74
Journal of Power Electronics, Vol. 8, No. 1, January 2008
JPE 8-1-7
Implementation of Direct Torque Control Method using Matrix
Converter Fed Induction Motor
Hong-Hee Lee, Hoang M. Nguyen* and Tae-Won Chun*
School of Electrical Engineering, University of Ulsan, Ulsan, Korea
ABSTRACT
This paper develops a direct torque control method (DTC) using a matrix converter fed induction motor. The advantages
of matrix converters are combined with the advantages of the DTC technique; under the constraint of the unity input power
factor, the required voltage vectors are generated to implement the conventional DTC method of induction motor. The
proposed DTC algorithm is applied to induction motors and the experimental results are given in steady-state and transient
conditions, while the discussion about the trend of the DTC method using the MC is also carried out. Furthermore, the
entire system of the matrix converter configuration using 7.5kW IGBT module is explained in detail.
Keywords: matrix converter, induction motor, direct torque control method
1. Introduction
In the past two decades, due to the need to increase the
quality and the efficiency of power supply and usage, the
three phase matrix converter has become a major modern
energy converter and has emerged from the previously
conventional energy conversion modules as one of the
best substitutions [1], [2]. It fulfills all the requirements of
the conventionally used rectifier/DC-link/ inverter
structures. Some advantages of the matrix converter can
be seen as follows: the use of a compact voltage source,
providing sinusoidal voltage with varying amplitude and
frequency besides the sinusoidal input current and unity
input power factor at the power supply side. As shown in
Figure 1, a matrix converter has a simple topology and a
Manuscript received Sep. 21, 2007; revised Nov. 26, 2007
Corresponding Author: [email protected]
Tel: +82-52-259-2187, Fax: +82-52-259-2187, Univ. of Ulsan
*
School of Electrical Eng., Univ. of Ulsan
compact design due to the lack of DC-link capacitor for
energy storage.
Since the Direct Torque Control (DTC) method has
been proposed in the mid 1980s, the DTC method has
become one of the higher performance control strategies
for AC machines to provide a very fast torque and flux
control[3]. There are no requirements for coordinate
transformation, PWM generation and current regulators.
The DTC method is widely known to produce a quick and
fast response in AC drives by selecting the proper voltage
space vector according to the switching status of the
inverter, which is determined by the error signal of
reference flux linkage and torque with their estimated
values and the position of the estimated stator flux.
For such advanced reasons, the combination of the
advantages of the matrix converter with those of the direct
torque control method is effectively possible according to
[4]
. However, this research only focuses on some
simulation results and the experimental results are not
Implementation of Direct Torque Control Method using
sufficiently given. Moreover, some drawbacks still exist in
the DTC method because the torque and the flux ripple
always exceed their bandwidths. Thus, many researchers
have paid much attention to solve these problems in the
DTC methods [5], [6].
Table 1
Fig. 1
Matrix converter schematic block diagram
In this paper, the DTC theory in the matrix converter
module to control the induction motor is first explained
and then the experimental results are provided in detail to
verify the effectiveness of the DTC theory.
2. DTC Principles by Matrix Converter
2.1 Matrix Converter Theory
The three-phase matrix converter module includes nine
bi-directional switches as shown in Figure 1. There are 27
switching configuration states, which mean 27 possible
space vectors can be used to control IM and can be split
respectively into 3 groups as shown in Table 1; in Group I,
two output lines are connected to one of the other input
lines; in Group II, all output lines are connected to a
common input line; while in Group III, each output line is
connected to a different input line. The corresponding
output line-to-neutral voltage vector and input line current
vector have fixed directions as represented in Figure 2.
However, Group III is not useful. Only 18 non-zero space
vectors in Group I (1, 2, , 9) and 3 zero space
vectors in Group II (0a, 0b, 0c) can be usually employed
in the modern control techniques for the matrix converter
(such as the Space Vector Modulation, DTC methods, etc.)
