gain of the compensator is
ω co
ωi = (17)
Aco K
(14)-(17) are also implemented in PSpice by using the .PARAM
statement.
1.4 .3 Compensator Block

The function of this block is to convert the values of poles and
block are ωp12, ωz12, and ωi that were obtained earlier from K-
zeros to component values of the error amplifier. The inputs to this
factor block. Besides, the user must provide reference voltage, Vref,
and Rbias.The conversions occur as follows:
C3 =
(ω i R3 f p12 )
f z12
(18)
⎛ f p12 ⎞
C 2 = C 3 ⎜⎜ − 1⎟⎟
⎝ f z12 ⎠
(19)
R2 =
2πf z12 C 2
1
(20)
R1 =
R3
⎛ f p12 ⎞
(21)
⎜⎜ − 1⎟⎟
⎝ f z12 ⎠
C1 =
2πf p12 R1
1
(22)
Rbias =
Vref
Vo − Vref
R3 (23)
(18)-(23) are also implemented in PSpice by using the .PARAM

statement. To make the values of C3, C2, C1, R1, R2, and Rbias
available in the schematic after the simulation finished, the
relevant parameters are represented by dependent voltage sources.
The three blocks above were implemented in a single
subcircuit. The PSpice netlist of the subcircuit and its symbol are
shown in Appedix 1.
1.4 APPLICATION EXAMPLES

Two examples are included in this section to demonstrate the
effectiveness of the proposed approach in designing feedback for
dc-dc converters. The first example is a buck converter operated at
the switching frequency of 100 kHz. The second example is a
similar buck converter, but the switching frequency is doubled and
the inductor is halved.
1.4.1 100 kHz Buck Converter

The buck converter is designed to operate in continuous
Vg = 10V to 15V, Vo = 5V, L = 30µH, R = 1.25Ω, fs = 100 kHz,

conduction mode. Following are the parameters of the converter:
Vp = 3V, Vref = 2.5V, Rbias = 10 kΩ. By knowing the converter

parameters, the design of feedback compensator can be performed
by using the developed subcircuit in PSpice. Before doing the
design, two parameters should be chosen. The first is to choose the
crossover frequency, fco. This parameter determines how quickly
the system responds to transient. In practice, fco lies between fs/10
and fs/3. In our design we choose fs = fs/6, as a compromise.
the overall loop. Typical PMs range 45° to 75°. Lower PM, like
Another parameter to choose is the phase margin, PM, desired in
45°, give good transient response at the expense of peaking of the
like 75°, give flat closed-loop transfer function and minimum

closed loop transfer function and output impedance. Higher PMs,
settling time. A good compromise is 60°.

peaking of output impedance, but at the expense of speed and
Using the above parameters, the developed subcircuit was

tested using the DC analysis in PSpice. The result is shown in Fig.
4. The time required to finish the simulation was just 0.13s, using
PSpice Version 9.2.
R = 1.25 VG = 10V VREF = 2.5V VP = 3V
C = 100uF RESR = 19m L = 30uH VO = 5V
24.42V K
134.2V PBOOST C1 4.526nV
106.5mV ACO Type-3-all C2 2.383nV
-164.2V PCO U14 C3 101.7pV
WI R1 427.0V
40.25KV
21.19KV WZ12 R2 19.81KV
517.5KV WP12 RBIAS 10.00KV
R3 = 10K
PM = 60 FCO = 16.6667KHZ
Figure 3 The result of DC analysis to give component values of the

compensator.
To check the developed subcircuit working properly or not, the

obtained component values of the compensator were used in the
average model. Frequency response of loop-gain was simulated by
From the bode plot, we can see that the crossover phase is –119.6°,
PSpice AC analysis. Bode plot of loop-gain is shown in Fig. 5.
indicating stable operation with PM of 60.4°. The crossover
(PM = 60.0° and fco = 16.667kHz) and the results of PSpice are
frequency is 16.623 kHz. The small differences between our target
due to round-up error.

Figure 4 Bode plot of loop-gain (gain in dB and phase in degrees)
1.4.2 200 kHz Buck Converter

The parameters of this converter are the same as the previous
and the inductor is halved (L = 15µH). It would be expected that

example, except the switching frequency is doubled (fs = 200kHz)
the transient response of this converter is faster than the previous

one. Without the developed subcircuit design procedure, any
changes in the parameters made the whole system must be
redesigned. With the developed subcircuit in hand, the process of
redesign doesn’t make any problem; the simulator can do it for us
and give the result as shown in Fig.6. The time taken to simulate
the whole process of redesign was 0.13s.
For comparison between the two examples for a transient
response, the switch model simulation (cycle-by-cycle) for a load
step was performed. The load steps were made to occur at 8.0 ms
(from 4A to 1A) and 8.3 ms (from 1A to 4A). The result is shown
in Fig. 7. As expected, the 200kHz buck converter has a faster
transient response due to the higher fco and the smaller inductor
value.
VG = 10V R = 1.25 VO = 5V L = 15uH
C = 100uF VREF = 2.5V VP = 3V RESR = 19m
17.40V K
126.1V PBOOST C1 1.877nV
55.37mV ACO Type-3-all C2 433.6pV
-156.1V PCO U14 C3 26.44pV
217.4KV WI R1 609.7V
50.21KV WZ12 R2 45.94KV
873.7KV WP12 RBIAS 10.00KV
R3 = 10K
PM = 60 FCO = 33.333kHz
Figure 5 DC analysis to compute compensator’s parameters for 200

kHz buck converter.
5.2V
5.1V
5.0V
4.9V
4.8V
4.7V
7.9ms 8.0ms 8.1ms 8.2ms 8.3ms 8.4ms 8.5ms
V(OUT
Time
Figure 6 Load steps response
1.5 CHAPTER SUMMARY

A new approach to using PSpice in designing feedback of dc-

dc converter system has been introduced. In this new approach, the
feedback design procedures are programmed and made as a
subcircuit in PSpice. The component values of the errror amplifier
have been easily obtained by means of PSpice DC analysis. The
extension of this approach to other converters and control schemes
like peak current mode (PCM), Average current control (ACC) and
power factor correction (PFC) is straightforward provided that
there exist small-signal models and design procedures.
REFERENCES
[1] Lloyd H. Dixon, “Closing the feedback loop,” Unitrode
Power Supply Design Seminar Handbook: SEM 700A, 1990.
[2] W. Tang, F. C. Lee and R. B. Ridley, “Small-Signal
Modeling of Average Current-Mode Control,” IEEE Trans.
Power Electronics, Vol. 8, no. 2, Apr. 1993, pp. 112-119.
[3] B. Holland, “Modeling, Analysis and Compensation of the
Current-Mode Converter,” Proceeding of the Powercon 11,
1984, pp. I-2-1-I-2-6.
[4] H. Dean Venable, “The k-factor: A New mathematical Tool
for Stability, Analysis, and Synthesis,” Proceeding of
Powercon 10, San Diego, CA, March 22-24, 1983.
[5] Lloyd H. Dixon, “Average Current Mode Control of
Switching Power Supplies,” Unitrode Application Note,
1999.
[6] J. Sun, R. M. Bass, “Modeling and Practical design issues for
Average Current Control,” Proc. of APEC, Vol. 2, pp. 980-
986,1999.
[7] S.A. Chickamenahalli et. al., “Effect of target impedance and
control loop design on VRM stability,” Proc. Of APEC,
2002, Vol. 1.
[8] Abraham I. Pressman, Switching Power Supply Design,
McGrawHill, 1998.
[9] N. Mohan, T. M. Undeland, W. P. Robins, Power
Electronics: Converters, Applications and Design, John
Wiley, 1995.
[10] C. M. Liaw, T. H. Chen, W.L. Lin, “Dynamic modelling and
control of a step up/down switching-mode rectifier,” IEE
Proc. - Electric Power Applications, Vol. 146 Issue: 3, May
1999, pp. 317–324.
APPENDIX 1
* source ERROR-AMP
.SUBCKT Type-3-all Aco C1 C2 C3 k Pboost Pco R1 R2
+ Rbias wi wp12 wz12 PARAMS:
+ Vg=12 Resr=20m R3=10k Vo=12V Vp=3V Vref=3V L=100uH fco=10kHz
+ PM=50 C=400uF R=5
.PARAM C2={C3*((wp12/wz12)-1)} w={2*pi*fco} R1={R3/(k-1)} C3=
+ {wz12/(wi*R3*wp12)} mag1={SQRT((1+(w*w)/(wzesr*wzesr)))} R2=
+ {1/(wz12*C2)} Pboost={PM-90-Pco} wi={w/(Aco*k)} wz12={w/sqrt(k)}
+ Rbias= {Vref*R3/(Vo-Vref)}
+ k={(Tan(((Pboost/4)+45)*pi/180))*(Tan(((Pboost/4)+45)*pi/180))}
+phase={-atan((w/(Q*wo))/(1-(w*w)/(wo*wo)))*180/pi}
+Mcomp={(1/mag)} Pco={Pcomp+phase1}
+ phase1={atan(w/wzesr)*180/pi} wo={1/SQRT(L*C)}
+ Aco= {Mcomp*mag1*Vg/Vp}
+ Pcomp={IF((1-((w*w)/(wo*wo)))<0,phase-180,phase)}
+ pi=3.14159 wp12={w*sqrt(k)} wzesr={1/(Resr*C)} Q={R/(wo*L)}
+ mag={SQRT(((w/(Q*w))*(w/(Q*w)))+(1-(w*w/(wo*wo)))*(1-
+(w*w/(wo*wo))))} C1={1/(wp12*R1)}
****************************************************************
***
E_ABM5 R1 0 VALUE { R1 }
E_ABM3B1 R2 0 VALUE { R2 }
E_ABM5B1 C1 0 VALUE { C1 }
E_ABM9B1 RBIAS 0 VALUE { Rbias }
E_ABM6 PBOOST 0 VALUE { Pboost }
E_ABM6B1 WI 0 VALUE { wi }
E_ABM7B1 WZ12 0 VALUE { wz12 }
E_ABM1X K 0 VALUE { k }
E_ABM4A1 ACO 0 VALUE { Aco }
E_ABM8B1 WP12 0 VALUE { wp12 }
E_ABM5A1 PCO 0 VALUE { Pco }
.ENDS Type-3-all
****************************************************************
***
Symbol R=5 RESR = 20M
VG = 12V L = 100UH
C = 400UF VO = 12V
VREF = 3V VP = 3V
PBOOST C1
ACO U6 C2
PCO Type-3-all C3
WI R1
WZ12 R2
WP12 RBIAS
R3 = 10K
PM = 50 FCO = 10KHZ
2
SIMULATION OF POWER CONVERTERS
WITH SLIDING MODE CONTROL USING
PSPICE
Nik Din Muhamad

Mohd Junaidi Abdul Aziz
2.1 INTRODUCTION
Design controllers for power converter systems present
interesting as well as difficult challenges. This is due to the fact
that power converters are non-linear time-varying systems. One
approach is to use a linear control. With this approach, non-linear
power converters have to be transformed into their linear
equivalent, so that classical control theory could be applied to the
design. The main disadvantage of this approach is that the stability
of control design is only guaranteed at a specific operating point.
Another approach is to use nonlinear control. One of these
approaches is sliding mode control. Sliding mode control uses
variable structure nature of power converter operation. This
approach improves the robustness against parameter, line and load
variation. This explains the wide use of sliding mode control in
power electronics converter systems lately.
The theory of sliding mode control to power converters has
widely investigated in literatures. Most of the papers have focused
on the theoretical aspect [1]-[3] and practical implementation [4]-
[6]. Hardware design engineers are discovering that increasing the
use of circuit simulation on their designs minimizes the number of
prototypes required, and eliminate many repetitive verification
Simulation of Power Converters with Sliding Mode Control Using 19
PSpice
procedures. Simulations therefore have a place in the analysis of

existing systems as well as the design of new systems.
One of the widely used circuit simulation program is PSpice.
However, the use of PSpice for the simulation of power electronics
systems is always not a simple task. This is because power
electronics circuits typically operate in a highly discontinuous
mode. The simulator typically could not follow the sudden
switching transitions and would become unstable and crash. This is
known as convergence problems in PSpice. Nevertheless, this does
not mean that PSpice could not be used for simulation of power
electronics systems. With the proper techniques and suitable
approaches, just about any power electronics systems could be
simulated with little or no convergence problems.
The aim of this paper is to demonstrate the use of PSpice for
simulation of power electronics converters with sliding mode
control. Two power converters with different sliding mode control
laws are given as examples. These include a buck converter and a
single phase inverter systems. Some guidelines on the use of
simulation to achieve sufficient accuracy are also given.
2.2 SLIDING MODE CONTROL
2.2.1 Background
In sliding mode control, the trajectory of the system is
constrained to slide along a predetermined sliding surface in the
state space. Such mode, known as sliding mode, is completely
robust and independent of parametric variations and disturbances.
Consider a system with two state variables and several control
inputs as described by the following state equations:
x& (t ) = f ( x, t ) + g ( x, t )u (t ) (1)
A common form of the sliding mode control for a converter adopts

a switching function:
⎧u max for S ( x) > 0
u=⎨
⎩u min for S ( x) < 0,
(2)
where S(x) is a scalar function of the state x and u switches

between two different inputs, umin and umax. If the switch could be
switched on and off infinitely fast the state vector of the plant
would stay on the sliding surface given by the equation S(x) = 0.
This mode is called sliding mode and S(x) = 0 describes the desired
motion of the plant.
The conventional way to select the sliding surface is by using a
linear combination of the state variables as:
S ( x) = c1 x1 + c 2 x 2 = C T x = 0 (3)
where CT = [c1, c2] is the vector of sliding surface coefficients.

Sometimes an integral term is added in (3) to eliminate the steady
state error. (2) and (3) are chosen so that both existence and
reaching conditions are satisfied. A detailed discussion of the
sliding mode control principle can be found in [1].
The common method of implementing the sliding mode control
is based on the control law described in (2) and (3). As the
switching frequency is limited, the implementation of (2) is easily
accomplished by refining it into:
⎧u max for S ( x) > κ

u=⎨
⎩u min for S ( x) < − k
(4)
where κ is an arbitrarily small value. The introduction of a

hysteresis band with the boundary conditions S(x) = κ and S(x) = -
κ provides a form of control to the switching frequency of the
converter. This also resolves the practical problem of a very high
frequency switching operation.
2.2.2 Simulation in PSpice

The sliding mode control system is shown in Figure 1. Inside
each block is given the related equations. The control law equation
described in (4) is a hysteresis comparator. In practice, this
comparator is commonly realized by using comparator IC such as
LM 311 with two different resistors connected in positive feedback
PSpice
configuration. Unfortunately, this simple configuration often

causes numerical instability in PSpice. First is due to fast
switching transition between the high and low states caused by
high gain of the comparator and the second is due to the positive
feedback configuration.
umax
u y
umin Eq. (1)
Eq. (4) Eq. (3)
Figure 1 Sliding mode control system under study
The hysteresis comparator, based on our experience, always

becomes a bottle neck in simulation of power electronics
converters with sliding mode control in PSpice. Therefore, careful
implementation of this part is extremely important. There are
several ways to realize the hysteresis comparator in PSpice. This
paper proposes a reliable hysteresis comparator model that can
simulate in a wide range of frequency with little or no convergence
problem. The high switching frequency, up to megahertz range, is
usually required to simulate a near ideal sliding mode principle. As
such, the theoretical aspect of the sliding mode control can be
carefully examined, without a second order effect. In this paper,
the realization of the hysteresis comparator model in PSpice is
based on behavior model. It is realized through ABM parts. Some
guidelines to avoid convergence problem will also be given. The
detailed development of the model will be discussed in the next
section.
The control law equation described in (3) computes the
instantaneous state-variable trajectory S(x). In practice, this is
realized through an analog or digital computer. In Pspice, this is
easily realized through an ABM part. The realization of (3)
through the ABM parts in PSpice rarely gives numerical instability
due to the continuity of the state variables.
The plant equation described by (1) is in general. For power
electronic circuits, the plant usually consists of a voltage source, a
resistor, an inductor, a capacitor, fully-controlled switches such as
MOSFET and diodes. There are several ways to implement the
plant in PSpice. The easy way is to implement it as it is. However,
this tends to raise convergence problem especially for power
electronics circuits that consist of more than one fully-controlled
switch, such as inverter circuits. In addition, it also requires a big
size of memory space to store the data and takes a long time to run
the simulation. In this paper, due to its simplicity, the buck
converter circuit will be implemented as it is. But, for the inverter
circuit, a system level simulation will be employed. This means
that neither fully-controlled switch nor diode is directly used in the
circuit. Therefore, the convergence problems related to the use of
fully controlled switch and diode are eliminated.
2.3 COMPARATOR
2.3.1 Comparator and Hysteresis Comparator

Figure 2 shows a basic comparator. A comparator works by
comparing the voltages at its two terminal inputs. When the
voltage at the positive terminal, v+, is greater than the voltage at
the negative terminal, v-, the output voltage is high and when it is
lower the output is low. This is due to large open-loop gain A in
the comparator. The mathematical equation that describes this
property is given by:
⎧+ V v+ > v−
v o = A(v + − v − ) = ⎨
⎩− V v+ < v−
(5)
PSpice
vo
+ +
o
- v+- v-
-V
Figure 2 A basic comparator and its characteristics

Figure 3 shows a comparator with hysteresis. Unlike the basic
comparator, the voltage v+ of the hysteresis comparator is not
independent. As we can see from the Figure 2, the voltage v+
depends on vo, R1 and R2. In particular, the positive terminal
voltage is
v+ =
R1
R1 + R 2
vo (6)
R2
R1
+
o
0 vi -
vo
+V
v+ - v-
-V
HB
Figure 3 A hysteresis comparator and its characteristics
Since the output voltage vo can assume any two values, so does
v+. When the output voltage vo = +V, the positive terminal voltage
is
v+ = V ≡ VUB
R1
R1 + R 2
(7)
When the output voltage vo = -V, the positive terminal voltage is
v+ = − V ≡ V LB
R1
R1 + R 2
(8)
(3) and (4) define the upper boundary (UB) and lower boundary
(LB) of the hysteresis band (HB). The hysteresis band, HB is given
by
VUB − V LB = 2V ≡ HB
R1
R1 + R 2
(9)
If we define R1/(R1+R2) = K and let V =10 Volt, we can find that K

= (1/20)HB = 0.05HB. Hence, the terminal positive of comparator,
v+, for both (7) and (8) as a function of HB can be written as
v + = 0.05 × HB × v o (10)
2.3.2 PSpice Implementation

There are several ways to implement the comparator in PSpice.
In this paper, we implement the comparator which is described in
(1) by using If-Then-Else function.
ABM2
+
-
IF(V(%IN1)>V(%IN2), 10, -10)
Figure 4 A basic comparator model in PSpice

PSpice
By using ABM2 part which has two inputs and one output, we
can write the expression for the comparator as shown in Figure 4.
As usual, we use the value of V = 10 Volt.
To implement the hysteresis comparator, the only difference is
that the terminal positive of the comparator is not independent; it
depends on the output voltage as in (10). However, this can be
easily added in PSpice by using ABM1 part which has a single
input and a single output as shown in Figure 5.
ABM1 ABM2
+
0.05*HB*V(%IN) IF(V (%IN1)>V(%IN2), 10, -10)
-
PA RAM ET E RS:
HB = 1
Figure 5 A hysteresis comparator model in PSpice
Notice that ABM1 part takes the output of ABM2 as its input,
does the computation and outputs the value to the positive terminal
of ABM2. Parameter HB is used in ABM2 and it is defined by
PARAM part. HB is the only parameter needs to be changed to suit
various applications.
2.3.3 Simulation
The above described behavioral model of the hysteresis
comparator has been simulated using PSpice to demonstrate its
validity. A sine wave signal with frequency of 500 kHz and
amplitude of 10 V is applied to the inverting input of the hysteresis
comparator. The hysteresis band HB is set as a default value i.e. 1.
The simulation parameters (.option) are set as their default values
too.
(a)
(b)
default maximum step size (b) Using maximum step size at 0.001 µs.
Figure 6 Simulation results of the hysteresis comparator (a) Using
The simulation results for the input and the output are shown in
Figure 6a. From the results we can see that the rise and the fall of
the output waveform are less sharp. It means that the accuracy of
the simulation result is quite poor when the default values of
simulation parameter (.option) are used. To make the accuracy of
the output waveform to be acceptable, we have to set just one
simulation parameter (.option) i.e. the maximum step size. From
our experience, the maximum step size should be set at least 1/100
output waveform when the maximum step size is set at 0.001 µs is

of the period of any fastest source in a circuit. The result for the
shown in Figure 6b.

