IEEE PEDS 2011, Singapore, 5 - 8 December 2011

A Modified Three Dimensional Space Vector

based PWM Method for Four-Leg Voltage
Source Inverter Fed Asymmetrical Two-Phase
Induction Motor
Heli Golwala, R. Chudamani
Electrical Engineering Department, Sardar Vallabhbhai National Institute of Technology, Surat, India
heligolwala@gmail.com, rc@eed.svnit.ac.in

Abstract- This paper presents a modified three dimensional

space vector PWM method for two-phase four-leg voltage source
inverter and its realization for asymmetrical two-phase
induction motor. The two-phase orthogonal reference voltages
are generated using the Volts/Hertz control technique in open
loop and are realized using a four leg voltage source inverter.
The two phase voltages are obtained by tapping the output from
terminal pairs ab and cd. The three dimensional space vector
modulation without null vector is used to generate the switching
signals for the inverter. This technique reduces the switching
losses in the inverter and the experimental and simulation
results obtained validate the algorithm used.
Keywords: 3 D SVM, Null Vector, 2 Induction motor, 4
Leg VSI

I.

waveforms with independently adjustable amplitudes on each output

phase.
Section II describes the mathematical model of the asymmetrical
2- I.M.. Section III and IV present a four-leg VSI topology with
proposed switching technique and V/Hz control characteristics,
respectively. Simulation and experimental results are then provided
in Section V.
II.

MATHEMATICAL MODEL OF ASYMMETRICAL

2- I.M.

The d-q axes equivalent circuits of an asymmetrical 2

I.M. in stationary reference frame is shown in Fig. 1(a & b).
The complete mathematical model equation of machine in the
stationary reference frame is represented in matrix form as
follows:

INTRODUCTION

n recent years, asymmetrical two-phase induction motor

(2 I.M.) have been extensively used in low power
applications [1]-[5] such as pumps, drill, compressor etc.
A three-leg voltage source inverter (VSI) fed vector and
scalar controlled asymmetrical 2 I.M. drives have been
widely developed. The most popular topology is the three-leg
VSI drives using pulse width modulation (PWM) control [6][9]. Based on this principle, the auxiliary and main winding
voltages of asymmetrical 2 I.M.s are required to have a
phase difference of 90 and independent magnitude with
respect to each other. The voltage magnitude relationship of the
two-phase depends on the turn ratio of windings. The auxiliary
winding phase voltage is higher than the main winding phase
voltage of the machine.
The sinusoidal PWM control strategy is used in [8, 9]. A simple
scalar volts/hertz control is used to generate the reference voltages.
The three leg VSI is used as Power converter. Though it offers good
steady state and dynamic performances, this topology suffers from
neutral point balance. In paper [1] the four-leg inverter with SPWM
technique is used. This technique has less DC bus voltage utilization
compared to the space vector pulse width modulation technique.
In this paper a novel three dimensional space vector pulse width
modulation technique (3D SVPWM) without the null vector is
proposed. The gating signals for the switches of the four leg inverter
are generated using TMS320F2812 DSP. The proposed control
algorithm can generate unbalanced two-phase output voltage

Fig. 1(a): d-axis equivalent circuit

Fig. 1(b): q-axis equivalent circuit

0

0
0
0

(1)

=
1

where vsd, vsq are the stator voltages, isd, isq, ird, irq are the stator
and rotor currents, Rsd, Rsq, Rrd, Rrq are the stator and rotor
resistances, Lsd, Lsq, Lrd, Lrq are the stator and rotor self inductances,
Lmd, Lmq are the mutual inductances, r is the rotor angular speed

978-1-4577-0001-9/11/$26.00 2011 IEEE

573

(expressed in electrical rad/s), p is the differential operator, a is the

windings turns ratio, defined as the ratio between number of turns
for the auxiliary and main stator windings, and d, q subscripts used
to represent auxiliary and main windings quantities, respectively.
The equations relating different currents and flux-linkages in the
stationary reference frame are
0

0
0

(2)

0
0

where, sd, sq, rd, rq are the stator and rotor flux-linkages.
The electromagnetic torque Te produced by the machine can be
written in terms of the stator and rotor currents as
=

(3)

where, P is the number of poles. The rotor angular speed r can

be determined by the electromechanical equation as follows:
=

(4)

voltages vsd and vsq are expressed as follows:

0
(6)
= 0
Under steady state conditions the differential operator in
(5) can be replace by js. Therefore, the steady-state voltage
equation can be written in vector form as
(7)
=
In order to maintain constant flux-linkage value of the
machine, the ratio of the stator voltage amplitude Vs to the
electrical frequency s should be kept constant. Thus, the
reference applied voltages vsd* and vsq* are given by
cos
(8)
= 1
cos
2
From (8), the d-q reference frame voltages are obtained.
For implementing the prism identification these voltages are
written - frame, as v(vdss) and v(vqss) which is used to
identify prisms. These vdss and vqss are further converted to
va*,vb*, and vc* given by (9) which is used to identify the
tetrahedron in the proposed method which is explained
elaborately in the following section.
1

where, J is the moment of inertia, and TL is the load torque.