Possible Switching Configurations of MC
Group Vector A B C
+1MC
75
vs
ii
2/3vab
2/3isa
-/6
-/6
-1MC
-2/3vab
-2/3isa
+2MC
2/3vbc
2/3isa
/2
-2MC
-2/3vbc
-2/3isa
/2
+3MC
2/3vca
2/3isa
7/6
-3MC
-2/3vca
-2/3isa
7/6
+4MC
2/3vab
2/3 2/3isb
-/6
-/6
-4MC
-2/3vab
2/3 -2/3isb
+5MC
2/3vbc
2/3 2/3isb
/2
-5MC
-2/3vbc
2/3 -2/3isb
/2
+6MC
2/3vca
2/3 2/3isb
7/6
-6MC
-2/3vca
2/3
7/6
+7MC
2/3vab
4/3 2/3isc
-/6
-/6
-2/3isb
-7MC
-2/3vab
4/3 -2/3isc
+8MC
2/3vbc
4/3 2/3isc
-8MC
-2/3vbc
4/3 -2/3isc
/2
+9MC
2/3vca
4/3 2/3isc
7/6
-9MC
-2/3vca
4/3 2/3isc
7/6
0a
0b
0c
II
III
/2
4,5,6
2,5,8
vs
0
1,2,3
7,8,9
Fig. 2
3,6,9
1,4,7
(a) The output line-to-neutral voltage vectors
(b) The input line current vectors
2.2 DTC Principles Using VSI
In the conventional DTC principles, there are three
main coefficients: a stator flux hysteresis, a torque
hysteresis comparator and the position of the stator flux as
76
Journal of Power Electronics, Vol. 8, No. 1, January 2008
shown in Figure 3. In order to maintain the estimated
stator flux and torque within their boundaries which are
determined by the two hysteresis bandwidths as shown in
figure 5a & 5b, at each sampling period, the torque and the
stator flux are estimated and compared with the
corresponding reference values before passing the
hysteresis comparator. The position of the stator flux is
detected, and the most suitable space vector among 8
space vectors generated by a VSI (6 non-zero space vector
and two zero space vector in figure 4) is selected from the
switching table given in Table 2 to compensate the load
torque and the stator flux. Therefore, it is possible to
implement the DTC schemes in having good and robust
performances.
Table 2
Sector of Flux
cT =-1
c = 0 cT = 0
cT = 1
cT =-1
c =+1 cT = 0
cT = 1
Basic DTC switching table using VSI
1
V2-vsi
V7-vsi
V6-vsi
V3-vsi
V0-vsi
V5-vsi
V3-vsi
V0-vsi
V1-vsi
V4-vsi
V7-vsi
V6-vsi
V4-vsi
V7-vsi
V2-vsi
V5-vsi
V0-vsi
V1-vsi
V5-vsi
V0-vsi
V3-vsi
V6-vsi
V7-vsi
V2-vsi
V6-vsi
V7-vsi
V4-vsi
V1-vsi
V0_vsi
V3-vsi
V1-vsi
V0-vsi
V5-vsi
V2-vsi
V7-vsi
V4-vsi
2.3 DTC Principles Using Matrix Converter
According to [4], the basic DTC principles using matrix
converters can be briefly described as follows: at each
sampling period, the proper switching configuration,
which allows the compensation of instantaneous errors in
the stator flux magnitude and torque, is selected under the
constraint of unity input power factor. This last
requirement of the input side of the matrix converter is
intrinsically satisfied if the average value of sin (i) is
maintained close to zero, where i is the displacement
angle between the input line voltage and input line current.
The hysteresis comparator shown in figure 5c directly
Fig. 3
controls this variable and the average value of sin (i) is
obtained by applying a low-pass filter to its instantaneous
The DTC diagram for VSI
value. The average value of sin (i) is controlled close to
zero because the input power factor is aimed close to
unity.