In order to use the model properly, the trade-off between
accuracy and simulation times is always to be made.
PSpice
2.4 SIMULATION EXAMPLES
2.4.1 Buck Converter

In the following, the simulation of a buck converter is
described to illustrate the modeling approach described previously.
Figure 7(a) shows the buck converter with the sliding mode
control. With the chosen values for the components, the converter
is operating in the continuous conduction mode. This circuit uses
only standard elements from the library of the student version.
The block called SMC accomplishes the task of sliding mode
controller. The inputs to this block are the output voltage (terminal
a) and capacitor current that was translated into voltage by the
element H (terminal b). The only output is the rectangular voltage
(terminal c). This output is fed to the MOSFET gate driver to turn
on and off the mosfet. The MOSFET gate driver is implemented by
the element E1.
IRF150 0.15 L1
out
110u F
RL IC = 0 V
V1 C1
+
24V D4 100uF Rload
-
Dbreak IC = 0 6
+
-
+
-
E1 H1 H
Ic
+
GAIN = 12 GAIN = 1 -
0
c b a
SMC
(a)
ABM2a a b
1
3
2
(1/(Beta*RL))*(Vref- Beta*V(%IN2)) -V(%IN1)
ABM1 ABM2b
0.05*HB*V(%IN)
+ c
IF(V(%IN1)>V(%IN2), 10, -10)
-
Hysteresis Comparator
PARAM ET ERS: PARAM ET ERS:

HB = 0.3 Beta = 0.275
Vref = 3.3
RL = 6
(b)
Figure 7 (a) Buck converter with sliding mode controller (b) Sliding
mode control.
Figure 7(b) shows the circuit that implements the sliding mode
controller (SMC). ABM2a part implements a standard sliding
mode control law for buck converter which is given by
S (v o , i c ) = (V ref − βv o ) − i c [3]-[4].
βR L
1
(11)
All parameters used in the schematic are defined by using PARAM

parts. Theoretically, with the control law in (11), neither the output
voltage nor the inductor current experiences overshoot.
In the schematic, the hysteresis band HB used for the hysteresis
comparator is 0.3, resulting in the switching frequency around 200
kHz. This parameter can be adjusted to produce the desired
switching frequency.
Figure 8 shows the simulation results for the buck converter using
the sliding mode control law as described in (11). It can be seen
PSpice
that no overshoot occurs during start-up event for both the inductor
current and the output voltage, verifying the theory.
Figure 8 Simulation results for start-up event
Figure 9 shows the simulation results for the converter when
equivalent load resistance was step-downed from 6 Ω to 3 Ω,

the load disturbances occur at 5 ms and 7 ms. At 5 ms the
while at 7 ms the reverse. Obviously, the output voltage response

follows the first order system during the sliding mode, agreeing
with the theory.
Figure 10 shows a phase plane plot of ic versus (Vref -βvo). A
point in phase plane, which is called a representative point,
completely defines the state of the system at one instant in time.
The steady state operating point is the origin, where the error is
zero. The control equation (11) represents a straight line through
the origin with slope -1/βRL. This line is known as a sliding line.
Figure 9 Simulation results for load disturbances
We can see that the representative point moves towards the

sliding line during start-up, slide on it to the origin, then stay at the
origin during steady-state.
We can also see that when a step change in load occurs, the
representative point will leave the sliding line (mode). The
representative point will follow two families of trajectories in the
phase plane, depending on whether the control system imposes u =
umin (corresponding to duty cycle, d = 0) or u = umax
(corresponding to duty cycle, d = 1). This condition will last till
the representative point reaches the sliding mode region again.
This explains the use of wide sliding mode region in the design.
The phase plane plot is a very important tool to study and design
sliding mode control, especially for second order systems.
PSpice
Figure 10 A phase plane plot of the buck converter
2.4.2 Single Phase Inverter

Figure 11 shows a conceptual single phase inverter system with
bipolar switching. The operation of the inverter is represented by
two voltage sources and a single pole double throw (SPDT) switch.
By operating the SPDT switch back and forth between the position
1 and 2, this conceptual inverter will ideally produce a bipolar
output voltage which is similar to a physical inverter. The LC low
pass filter at the output is used to reduce the output harmonics. We
choose to model this conceptual inverter in PSpice, instead of a
physical inverter, for system level simulation. Without
semiconductor switch, the model is so simple. Such a model
completely eliminates any convergence problems related to the use
of semiconductor switches.
Rl L
Rload
1 2
+ Resr
E
- - E
+
Figure 11 A conceptual single phase inverter
Figure 12 shows the single phase inverter with the sliding

mode control model in PSpice. As with the buck converter, this
circuit also uses standard elements from the library of the student
version.
The block called SMC accomplishes the task of sliding mode
controller. The inputs to this block are the output voltage (terminal
a) and capacitor current that was translated into voltage by the
element H (terminal b). The only output is the bipolar voltage
(terminal c). This output is multiplied by DC link voltage, VDC, to
produce the output voltage of the inverter before filtering.
Figure 13 shows the circuit that implements the sliding mode
controller (SMC). ABM2a part implements a common sliding
mode control law for the inverter, which is given by
S (v& c , v c ) = k (v& c _ ref − v& c ) + (Vc _ ref − v c ) [5]. (12)
The only design parameter in the (12) is k. By choosing a

suitable value for k, the inverter with sliding mode control can be
realized. As usual the hysteresis band HB is used to limit the
switching frequency. For the simulation purpose, we choose k =
0.001 and HB = 50. With these parameters the switching frequency
around 3 kHz can be achieved. In practice, it is not easy to
PSpice
determine the value of k for this system due to the fact that the load
step changes in this system results in unbounded disturbance. In
[5] k is determined solely by simulation study and experimental
investigation.
0.15 L1
Vc
6m
IC = 0
360*V(%IN+, %IN-) C1
E1 100uF
IN+ OUT+ IC = 0 RL
IN- OUT-
20
GAIN = 10k
10m
EVALUE Ic
+
- H
c b 0 H1 a
SMC
Figure 12 A single phase inverter with SMC
Figure 14 shows a phase plane plot of derivative of error versus

error as in (12). The control equation (12) represents a straight line
through the origin with slope -1/k i.e. the sliding line. We can see
that the representative point always stays on the sliding line under
steady-state condition. When a step change in load occurs, the
representative point will leave the sliding line. The representative
point will follow two families of trajectories in the phase plane as
with the buck converter system case, depending on whether the
control system imposes u = umin or u = umax. This condition will last
till the representative point reaches the sliding mode region again.
ABM1 ABM2b
+ c
0.05*HB*V(%IN)
IF(V(%IN1)>V(%IN2), 10, -10)
-
PARAM ET ERS:
HB = 50
PARAM ET ERS: ABM2a
k = 1m
b a
k*(V(Vc_dot_ref)-V(%IN1)) 1
3
+(V(Vc_ref)-V(%IN2)) 2
Vc_dot_ref Vc_ref
VAMPL = 0.891 V1 V2
FREQ = 50 VAMPL = 280
PHASE = +90 FREQ = 50
Figure 13 A schematic of SMC for the inverter
If the representative point is not on the switching line, its

motion is governed by a simple second order system. However, for
this inverter system the representative point always reaches the
sliding line and the representative point remains on it if it hits the
sliding line within the sliding mode region. This explains why the
sliding mode region should be as wide as possible.
Figure 15 shows a phase plane plot of the inverter system when
k = 0.05m was used. The system was subjected to the same load
disturbance as the previous example. The simulation result shows a
poor performance of the sliding mode control system. This
happens due to the fact that by reducing the value of k will result in
a smaller size of the sliding mode region, giving rise to several
second order motions in the phase plane. This shows that by only
adjusting the controller gain k, it is possible to adjust the size of
sliding mode region to obtain the desired performance.
PSpice
Figure 14 A phase plane plot for the inverter system with k = 1m
Figure 15 A phase plane plot for the inverter system with k =

0.05m
2.5 CHAPTER SUMMARY

Simulation of power electronics circuit with sliding mode
control has been presented. Two common sliding mode control
systems, i.e. a buck converter and a single phase inverter, have
been given as examples. The methodology to model the sliding
mode control systems in PSpice has been described. The
development of the hysteresis comparator model, i.e. the key
element in the sliding mode control system, has been described in

detail. The simulation results obtained with the models agree with
the theory. The models are thus well suited to the analysis and
design of power electronic circuits with sliding mode control.
REFERENCES
[1] V. Utkin, J. Guldner, and J. X. Shi, A Sliding Mode Control in

Electromechanical systems. London, U. K.: Taylor & Francis,
1999.
[2] H. Sira-Ramirez, “A geometric approach to pulse-width
modulated control in non-linear dynamical systems,” IEEE
Trans. Autom. Control, vol. 43, no. 2, pp. 234–237, Feb. 1998.
[3] P. Mattavelli, L. Rosetto, G. Spiazi, and P. Tenti, “General
purpose sliding mode controller for DC/DC converter
applications,” in Proc. IEEE PESC rec., Jun 1993, pp. 609–
615.
[4] S. C. Tan, Y. M. Lai, M. K. H. Cheung, and C. K. Tse, “On the
practical design of a sliding mode voltage controlled buck
converter,” IEEE Trans. Power Electron., vol. 20, no. 2, pp.
425–437, Mar. 2005.
[5] M. Carpita, and M. Marchesoni, “Experimental study of a
power conditioning system using sliding mode control,” in
Proc. IEEE Trans. Power Electron., vol. 11, no. 5, pp. 731-
742, Sept. 1996.
[6] M. Ahmed, M. Kuisme, K. Tolsa, and P. Silventoinen,
“Implementing sliding mode control for buck converter,” in
Proc. IEEE PESC Rec., Jun 2003, vol. 2, no. 2, pp. 634–637.
Dynamic Evolution Control for Step-down DC-DC Converter 37
3
DYNAMIC EVOLUTION CONTROL FOR
STEP DOWN DC-DC CONVERTER
Ahmad Saudi Samosir

Abdul Halim Mohd Yatim
3.1 INTRODUCTION
In recent years, the application of power electronics converter
has grown extremely. Some applications that are increasingly
being dominated by power electronics are: 1) switched-mode
power supplies; 2) adjustable speed motor drives; 3) efficient
control of heating and lighting; 4) efficient interface for
photovoltaic; 5) fuel cell and high voltage dc system for efficient
transmission of power, etc.
Much power electronic application operated with parameter
variation, non linearity, load disturbance, etc. Therefore, there is an
increasing need for a good controller design to perform tight
regulation under high unpredictable load variation. In designing of
classical control theory, e.g. PID controller, small signal linear
approximations have been applied to the nonlinear system. This
approach enables the designer to use a simple linear controller to
keep the system stable. But the main disadvantage using this type
of control is that it is applicable for operation only near a specified
operating point.
Since power converters are non-linear time-varying systems,
the design of controllers must have capability to cover up the
nonlinearity and time-varying properties of the system.
In this paper, a new approach for converter controllers

synthesis based on Dynamic Evolution Control theory is presented.
The Dynamic Evolution Control exploits the non-linearity and
time-varying properties of the system to make superior controller.
The Dynamic Evolution Control tries to overcome the mentioned
problem of linear control by explicitly using dynamic equation
model of the converter for control synthesis. The Dynamic
Evolution Control has several advantages, i.e. wide range stability,
zero steady state error, simple synthesis process and suitable for
digital control implementation, because it requires a quite low
bandwidth for the controller, because it requires simple
calculations, which can be realized digitally easily. Moreover, The
Dynamic Evolution Control is operated at constant switching
frequency, so that it causes less power filtering problems in power
electronics applications.
3.2 DYNAMIC EVOLUTION CONTROL THEORY
The objective of the Dynamic Evolution controller is to control

the dynamic characteristic of the system to operate on the target
equation, Y = 0. In Dynamic Evolution controller, the dynamic
characteristic of converter system is forced to make evolution by
following an evolution guideline. The selected evolution guideline
is an exponential function as shown in Fig. 1. The equation of this
exponential function can be written as:
Y = Ce − mt (1)
Where C is the initial value of Y, and m is proportional to the

initial decrease rate of Y.
In this exponential function, the value of Y is decrease
exponentially to zero as a function of time. The decrease speed of

Y is proportional to the decrease rate m.
Let Y represent the state error function of the converter. Then,
the state error function(Y) is forced to follow the evolution
guideline as show in fig.1, so the dynamic evolution of the state
error function (Y), with initial value Yo, will fixed according to
the equation (2).
Y = Yo.e− mt (2)
It means that the state error function (Y) is driven decrease to

zero exponentially with decrease rate m.
Fig. 1. Dynamic Evolution Guideline
Since, Y is the state error function of converter as describe in

equation (2), therefore, the derivative of Y is given as:
= − m.Yo.e − mt
dY
dt
= − m.Y
dY
dt
As a result, the Dynamic Evolution Function can be written as

equation (3).
+ m.Y = 0; m>0
dY
(3)
dt
Where, m is a design parameter specifying the rate of

evolution.
In order to synthesize the control law, the dynamic equation of
the converter system is analyzed and substituted into the Dynamic
Evolution Function (3) to generate the control law of controller,
which ensure the state error function (Y) of converter decrease to
zero follow the evolution guideline.
The obtained results work on the full non-linear system. Hence,
the designer does not need to make any simplification in the
modeling process to obtain a linear model as in classical control
theory.
3.3 SYNTHESIS OF SEP DOWN DC-DC CONVERTER

CONTROLLER
The Buck or step-down converter is one of the most common

topologies. It is used extensively in high efficiency power supplies.
The Buck Converter scheme is shown as Fig. 2. As already
mentioned, to synthesize the control law of the Dynamic Evolution
Controller, the dynamic equation of the converter system must be
analyzed and substituted into the Dynamic Evolution Function (3).
In this time, we use the average model of Buck Converter. The
dynamic equation is obtained as follows:
Vg .α = L + Vo; 0<α <1

diL
(4)
dt
Where Vg is input voltage, iL is the inductor current, Vo is

output voltage, α is the duty cycle and L is inductor inductance.
From (4), output voltage of converter can be written as:
Vo = Vg .α − L
diL
(5)
dt
The main part of the synthesis mission is to obtain the control

law α(iL,Vo) as a function of the state iL and Vo which give the
required values of converter output voltage for various operating
modes. In addition, the constraint of duty cycle (4) must be
satisfied in synthesis process.
Fig. 2. Step down DC-DC Converter scheme
The Dynamic Evolution synthesis of the controller begins by

defining the state error function (Y) as follows
Y = k .Verr (6)
Where k is a positive coefficient and Verr is error voltage

.
The derivative of Y is given by:
=k
dY dVerr
(7)
dt dt
Upon substitution of Y (6) and the derivative of Y (7) into the

Dynamic Evolution Function (3), yields
+ m.k .Verr = 0
dVerr
k
dt
+ (m.k − 1)Verr + Vref = Vo

dVerr
k (8)
dt
Directly substituting the converter voltage output Vo from (4)

into (8) we can get:
+ (m.k − 1)Verr + Vref = Vg .α − L

dVerr diL
k (9)
dt dt
Solving for duty cycle α, the control law is obtained which is

given by:
+ (m.k − 1)Verr + L + Vref

dVerr diL
α=
k
dt dt (10)
Vg
The expression for duty cycle α is the control action for the
converter controller.
Control law (9) forces the state error function (Y) to satisfy
equation (6). According to this equation, the state error function
(Y) forced to make evolution follow equation (2) and decrease to
zero (Y = 0) with a decrease rate m. so, the state error function (Y)
satisfy the equation
Y = k .Verr = 0
Consequently state error of the converter will converges to
zero.
Verr = 0 (11)
Substituting into (11), we can see that the

voltage output of converter converges to the converters steady
state:
Vo = Vref (12)
From the synthesis procedure, it is clear that the Dynamic

Evolution Controller works on the full nonlinear system and does
not need any linearization or simplification on the system model at
all as is necessary for application of traditional control theory.
Rearranging the control law equation (10), the control law can
be written by:
Vref ( m.k − 1)Verr

dVerr diL
α= + + dt + dt
k L
(13)
Vg Vg Vg Vg
It is interesting to note that the control law in (13) consists of

four distinct parts. The first part is the feedforward term, Vo/Vg,
which is calculated based on the duty cycle at the previous
sampling instant. This term compensates for variations in the input
and output voltages. The second and third terms consist of
proportional and derivative terms of the perturbations in the output
voltage respectively. The last term consists of the derivative terms
of the inductor current.
From (13), it is also seen that the input voltage, output voltage
and inductor current are involved in control output. It make the
Dynamic Evolution Control can compensate all of variation in the
input and output voltages also the change of inductor current. The
advantage is that it contributes better dynamic performance to the
controlled system.
The Dynamic Evolution Control law theoretically not requires
precise knowledge of the model parameters. The parameter which
required is only inductor inductance. This requirement places a
small limitation on the control system for two reasons: sometimes
it is not easy to identify the parameters; sometimes the system
parameters change with time. But, as will be demonstrated later,
this problem can be solved.
3.4 SYSTEM SIMULATION AND VERIFICATION
A comprehensive simulation analysis was conducted to verify

the controller performance. The performance of Dynamic
Evolution control law under step load variation condition is tested
through simulation by using MATLAB SIMULINK. The step-
down converter scheme and the Dynamic Evolution control law,
which is described by (10), are modelled in Simulink as shown in
Fig. 3. The model parameters are listed in Table I. The control goal
is to regulate output voltage Vo = 12V. The reference of the
output voltage is specified based on the desired output voltage,
which means that Vref = 12V. These reference values are not
varied during the operation. The selected parameters of the

controller are k = 0.1 and m = 6000.
Here the situation where the load changes suddenly from one
value of resistance to another is considered. This is particularly
interesting because it is a typical problem for power electronics,
where the power supply is supposed to compensate quickly for the
load variation.
At 50 ms, the load changes from nominal 5Ω to 2.5Ω, and
change from 2.5Ω to 5Ω at 110ms. The simulation waveform of
output voltage and inductor current are shown in Fig. 4. When a
step load variation occurs, the resistance of the load changes
suddenly, it is shown that the output voltage is not disturbed by the
step load variation. The controllers accomplish to regulate the
converter output voltage keep on steady-state at 12V reference.
This condition indicates that the state error of the converter goes
converge to zero and there is no steady-state error in the output
voltage.
TABLE I
SIMULATION MODEL PARAMETERS
Parameter Value
Nominal Input Voltage 20V

Reference Voltage 12V
Inductance L 500uH
Capacitance C 200uF
Initial Load 5Ω
Addition Load 5Ω
The simulation result of converter output voltage for variation

of m is shown in figure 5. From this picture we can see that the
converter output voltage when m=6000 is better than the other, and
trajectory of dynamic evolution control is depicted in figure 6.
Fig 3. Buck Converter Simulation Model
Output Voltage & Inductor Current

14
Volt
12
10
Output Voltage
8
Inductor Current
6
4
Ampere
0
0 0.02 0.04 0.06 0.08 0.1 0.12 0.14 0.16
Time (sec)
Fig 4. Simulation result of Vo and IL under Load change condition

Fig 5. Output Voltage for variation of m, with k = 0.1.
4
x 10 Trajectory of Dynamic Evolution Control
4
2
Start Point
1
change of error
-1
-2
-3
-4
-10 -5 0 5 10
error
Fig 6. Trajectory of Dynamic Evolution Control

3.5 CHAPTER SUMMARY
This chapter presents a Dynamic Evolution Control for step

down dc-dc power converter. From Simulation that have been
done, The Dynamic Evolution Control shown many advantages
over the traditional PID controller. It result a fast response and
better transient performance.
Simulation and experiment results show the Dynamic
Evolution Control Law is insensitive to the input and parameter
variations and the Dynamic Evolution Controller is robust for such
nonlinear dynamic system.
REFERENCES
[1] Raviraj, V.S.C.; Sen, P.C.; Comparative study of proportional
integral, sliding mode, and fuzzy logic controllers for power
converters, Industry Applications, IEEE Transactions on
Volume 33, Issue 2, March-April 1997 Page(s):518 – 524
[2] S. Sanders, J. Noworolski, X. Z. Liu, and G. C. Verghese,

“Generalized averaging method for power conversion
circuits,” IEEE Trans. Power Electron., vol. 6, pp. 251–258,
Apr. 1991.
[3] R. W. Erickson, S. Cuk, and R. D. Middlebrook, “Large-scale
modeling and analysis of switching regulators,” Proc. IEEE
PESC’82, pp. 240–250, 1982.
[4] A.S. Samosir ; A.H.M. Yatim, “Implementation of New
Control Method based on Dynamic Evolution Control with
Linear Evolution Path for Boost DC-DC Converter,” Proc.
IEEE PECon 2008, Johor Bahru, Malaysia, December 2008.
[5] A.S. Samosir ; A.H.M. Yatim, “Dynamic Evolution Control
of Bidirectional DC-DC Converter for Interfacing
Ultracapacitor Energy Storage to Fuel Cell Electric Vehicle

System,” Proc. AUPEC 2008, Sidney, Australia, December
2008.
[6] V. I. Utkin, “Variable structure system with sliding modes,”
IEEE Trans. Ind. Electron., vol. AC 22, pp. 212–222, Apr.
1977.
[7] A. Kolesnikov, G. Veselov, A. Kolesnikov, A. Monti, F.
Ponci, E. Santi, and R. Dougal, “Synergetic synthesis of DC-
DC boost converter controllers: Theory and experimental
analysis,” in Proc. IEEE APEC, Dallas, TX, Mar. 10–14,
2002, pp. 409–415.
4
DIGITAL PI CONTROLLER DESIGN
FOR POWER INVERTER USING
MATLAB-SIMULINK
Shahrin Md. Ayob
Naziha Ahmad Azli
Zainal Salam
4.1 INTRODUCTION
In dc-ac conversion equipments, it is utmost important for
inverter to be able to maintain a stable and a clean ac voltage
waveform regardless of type of loads connected to it. In addition,
the inverter should recover from disturbance with excellent transient
response and low distortion. Those requirements are oftenly
difficult to achieve due to more harmonics are injected into the
power line. These harmonics are commonly originated from the
modern loads, which employed high-efficiency power converters
such as computers and power supplies. To achieve such
requirements, feedback control technique is employed.
Proportional-Integral (PI) controller is one of prominent feedback
control technique used in industries [1, 2, and 3]. The controller
yields a very simple control structure and its development cost is
very low. It can be implemented either using analogue circuitry or
digital processors. In addition, the performance and the stability of
the controller are predictable via control metrics such as bandwidth,
phase margin and gain margin.
Despite of the advantages, PI controller is known to have several
limitations. It is a model-based controller where, its control
parameters are designed based on the mathematical model of the
Digital PI Controller Design for Inverter Regulation Using 51
Matlab-Simulink
inverter. Model mismatching the actual system will produce an

unacceptable control performance. In addition, the model itself is
derived based on single operating point. Excellent performance can
be only confirmed within small region around the specified
operating point. However, the neighbourhood of the region can be
widen by employing a multi-loop control structure [4].
Consequently, a robust PI controller can be achievable.
In this chapter, design of digital PI controller for power inverter
regulation is presented. Matlab-Simulink software is employed to
assist the design of the controller. In this design, a Matlab tool,
called sisotool is employed to design PI controller based on
frequency response approach.
4.2 INVERTER MODELLING

Figure 1 shows the typical single-phase inverter circuitry. It
comprised of a full-bridge inverter, an LC filter and a load. The
switches of the full-bridge are controlled using Pulse Width
Modulation (PWM).
L iL
s1 s3
iC iO
vdc va C vo R
s4 s2
Figure 1. Typical single-phase inverter circuitry
To obtain the model, it is assumed that the switching frequency of

PWM inverter is typically higher (over 10KHz) than the
fundamental frequency (50Hz). Due to that, the dynamics of the
PWM inverter switches can be ignored and can be modelled as a
voltage source va defined as,
va = k PWM * vcont . (1)

Where vcont . is a control signal from the regulator and k PWM is,
k PWM =
vdc
(2)
vˆtri
In equation (2), vdc is the dc voltage value of the inverter while vˆtri is
the peak value of the triangular carrier. It follows that the dynamic
of the system is now mainly determined by the LC filter. The
simplified equivalent circuit for the output filter for an inverter is
shown in Figure 2.
iL iO
+
L
iC
va
−
C vo R
Figure 2. Equivalent inverter circuit
Choosing the inductor current iL and output voltage vo as the state

variables, the state-space representation of the system can be written
as:
⎡ i&L ⎤ ⎡ 0 −1/ L ⎤ ⎡ iL ⎤ ⎡1/ L ⎤ ⎡ 0 ⎤

⎢& ⎥=⎢ ⎥ ⎢ ⎥ +⎢ ⎥ Va + ⎢ ⎥ iout (3)
⎣Vout ⎦ ⎣1/ C 0 ⎦ ⎣Vout ⎦ ⎣ 0 ⎦ ⎣ −1/ C ⎦
Based on equation (3), the dynamic model of inverter can be drawn

as in Figure 3. This model will be used to design digital PI
controller.
Matlab-Simulink
vcont . − iL
+ + −
va 1 1 iC vo
k PWM
Ls Cs
iO
1
R
Figure 3. Linear model of single-phase inverter system
4.3 DESIGN OF DIGITAL PI CONTROLLER

In general, there are two ways of designing digital PI controller,
namely direct and indirect digital design [5]. In direct digital
design, the digital controller is directly obtained using discrete
inverter model. For the second method, a continuous time-mode
controller is firstly designed using continuous time-mode inverter
model. Then using several transformation methods, its equivalent
digital controller is obtained.
In this work, since it is much familiar to design controller in
continuous time-mode, the second method is adopted rather than the
first one. Let consider the continuous time transfer function of a PI
controller,
⎡ ( K p / Ki )s + 1 ⎤
C (s) = Ki ⎢ ⎥
⎣ ⎦
(4)
s
The bilinear transformation, given in (5) is one of transformation

method used to find a discrete equivalent transfer function of a
continuous controller.
⎡ z − 1⎤
s= ⎢⎣ z + 1 ⎥⎦
2
(5)
Ts
Applying the bilinear transformation, an equivalent discrete form,