Therefore, the rotor speed nr can be easily written as
=

60

(5)

These model equations are used in the following section which

describes the volts/Hertz method of speed control of Induction
motor.
III.

VOLT/HERTZ CONTROL CHARACTERISTIC

Variable-frequency drives originally used open-loop V/Hz

control to regulate the flux-linkage of machine and maintain
its magnitude constant. Therefore, the dynamic equation in
(1) can be used to relate the stator voltages to the electrical
frequency in the steady-state term as follows: From (1) and
(2), the voltage equation can be rewritten in currents and fluxlinkages form as
0
0
0

0
0
0

0
0

0
0

2
3 0
0.5

IV.

1
2
3
2
0.5

1
2
3
2
0.5

(9)

PROPOSED SWITCHING TECHNIQUE

Fig. 2 shows the block diagram of the open loop control of two
phase induction motor for 4 leg VSI. Speed control is done using
V/F control method. Switching pulses for four leg VSI are generated
using modified 3D-SVPWM method. With the V/F profile of
induction motor we get the stationary voltages in d-q frame. The
same voltages are converted in to - frame and then into a-b-c
reference frame for prism and tetrahedron identification respectively.

0
0
0

0
0
0
0
0

0
0

0
0

(5)

0
0

Fig. 2: Block Diagram of the 2 I.M. with proposed method

The asymmetrical 2 I.M. is connected to 4 leg VSI as shown in

fig. 3. It is composed of four legs with two IGBT devices on each
leg.

From the first and the second rows in (5), the stator

574

A. Prism Identification
The prism identification is very similar to the sector
identification for 2-D SVM. Based on the projections of the
reference vector on the - plane, v and v, six prisms in the 3-D
space can be identified and numbered as Prisms I through VI.
Within the selected prism, there are six non-zero switching state
vectors and two zero switching state vectors which are eliminated in
proposed technique. Fig. 4 shows the physical positions of the
switching state vectors in 0-- coordinate system.
Fig. 3: Two Phase 4 Leg VSI

The 3D-SVPWM without considering null vector (proposed

method) is implemented in the following sequence: (i) Identification
of prism; (ii) Identification of tetrahedrons; (iii) Duty cycle
calculation; (iv) Generation of PWM sequence for the switches. The
first two steps for implementation of 3D-SVPWM, like
identification of prism and tetrahedrons are well elaborated in [4].
The most important step in 3D-SVPWM is to calculate the duty
cycle and to translate the information into four PWM waveforms for
the given sequencing scheme.
The major advantage of a four-leg inverter is that the neutral
point balancing is not required. In 3D-SVM, there are 16 possible
switching vectors: fourteen active nonzero vectors and two null
vectors. In 3D-SVM with null vector, there are 16 possible
switching vectors: fourteen active nonzero vectors and two null
vectors whereas in the proposed method two null vectors are
eliminated. The entire 3- D Space is divided into 6 prisms and 24
tetrahedrons. Each prism consists of four tetrahedrons. An
instantaneous reference input vector may lie in any of these
tetrahedrons at any point of time. Each operation is explained in a
detailed manner in following subsection.

The identification of prism is based on the value of the switching

vectors expressed in 0-- reference frame. An algorithm for prism
identification is explained with the help of a flowchart. Fig. 5 shows
the logic flow in prism identification.

Fig. 5: Sequence of Operations in Prism Identification

B. Tetrahedron Identification
Once the prism information is obtained the next step is to
identify the tetrahedron in which the reference vector is present. To
do this each prism is further subdivided into four tetrahedrons - thus
leading to a total of 24 tetrahedrons. Each tetrahedron is formed by
three non-zero switching state vectors and two zero switching state
vectors. The line-to-neutral voltage polarities produced by each non
zero switching state vectors are indicated as + or - or 0. It is
important to notice that within each tetrahedron, all the non zero
switching state vectors produce non-conflicting line-to-neutral
voltages, and thus they are adjacent vectors. Table I shows how
these tetrahedrons could be identified based on the sign of required
phase voltages at the output of the inverter [4].
C. Duty cycle Calculation and Pulse Generation
The computation of the duty cycles is based on calculation of
geometric projection of the reference vector along the three non-zero
switching state vectors which form the edges of the given
tetrahedron at a given instant. For each tetrahedron we can find a
3x3 projection matrix: St. The reference vector in 0-- reference
frame at any given time is denoted by a 3x1 column matrix Sr. Then
at any given instant, the duty cycle matrix D can be computed from
the simple matrix equation:

D=

= St-1 * Sr

(10)

Fig. 4: Physical Position of Switching State Vector

575

TABLE I: TETRAHEDRON IDENTIFICATION

PRISM

TETRAHEDRON
P1_T1
P1_T2
P1_T3
P1_T4
P2_T1
P2_T2
P2_T3
P2_T4
P3_T1
P3_T2
P3_T3
P3_T4
P4_T1
P4_T2
P4_T3
P4_T4
P5_T1
P5_T2
P5_T3
P5_T4
P6_T1
P6_T2
P6_T3
P6_T4

ACTIVE
VECTORS
V8,V9,V13
V8,V12,V13
V1,V9,V13
V8,V12,V14
V4,V12,V13
V4,V5,V13
V1,V5,V13
V4,V12,V14
V4,V5,V7
V4,V6,V7
V1,V5,V7
V4,V6,V14
V2,V6,V7
V2,V3,V7
V1,V3,V7
V2,V6,V14
V2,V3,V11
V2,V10,V11
V1,V3,V11
V2,V10,V14
V8,V10,V11
V8,V9,V11
V1,V9,V11
V8,V10,V14

CONDITION
Vaf > 0, Vbf < 0, Vcf < 0
Vaf > 0, Vbf > 0, Vcf < 0
Vaf < 0,Vbf < 0, Vcf < 0
Vaf > 0, Vbf > 0, Vcf > 0
Vaf > 0, Vbf > 0, Vcf < 0
Vaf < 0, Vbf > 0, Vcf < 0
Vaf < 0, Vbf < 0, Vcf < 0
Vaf > 0, Vbf > 0, Vcf > 0
Vaf < 0, Vbf > 0, Vcf < 0
Vaf < 0, Vbf > 0, Vcf > 0
Vaf < 0, Vbf < 0, Vcf < 0
Vaf > 0, Vbf > 0, Vcf > 0
Vaf < 0, Vbf > 0, Vcf > 0
Vaf < 0, Vbf < 0, Vcf > 0
Vaf < 0, Vbf < 0, Vcf < 0
Vaf > 0, Vbf > 0, Vcf > 0
Vaf < 0, Vbf < 0, Vcf > 0
Vaf > 0, Vbf < 0, Vcf > 0
Vaf < 0, Vbf < 0, Vcf < 0
Vaf > 0, Vbf > 0, Vcf > 0
Vaf > 0, Vbf < 0, Vcf > 0
Vaf > 0, Vbf < 0, Vcf < 0
Vaf < 0, Vbf < 0, Vcf < 0
Vaf > 0, Vbf > 0, Vcf > 0

As the null vector is eliminated the duration for each switching

period varies. The switching period is determined by (d1 + d2 + d3) *
Ts where Ts is the sampling period if the null vectors were
considered. This results in variable switching frequency operation.
Though the frequency of operation is not constant it varies over a
small range only. This is evident from the plot of (d1 + d2 + d3) as a
function of time as shown in Fig. 6. When reference vector is pure
to
i.e. 14
sine wave the switching frequency varies from
.
.
kHz to 16 kHz for a sampling frequency of 20 kHz.
1

0.8

d 1 +d 2 +d 3

0.7
0.6

0.4

0.2

0
0.052

0.054

0.056

0.058

0.06 0.062
Time (s)

0.064

0.066

0.068

Fig. 6: Plot of (d1+d2+d3) with Reference vector derived from balanced three phase
sinusoidal voltages

The next step is to choose an appropriate switching scheme and

to generate the four PWM pulses for the top switches of the inverter.
The three duty-cycle values d1, d2 and d3 are already known prior to
this step. The PWM pulses are generated by comparing the four
signals X1, X2, X3, and X4 with a triangular waveform of switching
frequency equals to the sampling frequency of the modulator, where
sampling frequency of modulator is given by (d1+d2+d3) * Ts. Let

X1, X2, X3 and X4 are the four signals calculated from the duty
cycles d1, d2, and d3 as given in the following matrix equation.
Xt =

= 2 * At

(11)

X4 = 1
X1 = 0

(12)
(13)

In (11), At is the peak value of the triangular waveform. The

period of this waveform is (d1+d2+d3) * Ts, which decides the
switching frequency of the inverter. The relation between the
elements of matrix Xt and the switching signals Sa, Sb, Sc, and Sf
would be different for different tetrahedrons. The assignment of the
pulses generated by the comparator is given in table II. Fig. 7 shows
the set of signals derived for tetrahedron P4_T3. It is seen from the
table that one device remains ON and one device remains OFF
throughout the switching period.