As a facultative example, after calculation at the first
time of each sampling period and considering the stator
flux vector lying in sector 1, the input voltage vector lying
in sector 2, the output of the torque hysteresis comparator,
the flux hysteresis comparator and the hysteresis
comparator of the average value of sin (i) are
V 3VSI
V 2VSI
V 4VSI
V 0VSI = V 7VSI
V 1VSI
s
s
V 5VSI
Fig. 4
V 6VSI
VSI output line-to-neutral voltage vector and
respectively cT = +1, c = 0 and c= +1. As shown in
corresponding stator flux variation in period t
Figure 6, first with cT = +1, c = 0 and the stator flux in
sector 1, the suitable voltage vector V6-vsi is the VSI output
voltage vector by the DTC algorithm in a given switching
period from Table 2. Then, with the chosen VSI voltage
vector V6-vsi, c= +1 and the input voltage vector in sector
2, the opportune matrix converter voltage vector is finally
selected as V-5MC from Table 3.
The schematic of the DTC method using the matrix
converter fed induction motor is represented in Figure 6.
The reference values of the torque and the stator flux are
cT
s = s S
Fig. 5
T = T T
BT
< sin( i ) >
(a) Flux hysteresis comparator
(b) Torque hysteresis comparator
(c) Hysteresis comparator of average sin(i)
Implementation of Direct Torque Control Method using
compared with the estimated values and coordinate with
the average value of the sin (i) hysteresis comparator. In
the lower part of the block diagram, the estimators of the
electromagnetic torque, stator flux and the average value
of sin (i) are represented. These estimators require the
knowledge of input and output of voltages and currents for
the matrix converter. However, only the input voltages
and the output currents of the matrix converter module are
measured by sensors, while other quantities such as the
input voltages of the induction motor and the input
currents of the matrix converter module are calculated
from the previous switching states, the input voltages and
the output currents of the matrix converter module.
Sector
of v i
c
V1-vsi
V2-vsi
V3-vsi
V4-vsi
V5-vsi
V6-vsi
Table 3
DTC Switching Table Using MC
+1
-1
+1
3
-1
-1
+1
5
-1
+1
6
+1
-1
-3MC
+1MC +2MC
-3MC
-1MC
+2MC +3MC
-1MC
-2MC +3MC
+1MC
-2MC
+9MC
-7MC
+9MC +7MC
-8MC -9MC
+7MC +8MC -9MC
-7MC
+8MC
-6MC
+4MC +5MC -6 MC -4MC
-8MC
+1
-1
+5MC +6MC
-4MC
-5MC +6MC
+4MC
-5MC
+3MC +1MC
-2MC -3MC
+1MC +2MC -3MC
-1MC
+2MC
-9 MC +7MC +8MC
-9MC
-7MC
+8MC +9MC
-7MC
-8MC +9MC
+7MC
-8MC
+6 MC -4MC
+6MC +4MC
-5MC -6MC
+4MC +5MC -6MC
-4MC
+5MC
+3 MC -1MC
-2MC
-5MC
vi
T
vi
< sini >
cT
T*
s*
is
c
s
s
T
vs
< sin i >
sin
ii
vi
is
77
The DSP board includes the CAN (Control Area Network)
function which is connected to a PC through a CAN card
to monitor the entire system. The sampling period is
200s, corresponding to a switching frequency of 5 kHz.
The control of the commutation process is performed by
a high-CMOS (electrically erasable) programmable logic
device (CPLD) board using EPM7128, manufactured by
Altera, running at 40MHz. The 4-step commutation time
is carried out within 2 s. A gate driver board (TC4420)
with six isolated power supplies is sufficient to drive the
MC power modules.
3.2 MC Power Circuit
Figure 7 shows the hardware block diagram of a matrix
converter, consisting of a 3-phase power supply and a
220V/60Hz while a 5HP induction motor is coupled with
an adjustable mechanical load in order to provide the load
torque. Motors parameters are as follows
Stator Resistance (Rs)
1.45875
()
Rotor Resistance (Rr)
Stator Inductor
(Ls)
Rotor Inductor
(Lr)
Mutual Inductor (Lm)
Number of Poles
0.9058333
0.1511383
0.1638935
0.1483087
4
()
(H)
(H)
(H)
A low pass filter including a 0.5mH inductor and 20F
delta-connected capacitors, corresponding to a cut-off
frequency of 1.6kHz, is connected at the power supply side.