C(z) for C(s) will be derived as:
mz + n
C ( z) =
z −1
(6)
Where the parameters m and n are given by:
⎡ K p Ts ⎤
m = Ki ⎢ + ⎥
⎣ Ki 2 ⎦
(7)
⎡T K p ⎤
n = Ki ⎢ s − ⎥
⎣ 2 Ki ⎦
(8)
In equation (7) and (8), Ts is a sampling time. Equation (6) can be

expressed in difference equation:
u& (k ) = (m + n)e(k ) − ne&(k ) (9)
Where e(k) is error signal sampled at kth interval, while e&(k ) is the
change of error, e(k ) , which can be computed as:
e&(k ) = e(k ) − e(k − 1) (10)
The digital PI controller block diagram of equation (9) can be drawn

in Figure 4.
m+n
e( k )
u& (k ) u (k )
−n
e&(k )
u (k − 1)
e(k − 1)
Matlab-Simulink
Figure 4. Discrete Linear PI control structure
Referring to Figure 4, e(k-1) is obtained by using unit time delay,

represented by 1/z. The output of the digital PI controller is an
incremental change of the control signal u& (k ) . Using digital
approximation for integration, the actual control signal u (k ) is
obtained,
u (k ) = u& (k ) + u (k − 1) (11)
4.4 DESIGN EXAMPLE FOR INVERTER REGULATION

This section provides a design example of digital PI controller for
inverter regulation purposes. Figure 5 illustrates the complete
inverter regulation system, which comprises of two feedback loops.
L
s1 s3
vdc va iL C
R vo
s4 s2
s1 s2 s3 s4 vref
iLref
PWM PI I PIV
Figure 5. Inverter with two feedback loops

The loops are arranged in a cascaded fashion, where inductor
current is used as the inner loop and output voltage as the outer
loop. Both loops are regulated using digital PI controller; subscript
‘1’ is used for current loop and subscript ‘V’ for voltage loop. Table
1 summarised the parameters value used in the design.
Table 1 Parameters values used in simulation
Parameter Value
Output Voltage (vo) 80V

Inductor filter (L) 250µH
Capacitor filter (C) 33µF
20K
Switching frequency (fs)
Hz
Current Loop Sampling Time (Ts1) 25µs
Voltage Loop Sampling Time
50µs
(Ts2)
4.4.1 Double-loop structure design

For double-loop control structure, several general guidelines
are recommended as follows [6]:
i. Controller design starts from the innermost loop and then

followed by the outer loop.
ii. To avoid interference in control loop from switching
frequency noise, the bandwidth of the innermost loop must at
least one or two order of magnitude smaller than the
switching frequency.
iii. For the outer loop design, the bandwidth must at least one of
two orders of magnitude smaller than bandwidth frequency
specified to the inner loop.
iv. For the outer loop design, the inner closed-loop is assumed
ideal and is represented as unity gain.
Bandwidth and phase margin are analysed using bode plot. Bode
plot is used since it offers a convenient way to design linear
controllers. For an efficient design, sisotool from Matlab is
employed. The tool provides an easy way of designing linear
controller via its Graphic User Interface (GUI). Figure 6 shows the
snapshot of sisotool dialogue box. Using this tool, one can
simultaneously analyse the effects of changing the poles and zeros
of the compensator on the system performance and stability. In
Matlab-Simulink
addition, the compensator, transfer function C(s) can be directly

obtained.
Transfer function of
the compensator
Figure 6. The snapshot of sisotool from MATLAB
4.4.2 Design for Inner Loop

Figure 7 shows the structure of the inner loop. From Figure 7,
KPWM is the PWM modulation model, which can be expressed as:
K PWM =
vdc
(12)
vˆtri
Where vdc is the input dc voltage and vˆtri is the peak value of
triangular carrier. In this work, vdc is 100V while vˆtri is 1 units.
Therefore, KPWM can be computed to have,
K PWM = = = 100
vdc 100
(13)
vˆtri 1
Using Mason rules, the physical model of the inverter can be

simplified as shown in Figure 4.8.
Inverter physical model
vo
iLref + vcont . va − iL + iC 1 vo
+
PI I K PWM 1
− −
Ls Cs
iL
iO 1
R
Figure 7. Inner current loop with inverter physical model
Inverter physical model
iLref + vcont . va Cs + R iL
LCs 2 + RLs + L
PI I K PWM
−
iL
Figure 8. Simplified block diagram of current loop using Mason rules

Matlab-Simulink
The open-loop transfer function for the inner current loop, GolI can
be written as:
⎡ K s + Ki ⎤
⎥ [ K PWM ] ⎢
⎡ Cs + R ⎤
Gol1 ( s ) = ⎢ p
⎣1424 3⎦ ⎣ LCs + RLs + L ⎥⎦
2
(14)
s
PI1
Substituting parameter values from Table 1 and KPWM =100 from

(13) into equation (14), the proportional gain Kp and integral gain Ki
can be obtained. For the innermost loop control design, the
bandwidth is designed to be one order of magnitude lower than the
switching frequency fs. As the fs is 20K Hz, the bandwidth has been
initially set at 2K Hz. However, it is difficult to design a
compensator that can achieve a phase margin in the range of 45o to
60o for this bandwidth. The controller is redesigned with the
bandwidth of 1K Hz. Using this bandwidth, the phase margin of 45o
is achievable. Figure 9 shows the close-loop bode plot for the inner
current loop controller.
Bode Diagram
0
-10 Closed-loop bandwith =1K Hz

Magnitude (dB)
-20
-30
-40
0
Phase margin = 45o

Phase (deg)
-45
-90
1 2 3 4 5
10 10 10 10 10
Frequency (Hz)
Figure 9. Magnitude and phase bode plots for current loop
Using sisotool, the transfer function of the current loop

compensator C(s)1, for a bandwidth of 1K Hz and phase margin of
45o is obtained as:
0.114s + 8628
C ( s) I = (15)
s
Next the bilinear transformation method, defined in (4.6) is

employed to transform C(s)1 to its equivalent discrete transfer
function, C(z)I. Substitute (5) into (15) with Ts1 = 25µs, the
equivalent discrete transfer function is obtained:
0.222z − 0.0063
C ( z )1 =
z −1
(16)
Comparing (6) with (16), m is obtained as 0.222 and n is -0.0063.
4.4.3 Design for Outer Loop

For outer control loop, its closed-loop control bandwidth is
selected to be one order of magnitude lower than the inner loop
bandwidth i.e. 100 Hz. The current loop is assumed to be ideal and
is represented as a unity gain as shown in Figure 4.10. Linear PI
controller, denoted as PIV is employed to regulate the loop. Figure
4.11 shows the simplified block diagram after the conversion using
Mason rules.
Matlab-Simulink
Current loop
vref + iLref iL + iC 1 vo
−
PIV
−
1
Cs
vo
io 1
R
Figure 10. Outer voltage loop structure
Current loop
vref + iLref iL 1 vo
Cs + R
PIV
−
1
vo
Figure 11. Outer voltage loop after simplification using Mason rules
The open-loop transfer function for the outer voltage loop, Gol2 is
written as:
⎡ K s + Ki ⎤
⎥ [{1] ⎢
⎡ 1 ⎤
GolV = ⎢ p
⎣1424 3⎦ loop ⎣ Cs + R ⎥⎦
(17)
s
current
PIV
Substituting parameter values from Table 1 into (17), the

proportional gain Kp and integral gain Ki for outer voltage loop can
be found using sisotool. The close-loop bandwidth is specified at
100 Hz and the phase margin is at 60o. The poles and zeros of the
compensator are adjusted until the phase margin is obtained. Figure

12 shows the resulting close-loop bode plot for magnitude and
phase.
Bode Diagram
5
0
Magnitude (dB)
-5
-10
Closed-loop bandwith =100 Hz
-15
-20
-25
0
Phase margin = 60o

Phase (deg)
-45
-90
1 2 3
10 10 10
Frequency (Hz)
Figure 12. Close-loop bode plot for outer voltage loop
From sisotool, the transfer function of the compensator for the outer
voltage loop, C(s)V is obtained as:
0.415s + 28000
C ( s )V = (18)
s
Then using bilinear transformation, defined in (5), the equivalent

discrete transfer function can be found. Substitute (5) with TsV =
50µs and into (18), the equivalent discrete transfer function is
obtained:
Matlab-Simulink
0.765z − 0.065
C ( z )V =
z −1
(19)
Comparing (19) with (6), m is obtained as 0.765 and n is -0.065.
4.5 SIMULATION RESULT

Figure 13 shows the Simulink block set for inverter system used
for simulation. A circuit-breaker is used to realize a step-load
disturbance (from no-load to rated load). The open and close state
of the circuit breaker is controlled by a pulse generator. To sample
the feedback signals at each sampling period Ts, Zero Order Hold
(ZOH) is used. Figure 14 depicts the Simulink block set for digital
PI controller.
Pulse
Generator
c
i 2
Universal Bridge +
- 1 +
v
-
g L Breaker
+ Vo
A
R
C
-
B Zero-Order
Vdc Hold1
Discrete Vref
PWM Generator
iLref
Pulses Signal(s) Controller Controller
Zero-Order
Hold
Figure 13. Simulink block set for inverter

1 m+n 1
Ref Control signal
1
-n z
Unit Delay
1
z
Unit Delay1
feedback
[A2]
Zero-Order
Hold1
Figure 14. Simulink block set for digital PI controller
Figure 15 shows the simulation result for the output voltage vo.
A step-load disturbance is subjected to the system at 5ms. As can
be seen, the transient revealed no oscillation and a zero steady-state
error. Thus, the result verified the design that has been conducted in
this work.
Matlab-Simulink
100
80
60
40
Output voltage (V)
20
-20
-40
-60
-80
-100
0 2 4 6 8 10 12 14 16 18 20
time(ms)
Figure 15. Output voltage waveform with disturbance at 5ms
REFERENCES
[1] Deng, H., Oruganti, R. and Srinivasan, D. 2005. Modelling and

Control of Single-Phase UPS Inverter: A Survey, The 6th IEEE
International Conference on Power Electronics and Drive
System, Kuala Lumpur, pp. 848 – 853.
[2] Habetler, T. G. and Harley, R. G. 2001. Power Electronic
Converter and System Control, Proceeding of IEEE, 89(6): pp.
913 – 925.
[3] Sanchis, P., Ursaea, A., Gubia, E. and Marroyo, L. 2005. Boost
DC-AC Inverter: A New Control Strategy, IEEE Transaction
on Power Electronics. 20(2): pp. 343-353.
[4] Ryan, M. J., Brumsickle, W. E. and Lorenz, R. D. 1997.
Control Topology Options for Single-Phase UPS Inverters.
IEEE Transaction on Industry Applications, 33(2): pp. 493 –
501.
[5] Duan, Y. and Jin, H. 1999. Digital Controller Design for
Switchmode Power Converters. The 14th Annual Applied
Power Electronics Conference and Exposition, United States of
America, pp. 967 – 973.
[6] Mohan, N. Electric Drive: An Integrative Approach. 2001.
John Wiley Publication.
Overview of Direct Torque Control of Induction Machines 67
5
OVERVIEW OF DIRECT TORQUE
CONTROL OF INDUCTION MACHINES
Nik Rumzi Nik Idris
5.1 MODELING OF INDUCTION MACHINE
In order to develop the Direct Torque Control (DTC) strategy, it is

necessary to construct a dynamic model of the induction machine.
The induction machine model used for a high performance control
systems must include all the dynamic effects that occur during the
transient states. For simplicity, a few assumptions and conventions
are adopted [1][2][3]. Despite the simplification, the model
adopted here is adequate for the design of a control system and is
valid for any arbitrary time variation of the machine voltages and
currents. The assumptions and conventions adopted are as follows :
• A constant radial length of air-gap is assumed to exist between

the smooth surfaces of the stator and rotor. These surfaces are
•
assumed to contain the 3-phase windings of negligible depth.
Permeability of the stator and rotor iron is assumed to be
infinity and the saturation and iron loss are assumed to be
•
negligible.
The symmetrical stator windings are Y-connected with the
neutral electrically isolated. The rotor is of a squirrel cage
•
type.
The space harmonics of the air-gap mmf and flux density are
negligible. The real axis of the stator winding coincides with
•
the phase A winding axis
The rotation is said to be positive in the counterclockwise
direction.
The cross-section of the single pole-pair machine with the above

assumptions is shown in Figure 2.1. D and Q are the real and
imaginary axis of the stationary stator, while d and q are the real
and imaginary axis of the rotor.
Figure 1 Cross-section of induction machine
5.1.1 Complex Space Vector Equations

The analysis of the symmetrical induction machine is more
conveniently carried out in terms of complex space vector, or
space phasor quantities [1][2][3][4]. The space phasor is a means
of representing any of the three phase variables (currents, voltages,
mmf or fluxes) by a single vector which rotates in space and is
valid for both steady state and transient operations. Once in space
phasor forms, the quantities can be easily expressed in their two-
axis forms (d-q axes).
In general if xa, xb, and xc are the three phase quantities, their space
phasor form is defined as:
x=
2
3
(xa + axb + a 2 xc ) , (1)
where a = ej2π/3
The coefficient 2/3 in (1) corresponds to the non-power invariant
or amplitude invariant definition of the space phasor [3]. In terms
of its d and q axes, the space phasor can be conveniently expressed
as follows:
x = xd + jxq (2)
The d and q axis components are related to the three phase

quantities by separating the imaginary and real terms of (1), i.e.,
⎡2
( )
xd = Re[x ] = Re ⎢ xa + axb + a 2 xc ⎥ = ⎜ xa − xb − xc ⎟
⎤ 2⎛ 1 1 ⎞
⎣ ⎦ ⎝ ⎠
(3)
3 3 2 2
⎡2
(
xq = Im[x ] = Im ⎢ xa + axb + a 2 xc ⎥ =
⎤
)
1
( xb − x c )
⎣3 ⎦
(4)
3
With the definition of space phasor given above, the complex

space phasor quantity expressed in the general reference frame of
the stator and rotor voltage equations are given by [3] :
dψ sg
v sg = Rs isg + + jω g ψ sg (5)
dt
dψ rg
0 = Rr irg + + j( ω g − ωr )ψ rg (6)
dt
Where ψ sg and ψ rg are the stator and rotor flux linkages

respectively and are given by:
ψ sg = Ls isg + Lm irg (7)

ψ rg = Lr irg + Lm i sg (8)
The super-script ‘g’ in the above equations denotes that the

quantity is referred to the general reference frame. The torque and
mechanical dynamics of the machine are modelled by the

following:
Te = ψ s × is
3p g g
(9)
dωm 2 dωr
22
J =J = Te − Tload (10)
dt p dt
5.1.2 d-q axis Equations

For the purpose of simulation and microprocessor implementation,
the complex space phasors representation of the induction machine
is converted to its equivalent d-q axis form. Transforming
reference frame (ωg = 0), and re-arranging them into matrix form,
equations (5) – (8) to their equivalent d-q axis forms in stationary
the following is obtained:
⎡v sd ⎤ ⎡ Rs + sLs 0 ⎤ ⎡isd ⎤
⎢ ⎥ ⎢ ⎢ ⎥
sLm ⎥⎥ ⎢isq ⎥
0 sLm
⎢ vsq ⎥ = ⎢ 0 Rs + sLs
⋅
⎢v ⎥ ⎢ sLm ωr Lr ⎥ ⎢ird ⎥
0
ωr Lm Rr + sLr
,
⎢ rd ⎥ ⎢ ⎥ ⎢ ⎥
⎢⎣ vrq ⎥⎦ ⎣ − ωr Lm sLm − ωr Lr Rr + sLr ⎦ ⎢⎣irq ⎥⎦
(11)
where ‘s’ represents the derivative operator d/dt. The complex

space phasor equations (5) – (8) can also be put into state space
forms with the choice of flux linkages or currents as state
variables. For example, in the case of the stator and rotor currents
chosen as the state variables, equation (11) can be re-arranged (11)
to give (12) [2][5].
⎡i&sd ⎤ ⎡ R s Lr − ω r L2m i sq − Rr Lm − ω r Lm Lr ⎤ ⎡i sd ⎤ ⎡− Lr 0⎤
⎢& ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ 0
⎢i sq ⎥ = ⎢ ω r Lm ω r Lm L r − Rr Lm ⎥ ⎢i sq ⎥ ⎢ Lr ⎥⎥ ⎡v sd ⎤
⋅ + ⋅⎢ ⎥
2
⎢i&rd ⎥ L − L L ⎢ − R L ω r Lr Ls ⎥ ⎢i rd ⎥ L2m − Lr Ls ⎢ Lm 0 ⎥ ⎣ v sq ⎦
1 R s Lr 1
ω r Lm L s
⎢ ⎥ ⎢ ⎥ ⎢ ⎥ ⎢ ⎥
2
Rr Ls
⎣⎢i rq ⎦⎥ ⎣⎢− ω r Lm Ls − R s Lm − ω r Lr L s Rr Ls ⎦⎥ ⎣⎢i rq ⎦⎥ ⎣ 0 Lm ⎦
m r s s m
&
(12)
From equation (9), the torque equation in stationary reference

frame can be written as :
Te =
3 p
ψ s × is =
3p
(Ls is + Lm ir )× is (13)
22 22
Noting that the cross product is × is = 0 , (13) can be expressed in

its 2-axis form given by (14)
Te = Lm (ird isq − irq isd )

3p
(14)
22
5.2 PRINCIPLES OF DIRECT TORQUE CONTROL
This section reviews the principles of DTC. The 3-phase VSI and
its voltage space vectors are briefly reviewed. It will be shown that
by utilising the eight possible switches configurations in the 3-
phase VSI, the direct flux and torque control of induction machine
can be established. The independent control of stator flux and
torque is obtained by selecting voltage vectors that will satisfy both
the flux and torque demand. Various methods used to estimate the
stator flux are also discussed.
5.2.1 3-Phase Voltage Source Inverter Space Vectors

In 3-phase voltage source inverter (VSI), the upper and lower
switches are always complimentary to each other. Since there are
three legs of upper and lower switches, there are 8 ( i.e., 23)
possible switch configurations. It can be easily shown that the
( )
phase voltage space vectors of the 3-phase VSI are given by (15).
v s ,k = 2 3 Vd S a + S b a + S c a 2 where a = e and k = 0 ,1,2 ,....7.

j 2 3π
(15)
Sa, Sb and Sc corresponds to the switching states of the phases a, b,

and c of the VSI which depend on k and defined as in Table 1
Table 1 Switch pattern of voltage vectors
Sa Sb Sc k
0 0 0 0
1 0 1 1
1 0 0 2
1 1 0 3
0 1 0 4
0 1 1 5
0 0 1 6
1 1 1 7
Sa, Sb or Sc equals 1 when the upper switch of the leg is ON and

equals 0 when the lower switch of the leg is ON. The phase voltage
space phasors are shown in Figure 2 with the switching states
shown in square brackets. In general, the magnitude of each non-
( k − 2 )π
zero space phasor is the same (i.e. 2/3Vd) with a phase of
(where k = 1, 2,…6).
3
Figure 2 Voltage space vectors of a 3-phase VSI
5.2.2 Direct Flux Control

In stationary reference frame, over a small period of time and
neglecting the voltage drop across the stator resistance, equation
(5) can be written as
∆ψ s = vs ∆t (16)
Equation (16) shows that the change in stator flux linkage is

directly controlled by the stator voltage space vector. If the
magnitude and orientation of the stator flux are known, its locus
can be controlled by selecting appropriate voltage vectors. In DTC,
the stator flux is forced to follow a circular locus by limiting its
magnitude within its hysteresis band. When the stator flux touches
its upper or lower hysteresis band, a suitable voltage vector is
selected to reduce or to increase it respectively. A block diagram of
such a hysteresis comparator is shown in Figure 3. The output of
the hysteresis comparator, named as stator flux error status,
indicates whether the stator flux needs to be increased or

decreased. Stator flux error status of 1 indicates that the stator flux
error touches its upper band which means that the actual flux needs
to be increased and vice versa. Figure 4 illustrate typical
waveforms of stator flux, stator flux error and stator flux error
status.
Figure 3 Stator flux hysteresis comparator
Figure 4 Typical waveforms of stator flux and stator flux error

for the hysteresis comparator of Figure 3
In order to select the appropriate voltage vectors, the knowledge of

stator flux orientation or position must be known. For this purpose,

the stator flux plane is divided equally into six sectors as shown in
Figure 5
Figure 5 Six sectors of the stator flux plane

Each sector will have a different set of voltage vectors to increase
or to decrease the flux. For example, Figure 6 illustrates how the
voltage vectors vs,2 and vs,3 are used to increase and decrease the
stator flux when it occupies sector I and voltage vectors vs,3 and
vs,4 are used to increase and decrease the flux when it is in sector II
. Generally, there are more than one possible voltage vectors which
can be selected either to increase or decrease the flux in all the
sectors. For example in sector I, voltage vectors vs,1 and vs,4 can
also be used to increase and to decrease the flux. However, the use
of these voltage vectors which results in a very fast increase or
decrease in stator flux is usually avoided [6]
Figure 6 Voltage vectors selection to control the stator flux

locus within its hysteresis band
5.2.3 Direct Torque Control

The instantaneous torque expression of equation (9) can be
expressed in terms of stator and rotor flux linkages, which is given
by (17).
Te = − ψ sg × ψ rg = ψ sg ψ rg sin θ sr
3 Lm 3 Lm
2 σL s Lr 2 σL s Lr
(17)
The relation between the rotor and the stator flux linkages in the
rotor flux reference frame, is given by (18).
ψ rr = ψ sr
Lm
1 + pστ r
Ls
(18)
constant στr. Owing to the switching characteristic of the VSI, the

Equation (1.18) shows the rotor flux lags the stator flux with a time
instantaneous angular frequency of the stator flux is irregular.

However, due to the low pass filtering action as given in (1.18) –

which has a time constant much larger than the switching period –
the angular frequency of the rotor flux is continuous. Figure 7
+ ∆t. As shown in the figure, the rotor flux moves continuously as

shows the rotor and stator fluxes space phasor at the time t1 and t1
oppose to the irregular movement of the stator flux which is

directly influenced by the applied voltage vectors. If a forward
applied, the angle θsr is increased. Applying zero voltage vectors

voltage vector of the same direction as the rotor flux rotation is
continuously, this will decrease the angle θsr. If a voltage vector in

ideally stop the stator flux rotation. Since the rotor flux rotates
the opposite direction is applied, the angle decreases rapidly.
Figure 7 The variation θsr with application of active and zero

voltage vectors
Within a small switching period, the magnitude of the stator flux

and rotor flux of Equation (17) can be assumed constant.
Consequently, the torque equation given by (17) is mainly
controlled by the angle θsr which is directly controlled by the
decreases θsr
selected voltage vectors. Choosing voltage vectors that increases or
causes the torque to increase or decrease
respectively.
Assuming that rotor is rotating counterclockwise with the stator
the angle θsr.

flux in sector k, four cases can be considered which directly affect
Case 1. Voltage vectors vs,k+1 and vs,k+2 , which are

referred as active forward voltage vectors, are chosen.
increasing the angle θsr. The torque will increase.

This will cause stator flux to advance forward thus
Case 2. Voltage vectors, vs,0 and vs,7, which are

referred as zero voltage vectors, are chosen. From
(12), the change in stator flux voltage is zero. Ideally,
stator flux freezes, however, due to the small stator
resistance drop, the stator flux weakens. According to
counterclockwise direction, thus θsr reduces. The

(1.18), rotor flux continues to rotate in
torque will decrease.

Case 3. Voltage vectors, vs,k and vs,k+3 which are
referred as radial voltage vectors, are chosen. Stator
flux magnitude increases or reduces rapidly. Its
forward, thus decreasing θsr. The torque will

angular velocity almost halt and rotor flux advanced
decrease.
Case 4. Voltage vectors, vs,k-1 and vs,k-2, which are
referred as reverse active voltage vectors are chosen.
causes θsr to decrease rapidly. According to (1.18),

The stator flux rotates in reverse direction. This
direction, thus θsr reduces. The torque will decrease

rotor flux continues to rotate in counterclockwise
rapidly.
The above cases show that the torque can be increased or

decreased by selecting suitable voltage vectors provided that the

instantaneous stator flux orientation and the torque itself are
known. In DTC, the torque is limited within its hysteresis band. It
is reduced when it touches its upper band and increased when it
touches its lower band. This is done by selecting suitable voltage
vectors. In order to reduce the steady state switching frequency, a
rapid change in flux is normally avoided [6][7][8][9]. For example,
to reduce the torque for counter-clockwise (or positive) rotation,
zero vectors rather than k or k+3 vectors are chosen. By employing
three-level hysteresis torque comparator [7] (as shown in Figure 8),
zero voltage vectors will be selected to reduce the torque for
counter-clockwise (or positive) shaft rotation in steady state
operation. The output of the comparator, referred to as torque error
status, indicates whether the actual torque needs to be increased or
decreased. If the torque error is too large, such as during speed
reversal or braking, torque error status becomes –1 and reverse
voltage vectors (k-1 or k-2) will be selected. For clockwise rotation
(operation in third quadrant), zero vectors rather than k+1 or k+2
will be used to increase the torque.
Figure 8 Torque hysteresis comparator
An ideal waveform of torque, torque error and torque error status

for a step reduction (negative) followed by a step increment
(positive) in torque reference for a counter-clockwise (or positive)
rotation is shown in Figure 9. The load torque is assumed constant.

When a negative reference torque is applied, the motor is operating
in the second quadrant braking mode. During this mode, the torque
error status alternates between –1 and 0, which means reverse
voltage vectors and zero vectors are chosen. A positive reference
torque is again applied, which brings back the operation into the
first quadrant motoring mode, provided that the speed is in positive
rotation.
Figure 9 Typical waveforms of the torque, torque error, speed

and torque error status for the hysteresis comparator of Figure 8
5.2.4 De-Coupled Control of Torque And Flux : Switching

State Selector
In order to establish a high performance torque control in induction
motor drives, it is necessary to have independent control of torque
and flux components. It is shown previously that the torque and
stator flux can be controlled by selecting appropriate voltage
vectors and can be made to follow a desired locus or value within
their hysteresis bands. A de-coupled control of flux and torque in
DTC is obtained by selecting voltage vectors that can
simultaneously satisfy the torque and the stator flux demands.
These voltage vectors are tabulated in a three dimensional look-up
table; it is accessed by stator flux error status (whether to increase
or to reduce), torque error status (whether to increase or to reduce),
and stator flux orientation (sector in which stator flux occupies). If
the position of the stator flux is assumed to be in sector k, there are
7 different voltage vectors that can be selected: 6 non-zero and 2
zero-voltage vectors. Each of these voltage vectors gives different
effects to the torque and stator flux of the motor. Based on the
preceding discussion, Table 2 summarises the effect of application
of all possible voltage vectors when stator flux is in sector k for the
clockwise and counterclockwise rotation of the rotor shaft. The
thicker arrows indicate that the torque or flux increases or
decreases rapidly.
Table 2 indicates that any possible combination of torque and
stator flux demands at any sector can be fulfilled simultaneously
by selecting one of the 7 possible voltage vectors. A selection table
can now be constructed as given by Table 3, and proposed by [7].
The table is accessed by the torque and stator flux error status, and
sector in which the stator flux occupies. To reduce the switching
frequency, radial voltage vectors vs,k and vs,k+3, are avoided.
Instead of choosing these vectors to increase or decrease the flux
during which the torque needs to be reduced, zero-voltage vectors
are selected which effectively stop the flux angular frequency.
Table 2 Possible application of voltage vector
Table 3 Voltage vectors selection table as proposed in [7]
5.2.5 Stator Flux Estimation

To control the stator flux and torque precisely, the estimation of
the stator must be accurate. Also, the torque calculation , as given
by equation (9) depends on the accuracy of the estimated stator
flux. The simplest form of stator flux estimator, also referred to as
voltage model [10][9][11], is based on an open-loop integration of
the stator back emf and requiring only stator resistance and
terminal quantities (i.e. stator current and voltage). This is given by
equation (19) which is expressed in stationary stator reference
∫
frame.
ψ s = ( v s − Rs i ) dt (19)
This estimation technique, however, suffers from initial value and
drift problems [12]-[14]. Even a small dc offset present in the
measured current or voltage can cause the integrator to saturate. At
low to zero speeds, with low stator back emf, the stator resistance
drop becomes significant and a small deviations in its value can
third harmonics voltage are used to calculate the air−gap flux −

give significant error in the estimated flux [10][15][7]. In [16],
rather than stator flux − for the Direct Self Control (DSC) scheme
of induction motor drive. In this way, the problem of stator
frequency of the air−gap, particularly during transients states is a

resistance variation is avoided, however to estimate the angular
problem. Other approches that are proposed to overcome the

problem of stator resistance variation due to temperature increase
include the use of the difference between the actual and reference
stator current vectors to correct or tune the stator resistance by
using a fuzzy controller or PI controller [17][18] and ANN[19]. In
[20], a quasi-fuzzy method of online stator-resistance estimation of
an induction motor for stator flux FOC is proposed. The stator
winding temperature is estimated as a function of stator current and
frequency through an approximate dynamic thermal model of the
machine. The stator resistance is then derived from this estimated
stator-winding temperature.
The poor performance of the voltage model stator flux estimator at
low speed results in a sluggish torque response and is unsuitable
for some applications. This has led to the development of another
stator flux estimator, referred to as current model. To perform the
estimation based on current model, the knowledge of the speed or
position of the rotor and the stator current are required [21]. From
the measured rotor speed and stator current, rotor flux linkage is
first estimated using (20). The stator flux is obtained from the rotor
flux using (21). Both equations (20) and (21) can be derived from
the induction machine equations (5) - (8)
∫
⎛ L i − Ψr ⎞
ψ r = ⎜⎜ m s − jωr Ψr ⎟⎟ dt
⎝ τr ⎠
(20)
ψs = ψ r + σLs is
Lm
(21)
Lr
In discrete implementation, the integrator in (20) uses the previous
value of rotor flux, i.e. a closed-loop integrator, as opposed to the
open-loop integrator used in (19). For this reason, the stability and
At high rotor speed − in particular during field weakening

drift problems experienced by the voltage model are avoided [22].
operation − where the slip frequency is relatively small, the

performance of this estimator degraded. Even a small error in the
measured rotor speed can cause a significant error in slip frequency
and hence the estimated flux [15]. In [10], an investigation on the
sensitivity of the voltage and current model estimators on the
−which is valid for steady state conditions. It was found that the
parameter variations utilising the frequency function response
current model is relatively less sensitive to parameter variations at

low speed and thus is suitable to be used at low speed, or even zero
speed, operations. Voltage model on the other hand was shown to
perform well at high rotor speed where it has small sensitivity to
parameter variations.
To take advantage of both type of estimators, some researchers
have combined both types of estimators. In [7], a simple method of
combining both models, i.e. by using a first order lag network, is
proposed. The transition frequency between voltage and current
models occurs at around 1/Tc, where Tc is the time constant of the
closed−loop observer. The current model is used as an implicit flux

lag network. In [10], both models are combine to form a
reference to the voltage-model –based estimator and the smooth

transition between the models is accomplished via a linear PI
controller. An improvement in flux accuracy is obtained in [11] by
replacing the PI controller with a speed dependent gain controller
(the rotor flux is estimated and used for the direct field orientation
of an induction motor drive). Below and beyond the closed-loop
bandwidth, the current model and the voltage model dominates the
estimation, respectively. The closed-loop bandwidth can be set
using the controller parameters. In [15], the voltage and current
models are used alternately and the stator and rotor resistance are
periodically updated to improve the estimation accuracy. Other
variations of closed-loop observer includes Gopinath’s reduced
order observer [23], Luenberger observer [13] and Kalman
filter[24].
Although the performance of the closed-loop flux observer is
attractive, it has a drawback of requiring the knowledge of speed or
position of the rotor. This is regarded as unsuitable in some
applications where the installation of the speed or position sensors
are not favourable; speed sensors results in higher cost, increased
size of the drive machine, the drive is less robust and reduced
reliability. To resolve this problem, some researchers have used an
estimated speed to calculate or estimate the flux. In [25], the speed
is estimated using the Model Reference Adaptive System (MRAS)
incorporating the mechanical system model. In a more recent
sensorless DTC drive [26], the rotor position used in the current
model is obtained from the estimated stator flux which is used as
the implicit reference for the closed loop flux observer. The speed
is estimated using the MRAC. However as pointed out in [22], the
zero or low speed operation problems remain unsolved, because by
doing so, one actually attempts to solve three unknowns using two
equations. Without heavy filtering, the positive feedback inherent
in this type of control can result in instability. The three unknowns
are the two flux estimates and the rotor speed while the two
equations are the voltage and current models. It is shown in [22]
that comparable speed-sensorless flux estimation can be achieved
by simply modifying the pure integrator used in the voltage model
estimator. The method replaces the pure integrator with a low-pass
filter with its cutt-off frequency implemented entirely using
software. The suitable cut-off frequency therefore can be
dynamically change depending on the operating conditions. The
modified voltage model-based estimator was also applied to a
SVM based DTC drive of an induction machine [27] and has been
shown to give good results even under locked-rotor conditions

with a near zero frequency operation.
`
5.3 EFFECT OF HYSTERESIS BANDS ON THE DTC
DRIVE PERFORMANCE
The width of the torque and flux hysteresis bands in DTC

induction motor drive influence the drive performance in terms of
the current harmonics, torque ripple and switching frequency of the
device [28][29][30]. This section looks at how the bands affect
these performance criteria.
5.3.1 Total Harmonic Distortion in Phase Current

In [29], an investigation on the effect of torque and flux hysteresis
bands on the performance of the DTC drive was carried out. It was
shown that the flux hysteresis band strongly influences the Total
Harmonic Distortion (THD) of the phase current, which is defined
as:
I phase − I1, phase

THD =
I12, phase
Where Iphase and I1,phase are the r.m.s values of the phase current and
the fundamental harmonic component of the phase current,
respectively.
For a fixed torque hysteresis band, the THD increases with the flux
hysteresis band. As the flux band is further increased, the flux
locus approaches that of hexagonal shape, similar to that of the six-
step inverter-fed drive [29]. A simulation of the DTC drive was
carried out (machine parameters as given in Chapter V) with the
torque hysteresis band varied from 5% to 25% of the torque limit
allowed during acceleration/deceleration; the torque limit in this
case is set to 1 N-m. The stator flux band is also varied from 2.5%
to 15% of the command flux which is set to the rated value of 0.45
Wb. The THD of the phase current is calculated and the results are
plotted as shown in Figure 10 which indicates that the THD is

affected by the variation in stator flux hysteresis band more than
by the torque hysteresis band.
Figure 10 Variation of THD in phase current with torque and

flux hysteresis bands
5.3.2 Torque Ripple

It is shown in [29] that the torque ripple is only affected by the
torque hysteresis band and is independent of the flux hysteresis
band. It is also highlighted that the relation between the torque
hysteresis band and the torque dispersion is linear. In practical
microprocessor implementation, there is a delay between the
instant the variables are sampled and the instant the voltage
switching pattern is determine and pass to the inverter. Because of
this, there is an overshoot or undershoot in the estimated torque
above or below the torque hysteresis bands. As a result, the ripple
is non-zero even a zero torque hysteresis is applied. In three-level
hysteresis torque implementation, the width of the hysteresis
torque band must be selected appropriately. If the band is too
small, a torque overshoot which may cause the torque error to
exceed the hysteresis band will occur. This will result in a reverse
voltage vector to be selected (instead of zero vector) to reduce the

torque [9][31]. A reverse voltage vector will reduce the torque
rapidly and hence may in turned causes a torque undershoot and
a typical waveform − which is obtained from the constructed DTC

consequently, the torque ripple is increased. Figure 11(a) illustrates
drive as discussed in Chapter V − of the torque controller with

undershoot and overshoot but does not result in a reverse voltage
vector selection. The torque hysteresis band for Figure 11(a) is set
to 0.12 N-m. Since the oscilloscope is set to display 500 mV/div
equivalent to 600 mV − as marked by the vertical markers of the

which correspond to 0.1 N-m/div, the width of the band is
oscilloscope. For Figure 11(b), the torque hysteresis band is set to

0.08 N-m. Again, the band is marked by the vertical markers of
the oscilloscope. When the torque error exceeds the lower band, a
reverse voltage vector is selected which causes a rapid reduction
in torque and hence an increase in the ripple.
(a)
(b)
Figure 11 Torque and torque error (a) No reverse voltage

vector is selected, (b) Reverse voltage vectors are selected
because of the overshoots in torque
The amount of overshoot (or undershoot) can be reduced by

reducing the sampling time. In other words, the undesired
overshoot or undershoot in torque can be minimised by employing
a faster sampling period, which will require fast processors and
expensive devices.
A predictive control scheme is used in [32] and [28] for the
purpose of torque ripple reduction. The method determines the
instant at which a zero voltage vector should be applied within one
sample period to give a zero torque error. In [32], the zero torque
error is calculated based on positive and negative torque slope,
whereas in [28] it is based on the minimum torque ripple equation.
In [33], the duty ratio of the zero and non-zero voltage vectors
within one sampling period is determined by a fuzzy controller.
The duration of the non-zero voltage vectors required to force the
torque to its reference value is a function of torque error, flux error
and flux position, hence are treated as the inputs to the fuzzy
controller. The rules used by the fuzzy controller are formulated
based on the simulation of the system. From the simulation work
carried out, this method has shown to be capable of reducing the

steady state torque ripple significantly. A similar approach is
implemented in [34] with the input to the fuzzy controller being
the torque error and its derivative. Experiments were conducted to
validate the proposed scheme.
5.3.3 Switching Frequency

Despite its simplicity, DTC drive based on hysteresis comparator
suffers from switching frequency which varies with operating
conditions, namely the stator and rotor fluxes, stator voltage and
rotor speed [29][30][35]. It was shown in [29] that the switching
frequency is mainly affected by the torque hysteresis band and
increases with the width of the band. For a fixed torque hysteresis
band controller, it is therefore necessary to set the band to the
maximum (or worst) condition. This will ensure that the switching
frequency does not exceed its limit which is determined by the
thermal restriction of the power devices.
To look at the effect of flux and torque hysteresis bands on the
switching frequency, a simulation on the DTC drive was carried
torque limit allowed during acceleration or deceleration − the

out with the torque hysteresis band varied from 5% to 25% of the
torque limit is set to 1 N-m. The stator flux band also varied from
2.5% to 15% of the command flux which is set to the rated value
of 0.495 Wb. The motor speed is set to run at 25 rad/s, 100 rad/s,
and 301 rad/s. The switching frequency is calculated and a three
dimensional graph is plotted (Figure 12) which shows that the
switching frequency is mainly affected by the width of the torque
hysteresis band.
(a)
(b)
(c)
Figure 12 Switching frequency variation with torque and flux
hysteresis bands : (a) at 25 rad/s, (b) at 100 rad/s and (c) at 301
rad/s
As shown by Figure 12, the switching frequency also varies with

the motor’s speed. The variation in the switching frequency at
different operating conditions is due to the fact that the torque
slope varies with operating condition. As the toque slope varies,
the time taken for the estimated torque to touch the upper and
lower bands also varies, hence the torque switching frequency, i.e.
the frequency of the torque error status, also varies with operating
conditions. Since the selection of the voltage vectors is determined
from the torque error status, the torque switching frequency
directly influenced the device switching frequency. The slope of
induction machine as given in [36]. Over a small period ∆t, the

the torque can be approximated by considering the discrete form of
stator and rotor flux at the (n + 1)th sampling instant in stationary

reference frame can be written as:
dψ s
ψ s ,n+1 = ψ s ,n + ∆t (22)
dt n
dψ r
ψ r ,n+1 = ψ r ,n + ∆t (23)
dt n
In stationary reference frame fixed to the stator, the rate of change

of stator and rotor flux at the nth sampling instant can be obtained
by re-arranging equations (5) to (8) and are given by:
dψ s ,n
= − s ψ s ,n + s m ψ r ,n + v s ,n
R RL
σLs σLs Lr
(24)
dt
dψ r ,n Rr Lm ⎛ ⎞
= ψ s ,n + ⎜⎜ jωr − r ⎟⎟ψ r ,n
R
σLs Lr ⎝ σLr ⎠
(25)
dt
From equation (17), the (n+1)th sampling instant torque can be
[ ]
written as:
Te,n+1 = ψ s ,n+1 • j ψ *r ,n+1

3 p Lm
2 2 σL s Lr
(26)
Substituting (24) to (25) into (26) and simplifying gives:
[(v ]
)• jψgr,n ∆t
⎛ 1 1 ⎞
Te,n +1 = Te,n − Te,n ⎜ + ⎟∆t + − jω r ψs,n
3 p Lm
⎝ στ s στ r ⎠ 2 2 σL sL r
g
s,n
Thus, the torque slope can be approximated as:
[
(vs,n − jωr ψs,ng )• jψgr,n ]
Te,n +1 − Te,n ⎛ 1 1 ⎞ 3 p Lm
= −Te,n ⎜ + ⎟+
∆t ⎝ στs στ r ⎠ 2 2 σL sL r
(27)
Equation (27) indicates that the torque slope is dependent on the

stator flux, rotor flux, the speed and the stator voltage. The positive
and negative torque slope can be approximated by equations (28)
and (29); where for the negative slope, the voltage vector is set to
zero.
[(vs − jωr ψ s )⋅ jψ r ]
Te ,n+1 − Te ,n
slope + ≡ = − e ,n +
T 3 p Lm (28)
∆t στ sr 4 σLs Lr
[(− jωr ψ s )⋅ jψ r ]
Te ,n+1 − Te ,n
slope − ≡ = − e ,n +
T 3 p Lm
∆t στ sr 4 σLs Lr
(29)
A number of technical papers appeared in literature to produce a
predictive control scheme which calculates the stator voltage −

constant device switching frequency for DTC scheme. In [37][38],
based on torque and stator flux errors− required to drive the torque
and stator flux to their reference values within one sampling instant
was presented. The demand stator voltage was implemented using
SVM technique. Consequently, the switching frequency of the
device is totally controlled by the sampling frequency. The method
however increases the complexity of methematics involved, thus
require faster and powerful prosessor. Other works which are
based on predictive control scheme were presented in [32][28].
The duty ratio of the non-zero and zero voltage vectors was
calculated for every sampling instant and hence results in a
constant switching frequency. The non-zero voltage vectors are
selected based on the torque and stator flux errors, as determine in
the conventional DTC scheme. The approach which include the
use of fuzzy controller to reduce the torque ripple are given in
[34][33], which at the same time resulted in a constant switching
frequency.
5.4 CHAPTER SUMMARY
This chapter has presented the overview of DTC of induction

machines. The theory and modeling of induction machines in their
space vector and d-q components have been presented. It has been
shown that the stator flux and torque of the induction machines can
be directly controlled by the applied voltage vectors. Various
methods of stator flux estimations and their associated problems
and drawbacks have been reviewed. The performance of the DTC

drive in terms of the THD of phase current, torque ripple and
switching frequency have been given. Finally, the derivation of the
positive and negative torque slopes have been presented.
REFERENCES
[1] Leonhard,W. (1984). “Control of Electrical Drives.” Berlin

:Springer-Verlag.
[2] Bose, B.K. (1986). “Power Electronics and AC Drives”,
Prentice-Hall, New Jersey.
[3] Vas, P. (1990). “Vector Control of AC Machines”, Oxford
University Press, New York
[4] Murphy, J. M. D. and Turnbull, F. G.(1987). “Power
Electronic Control of AC Motors”, Pergamon Press,
Oxford
[5] Wade, S., Dunnigan, M. W. and Williams, B. M. (1994).
“Simulation of induction machine vector control and
parameter identification”, In Conf. Rec. Power Electronics
and Variable Speed Drives, No. 399, pp. 42-47
[6] Purcell, A. and Acarnly, P. (1998). “Device switching
scheme for direct torque control”, Electronics Letters, Vol.
34, No.4, pp. 412-414.
[7] Takahashi, I.and Noguchi, T. (1986). “A new quick-
response and high-efficiency control strategy of an
induction motor”, IEEE Trans. Ind. Appl., Vol. IA-22, No
5, pp. 820-827
[8] Bird, I.G. and Zelaya De La Parra, H. (1997). “Fuzzy logic
torque ripple reduction for DTC based AC drives”,
Electronic Letters, Vol. 33, No.17, pp. 1501-1502.
[9] Casadei, D., Grandi, G., Serra, G. and Tani, A.(1994).
“Switching Strategies in direct torque control of induction
machine”, In Con. Rec. International Conf on Electrical
Machines, Paris, France, pp. 204-209
[10] Jansen, P.L., and Lorenz, R.D (1994). “A physically
insightful approach to the design and accuracy assessment
of flux observers for field oriented induction machine

drives”, IEEE Trans Ind Appl., Vol. 30, No. 1, pp. 101-110
[11] Elbuluk, M.E., Langovsky, N. and Kankam, M. D. (1998).
“Design and implementation of a closed-loop observer and
adaptive controller for induction motor drives”, IEEE Trans
Ind Appl., Vol. 34, No. 3, pp.435-443.
[12] Hu, J. and Hu, B. (1998). “New integration algorithms for
estimating motor flux over a wide speed range”, IEEE
Trans Power Electronics., Vol. 13, No. 5, pp. 969-976.
[13] Bird, I.G. an Zelaya De La Parra, H. (1996). “Practical
evaluation of two stator flux estimation technique for high
performance direct torque control”, In Conf. Rec., Power
Electronics and Variable Speed Drives, No. 429, pp. 465-
470.
[14] Monti, M., Pironi, F., Sartogo, F. and Vas, P. (1998). “A
new state observer for sensorless DTC control”, in Conf.
Rec., Power Electronics and Variable Speed Drives, No.
456, pp. 311-317.
[15] Habetler, T.G., Profumo, F., Griva, G., Pastorelli, M.and
Bettini, A. (1998). “Stator resistance tuning in stator flux
field-oriented drive using an instantaneous hybrid flux
estimator”, IEEE Trans Power Electronics, Vol. 13, No.1,
pp 125-133
[16] Bonanno, F.; Consoli, A.; Raciti, A. and Testa, A. (1997).
“An innovative direct self-control scheme for induction
motor drives”, IEEE Trans. Power Electronics, Vol. 12, No.
5, pp. 800-806
[17] Byeong-Seok Lee; Krishnan, R (1998). “Adaptive stator
resistance compensator for high performance direct torque
controlled induction motor drives”, in Conf. Rec. IEEE-
IAS Annual Meeting, pp. 423-430
[18] Mir, S., Elbuluk, M.E. and Zinger, D.S. (1998). “PI and
fuzzy estimators for tuning the stator resistance in direct
torque control of induction machines”, IEEE Trans. Power
Electronics, Vol. 13, No.2, pp. 279- 287
[19] Cabrera, L.A.; Elbuluk, M.E.; Husain, I. (1997). “Tuning
the stator resistance of induction motors using artificial

neural network”, IEEE Trans. Power Electronics, Vol. 12,
No.5, pp. 779-787
[20] Bose, B.K. and Patel, N.R (1998). “Quasi-fuzzy estimation
of stator resistance of induction motor”, IEEE Trans. Power
Electronics, Vol. 13, No.3, pp. 401-409.
[21] Jansen, P.L., Lorenz, R.D. and Novotny D. (1993).
“Observer based DFO: Analysis and comparison of
alternative methods”, in Proc. IEEE-IAS Annual Meeting,
pp 536-543.
[22] K.D. Hurst, T. G. Habetler, G. Griva, F. Profumo, “Zero
speed tacholess IM torque control: Simply a matter of
stator voltage integration”, IEEE Trans Ind Appl., Vol 34,
No. 4, pp. 790-795.
[23] Tajima, H. and Hori, Y. (1993). “Speed sensorless field
orientation control of the induction machine”, IEEE Ind.
Appl., Vol 29, No. 1, pp. 175-180.
[24] Dell'Aquila, A.; Cupertino, F.; Salvatore, L. and Stasi,
S.(1998). “Kalman filter estimators applied to robust
control of induction motor drives”, IEEE-IECON’98,
Vol.4, pp. 2257-2262
[25] Jansen, P.L. and Lorenz, R.D (1991). “Accuracy limitations
of velocity and flux estimation in direct field oriented
induction machines”, in Conf. Rec. EPE’91, pp. 312-319.
[26] Lascu, C., Boldea, I. And Blaabjerg, F. (1998). “A
modified direct torque control for induction motor
sensorless drive”, IEEE-IAS Annual Meeting, pp. 415-422.
[27] Hurst, K.D. and Habetler, T.G. (1997). “A simple, tacho-
less, IM drive with direct torque control down to zero
speed”, IECON’97, Vol. 2, pp. 563-568.
[28] Kang, J.K. and Sul, S.K. (1998). “Torque ripple
minimisation strategy for direct torque control of induction
motor”, in Conf. Rec. IEEE-IAS Annual Meeting, pp. 438-
443.
[29] Casadei, D, Grandi, G., Serra, G. and Tani, A. (1994).
“Effect of flux and torque hysteresis band amplitude in
direct torque control of induction motor”, in Conf. Rec.