A 7.5kW common-emitter IGBT MC module
(FM35R12E3) by Eupec is used in this experiment. At the
stop mode, the induction motor currents have no closing
path and can cause over-voltage across the IGBT switches.
The most common solution for this problem is to add 2
3-phase full diode bridges between the input and output
sides, and then the energy can be stored in the small
capacitor of the clamp with the acceptable over-voltage.
Fig. 6 Block diagram of the DTC with MC
3. Hardware Design for MC System
3.1 MC Control Board
The input voltages and output currents are measured by
a 12-bit A/D converter (AD7891). The calculation of the
DTC method is executed on a floating-point 32-bit DSP
TMS320C32 by Texas Instrument, running at 60MHz.
Fig. 7
Experimental block diagram of matrix converter
78
Journal of Power Electronics, Vol. 8, No. 1, January 2008
4. Experimental Results
Experiments are carried out to verify the validity of the
DTC method for the matrix converter. The steady state
performance of the matrix converter has been tested at low
and high speeds. Figure 8, 9 and 10 show the motor
currents for the different speed commands. The dynamic
behavior has been tested under low speed changes in
Figure 11 and 12, and under high speed changes in Figure
13.
Figure 14 shows the rotor speed transient state at 500
rpm as the load torque changes from 1N.m to 1.5N.m.
Figure 15 shows the electromagnetic torque at the low
speed of 100 rpm as the load torque command is abruptly
changed. In these experimental results, the stator current
waveforms are almost sinusoidal.
Furthermore, Figure 16 shows the experimental results
of the motor speed at both forward and reverse directions
as the speed reference changes between 200 rpm and
-200rpm.
Finally, Figure 17 shows the unity input power factor at
the power supply side, and the harmonic spectrum analysis
of the input current shows the current harmonic
components, which are filtered below the cut-off
frequency, 1.6kHz. However, the input current still
contains many high harmonics due to the hardware
limitation; the sampling frequency must be much higher to
achieve an efficient implementation of DTC technique.
Fig. 10
Fig. 11
Fig. 8
Stator current at 100 rpm with 1N.m load torque
Stator current at 1000 rpm with 1.5 N.m load torque
Stator current with the speed change from 400rpm to
200rpm with 1 Nm
(rpm) 1000
IM's Speed
Ref. Speed
800
600
400
200
0
(a)
5 (s/div)
(Wb) 0.4
0.37
0.3
Fig. 12
Fig. 9
Stator current at 500 rpm with 1.5 N.m load torque
(b)
Stator Flux
Ref. Flux
5 (s/div)
(a) Rotor speed (b) Stator flux magnitude with
the speed command between 200rpm and 400rpm
with 1Nm
Implementation of Direct Torque Control Method using
79
(rpm) 1500
Rotor Speed
Ref. Speed
Input Phase Voltage
Input Phase Current
1000
500
0.1 (s/div)
Hamornic Spectrum of Input Phase Current
0
5 (s/div)
Fig. 13
Rotor speed as the reference speed change from
500 rpm to 1000 rpm with 1 Nm
0
5 kHz
(rpm) 1500
Fig. 17 Input line to neutral voltage (100V/div)and input line
current (10A/div) waveforms with 3.5 Nm torque
command
IM's Speed
Ref. Speed
1000
5. Conclusions
500
10 (s/div)
Fig. 14 Rotor speed at 500 rpm when the load torque changes
from 1Nm to 1.5Nm
(rpm) 400
IM speed
Ref. speed
300
200
100
0
5 (s/div)
(N.m) 5
Electromagnetic Torque
4
3
2
This paper deals with the experimental implementation
with the DTC method using the matrix converter fed
induction motor. This inherits the advantages of the DTC
method and the advantages of the matrix converter
characteristics. The experimental environment is
explained in detail. The steady and transient experimental
results are shown to validate the stability and the
robustness of the DTC method in the practical applications.