IEEE-IECON ’94, pp. 299-304.
[30] Jun-Koo Kang; Dae-Woong Chung; Seung-Ki Sul (1999).
“Direct torque control of induction machine with variable
amplitude control of flux and torque hysteresis bands”,
International Conference IEMD'99, pp. 640 642.
[31] Purcell, A. and Acarnly, P. (1998). “Multilevel hysteresis
comparator forms for direct torque control schemes”,
Electronics Letters, Vol. 34, No.6, pp. 601-603.
[32] Li., Y, Shao, J. and Si, B. (1997). “Direct torque control of
induction motors for low speed drives considering discrete
effect of control and dead-time timing of inverters”, in
Conf. Rec. IEEE-IAS Annual Meeting, pp. 781-788.
[33] Mir, S. and Elbuluk, M.E. (1995). “Precision torque control
in inverter-fed induction machines using fuzzy logic”, In
Conf. Rec. IEEE-IAS Annual Meeting, pp. 396-401.
[34] Bird, I.G. and Zelaya De La Parra, H. (1997). “Fuzzy logic
torque ripple reduction for DTC based AC drives”,
Electronic Letters, Vol. 33, No.17, pp. 1501-1502.
[35] Kazmierkowski, M.P. and Kasprowicz, A.B. (1995),
“Improved direct torque and flux vector control of PWM
inverter fed induction motor drives”, IEEE Trans. Indus.
Elec., Vol 42, No.4, pp. 344-350.
[36] Casadei, D. and Serra, G. (1997). “Analytical investigation
of torque and flux ripple in DTC schemes for induction
motors”, in Proc. IEEE-IECON’97, pp. 552-556.
[37] Habetler, T.G., Profumo, F., Pastorelli, M. and Tolbert,
L.M. (1992). “Direct torque control of induction machines
using space vector modulation”, IEEE Trans. Ind. Appl.,
Vol. 28, No. 5, pp. 1045-1053
[38] Casadei, D., Grandi, G. and Tani, A. (1996). “Constant
frequency operation of DTC induction motor drive for
electric vehicle”, Proc. ICEM’96, Vol. 3, pp. 224-229.
Study on Stability and Performances of DTC Due to Stator Resistance 99
6
STUDY ON STABILITY AND
PERFORMANCES OF DTC DUE TO
STATOR RESISTANCE VARIATION
Auzani Bin Jidin

Nik Rumzi Bin Nik Idris
Abdul Halim Mohd Yatim.
6.1 INTRODUCTION
High performance torque controlled induction machine drives can

be achieved when the proper control scheme utilizes accurate
estimated flux. In general, there are two methods for flux
estimation. One is based on current-based method and the other is
voltage-based method. In the current based method, the flux
estimation can be made at any speed, including zero speed with the
help of speed and current signals. Because of that, this model
requires a speed encoder which is not suitable for some
applications. Furthermore, the estimation accuracy is strongly
affected by the variation of machine parameters, in which some of
the parameters are difficult to compensate because of
inaccessibility. Alternatively, the voltage-based method is mostly
preferred since it only requires one parameter such that a stator
resistance, Rs , and moreover the motor speed sensor is not
required. Thus, this simple structure of flux estimation offers less

sensitivity to parameter variations. However, implementation of
flux estimation that utilizes an integrator in this model is no easy
task particularly at low speed operation. The major problems such
that a dc drift and initial value problems can occur in a pure
integrator with small offset in the measurement system utilized in
estimating the flux [J. Hu 1997 and I.G. Bird 1996]. The problems
have been tackled using many strategies in order to improve the
flux estimate so as to make the control scheme employed, able to
operate for a wide speed range, as proposed in [B.K. Bose 1997
and K.D. Hurst 1998].
Instead of having the problems due to the inaccurate integration
technique or improper sensor as mentioned above, other problem
that has also shown significant interests to be solved, is errors
introduced in flux estimation due to variation in stator resistance
[R. Krishnan 1998 and T.G. Habetler 1993]. In real practice, a
large extent in temperature variation can cause a wide variation in
stator resistance changes in induction machine [R. Krishnan et. al
1998]. Consequently, it will deteriorate the performance of DTC
by introducing errors in the stator flux estimation and also in the
electromagnetic torque estimation particularly at low speeds. The
sensitivity of the estimator to variation in stator resistance changes
is extremely affected in low speed region wherein the stator
resistance voltage drops become dominant as the stator voltages
become very small. In this case, the adjustment of the estimated
stator resistance to track a wide variation of actual stator resistance
changes is mandatory for accurate flux and torque estimations.
Numerous papers have been presented on tuning of stator
resistance [R. Krishnan 1998 and T.G. Habetler 1993]. In general,
the tuning of estimated stator resistance can be classified into two
purposes. One is for the thermal monitoring purpose which is one
of the fundamental protections required for induction motors [D.A.
Paice 1980 and Sang-Bin Lee 2002], and the second one is for the
flux estimation.
Commonly, there are two approaches for estimating Rs , namely

induction machine model-based Rs estimation [R. Beguanne 1999
and Sang-Bin Lee 2002] and signal injection-based Rs estimation
[D.A. Paice 1980 and T.G. Habetler 1993]. The approaches are
widely been implemented for the purpose of flux estimation to
obtain excellent drive performance. Ideally, the stator resistance
estimation should be able to operate over the entire speed range
and independent of the motor control method. For example, the
stator resistance tuning has been proposed using hybrid flux
estimation [Takahashi 1986 and T.G. Habetler 1993]. The hybrid
flux estimation utilizes a combined model which has smooth
transition from the stator voltage to the rotor voltage based flux
estimations from low speeds to high speeds, and vice versa. In
[D.A. Paice 1980 and L. A. S.Ribeiro 1999], the stator resistance is
calculated by means of dc components of the voltage and current
measurements. In this method, a dc bias is injected on-line, into the
stator supply voltage. Although these methods are capable of
giving a more reliable in Rs estimation, the motor torque
performance may be disturbed due to the injected signal on-line. It
is also possible to synthesize the tuning of stator resistance using
intelligent control technique. For instance, S. Mir et. al. [S. Mir et.
al. 1998] proposed Fuzzy estimator which employs stator current
phasor error to adjust the stator resistance. It has been shown that,
among the other variables of the machine, the stator current is
highly affected by the resistance changes. From that point of view,
the PI adaptive control using stator current phasor error was
presented in [B. S. Lee and R. Krishnan 1998]. The stator current
command was derived with stator flux and torque references as
inputs. However, the derived stator current command does not
consider other inputs such as load torque and operating
frequencies, in case if the stator flux locus tends to become
hexagonal shape at low speeds. In addition, the way to limit the
speed is not mentioned and questionable.
In this chapter, the effect on stability and performances on

DTC will be discussed with the help of simulations. As reported in
[B. S. Lee and R. Krishnan 1998], the instability occurs
particularly when estimated R s is greater than its actual value.
However, the explanation about this matter is only limited to the
mathematical analysis based on induction machine equations. This
chapter will present some analysis on performances of the drive
control quantities (i.e stator current, torque, flux, stator voltage
vectors) when a step stator resistance change is applied particularly
at low speeds operation. The significance of the study, is to give
more insight to the instability which occur when there is a
mismatch in the stator resistance.
6.2 PINCIPLE OF DIRECT TORQUE CONTROL
The induction machine equations in terms of space vectors, written

in a stator reference frame, are
dψ s
v s = Rs i s + (1)
dψ r
dt
0 = Rr i r + − jω r ψ r (2)
dt
where ω r is the rotor angular speed in electrical radians and v s is

the stator voltage space vector. i s and i r are the stator and rotor
current space vectors, while Rs and Rr are the stator and rotor
resistances, respectively. ψ s and ψ r are the stator and rotor flux
linkages, respectively and are given by
ψ s = Ls i s + Lm i r (3)
ψ r = Lr i r + Lm i s (4)
where Ls and Lr are the stator and rotor self-inductance,

respectively.
Unlike in FOC, the DTC scheme offer simple control structure

wherein the torque and flux can be separately controlled using
hysteresis comparators as shown in Figure 1. The stator flux can be
directly calculated from the stator voltage equation in the stator
( )
reference frame that
ψˆ sd = ∫ vsd − Rˆ s isd dt (5a)
(
ψˆ sq = ∫ vsq − Rˆ s isq dt ) (5b)
where vsd ( isd ) and vsq ( isq ) are the respective d and q-axis of
stator voltage (current) components. R̂s is the estimated stator
resistance. Thus the magnitude of stator flux is calculated by
ψˆ s = (ψˆ sd2 + ψˆ sq2 ) (6)
and the stator flux angle, φ is calculated by
⎛ ψˆ sq ⎞
φ = arctan⎜⎜ ⎟⎟
⎝ ψˆ sd ⎠
(7)
By considering that, ψˆ s = ψˆ s e jφ and i s = i s e jα , where α is the

angle of the stator current with respect to the direct-axis of the
stator reference frame, the torque can be calculated as written into
the following form
Tê = ψˆ s i s sin (α − φ )
3 p
(8)
22
where p is the number of pole and (α − φ ) is the angle between the
the command values of flux, ψ s* and torque, Te* are compared to

stator flux linkage and stator current space vector. In this scheme,
their respective estimated values. The errors are fed to a two level
comparator, in the case of flux, and a three level comparator in the
case of torque. The flux comparator output is defined as
∆ψ s
dψ s = 1 for ψˆ s ≤ ψ s −
*
(9a)
∆ψ s
2
dψ s = 0 for ψˆ s ≥ ψ s +
*
(9b)
2
* dTe
Te
dψ s
ψs
*
φ
ψˆ e
Tê
Figure 1 Structure of Basic DTC Scheme
and torque comparator output as
dTe = 1 for Tê ≤ Te* − ∆Te (10a)

dTe = 0 for Tê ≥ Te* (10b)
for counter-clockwise (CCW) rotation and
dTe = 0 for Tê ≥ Te* + ∆Te (11a)

dT = −1 for Tˆ ≤ T *
e e e (11b)
for clockwise rotation (CW).
The output of the comparators and the stator flux angle are used
to index a look up table of optimum voltage vectors as proposed in
[Takahashi and T. Noguchi 1986], in order to determine the
suitable voltage vectors, which is tabulated in Table 1(a). The
sector of the stator flux as illustrated in Figure 2 is divided into six
sectors. Figure 2, indicates that, the appropriate voltage vector (is
taken from the table of optimum voltage vectors) should be chosen
in a particular sector, either to increase stator flux or to decrease
stator flux and either to increase torque or to reduce torque. The
selection of voltage vector is made so as to restrict the errors of the
stator flux and torque within their respective hysterisis bands.
Consequently the fastest torque response and highest efficiency at
every instant can be obtained [Takahashi and T. Noguchi 1986].
Table 1(a) Voltage Vectors Look-up Table
Counter Sector
dψ s
clockwise
dTs I II III IV V VI
1 110 010 011 001 101 100

1 0 111 000 111 000 111 000
-1 101 100 110 010 011 001
1 010 011 001 101 100 110
0 0 000 111 000 111 000 111
-1 001 101 100 110 010 011
IV III
010 110 010 110
011 100 011 100
001 101 001 101
010 110 010 110
V 011 100 011 100 II

001 101 001 101
010 110 010 110
011 100 011 100
001 101 001 101
VI I
Figure 2 Six sectors of stator flux plane
6.3 EFFECT OF STATOR RESISTANCE
In DTC, the accurate estimated stator flux is very important to

achieve high performance of torque control. To obtain the high
degree accuracy of the stator flux estimation, the stator resistance
value, which is used in the controller, must match to its real value.
The mismatch between the estimated stator resistance value and its
actual value somehow can deteriorate the torque and flux control
especially when the estimated stator resistance is higher than its
actual value as reported in [B. S. Lee and R. Krishnan 1998].
However, the explanations on how the unstable of the performance
occurs were not discussed in details. This section will perform a
study on the effect of the higher estimated stator resistance than its
actual value on DTC. On the other hand, the discussion of the root
cause of the instability, will also be presented. All the analysis will
be performed through simulations using induction machine
parameters as given in Table 1(b).
Table 1(b) Induction Machine Parameters
Stator resistance, Rs 5.5 Ω

Rotor resistance, Rr 4.51 Ω
Stator self inductance, Ls 306.5 mH
Rotor self inductance, Lr 306.5 mH
Mutual inductance, Lm 291.9 mH
Combined inertia, J 0.0165 kg-m2
No. of pole, p 4
Rated stator flux, Ψs,rated

Rated torque, Trated 10 Nm
1.2 Wb
For the sake of discussion, a mismatch between the estimated

stator resistance and actual stator resistance is performed with a
step actual stator resistance change from 100% to 75% of its
nominal value at 2 s. A half of rated torque command is applied at
0.1 s and a rated stator flux command is kept constant. The motor
is kept running at low speed which is about 20 rad/s by adjusting
the load torque to the appropriate value. From Figure 3, it can be

observed that, the system becomes unstable when estimated stator
resistance is higher than its actual value. In general, the reason of
the instability of the system is due to the opposite effects between
the controller and the motor. Reference [B. S. Lee and R. Krishnan
1998], reported that, the increased current due to lower actual
stator resistance value, cause increased stator resistance voltage
drops in the estimator resulting in lower flux linkages and
electromagnetic torque estimations. The lower estimated flux
linkage and electromagnetic torque result in larger errors when
they are compared with their command values. As a result, the
commanding larger voltage is selected to compensate the errors. In
this case, the commanding larger voltage to compensate the errors
as stated in [B. S. Lee and R. Krishnan 1998], is cannot totally be
acceptable. The next section will study the reasons of instability
problems by examining the effects on flux, torque, stator voltage
vectors and stator current due to a step stator resistance change
with the aid of simulations.
Rs
is
Te
ψs
T̂e
ψˆ s
Figure 3 Instability due to parameter mismatch for a step stator

resistance change
6.3.1 Effect on Stator Flux
Using simulations, the actual and estimated stator flux locus

before and after a step stator resistance change are carried out as
depicted in Figure 4. The figure shows the simulation results for
one cycle of rotation of stator flux trajectories. The trajectories of
estimated and actual stator flux are rotated in anticlockwise. The
figure also depicts that the locus of the estimated and actual stator
flux tends to become hexagonal shape because of the flux
weakening at every sector transition. Note that, the locus of
estimated and actual stator flux are deviated each other when
estimated stator resistance is higher than actual stator resistance.
The divergence of the actual stator flux locus from the desired
stator flux locus is the root on why the unstable condition
occurred. It can be observed that in Figure 4(b), particularly at the
dotted line in sector 3, the actual stator flux start to deviate from
the estimated stator flux, just after a step actual stator resistance
change is applied at t=2 s.
The divergence between the actual and estimated stator flux may
be reasoned as follows. Firstly, let us study the change of the
actual and estimated stator flux vectors in the particular sector as
illustrated in Figure 5. Using (1), the change of the estimated stator
flux vector, ∆Ψ ˆ s and actual stator flux vector, ∆Ψ can be
s
( )
expressed as given in (12a) and (12b), respectively.
∆Ψ
ˆ = V s , k − I s Rˆ .∆T
= (V R ).∆T
(12a)
∆Ψs −Is
s s
s,k s (12b)
where, I s is the stator current vector and ∆T is the time interval

for the same stator voltage vector V s ,k is to be applied either to
increase or decrease the stator flux vector.
ψs
t ≥ 2s
ψˆ s
t ≤ 2s t ≥ 2s
ωe
ωe
ψˆ s =ψ s
ψs
ψˆ s
(a) (b)
Figure 4 Actual stator flux ψ s and estimated stator flux ψˆ s , for

θ ≤ φ ≤ θ + 2π , (a) before stator resistance mismatch and (b) after
stator resistance mismatch.
In such case, to increase the stator flux an appropriate active
voltage vector <010> is selected while to decrease the stator flux a

zero voltage vector <000> is chosen. Note that, the step change of
the actual stator resistance is set at t=2 s (in sector 3). From Figure
5, the switching actions kept the increasing and decreasing of the
estimated stator flux within its hysterisis band approximately. It
can be considered that, the estimated stator flux is well regulated at
the beginning, since the magnitude and phase errors of estimated
stator flux can be neglected; hence no incorrect voltage vector is
selected. Taking into account the fact that, the switching is
strongly affected by the torque regulation. Therefore, the larger
commanding voltage vector to compensate the sudden larger errors
in the estimated flux and torque as stated in [B. S. Lee and R.
Krishnan 1998], is questionable. It can be observed that the
divergence of the actual stator flux from estimated stator flux is
mainly caused by the difference between the actual and estimated
stator flux change vectors.
As given in (12), the reduction of estimated stator flux vector
during the application of zero vectors is significantly higher than
that of the actual stator flux vector. Note that, the ohmic drop in
(12a) is greater than that in (12b) when the same zero voltage
vector <000> is applied to decrease these stator flux vectors. While
the increasing of the actual and estimated stator flux vectors are
almost similar since the ohmic drops in both (12a) and (12b) for
the same applied active voltage vector <010> become less
significant. In this case, the higher change of stator flux vector
means the higher change of stator flux magnitude and hence a
steeper slope of the stator flux vector.
6.3.2 Effect on Torque
Based on (8), the accurate of stator flux estimation is mandatory in

estimating the torque. The accurate estimated torque is so
important for high performance and proper control in DTC since
the switching action is strongly affected by the torque controller.

From Figure 6, the error of estimated torque is restricted within its
hysterisis band since the estimated stator flux is considered well-
regulated in this particular sector. However, the actual torque
deviates from its hysterisis bands since it is strongly influenced by
the actual stator flux.
6.3.3 Effect on Voltage Vector Selection
To study the effect on voltage vector selection, let us examine

together with the effect on the stator current and stator flux, at the
first transition between sectors after a step stator resistance change
is applied. From Figure 7, it can be seen that the actual and
estimated sectors are different at the transition between sectors.
Figure 8, indicates that the magnitude of the actual stator flux and
stator current are steeply decreased due to the incorrect voltage
vector <001> is selected in sector 3 of the actual stator flux.
Suppose that, the voltage vector <111> or <011> is selected during
flux weakening at this particular sector transition.
In practice, the stator resistance does not change in a step
manner. A linearly increasing estimated stator resistance is
simulated and the performance is shown in Figure 9. Even for such
a gradual change of stator resistance, note that the system becomes
unstable. The reason of the instability is similar as mentioned for a
step stator resistance change.
Figure 5 Effect of estimated and actual stator flux magnitude due to

parameter mismatch for a step stator resistance change.
Figure 6 Effect of estimated and actual torque due to parameter

mismatch for a step stator resistance change.
Figure 7 Effect to the voltage vector selection at sector transition. (a)

Actual and estimated sector, (b) switching of phase A, (b) switching of
phase B, (c) switching of phase C.
Figure 8 Effect to the stator current, actual and estimated stator fluxes
corresponds to the incorrect voltage vector selection as shown in
Figure 7.
Rs
is
Te
ψs
T̂e
ψˆ s
Figure 99 Instability due to parameter mismatch for a linear stator

resistance change
6.4 CHAPTER SUMMARY
This chapter has discussed on the effects of the mismatch values

between actual stator resistance and estimated stator resistance on
the performance of DTC. A step change in actual stator resistance
is used to show how the instability of the system occurred. The
simulation results are used to study the performances of the drive
system particularly at low speed in which the stator flux is weaken
at every sector transition. This study reveals that, the mismatch in
the stator resistances result in divergence on the actual from the
estimated flux, consequently introduced phase and magnitude
errors. These errors lead to the incorrect voltage vector is selected
especially at every transition between sectors.
REFERENCES
J. Hu and B. Wu, “New integration algorithms for estimating

motor flux over a wide speed range,” in Proc. IEEE PESC
Rec., 1997, pp. 1075–1081.
Bird, I.G.; Zelaya de La Parra, H., "Practical evaluation of two
stator flux estimation techniques for high performance
direct torque control," Power Electronics and Variable
Speed Drives, 1996. Sixth International Conference on
(Conf. Publ. No. 429), Vol., Iss., 23-25 pp. 465- 470, Sept.
1996
K.D Hurst, T.G. Habetler, G. Grive, F. Profumo, "zero speed
tacholess IM torque control: simply a matter of stator
voltage integration", IEEE Trans Ind. Appl., Vol 34, No. 4,
1998, pp. 790-795.
B. K. Bose and N. R. Patel, “A programmable cascaded low-pass
filter-based flux synthesis for a stator flux-oriented vector-
controlled induction motor drive,” IEEE Trans. Ind.
Electron., vol. 44, no. 1, pp. 140–143, Feb. 1997.
N.R.N. Idris and A.H.M Yatim, “An improved stator flux
estimation in steady state operation for direct torque control
of induction machines,” in Conference Record of the IEEE
Industrial Applications Conference, Rome, Italy, 8-12 Oct.
2000, vol.3 pp. 1353-1359.
B. S. Lee, R. Krishnan, “Adaptive stator resistance compensator
for high performance direct torque controlled induction
motor drives,” Conf. Record of IEEE-Industry Applications
Society Annual Meeting, pp. 423-430, Oct. 1998.
Sang-Bin Lee Habetler, T.G. Harley, R.G. Gritter, D.J. "An
evaluation of model-based stator resistance estimation for
induction motor stator winding temperature monitoring,"
IEEE Transactions on energy conversion, Vol. 17, No. 1,
pp. 7-15, March 2002
R. Beguanne and M. E. H.Benbouzid, “Induction motors thermal
monitoring by means of rotor resistance identification,”
IEEE Trans. Energy Conv., vol. 14, pp. 566-570, Sept.
1999.
D. A. Paice, “Motor thermal protection by continuous monitoring
of winding resistance,” IEEE Trans. Ind. Electron. Instr.,
vol. IECI-27, pp. 137-141, Aug. 1980.
L. A. S.Ribeiro, C. B.Jacobina, and A. M. N.Lima, “Linear
parameter estimation for induction machines considering
the operating conditions,” IEEE Trans. Power Electron.,
vol. 14, pp. 62-73, Jan. 1999.
S. Mir, M. E. Elbuluk, and D. S. Zinger, “PI and fuzzy estimators
for tuning the stator resistance in direct torque control of
induction machines,” IEEE Trans. On Power Electronics,
Vol. 13, No.2, pp.279-287, March 1998.
Bose, B. K. and Patel, N.R. (1998). "Quasi-fuzzy estimation of
stator resistance of induction motor," IEEE Trans. Power
Electronics, Vol. 13, No. 3, pp. 401-409.
T. G. Habetler, F. Profumo, G. Griva, M. Pastorelli, and A. Bettini,
“Stator resistance tuning in a stator flux field oriented drive
using an instantaneous hybrid flux estimator,” Conf.
Record, EPE Conf., Brighton, UK, V01.4, pp.292-299,
1993.
Takahashi and T. Noguchi, “A new quick-response and high
efficiency control strategy of an induction machine”, IEEE
Trans. Ind. Applicat. 22, 820–827 (1986).
7
A QUICK DYNAMIC TORQUE
CONTROL FOR DIRECT TORQUE
CONTROL HYSTERESIS-BASED
INDUCTION MACHINES
Auzani Bin Jidin

Abdul Halim Bin Mohd Yatim
7.1 INTRODUCTION
The dynamic torque performance is so important especially in

traction and electric vehicle applications. The dynamic torque
control of vector controlled of induction machine drives can be
improved by increasing the inverter output voltage capability
through overmodulation. Apparently, a fast torque response can be
obtained by operating the inverter in overmodulation mode under
dynamic operating condition. The operation in overmodulation
mode under dynamic operating condition is called dynamic
overmodulation [A. M Hava et al. 1999].
It is well known that, the DTC of induction machine drives
offer fast torque and flux control due to its inherently decoupled
control structure of torque and flux, which is directly controlled
based solely on the instantaneous errors in torque and flux. For the
past two decades, many researchers have shown great interests in
proposing various solutions to improve the torque response of
induction motor drives by applying the dynamic overmodulation
strategies. For example, Habetler et al. proposed DTC using space
A Quick Dynamic Torque Control for Direct Torque Control Hysteresis 119
based Induction Machines
vector modulation (SVM) to improve dynamic torque response [T.

Habetler et al. 1992 and 1995]. With this proposed method, a
suitable voltage reference which is normally used to define the
mode of overmodulation is computed and it is then modified to
give the maximum possible voltage so that larger torque producing
component as well as a faster torque transient can be achieved.
However, this scheme requires several complicated equations used
to generate the reference voltage in real-time and it is not capable
to achieve torque control in six-step region. Tripathi et. al.
proposed a scheme for dynamic torque in the overmodulation and
field weakening region using DTC-SVM of induction machine
[Tripathi et al. 2006]. This scheme is able to achieve a fast torque
response in field weakening region with six-step mode. However,
the control structure is complicated since accurate synchronous
angular velocity needs to be estimated. Moreover, a fast digital
signal processor is required to perform the SVM algorithm.
This chapter presents a simple dynamic overmodulation
strategy which can be incorporated in DTC hysteresis-based
induction motor drives. The operation of the dynamic
overmodulation is similar with that proposed in [Tripathi et al.
2006] such that in selecting only single voltage vector to produce
the largest tangential flux component to obtain the fastest torque
response. However, Tripathi et al. uses DTC-SVM utilizing
predictive stator flux vector control which is complicated and
moreover the computation to calculate precise stator flux vector
used to define overmodulation mode is problematic [Tripathi et al.
2006]. In the proposed dynamic overmodulation, the flux error
status produced from the flux hysteresis comparator is modified
before it is being fed to the look-up table. In such manner, a fast
torque response can be achieved with six-step mode under
dynamic operating condition. This paper will also discuss the
effects of some operating conditions on the dynamic torque
performance in DTC hysteresis-based system. It will be shown
that, the proposed dynamic overmodulation is also capable of
producing a fast torque response in field weakening region with
six-step operation.
7.2 PRINCIPLE OF DTC
The behavior of induction machine in DTC drives can be described

in terms of space vectors by the following equations written in
stator stationary reference frame.
dψ s
v s = rs i s + (1)
dψ r
dt
0 = rr i r − jω r ψ r + (2)
dt
ψ s = L s i s + Lm i r (3)
Te = ψ s i s sin δ
3 p
(5)
22
where p is the pole pairs’ number, ωr is the rotor angular speed in

electric radians, Ls, Lr and Lm are the motor inductances and δ is
the angle between the stator flux linkage space vector and stator
current space vector. Unlike FOC, DTC offers simple control
structure (as shown in Figure 1), since there no frame
transformation involved and only requires stator resistance for
stator flux estimation [Takahashi I. and Naguchi T. 1986]. The
estimated stator flux and electromagnetic torque are controlled
using 2-level hysteresis comparator and 3-level hysteresis
comparator, respectively. The output of the comparators and the
stator flux angle are used to index a look-up table to select suitable
voltage vectors to control both the stator flux and torque.
σT
Torque Hysteresis
comparator
+ q(t)
Te* 0
1
Voltage
Sa
-1
- Sb Voltage
σψ
Vector
ψ
Flux Hysteresis Source
+
Selection IM
ψ s*
Comparator
Table Sc Inverter
+
θs
1
0
-
ψ + Vdc -
Stator Flux and
Te Torque Estimator
Figure 1 Basic DTC Structure
It is well known that, a fast instantaneous torque control can be

achieved in DTC-hysteresis based structure since the torque and
flux are directly controlled based solely on the instantaneous errors
in torque and flux. However, the dynamic torque control employed
in the DTC-hysteresis based structure may become slow for some
operating conditions which will be described later.
7.3 DYNAMIC TORQUE PERFORMANCE IN DTC
This section will discuss the performance of dynamic torque

utilizing the basic DTC algorithm with speed control as the speed
reference is subjected to change for different operating conditions
as follows
7.3.1 A Step Change of Speed Reference with the Stator Flux

Angle within a Sector at θ =0, π/12, π/6 or π/4 rad
A step change of speed reference is applied (from 0.85 to 1.0 p.u)

as the trajectory of stator flux in counter-clockwise rotation
reaches at either one, θ =0, π/12, π/6 or π/4 rad. as indicated in
Figure 2. The simulation results of dynamic torque performances
for each of the stator flux angle are depicted in Figure 3.
Apparently, a faster torque transient response is achieved when
a step change of speed reference is applied at θ =0 or θ = π/4 rad.

This is because at these particular stator positions, the duration of
the applied voltage vector that will give larger stator flux
tangential components (hence faster torque response) is longer. In
this case, the applied voltage vector for θ = 0 rad is u1, while the
applied voltage vector for θ = π/4 rad is u2 (see Figure 2). On the
other hand, the duration in which these voltage vectors applied is
shorter and are alternately switched more often during the transient
condition for the case of θ =π/12 or π/6 rad, thus results in a slower
change of stator flux position as well as torque transient.
Figure 2 Trajectory of stator flux within a sector 1

(a) (b)
1.0 1.0
Torque (p.u)
Torque (p.u)
0.5 0.5
0 0
t1 t2
Time (s) Time (s)
(c) (d)
1.0 1.0
Torque (p.u)
0.5 Torque (p.u) 0.5
0 0
t3 t4
Time (s) Time (s)
Figure 1.3 A step change of output torque at a different stator flux

angle, (a) θ = 0 rad (b) θ = π/12 rad (c) θ = π/6 rad (d) θ = π/4 rad
7.3.2 A Step Change of Speed Reference from Different Level

Speed Operations
In this study, several step changes in speed reference is applied and

the dynamic torque response are observed and compared. The
simulations were performed for the following step changes in
speed reference: 0.25p.u to 0.5 p.u, 0.5p.u to 0.75 p.u and 0.75p.u
to 1.0 p.u. For all the cases, the stator flux angle is at θ = π/6 rad
(the worst case). The simulation results obtained are plotted in the
same graph (for convenience of comparing the dynamic torque
performances) as depicted in Figure 4. Apparently, the fastest
torque response is achieved when the speed reference is changed
from the lowest speed operation that is from 0.25 p.u to 0.5 p.u.
This is due to the fact that the positive rate of change of torque
decreases with speed. Thus, the dynamic torque performance
deteriorates at high speed operation. It is still a great interest for
researchers to propose dynamic overmodulation in vector-

controlled motor drives having a quick torque response particularly
in field weakening region with six-step operation.
1.0
(a)
Torque (p.u)
(b)
0.5
(c)
t3
Time (s)
Figure 4 Waveforms of output torque for a different step change of

speed reference, (a) 0.25 p.u to 0.5 p.u, (b) 0.5 p.u to 0.75 p.u, or (c)
0.75p.u to 1.0 p.u, when stator flux angle at θ = π/6 rad.
7.4 PROPOSED DYNAMIC OVERMODULATION
The proposed dynamic overmodulation used a simple control

algorithm which can be easily incorporated in the DTC hysteresis-
based structure. No generation of voltage reference as commonly
used in DTC-SVM [T. Habetler et al. 1992 and 1995] is needed. In
the proposed dynamic overmodulation strategy, the stator flux
error status is modified as the dynamic condition is encountered,
before it is being fed to the look-up table. In such manner, only a
voltage vector that produces the largest tangential flux component
will be selected, and hence the stator flux position can be quickly
changed to obtain the fastest torque response. The proposed
structure of DTC hysteresis-based with flux error modification is
shown in Figure 5. Notice that, most of the component parts in the

simple structure of DTC hysteresis-based are retained, except the
addition of the ‘modified flux error status’ block which is
responsible in performing the dynamic overmodulation strategy.
The operation of the proposed dynamic overmodulation is
described in Figure 6. Figure 7 shows the trajectory of stator flux
during the dynamic torque condition. The proposed dynamic
overmodulation is activated when the torque error, σT is larger than
150% of torque hysteresis band amplitude, ∆T. It can be shown
that the fastest torque response can be obtained for a wide speed
operations including in field weakening range with six-step
operation.
σT
Torque Hysteresis
comparator
+ q(t)
Te* 0
1
Voltage Sa
ψ
-1
- Voltage
σψ
Vector Sb
−
Selection IM
ψ
Comparator
Inverter
+
Modified Table Sc
ψ+
flux error
θs
* 1
s 0 status
-
ψ + Vdc -
Stator Flux and
Te Torque Estimator
Figure 5 The proposed structure DTC Hysteresis-based with flux

error modification
Figure 1.6 The operation of the proposed dynamic

overmodulation
r3
Se
cto
ct o
r4
Se
6
tor
Su
Sec
b s ec
to
r
Figure 7 The trajectory of stator flux during dynamic overmodulation

7.5 SIMULATION RESULTS
This section presents some simulation results to verify the

effectiveness of the proposed dynamic overmodulation in
improving of the dynamic torque performance. Figure 8 compares
the output torque performances during dynamic condition when a
step change of speed reference is applied (0.85 p.u to 1.0 p.u).
From the Figure 8, it can be seen that a faster torque response is
achieved in the DTC with the proposed dynamic overmodulation
than that obtained in the DTC without dynamic overmodulation.
Figure 9 shows the phase stator voltage obtained from these
methods (correspond to the results obtained in Figure 8). Notice
that only single active voltage vector (that produces largest
tangential flux component) is implemented in the proposed method
during the transient condition.
The proposed dynamic overmodulation is also capable to
have a quick dynamic torque response in field weakening region.
Figs. 1.10 and 1.11, show the performances of the stator flux and
the torque when a step change of speed reference is applied from
1.25 p.u to 1.5 p.u. From Figure 10, it is shown that a slower
dynamic torque performance is observed when a step change of
speed reference employed in the DTC without dynamic
overmodulation. In this case, the stator flux is well regulated to
track its reference even in the dynamic condition. Conversely, a
sharp flux weakening is detected during the dynamic condition in
the proposed method as shown in Figure 11. The sharp flux
weakening occurs since the proposed algorithm selects single
active voltage vector to produce the largest tangential flux
component to get the fastest torque response.
∆ta
∆tb
1.0
(b)
(a)
Torque (p.u)
0.5
0.214 0.2145 0.215 0.2155 0.216 0.2165 0.217

Time (s)
Figure 8 Output torque transient when a step change of speed

reference is applied for (a) DTC without overmodulation (b) DTC
with the proposed dynamic overmodulation.
2/3Vdc ∆ta
Phase voltage (V)
0
(a)
-2/3Vdc
0.19 0.2 0.21 0.22 0.23 0.24 0.25

Time (s)
2/3Vdc ∆tb
Phase voltage (V)
(b) 0
-2/3Vdc
0.19 0.2 0.21 0.22 0.23 0.24 0.25

Time (s)
Figure 9 Waveforms of phase stator voltage due to the dynamic

torque condition for (a) DTC without overmodulation (b) DTC with
the proposed dynamic overmodulation
The proposed dynamic overmodulation is also can be

incorporated in [Jidin A. et al. 2008], to obtain higher torque
capability in field weakening region. For the same test condition
(as performed in Figs. 1.10 and 1.11) where a step change of speed
reference is applied from 1.25 p.u to 1.5 p.u, a higher torque
capability is achieved during the speed acceleration as illustrated
in Figure 12. A fast dynamic torque response is achieved since a
phase stator voltage is suddenly changed from PWM to six-step
operation when the dynamic condition is encountered. In this case,
the phase stator voltage performs six-step mode to obtain
maximum torque capability to accelerate the rotor speed at a
higher rate. Thus, the proposed dynamic overmodulation is capable
of achieving fast dynamic torque response for a wide speed
operation including the field-weakening region with six-step
waveform.
1.00
Stator flux (p.u)
0.75
0.3 0.305 0.31 0.315 0.32 0.325 0.33 0.335
Time (s)
1.0
T e*
Torque (p.u)
0.5
Te
0
0.3 0.305 0.31 0.315 0.32 0.325 0.33 0.335

Time (s)
Figure 10 Waveforms of stator flux and output torque during

dynamic condition without overmodulation strategy
1.00
Stator flux (p.u)
0.75
0.3 0.305 0.31 0.315 0.32 0.325 0.33 0.335
Time (s)
1.0
T e*
Torque (p.u)
0.5
Te
0
0.3 0.305 0.31 0.315 0.32 0.325 0.33 0.335

Time (s)
Figure 11 Waveforms of stator flux and output torque during

dynamic condition with the proposed dynamic overmodulation
strategy
1.0
T e*
Torque (p.u)
0.5
Te
0
0.26 0.28 0.3 0.32 0.34 0.36 0.38 0.4 0.42 0.44
Time (s)
Phase voltage (V)
2/3Vdc
-2/3Vdc
0.26 0.28 0.3 0.32 0.34 0.36 0.38 0.4 0.42 0.44
Time (s)
Figure 12 Waveforms of output torque and phase stator voltage

during dynamic condition for the proposed dynamic overmodulation
in [Jidin A. et al. 2008]
7.6 CHAPTER SUMMARY
A poor dynamic torque performance may result when the dynamic

condition occurs at high speed operation and/or the flux position is
at around the middle of any flux sector. Through the proposed
dynamic overmodulation, a quick dynamic torque response can be
obtained by selecting only voltage vectors that produce the largest
tangential flux component. The proposed dynamic overmodulation
is capable of producing fast dynamic torque in field-weakening
region with six-step operation. The main benefit of the proposed
method is its simplicity since most components of the basic DTC
hysteresis-based structure are retained and the results obtained are
comparable with that obtained in the complex structure of DTC-
SVM [Tripathi et al. 2006].
REFERENCES
A. M. Hava, S.-K. Sul, R. J. Kerkman, and T. A. Lipo, “Dynamic

overmodulation characteristics of triangle intersection pwm
methods,” IEEE Trans. Ind. Appl., vol. 35, no. 4, pp. 896–
907, Jul./Aug. 1999.
T. Habetler, F. Profumo, M. Pastorelli, L. Tolbert. “Direct Torque
Control of Induction Machines Using Space Vector
Modulation”. IEEE Trans. on Industry Applications,
28(5):1045–1053, September/October 1992.
T. G. Habetler, F. Profumo, and G. Griva, “Performance evaluation
of a direct torque controlled drive in the continuous PWM-
square wave transition region,” IEEE Trans. Power
Electron., vol. 10, no. 4, pp. 464-471, 1995.
Tripathi, A. M. Khambadkone, S.K. Panda, “Dynamic Control of
Torque in Overmodulation and in the Field weakening
region,” IEEE Transactions on Power Electronics, vol. 21,
No. 4, pp. 1091-1098, July 2006.
Takahashi I, Naguchi T, “A new quick-response and high

efficiency control strategy of an induction motor,”
IEEE Trans. Industry Applications, vol. IA-22, no. 5, pp.
820–827, 1986.
Jidin, A.; Idris, N. R. N.; Yatim, A. H. M.; Elbuluk, M., "A Novel
Overmodulation and Field Weakening Strategy for Direct
Torque Control of Induction Machines," Industry
Applications Society Annual Meeting, 2008. IAS '08.
IEEE, vol., no., pp.1-8, 5-9 Oct. 2008
Improved Torque Capability through Overmodulation for Direct 133
Torque Control Hysteresis-based Induction Machines
8
IMPROVED TORQUE CAPABILITY
THROUGH OVERMODULATION FOR
DIRECT TORQUE CONTROL
HYSTERESIS-BASED INDUCTION
MACHINES
Auzani Bin Jidin
Abdul Halim Bin Mohd Yatim
8.1 INTRODUCTION
In many applications such as spindle, traction and electric vehicle,

the full utilization of DC bus voltage is very important to obtain
maximum torque capability for a wider range of speed operation.
Without any extra hardware, the available DC bus voltage can be
fully utilized through overmodulation method. It is a common
approach to use the DTC associated with space vector modulation
(referred as DTC-SVM) to operate in overmodulation mode.
However, the DTC-SVM results in complex control structure since
it requires the computation of stator voltage vector reference as
proposed in [T. G. Habetler et al. 1992 and 1995] or stator flux
vector reference in [Tripathi et al. 2006] in order to define the
mode of overmodulation. In such ways, [T. G. Habetler et al. 1992
and 1995] used dead-beat control with several complicated
equations to generate the reference voltage in real-time and
[Tripathi et al. 2006] utilized predictive control with complex
algorithm to estimate a proper stator flux vector reference.
Until now, no study has been reported to carry out the

overmodulation strategy in hysteresis-based structure since no
voltage reference is generated as performed in the complex DTC-
SVM structure. This paper presents a simple overmodulation
strategy based on stator flux vector control. It will be shown that
by changing the stator flux locus from circular to hexagonal shape,
the phase stator voltage can be operated into the overmodulation
mode. It is well known that the capability of torque may decrease
as the back-emf voltage increases approaching the stator voltage
limit. Furthermore, the output torque demand can be satisfied only
if the synchronous angular velocity is higher than the rotor angular
velocity [D. Casadei et al. 2007]. From this point of view, it is
therefore possible to extend the constant torque region by
increasing the stator voltage limit through overmodulation. By
doing so, the maximum torque capability can be retained since the
back-emf voltage can be further increased to reach a new stator
voltage limit at nearly six-step mode. Then, a conventional flux
weakening method is applied to enhance the torque capability
beyond the base speed.
The main benefit of the proposed overmodulation strategy is
its simplicity at which most components in the basic DTC
hysteresis-based structure is retained. It is also suitable to be
implemented in many electric drive applications to provide
maximum torque capability for a wider speed range operation.
8.2 PRINCIPLE OF DTC
The behavior of induction machine in DTC drives can be described

in terms of space vectors by the following equations written in
stator stationary reference frame.
dψ s
v s = rs i s + (1)
dt
dψ r
0 = rr i r − jω r ψ r + (2)
dt
ψ s = Ls i s + Lm i r (3)
Te = ψ s i s sin δ
3 p
(5)
2 2
where p is the number of pole, ωr is the rotor angular speed in

electric radians, Ls, Lr and Lm are the motor inductances and δ is
the angle between the stator flux linkage space vector and stator
current space vector. The structure of basic DTC scheme is shown
in Figure 1. It has a simple control structure where no frame
transformation is required and only stator resistance information is
used to estimate the stator flux and the motor torque. The torque
and the stator flux are independently controlled using three-level
and two-level hysteresis comparators, respectively. The output of
the comparators and the stator flux angle are used to index a look-
up table of optimum voltage vector as proposed in [Takahashi I.
and Naguchi T. 1986], to determine the suitable voltage vectors to
control both flux and torque. The table of voltage vector selection
is given in Table 1(a).
In DTC, the stator flux space vector as well as
electromagnetic torque can be simultaneously controlled by
selecting appropriate voltage vectors. Figure 2 shows the control of
stator flux space vector in sector 2, using some suitable voltage
vectors as given in Table 1(a). For convenience, only one sector of
stator flux trajectory is considered to understand the principle of
direct flux and torque control. Assuming that, the rotation of stator
flux vector is in counter clockwise and both torque and stator flux
are in steady state operation. With this assumption, the direction of
stator flux vector changes (either to increase or to decrease) is
directly controlled by the appropriate voltage vectors since the

stator resistance voltage drop can be assumed negligible. For
example in this sector, voltage vector u2 is selected to increase the
stator flux while voltage vector u3 is applied to decrease the stator
flux. In this way, the magnitude of stator flux is restricted within
the flux hysteresis band by switching alternately between u2 and
u3.
In the case of electromagnetic torque control, it is very useful to
consider another form of expression for the electromagnetic torque
to describe the operation. Equation (5) is also can be written in the
following form
Te = P ' m ψ r ψ s sin γ
3 L '
(6)
2 Ls Lr
where γ is the angle between the stator and rotor flux space
vectors. Since the rotor flux vector rotation is continuous in
counter clockwise (lagging the stator flux vector), the
instantaneous output torque can be quickly controlled by changing
the position of stator flux vector. Thus, the angle γ in (6) plays a
vital role in controlling the output torque. For instance, an active
voltage vector (either u2 or u3) is selected to increase the angle γ as
well as output torque to reach its demand. When the torque error
touches the upper band, zero voltage vector (either u7 or u0) is
selected, which ideally stops the stator flux. This in turn reduces
the output torque as the angle γ decreases. In such ways, the output
torque and stator flux magnitudes are restricted within their
respective hysteresis bands.
σT
Torque Hysteresis
comparator
* + q(t) Sa
Te 0
1
Voltage
-1
- Voltage
σψ
Vector Sb
ψ +
Selection IM
ψ
Comparator
Table Sc Inverter
* + 1
θs
s 0
-
ψs
+ Vdc -
Stator Flux and
Te Torque Estimator
Figure 1 Structure of DTC Hysteresis-based
u3 (FD, TI) u2 (FI, TI)
ψs
Sector 3
γ ψr
u0,u7 (FH, TS)
Sector 2
u5 (FD, TD) u6 (FI, TD)