With an improved hardware design, the DTC method
using the matrix converter may become the most
appropriate solution to build a compact drive fed induction
motor.
1
0
5 (s/div)
Acknowledgment
Fig. 15
Electromagnetic torque with the load torque command
between 3.5 Nm and 1 Nm
rpm
Speed Direction
Rotor Speed
Ref. Speed
+400
The authors would like to thank Ulsan Metropolitan
city, the MOCIE and the MOE of the Korean Government
which partly supported this research through the
Network-based Automation Research Center (NARC) and
the Post BK 21 Program at University of Ulsan.
+200
References
-200
[1]
-400
5 (s/div)
Fig. 16
Rotor speed in the forward and reverse directions
E. Watanabe, S. Ishii, E. Yamamoto, H. Hara, Jun-Koo
Kang & A.M. Hava, High performance motor drive using
matrix converter, Advances in Induction Motor Control
(Ref. No. 2000/072), IEE Seminar, May 2000.
80
[2]
[3]
[4]
[5]
[6]
Journal of Power Electronics, Vol. 8, No. 1, January 2008
T. Matsuo, S. Bernet, R.S. Colby & T.A. Lipo,
Application of the matrix converter to induction motor
drives, Thirty-First IAS Annual Meeting Conference
Record, Vol.1, pp.60 67, Oct. 1996.
Romeo Ortega, Nikita Barabanov & Gerardo Escobar
Valderrama, Direct Torque Control of Induction Motors:
Stability Analysis and Performance Improvement, IEEE
transactions on automatic control, Vol. 46, No. 8, August
2001.
D. Casadei, G. Serra & A. Tani, The use of matrix
converters in direct torque control of induction machines,
Conference Proceedings of IECON '98, Vol. 2, pp.744
749, Sept. 1998.
Hong-Hee Lee & Minh-Hoang Nguyen, Matrix converter
fed induction motor using a new modified direct torque
control method, Conference Proceedings of IECON
2004, Vol. 3, pp.2301 2306, Nov. 2004.
Guo qiangang, Li yaohua, Meng Yanjing & Liu Weiguo,
Modeling and simulation study on matrix converter fed
induction motor drive system implemented by direct
torque control, Proceedings of the Eighth International
Conference on Electrical Machines and Systems, Vol. 2,
pp.1069 1074, Sept. 2005.
Hong-Hee Lee received his B.S., M.S., and
Ph.D. degrees from Seoul National University,
Seoul, Korea. Currently, he is a Professor at
the School of Electric-Electronic Information
System Engineering, University of Ulsan,
Ulsan, Korea. He is also Director of the
Network-based Automation Research Center (NARC), which is
sponsored by the Ministry of Commerce, Industry and Energy
(MOCIE). His research interests are power electronics,
network-based motor control, and control network. He is a
member of IEEE, KIEE, KIPE, and ICROS.
Hoang M. Nguyen was born in Nha Trang,
Vietnam, in 1979. He received his B.S.
degree in Electrical Engineering from the
University of Technology, HoChiMinh,
Vietnam in 2002 and his M.S. degree from
the University of Ulsan, Korea in 2005. He is
currently a Ph.D. student at the University of Ulsan, Ulsan,
Korea. He is engaged in research on motor control, especially
matrix converters and industrial networks.
Tae-Won Chun was born in Korea in
1959. He received his B.S. degree in
Electrical Engineering from Pusan National
University in 1981, and received his M.S.
and Ph.D. degrees in Electrical Engineering
from Seoul National University in 1983 and
1987, respectively. Since 1986, he has been a member of the
faculty of the Department of Electrical Engineering, Ulsan
University, where he is currently a Full Professor. His current
research interests are control of electrical machines, power
converter circuits, and industrial applications.