Sector 1
Flux Hysteresis band

FI: Flux increase; FD: flux decrease; TI: Torque increase;
TD: Torque decrease; TS: Torque satisfied; FH: Flux halt
Figure 2 Control of stator flux with optimum switching voltage

vectors
Table 1(a) Voltage vector selection table
Ψ+ q(t) Sec Sec Sec Sec Sec Sec

(1) (2) (3) (4) (5) (6)
1 u1 u2 u3 u4 u5 u6
1 0 u0 u7 u0 u7 u0 u7
-1 u5 u6 u1 u2 u3 u4
1 u2 u3 u4 u5 u6 u1
0 0 u7 u0 u7 u0 u7 u0
-1 u4 u5 u6 u1 u2 u3
8.3 TORQUE CAPABILITY IN DTC

In practice, a rated stator flux and inverter current limit are used to
obtain maximum torque capability in DTC of induction machine
drive. The inverter current limit is usually 150-200% of rated
current machine in order to achieve higher acceleration during
transient condition. Figure 3, shows phasor diagrams of back-emf
voltages, stator voltages and stator resistance voltage drops for two
different of speed operations. The phasor diagrams were drawn
At low speed operation, the back-emf, jωe1ψ s is much lower

based on (1).
than the stator voltage limit, v s , lim . In this situation, the DTC
algorithm produces sufficient and appropriate voltage vectors v s1
to control both stator flux and torque, simultaneously. At base
and the back-emf voltage increases to jωe 2ψ s . The base speed is

speed operation, the magnitude of stator voltage touches its limit
defined as the highest speed at which the maximum torque

capability for a rated stator flux can be retained. Notice that, the
torque and flux producing current components (in stator flux
reference frame), iψsds and iψsqs are under controlled for below base
speed operation. Thus, the maximum output torque below base
speed operation is determined by the current limit of the inverter,
assuming no current saturation. In other words, the induction
machine can be considered to be controlled by a current source, for
below base speed operation. This is also true in the hysteresis-
based DTC with decoupled control structure and no current control
loop present as oppose in Field Oriented Control (FOC) system.
The capability of torque may reduce as the back-emf voltage
increases approaching the stator voltage limit as illustrated in
Figure 4. For convenience, the phasor diagram of stator voltage,
back-emf and stator resistance voltage drop at base speed
operation is presented again. Obviously, the torque producing
ψ
current component, i sqs decreases as the back-emf increases
beyond the jωe 2ψ s (see vector components represented by arrow
with dashed line). On the other hand, the flux producing current
component iψsds is still under controlled in regulating the stator flux
at its rated value.
To enhance the output torque capability, a flux weakening
method is commonly applied. For example, [S. H. Kim 1995 and
G. Griva et al. 1998] consider voltage and current limits condition
to obtain maximum torque capability in the field weakening
region. However, this strategy requires a precise current
regulation, reference frame transformation and knowledge of
machine parameters to derive the optimal controllable currents iψsds
and iψsqs . Recently, a robust field weakening strategy utilizing
hysteresis-based direct torque control was proposed in [D. Casadei
et al. 2007]. The main idea of this scheme is to adjust the flux
reference on the basis of torque error. In this way, the appropriate
flux level is spontaneously determined and maximum torque
capability is obtained as well. This proposed method is less
dependent on machine parameters, allowing a satisfying operation
in the whole speed range.
jω e 2ψ s
jωe1ψ s
i s Rs i s Rs
vs2
ψs v s1
isd is
v s ,lim
ψs
ψs isq
Figure 3 Capability of torque retained for below base speed

operations
jω e3ψ s i ' R
jω e 2ψ s s s
i s Rs
v s2
ψs '
isd i s is
v s ,lim
ψs
ψs isq
Figure 4 Poor torque capability as the back-emf increases

approaching the stator voltage limit
In this chapter, a simple overmodulation method employed in

hysteresis-based DTC is proposed to extend a constant torque
region until the increased stator voltage limit reaches at nearly six-
step mode. Then, a conventional flux weakening method [R.
Joetten and H. Schierling 1983] is applied when the output torque
is considered no longer satisfied its demand. By doing so, the
simple control structure of basic DTC scheme can be retained and
the maximum torque capability for a wider speed range operation
can be achieved, which is an important requirement in many
electric drive applications.
8.4 IMPROVED TORQUE CAPABILITY THROUGH THE

PROPOSED OVERMODULATION STRATEGY
This section presents a proposed overmodulation strategy

employed in DTC hysteresis-based structure. It uses a
straightforward strategy by changing the stator flux locus from
circular to hexagonal shape. This in turn will transform the stator
voltage from PWM to almost six-step operation. This can be done
by modifying the original stator flux error status + before the

‘modified stator flux error status’, - is being fed into the look-up
table. The proposed structure of DTC hysteresis-based with flux
error modification is shown in Figure 5. Notice that, most of the
component parts in the simple structure of DTC hysteresis-based
are retained, except the addition of the ‘modified flux error status’
block which is responsible to perform the overmodulation mode.
As explained earlier, the output torque capability may
decrease as the back-emf increases approaching the stator voltage
limit. It is possible to retain the maximum torque capability over a
wider speed range operation, when the stator voltage can be
increased up to nearly six-step mode using the proposed
overmodulation method. The proposed overmodulation strategy is
activated when the output torque is considered no longer capable
to satisfy its demand. This condition is activated as the hysteresis
torque controller continuously produce torque error status, q(t)=1,
to increase the output torque but without succeeding for π/12
radian rotation of stator flux vector. When the proposed
overmodulation is started, the locus of stator flux is suddenly
status, ψ + before ‘the modified flux error status’, ψ − is being fed

changed to hexagonal shape by modifying the original flux error
into the look-up table. The activation of the proposed

overmodulation method and the operation of modification of flux
error status can be described as illustrated by the flowchart shown
in Figure 6.
Using this method, the switching pattern of stator voltage can
operate under overmodulation mode since the application of two
appropriate active voltage vectors for the stator flux to track the
circular path is no longer employed. However, the application of
zero voltage vectors is forced to occur in retaining the maximum
torque capability (wider constant torque region). The number
application of zero voltage vectors will then decrease as the speed
increases. Thus, the modulation index of stator voltage with
hexagonal shape flux locus is increased gradually (in natural) until
it reaches nearly six-step mode.
Therefore, no complex computation is required to control

stator voltage as well as output torque in overmodulation range. In
fact, the proposed overmodulation strategy employed in DTC
hysteresis-based uses decoupled control structure, which does not
require the computation of voltage reference.
In contrast, the overmodulation strategy implemented in DTC-
SVM requires a precise estimated voltage reference to define the
mode of overmodulation in controlling both flux and torque,
simultaneously. Moreover, a smooth transition of stator voltage
from PWM to six-step mode by gradually changing the holding
angle from 0 to π/6 rad., is a must to avoid any large sudden
change in the output torque [T. G. Habetler et al. 1992 and 1995].
As consequences, these requirements will lead to complex control
structure in DTC-SVM of induction machine drive.
In the proposed overmodulation method, a small sudden
change in the output torque occurs (which can be neglected) due to
rapid change in stator flux locus to hexagonal shape. As the
increased stator voltage reaches nearly six-step mode, a basic flux
weakening method (referred as ‘1/ωr method’) [R. Joetten and H.
Schierling 1983] is utilized to enhance the output torque capability
of induction machine. In this field-weakening region, the
maximum output torque is estimated as proposed in [D. Casadei et
al. 2007] and the stator voltage will continuously be operated near
the six-step operation. It is expected that, the capability of torque is
higher than that obtained in [D. Casadei et al. 2007], since the
proposed overmodulation strategy results in a wider constant
torque region and operates near the six-step mode in field
weakening region.
σT
Torque Hysteresis
comparator
+ q(t)
Te* 0
1
Voltage Sa
ψ−
-1
- Voltage
σψ
Vector Sb
Selection IM
ψ s*
Comparator
Inverter
+
Modified Table Sc
ψ+
flux error
θs
1
0 status
-
ψ + Vdc -
Stator Flux and
Te Torque Estimator
Figure 5 The proposed structure DTC Hysteresis-based with flux

error modification
Get torque error

status, q(t)
Get stator flux

angle, θs
∑(∆θs) is total summation

Total Summation: of difference angle stator flux
(∑(∆θs)) x q(t) between current sampled value
and previous sampled value.
No Yes
Ψ- = Ψ+ Overmodulation? Ψ- ≠ Ψ+
q(t) always 1 for
∑(∆θs)=π/12 rad.
Get stator flux
Get flux error angle, θs
status, Ψ+
Calculate θs within a
Stator flux increases/ sector,
decreases 0 ≤ θs ≤ π/3
Ψ- = Ψ+
Stator flux Yes No Stator flux

θs greater than
decreases, increases,
π/6 rad ?
Ψ- = 0 Ψ- = 1
Figure 6 The activation of the proposed overmodulation and flux

error modification operations
8.5 SIMULATION RESULTS
The simulation of the DTC induction motor drive with the

proposed overmodulation strategy was performed using
MATLAB/SIMULINK simulation package. The parameters of
induction machine as tabulated in Table 1(b) were used in the
simulation.
Figure 7, compares the performances obtained in DTC
without overmodulation with that obtained in DTC with the
proposed overmodulation method, when a step change of speed
reference is applied from 0.75 p.u. to 1.5 p.u. In this case, a
conventional flux weakening method (a flux reference is varied in
proportional to the inverse of rotor speed) is applied in field
weakening region.
It can be observed that an extension of constant torque region
and higher capability of torque in field weakening region are
obtained in DTC with the proposed overmodulation method. Thus
a shorter acceleration time is obtained during the speed transient
response with the proposed overmodulation method. In general,
the overmodulation mode in DTC produces higher current
harmonics content as a result of the hexagonal shape of the stator
flux.
Table 1(b) Induction machine parameters
Stator resistance 5.5 Ω

Rotor resistance 4.51 Ω
Stator self inductance 306.5 mH
Rotor self inductance 306.5 mH
Mutual inductance 291.9 mH
Moment of inertia 0.01 kg.m2
Number of poles 4
Rated speed 1410 rpm
DC-link voltage 654 V
Load torque 1 Nm
Apparently, the stator current distortions only occur when the

proposed overmodulation method is applied in DTC to achieve
higher torque capability and hence rotor to accelerate faster. The
PWM mode of stator voltage and current will be operated again as
the speed reaches its reference.
Figure 8, shows the stator flux locus correspond to the results
obtained in Figure 7. In the proposed overmodulation method, the
stator flux locus is suddenly changed to the hexagonal shape. In
such manner, the stator flux angular velocity increases as the
hexagonal stator flux locus shrinks (flux weakening) as shown in
Figure 8.
8.6 CHAPTER SUMMARY
This chapter presents a simple overmodulation method employed

in DTC hysteresis-based scheme. It is shown that, the stator
voltage can perform under overmodulation mode by changing the
stator flux locus from circular to the hexagonal shape. An
extension of constant torque region and higher capability of torque
in field weakening region can be obtained with the proposed
overmodulation method. The main benefit in the proposed strategy
is its simplicity and it can be well-implemented in many electric
drive applications to obtain maximum torque capability for a wider
speed range operation.
1.5
ωreference 1.5
ωreference
Speed (p.u)
Speed (p.u)
1
ωactual ωactual
1
0.5 0.5
0.2 0.25 0.3 0.35 0.4 0.45 0.2 0.25 0.3 0.35 0.4 0.45
Time (s) Time (s)
1.0
Treference 1.0
Treference
Torque (p.u)
Torque (p.u)
0
Tactual 0
Tactual
0.2 0.25 0.3 0.35 0.4 0.45 0.2 0.25 0.3 0.35 0.4 0.45
Time (s) Time (s)
2.0 2.0
Stator currents (p.u)
Stator currents (p.u)

1.0 1.0
0 0
-1.0 -1.0
-2.0 -2.0
0.2 0.25 0.3 0.35 0.4 0.45 0.2 0.25 0.3 0.35 0.4 0.45
Time (s) Time (s)
2/3Vdc 2/3Vdc
Phase voltage (V)
Phase voltage (V)
0 0
-2/3Vdc -2/3Vdc
0.2 0.25 0.3 0.35 0.4 0.45 0.2 0.25 0.3 0.35 0.4 0.45
Time (s) Time (s)
(a) (b)
Figure 7 Waveforms of rotor speed, output torque, d-q stator current
components, phase stator voltage during speed acceleration, for (a)
DTC without overmodulation (b) DTC with the proposed
overmodulation
1.5 1.5
Speed at Speed at
1 0.75 p.u 1 0.75 p.u
q-Stator flux axis (p.u)
q-Stator flux axis (p.u)
0.5 0.5
Speed at Speed at
0
1.5 p.u 0
1.5 p.u
-0.5 -0.5
-1 -1
ω ω
-1.5 -1.5
-1.5 -1 -0.5 0 0.5 1 1.5 -1.5 -1 -0.5 0 0.5 1 1.5
d-Stator flux axis (p.u) d-Stator flux axis (p.u)
(a) (b)
Figure 8 Stator flux locus during acceleration speed from 0.75 to 1.5
p.u. for (a) DTC without overmodulation (b) DTC with the proposed
overmodulation
REFERENCES
T. G. Habetler, F. Profumo, and G. Griva, “Performance evaluation

of a direct torque controlled drive in the continuous PWM-
square wave transition region,” IEEE Trans. Power
Electron., vol. 10, no. 4, pp. 464-471, 1995.
T. Habetler, F. Profumo, M. Pastorelli, L. Tolbert. “Direct Torque
Control of Induction Machines Using Space Vector
Modulation”. IEEE Trans. on Industry Applications,
28(5):1045–1053, September/October 1992.
D. Casadei, G. Serra, A. Tani, “Constant frequency operation of a
DTC induction motor drive for electric vehicle”, Proc. of
ICEM Conf., Vol. 3, 1996, pp. 224-229.
Tripathi, A. M. Khambadkone, S.K. Panda, “Dynamic Control of
Torque in Overmodulation and in the Field weakening
region,” IEEE Transactions on Power Electronics, vol. 21,
No. 4, pp. 1091- 1098, July 2006.
D. Casadei, G. Serra, A. Stefani, A. Tani, L. Zarri, “DTC drives
for wide speed range applications using a robust flux-
weakening algorithm,” IEEE Trans. On Industrial
electronics, Vol. 54, no. 5, pp. 2451-2461, October 2007
Takahashi I, Naguchi T, “A new quick-response and high
efficiency control strategy of an induction motor”, IEEE
Trans. Industry Appl 22:820–827, 1986
S. H. Kim and S. K. Sul, “Maximum torque control of an induction
machine in the field weakening region,” IEEE Trans. Ind.
Appl., vol. 31, no. 4, pp. 787–794, Jul./Aug. 1995.
G. Griva, F. Profumo, M. Abrate, A. Tenconi, and D. Berruti,
“Wide speed range dtc drive performance with new flux
weakening control [for induction motor drives],” in Proc.
29th Annu. IEEE Power Electron.Spec. Conf. (PESC’98),
May 1998, vol. 2, pp. 1599–1604.
R. Joetten and H. Schierling, “Control of the induction machine in
the field weakening range,” Proc. IFAC, 1983, pp. 297-
304.
9
THE MODELING AND SIMULATION
OF IMPROVED DTC OF PMSM DRIVES
USING MATLAB/SIMULINK
Tole Sutikno
Nik Rumzi Nik Idris
9.1 INTRODUCTION
The Permanent Magnet Synchronous Motor (PMSM) offers
many advantages over the induction machine, DC motor and
synchronous motor. The PMSM has lower inertia, higher
efficiency and power density, smaller losses and more compact
motor size. Because of the advantages, PMSM are indeed excellent
for use in many applications.
Usually, the AC machine control technique can be classified as
scalar control and vector control. The AC machine vector
controlled drives are widely used in many drive applications since
they offer high dynamic performance. The most popular vector
controlled are known as field oriented control (FOC) and direct
torque control (DTC) [1], [2], [3]. The DTC method has more
advantages such as lower parameter dependency and lower
complexity of control structure compared to the FOC.
The DTC method was proposed by Takahashi [4] and
Depenbrock [5] for induction motor drives in the middle of 1980’s;
it also was developed in 1990’s for PMSM drive [6]. The DTC is
not use current controller and not depend motor parameters. The
DTC method provides a very quick and precise torque response
without any complex field orientation system and current
controller.
The modeling and Simulation of Improved DTC of PMSM Drives 149
Using Matlab/Simulink
Nowadays, the DTC of PMSM is shown great attention of

researches since the simple control structure of DTC implemented
in PMSM produces high efficiency such that high rating of torque
and speed compared to that produced in DTC of induction machine
drives. Similarly, the common problem due to hysteresis torque
controller in basic DTC scheme such that high torque ripples still
occurs. To reduce the torque ripple, a fast switching frequency that
optimized the capability of power switch devices can be applied.
However this is limited by processor used. A very common way to
minimize the torque ripple is the use of Space Vector Modulation
technique (SVM) [1]. It offers constant switching frequency so that
the filtering of unwanted harmonics of sensed currents is simpler
since the dominant harmonic content can be predicted. However,
the PMSM utilizing with SVM results in complex control structure
and somehow it may degrade the high performance of DTC.
Furthermore, to solve the SVM itself, it requires fast speed of
processor so that it can produce fast switching frequency to reduce
the torque ripple.
This research presents the constant switching frequency torque
controller (CSFT) proposed in [12] to be employed in DTC of
PMSM to decrease the torque ripple. The simple control structure
in basic DTC scheme is retained since only minor modification is
made such that only the hysteresis torque controller is replaced by
the CSFT. By doing so, a constant and high switching frequency
can be achieved which can greatly reduce the torque ripple. The
effectiveness of the CSFT of PMSM to minimize the torque ripple
was verified through simulation results.
9.2 THE BASIC DTC SCHEMES OF PMSM

The structure of the basic DTC for PMSM drive is shown in
Fig.1, which is general for all schemes.
Figure 1 Basic DTC scheme of PMSM drive system
However, they differ in the switching tables design as indicated

below [8], [9]:
a. Eight-state table (switching table 1)
The switching table has been used in this research is given in
Table 1, where (cT and cφ) are the discrete torque and flux error
signals respectively and the resultant vector follows the notation
given in Figure 2. The switching table submitted in [10]. This
table provides zero inverter state whenever the torque error
signal is low.
Table 1 Eight-state table (switching table 1)

Cψ CT Sector Sector Sector Sector Sector Sector
1 2 3 4 5 6
1 1 V2 V3 V4 V5 V6 V1
1 0 V7 V0 V7 V0 V7 V0
0 1 V3 V4 V5 V6 V1 V2
0 0 V0 V7 V0 V7 V0 V7
Figure 2 Voltage vectors of the 3-phase VSI and the six stator
flux position sectors
b. Six-state table (switching table 2)

The switching table has been used in this research is given in in
Table 2. This design has slashed the zero states. This scheme
has been adopted in studies [6, 10, 11].
Table 2 Six-state table (switching table 2)

1 2 3 4 5 6
1 1 V2 V3 V4 V5 V6 V1
1 0 V6 V1 V2 V3 V4 V5
0 1 V3 V4 V5 V6 V1 V2
0 0 V5 V6 V1 V2 V3 V4
c. Bipolar torque/eight-state table

The switching table produces a zero state only when the torque
controller output is in the middle level (cT=0), which can be
interpreted as no significant torque change is required. This is
shown in Table 3. In the study had been presented in [1] not
only the switching table has been modified, but also the
hysteresis controller (a two-level controller) has been replaced
by a three-level controller. The characteristics of the 2- and 3-

level hysteresis controller are shown in Figure 3.
Table 3 Bipolar torque / eight-state table

1 2 3 4 5 6
1 1 V2 V3 V4 V5 V6 V1
1 0 V7 V0 V7 V0 V7 V0
1 -1 V6 V1 V2 V3 V4 V5
0 1 V3 V4 V5 V6 V1 V2
0 0 V0 V7 V0 V7 V0 V7
0 -1 V5 V6 V1 V2 V3 V4
Figure 3 Two and three level torque hysteresis controller

characteristics
Although the DTC scheme is an attractive proposition in its

own right, this basic DTC for PMSM drive has some drawback. It
still lefts some problems [6], such as: variable switching
frequency, torque and flux ripples, require of high sampling time,
current and torque distortion and drift in flux estimator.
In order to get an excellent performance, many modifications
of the basic DTC (similar techniques those used in induction
machine) have been proposed during last decade, as like: using
new switching table, modification of structure of inverter, and
modification of hysteresis controller.
Most modification of DTC PMSM drive appointed above

allow performance to be improved, but at the same time they lead
to more complex schemes. In [12] a novel DTC scheme for
induction motor with a pair of torque and flux controllers to
replace the hysteresis-based controllers can significantly reduce
torque and stator flux ripples, with the switching frequency is fixed
at 10.4 kHz and a more sinusoidal phase current. This DTC
method is fixed simple and quick response, so it can support
operation of high performance induction machine drive.
This paper proposes a constant torque controller with any
modification from [12] to improve DTC scheme for PMSM in
order to produce a better constant torque.
9.3 THE MODELING AND SIMULATION OF IMPROVED

DTC OF PMSM DRIVE
The proposed DTC is shown in Figure 4. The proposed torque
controller consists of two triangular waveform generators, two
comparators, and a proportional integral (PI) controller as shown
in Figure 5. The two triangular waveform (Cupper and Clower) are
1800) out of phase with each other.
Figure 4 Proposed DTC scheme of PMSM drive system

In this research, simulation of four different DTC scheme sets

have been carried out as follows: basic DTC was using switching
table 1, table 2, table 3, and proposed DTC scheme was using
switching table 3.
Figure 5 Proposed torque controller, with any modification

from [12]
speed_ref
Speed Set
Novel DTC of IPMSM Drive
u*
Continuous
y [Torque _ref ]
[speed ] u Goto 2 powergui DC Volt age

Source
From 1
speed Regulator
Torque _ref s
-
+
Ref Torque Ref Torque 1 Universal

Bridge
C
A
B
g
[Torque _ref ] Tq_err Tq_cont torque_error vector

Tm
From 2
New Torque
Controller
B
C
Tm
A
Flux _ref flux _error ud
flux _r
PMSM
Flux is_abc
theta uq
m
Hysteresis
Ref Flux
m wm
Flux _ref
Switching Table
angle (0-2*phi) Te
id
Est Flux
d-axis Machines Scope 1
Flux _est stator current Measurement
Theta id id Demux Torque
Iabc To Workspace4
Flux_est
Iq iq
Torque_est
d-axis flux [speed ]
iq ABC-DQ
Est Torque ud ud Goto
fd Flux_d q-axis
q-axis flux d-axis speed_act
Torque _est stator current
voltage
fq Flux_q uq uq
q-axis
Torque & Flux
voltage
estimator
Figure 6 A block diagram of a proposed DTC system

The PMSM DTC simulation system is based on the

Matlab/Simulink 6.6 (R2007a). A block diagram of a proposed
DTC system for PMSM is shown in Figure 6, which includes the
ABC to DQ transformation, torque and flux estimator subsystem,
etc. The ABC to DQ transformation and torque and flux estimator
each are shown in the Figure 7 and Figure 8.
Figure 7 The ABC to DQ transformation
Figure 8 The torque and flux estimator

In the flux linkage estimator in Figure 8, we shouldn’t set the

linkage initial value equal to zero, because the program can not run
successfully in Simulink. So, in the flux estimator is should set an
initial value of integrator block gain [7]. In this paper, it is chosen -
0.553. The torque and flux estimator are calculated with equation
followed:
(1)
(2)
(3)
(4)
where: Ud = 0.66667 Udc (Sa-0.5Sb-0.5Sc)

Uq = 0.57735 Udc (Sb-Sc)
In the DTC system, the command torque is obtained from the

speed PI regulator as is shown in Figure 6 and in detail is shown
in Figure 9.
Figure 9 The command torque

The proposed torque controller in this paper can be seen in

Figure 10, and determining switching table in Figure 11.
Figure 10 The proposed torque controller
Figure 11 Determining the switching table

9.4 SIMULATION RESULTS AND DISCUSSIONS

The comparison of the electromagnetic torque response
simulation under the basic DTC was using switching table 1, table
2, table 3 and under proposed DTC scheme is shown in Figure 12.
It is seen, that the proposed DTC system for PMSM can reduce
torque ripples.
Figure 12 Comparison of the electromagnetic torque response

under the basic DTC using switching table 1, using switching
table 2, using switching table 3, and under proposed DTC scheme
Next, the stator flux linkage trace from four different sets of
DTC simulations for PMSM is show in Figure 13. The
comparison of the flux response simulation under the basic DTC
was using switching table 1, table 2, table 3, and under proposed
DTC scheme. It is shown that the proposed DTC system for
PMSM can also reduce flux ripples, so the stator flux linkage is
controlled better at the required value.
Fig.13. Comparison of the flux response under the basic DTC

using switching table 1, using switching table 2, using switching
table 3 and under proposed DTC scheme
The parameters for IPMSM be used in this simulation are as

followed:
d-axis stator inductance (Ld) = 0.0446 H
q-axis stator inductance (Lq) = 0.1027 H
number of pole-pairs (np) = 2
stator resistance (Rs) = 5.8 ohm
flux linkage of magnet (fluxm) = 0.533 Wb
reference flux linkage (flux_r) = 0.55 Wb
dc voltage of VSI (vdc) = 340 V
band of torque (bdwt) = 0.2
band of flux (bdwf) = 0.0055
9.5 CHAPTER SUMMARY
This chapter presents simulation model of a new torque

controller for DTC of PMSM drives. The comparisons through
simulations with basic DTC have been done. The simulation based
on Matlab/Simulink in this research represented the behaviour of
DTC clearly and correctly. The simulation result has shown that
the proposed method can reduce the ripple of torque and flux
greatly. By this way, torque and flux ripple is smaller than those of
basic DTC.
REFERENCES
[1]. G. Diamantis and J. M. Prousalidis, "Simulation of a ship
propulsion system with DTC driving scheme," in Power
Electronics, Machines and Drives, 2004. (PEMD 2004).
Second International Conference on (Conf. Publ. No. 498),
2004, pp. 562-567 Vol.2.
[2]. B. Bose, "Power Electronics and Motor Drives: Advances
and Trends," Elsevier Inc., 2006.
[3]. C. L. Ferreira and R. W. G. Bucknall, "Modelling and real-
time simulation of an advanced marine full-electrical
propulsion system," in Power Electronics, Machines and
Drives, 2004. (PEMD 2004). Second International
Conference on (Conf. Publ. No. 498), 2004, pp. 574-579
Vol.2.
[4]. I. Takahashi and T. Noguchi, "A New Quick-Response and
High-Efficiency Control Strategy of an Induction Motor,"
IEEE Transactions on Industry Applications, vol. Vol.IA-22,
No.5, pp. 820-827, Sept/Oct 1986.
[5]. M. Depenbrock, "Direct self control (DSC) of inverter-fed
induction machine," IEEE Trans. on Power Electronics, vol.
3 (4), pp. 420–429, 1988.
[6]. L. Zhong, M. F. Rahman, W. Y. Hu, and K. W. Lim,
"Analysis of direct torque control in permanent magnet
synchronous motor drives," Power Electronics, IEEE
Transactions on, vol. 12, pp. 528-536, 1997.
[7]. L. Zhuqiang, S. Honggang, H. L. Hess, and K. M. Buck,
"The modeling and simulation of a permanent magnet
synchronous motor with direct torque control based on
Matlab/Simulink," in Electric Machines and Drives, 2005
IEEE International Conference on, 2005, p. 7 pp.
[8]. Y. Hu, C. Tian, Y. Gu, Z. You, L. X. Tang, and M. F.

Rahman, "In-depth research on direct torque control of
permanent magnet synchronous motor," in IECON 02
[Industrial Electronics Society, IEEE 2002 28th Annual
Conference of the], 2002, pp. 1060-1065 vol.2.
[9]. M. N. Abdul Kadir, S. Mekhilef, and W. P. Hew,
"Comparison of Basic Direct Torque Control Designs for
Permanent Magnet Synchronous Motor," in Power
Electronics and Drive Systems, 2007. PEDS '07. 7th
International Conference on, 2007, pp. 1344-1349.
[10]. M. F. Rahman, L. Zhong, and K. W. Lim, "A comparison of
two high performance, wide speed range drive techniques for
interior magnet motors," in Power Electronic Drives and
Energy Systems for Industrial Growth, 1998. Proceedings.
1998 International Conference on, 1998, pp. 276-281 Vol.1.
[11]. L. Zhong, M. F. Rahman, W. Y. Hu, K. W. Lim, and M. A.
Rahman, "A direct torque controller for permanent magnet
synchronous motor drives," Ieee Transactions on Energy
Conversion, vol. 14, pp. 637-642, Sep 1999.
[12]. N. R. N. Idris, C. L. Toh, and E. Elbuluk, "A New Torque
and Flux Controller for DTC of Induction Machine," IEEE
Transactions on Industry Applications, vol. Vol.42, No.6, pp.
1358-1366, Nov/Dec 2006. 2006.
10
MODELING AND SIMULATION OF A
NETWORK OF ADJUSTABLE SPEED
DRIVES
Makbul Anwari
M. Imran Hamid Taufik
10.1 INTRODUCTION
Adjustable Speed Drives (ASDs) are power electronic circuits used

to control electric motors behaviors. It is a further development of
constant speed drives that most used long time ago. In practice,
many ASDs are connected together to form a network of
adjustable speed drives to control the speed of motors in
manufacturing lines, buildings (for HVACs), agricultural sectors
(for irrigation pumps), and house-hold applications. Due to the
advent in power electronics, adjustable speed drives employing
solid-state switches have become popular in motor applications
with the significant energy saving that they offer [1,2]. However,
adjustable speed drives do possess inherent drawbacks, mainly
power quality issues due to the injection of harmonics into the
power systems. The problem has become more prevalent as the use
of adjustable speed drives becomes even more wide spread.
Research about harmonics product by ASDs has been explore by

many researchers, most of them devoted to explore the harmonics
product by single ASDs, in fact in real application, some times
ASDs is setting up as a network to role an integrated functional
system. In such a system, harmonics produced by an ASD interact
Modeling and Simulation of a Network of Adjustable Speed Drives 163
to other ASDs and create new shape of harmonics on point of

common coupling (PCC). In supply line, harmonics is viewed as
accumulation of individual contribution from these ASDs and its
may shows addition or subtraction shape of these individual
contribution. Harmonics on supply point vary with variation of
loading condition (speed and torque) of each ASDs during
operation time of the functional system.
Identification of harmonics on point of common coupling is
important in designing harmonics filter for the network. This paper
describe model and simulation of a method to find and reduce
harmonics distortion in a network of adjustable speed drives.
Individual harmonics produced by three type of adjustable speed
drives; single, three phase and multi phase input are reviewed.
Simulation using MATLAB/Simulink is then use to harmonics
distortion when the adjustable speed drives operate as a network in
a drive system.
10.2 ADJUSTABLE SPEED DRIVES AND HARMONICS

PRODUCED
The general configuration of ASD is shown in Fig. 1, The ASD

contains set of rectifier that used to convert the input voltage and
frequency of the system to dc voltage on dc link. The dc voltage
then converted back to different variable ac voltage and frequency
to be applied to the electric motor. In the practice, the above basic
configuration then spread and developed to various type of ASD to
follow the need of performance, efficiency, mechanical output
behavior and input side power requirement. In power quality
aspect, application of PWM methods on their switching control
that synthesize the motor currents as near to the sinusoidal has
made the ASDs impact to power quality are improved.
Figure 1 General configuration of ASD
Nevertheless, further development still continuing improved

including the input side of ASD that consisting of rectifier and
their input voltage. According to the input voltage, the ASD can
be categorized as single phase, three phase and multi-phase input
that create three phase on output side of ASD (motor side). A brief
description of these ASDs as follows:
The single phase input ASD employs single phase rectifier mainly
in bridge connection using controlled or uncontrolled switch. This
ac/dc converter is well known creates rich odd and triple
harmonics current. The input voltage and current shape of this type
of ASD using full bridge diode rectifier are shown in Fig. 2. For
each half cycle of input voltage it is created one pulse on their
input current. A distortion on voltage shape appears during
commutation period between rectifier switch. For input current as
5.54 A, its draw distortion as 3.59 % fundamental voltage total
harmonics distortion (THDV) and 58.38 % fundamental current
total harmonics distortion (THDI) with dominated by the 3rd, 5th,
7th, 11th and 13th harmonic order.
Figure 2 Input voltage and current of single phase input ASD
Three-phase input ASD uses three phase rectifier in their ac/dc

converter side to allow high density power from the line. As the
single phase input ASD, this type of ASD also employ controlled
or uncontrolled static switch. The voltage and current shape of this
ASD are shown in Fig. 3. It is shown that for each half cycle of
input voltage, it is created two pulses on input current. The figure
was taken when the ASD is loaded as 1.98 A that creates 2.23 %
fund of THDV and 45.84 % fund of THDI. It is known that the 5th,
7th, 11th, 13th … 6n±1 order dominated harmonics spectrum. Here,
n = 1, 2,3,…
Figure 3 Input voltage and current of three phase input ASD
Multi-phase ASD is further development of three phase input

ASD. Multi-phase voltage are usually created using special design
rectifier transformer that allow to transform three phase input to
several phase on its output. Fig. 4 shows voltage and current on
input side of an ASD employing 18 pulse rectifier. The 18 pulse is
resulted from rectifying the nine phase voltage at secondary side of
a rectifier transformer. The ASD is loaded as 12.88 A that gives
3.50 % fund of THDV and only 5.69 % fund of THDI. The 5th, 7th,
11th, 13th … 6n±1 order are also dominated their harmonics
spectrum.
Figure 4 Input voltage and current of Multi-phase input ASD
The above three types of ASDs have widely used and tend to
increase along with the need of drive demand in many application.
As have shown, during operation each type of ASD draw their
typical harmonics content and distort the line where they are
connected. Along with the increase of number ASD used,
distortion on a certain location on the line (PCC) is also increase
which harmonics content depend on ASD type that used. For
example if there are many single phase input ASD are used, then
the line will rich with the 3rd, 5th, 7th, 11th and 13th harmonic order.
The magnitude of these harmonics increase as the number of ASDs
connected. In this condition, if harmonics filter will be placed to
reduce harmonics problem, relatively high capacity with high cost
filter must be placed.
10.3 SIMULATION RESULTS AND DISCUSSION
As describes previously, each type ASD will give typical and different
harmonics order on their voltage and current spectrum. Figure 2 – 4
show that current shape are more significantly different between ASD,
meanwhile the voltage shape relatively identical with small distortion in
form of notch caused by switching commutation. The current
magnitude distortion will become greater when several ASD from same
type are operated together. One method to reduce the harmonics
distortion created from several ASDs is by operating some ASDs in
different type in a same network. Basically the method based on
harmonics reduction method with addition of nonlinear component to a
distorted network. By this method, several ASDs in different type are
connected on a PCC of the network so that the harmonics from each
type of ASD are interact each other and give other voltage and current
shape with lower distortion content.
In practice, this method is suitable to be implemented on new drive

network or where a network of ASD that contains same type of ASD
have been existed but intend to improve their power quality behavior.
To show this method, a simple simulation circuit in MATLAB/Simulink
as shown on Fig. 5 is developed. The circuit contains three ASDs with
different type as was describing before and a resistive load.
VL
Load
Inverter Gate
Circuit 3
Rpm _1
Gate
bc [ASD1]
A L1
ScopeA
PWM
multi winding transformer 2 IGBT Inverter 4
a bc IL 1 1
com
A A1
A a g Tm
5 2
10
rpm
9 +
3
b aB -K-
A <Rotor speed (wm)> (rpm)
A2
7 4 A A
C8
B B b
2 5 m
c b B 6 6
A4 <Electromagnetic torque Te (N*m)>
B B
4 7
11
C C c 3 -
8 OutputMech 1
C C
n2 c A0
C C
8 9
M4 rectifier -bridge 2
Tm _1
LINE SUPPLY
MULTI-PULSE ASD Inverter Gate
Load 1
Circuit 1
Rpm _
Gate
C2
PWM
Rectifier 2 IGBT Inverter 2
g Tm
[ASD2]
A + rpm 1
+
c
-K-
2 <Rotor speed (wm)> (rpm)
1 A A
C
Breaker
m
B B <Electromagnetic torque Te (N*m)>
c B -
2 -
OutputMech 2
1
C C
Breaker1
Tm _2
SINGLE PHASE FED ASD

Load 2
Inverter Gate
Circuit 2
Rpm _2
[ASD1] Gate
C3
PWM
Rectifier 3
IGBT Inverter 1
com
a g
A Tm
+ rpm 2
+
A -K-
<Rotor speed (wm)> (rpm)
A A
b B C1
m
B
B <Electromagnetic torque Te (N*m)>
B
-
C
c -
C OutputMech 3
C
C
Tm _
THREE PHASE ASD
Scope 3
i
+ - c
A
THREE PHASE RESISTIVE LOADLOAD

A
aB
B
b
C
c
C
node 10
Figure 5 Simulation circuit of a network contain different type ASD
The ASDs network is supplied from line and connected to a PCC

where the voltage and current shape are observed. On input line of
each ASD, voltage and current are measured in order to observe
their individual harmonics contribution. Each ASD is then
operated on their nominal load, voltage and current shape of each
ASD as shown on Fig. 2, 3 and 4.
The voltage, current shape and their harmonics spectrum on PCC
in this condition are described on Fig. 6. It is shown that by
operating several type of ASDs in the network, the current shape
has improved, for 35.96A total load current, it only draws current
distortion THDI as 12.14 % of fundamental. This result indicates
a significant improvement of distortion reduction compare if for

example total load are supplied by several single phase input or
three phase input ASDs. On the voltage shape it is shown that
more notching are appear as caused by diversity event of
commutation in converter switch between ASDs.
Figure 6 and 7 also show harmonics spectrum of voltage and
current on PCC. It shown that both voltage and current spectrum
contains all harmonics order of each connected ASD, existence of
these harmonics order in PCC voltage / current spectrum make
their shape tend to near sinusoidal that mean lower distortion.
Figure 6 Voltage shape and their harmonics spectrum of a

network of different type ASD
Figure 7 Current shape and their harmonics spectrum of a

network of different type ASD
Table 1 shows current harmonics order created by each ASD and

harmonics appear on PCC when they are operate simultaneously as
a network contain 6.218 A resistive load.
Harmonics ASD 1 ASD 2 ASD 3 Network

order
% of Fundamental
1 100 100 100 100
2 0.03 0.16 0.62 0.17
3 55.78 0.02 0.66 10.32
4 0.02 0.01 0.27 0.09
5 13.78 38.45 2.16 4.96
6 0.01 0.01 0.5 0.17
7 8.36 20.88 1.92 3.23
8 0.01 0.01 0.47 0.17
9 3.9 0 0.17 0.64
10 0.1 0 0.19 0.08
11 3.27 8.57 1.57 1.44
12 0.01 0.01 0.14 0.05
13 1.96 6.16 1.18 0.8
14 0 0 0.04 0.01
15 1.64 0 0.22 0.16
16 0 0 0.07 0.03
17 1.24 4.74 3.59 1.32
18 0 0 0.06 0.03
19 0.96 3.62 2.2 0.7
Fund 5.504 1.93 12.88 25.43

THD-I 58.38 45.84 5.69 12.14
Table 1 ASD Current harmonics individually and as a network
10.5 CHAPTER SUMMARY
Simulation results based on MATLAB/Simulink of harmonics

distortion produced by several ASDs with different type operate
individually and in a network of ASD were presented. Simulation
shows that by combining several ASD in different type to form a
network of ASD to drive a load system can give significant
reduction on harmonics distortion compared if several ASD in
same type are employed. The method can be considered as an
alternative method in harmonics reduction and power quality
improvement especially on the existing drive system.
REFERENCES
[1] Jimmie J. Cathey, “A Matlab-Based graphical Technique for

Amortization study of Adjustable speed drives”, IEEE Trans. On
Ed. Vol. 45 No. 2 May 2002
[2] David E. Rice, “A Suggested Energy-Savings Evaluation Method
for AC Adjustable-Speed Drive Applications”, IEEE Trans. On
Ind. App. Vol. 24 No. 6 Nov 1988
[3] K.S. Smith, L. Ran, “PWM drives: Voltage-type harmonic sources
in power systems”, IEE Proc.-Genrr. Trun.vm. Dbstrih., Vol. 145,
No. 3, May 1998
[4] D.A. Jarc and D.P. Connors, “Variable frequency drives and
power factor” IEEE Trans. on Ind. App. Vol. 1A-21 No.4
May/June 1985.
[5] P. Caramia, A. Russo and R. Carbone , “Attenuation of Harmonic
Pollution due to the Adjustable Speed Drives in the Electric
Circuits of the Power Plant Auxiliary Services”, 0-7 803- 59 35 -
6/00/$10.0(0c) 2000 IEEE.-
174
INDEX
Adjustable speed drives d-q equations, 70

network, 163, 164 field weakening region, 119,
Point of common coupling, 125, 129
164, 168-172 Inverter
Power quality, 163, 165, 169, model, 51– 53
173 control, 31– 36, 52 – 58
THD, 166 – 39 Maximum torque capability, 134,
Bilinear transformation, 54,62,65 139, 142, 147
Buck converter Overmodulation
dynamic evolution control, 41 – inverter, 119
43, 46 six-step, 119, 125, 126, 129,
sliding mode control, 27 – 28 133, 134, 139, 142
voltage mode control, 12 –18 space vector modulation, 119
Comparator Phase plane plot, 29, 31, 35
basic, 23 – 25 Predictive control scheme, 87, 90
hysteresis, 23 – 25 PSpice
Digital PI, 51–57, 66 ABM, 7, 21 – 23
Double-loop control, 57 convergence problem, 2
DTC if-then-else function, 25 – 26
direct flux, 73 simulation, 3 – 6, 26 – 33
direct torque, 76 Torque Ripple, 87
hysteresis-based, 119, 126 Type-3 error amplifier
Dynamic evolution control circuit, 5
linear, 38 – 39 transfer function, 5– 6
nonlinear, 37 – 40 k-factor, 9
Flux estimation, 111 Sisotool, 51, 58, 62,65
Hysteresis band, 73, 76, 86, 87, 90 Sliding mode control
Induction machine buck converter, 27 – 28
inverter, 31 – 33

The Modeling and Simulation of Improved

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

The Modeling and Simulation of Improved

Uploaded by

Copyright:

Available Formats

First Edition 2008

© NIK DIN MUHAMAD 2008

Perpustakaan Negara Malaysia Cataloguing-in-Publication Data

Editor: Nik Din Muhamad

Diatur huruf oleh / Typeset by

Diterbitkan di Malaysia oleh / Published in Malaysia by

Dicetak di Malaysia oleh / Printed in Malaysia by

Chapter 1 Control Loop Design of DC-DC Converter 1

Chapter 2 Simulation of Power Converters with 18

Chapter 3 Dynamic Evolution Control for Step-down 37

Chapter 4 Digital PI Controller Design for Power 50

Chapter 5 Overview of Direct Torque Control of 67

Chapter 7 A Quick Dynamic Torque Control for 118

Chapter 8 Improved Torque Capability through 133

Chapter 9 The Modeling and Simulation of Improved 148

Chapter Modeling and Simulation of a Network of 162

Power electronics technology is used for efficient control of

A vast amount of technology in switching power supply and

Part I : Modeling and Control of Power Converters and Drives

ENCON will strive to maintain, update and expand the series.

Nik Din Muhamad

Nik Din Muhamad

the design procedure is treated as a library component, which

1.2 DESIGN BASED ON SMALL SIGNAL MODEL

1.2.1 Buck Converter Power Stage

Figure 1 Buck converter with PWM modulator

It can be seen from (1) that the control-to-output transfer function

1.2.2 K-factor Approach

many researchers [6-10]. The type-3 K-factor compensator is

frequency, ωco, the phase boost that the compensator should be

It is important to point out here that this design procedure is

1.2.3 Type-3 K-factor Error Amplifier

Figure 2 Type-3 K-factor compensation network

The small-signal transfer function for the compensation network is

(1 + sR1C1 )⎜⎜1 + sR2 ⎜⎜ 2 3 ⎟⎟ ⎟⎟

By matching (5a) and (5b) we can obtain

1.3 PROPOSED APPROACH

compensation networks. The outputs of this block are: ωi,

ωp12 and ωz12.

1.4 PSPICE IMPLEMENTATION

mathematical problems by translating them to an electrical circuit.

1.4.1 Power Stage Block

previously given in (1). With s = jω, we see that transfer function

is a complex number containing a real part and an imaginary part.

The phase of the control-to-output transfer function can be

1.4.2 K-factor Block

function at the crossover frequency, fco (or ωco). By knowing Aco

Then, K-factor is calculated as

The two poles of the compensator are located at

1.4 .3 Compensator Block

(18)-(23) are also implemented in PSpice by using the .PARAM

1.4 APPLICATION EXAMPLES

1.4.1 100 kHz Buck Converter

Vg = 10V to 15V, Vo = 5V, L = 30µH, R = 1.25Ω, fs = 100 kHz,

Vp = 3V, Vref = 2.5V, Rbias = 10 kΩ. By knowing the converter

45°, give good transient response at the expense of peaking of the

like 75°, give flat closed-loop transfer function and minimum

settling time. A good compromise is 60°.

Using the above parameters, the developed subcircuit was

R = 1.25 VG = 10V VREF = 2.5V VP = 3V

C = 100uF RESR = 19m L = 30uH VO = 5V

134.2V PBOOST C1 4.526nV

106.5mV ACO Type-3-all C2 2.383nV

-164.2V PCO U14 C3 101.7pV

(1/(BetaRL))(Vref- Beta*V(%IN2)) -V(%IN1)