Fig. 7: Principle of PWM waveform Generation

TABLE II: PULSE ASSIGNMENT FOR EACH TETRAHEDRON
NON ZERO
SWITCHING
PRISM
TETRAHEDRON
Sa
Sb
Sc
STATE
VECTORS
P1_T1
V8,V9,V13
0
X3
1
P1_T2
V8,V12,V13
0
X2
1
1
P1_T3
V1,V9,V13
X2
X3
1
P1_T4
V8,V12,V14
0
X2
X3
P2_T1
V4,V12,V13
X2
0
1
P2_T2
V4,V5,V13
X3
0
1
2
P2_T3
V1,V5,V13
X3
X2
1
P2_T4
V4,V12,V14
X2
0
X3
P3_T1
V4,V5,V7
1
0
X3
P3_T2
V4,V6,V7
1
0
X2
3
P3_T3
V1,V5,V7
1
X2
X3
P3_T4
V4,V6,V14
X3
0
X2
P4_T1
V2,V6,V7
1
X2
0
P4_T2
V2,V3,V7
1
X3
0
4
P4_T3
V1,V3,V7
1
X3
X2
P4_T4
V2,V6,V14
X3
X2
0
P5_T1
V2,V3,V11
X3
1
0
P5_T2
V2,V10,V11
X2
1
0
5
P5_T3
V1,V3,V11
X3
1
X2
P5_T4
V2,V10,V14
X2
X3
0
P6_T1
V8,V10,V11
0
1
X2
P6_T2
V8,V9,V11
0
1
X3
6
P6_T3
V1,V9,V11
X2
1
X3
P6_T4
V8,V10,V14
0
X3
X2

Sf
X2
X3
0
1
X3
X2
0
1
X2
X3
0
1
X3
X2
0
1
X2
X3
0
1
X3
X2
0
1

576

Using the proposed method the open loop speed control of 2-ph
induction motor is implemented in hardware and the experimental
results are presented in the following section. Add this sentence at
the end of this section.

phase currents as shown in 9(c). Fig. 9(f) and 9(g) shows the
electromagnetic torque and the rotor speed. It is seen that the
rotor speed settles to a steady state value of 1800 rpm in 0.4 s.

V.

SIMULATION AND EXPERIMENTAL

RESULTS
The modeling of an asymmetrical 2 I.M. in this paper
has been developed in Matlab/Simulink environment. The
four leg inverter topology shown in Fig. 3 is used for
simulation studies. The motor parameters used for the
simulation are listed in Table III. To demonstrate feasibility
of the proposed open loop V/Hz SVPWM algorithm for the
two-phase four-leg VSI, the inverter is supplied a constant
dc-link voltage at 500 V.

(a)

(b)

TABLE III: SIMULATION PARAMETERS FOR ASYMMETRICAL 2

I.M.
Rsd = 45.24
Lsd = 1.06H
Lmd = 0.9174H
Rsq = 9.04
Lsq = 0.7916H
Lmq = 0.7479H
Rrd = 26.76
Lrd = 0.9887H
J = 2.8 e -3 kg-m2
Rrq = 7.56
Lrq = 0.7698H
a = 1.71
HP = 0.5
Vs = 110 V

(c)

The ratings of the asymmetrical 2 I.M are 1/2 hp, 220 V,

50 Hz. Fig. 8 shows the simulated waveforms for steady-state
output voltages vsd, vsq of the unbalanced two-phase four-leg
VSI when the asymmetrical 2 I.M. utilizing the proposed
method is operated. It can be seen that the output voltages are
identical in amplitude at 50 Hz, while the voltages are
displaced at 90.

(d)

(e)

Fig. 8: Two-phase voltage waveform of the auxiliary and the main winding.
vsd, vsq

Fig. 9 (a-g) shows the simulated waveforms for the

asymmetrical 2 I.M. driven by a four leg VSI, using the
proposed method in which the magnitude commands of two
output voltages were unbalanced because of the windings
turn ratio of machine.
In Fig. 9(a), the main and auxiliary supply voltages are
identical in amplitude at 50 Hz and orthogonal in phase. The
voltage commands were selected according to vaux = av
main[6]. The supply voltage vaux was fixed at 1.71vmain since the
winding turn ratio a was estimated to be approximately 1.71.
The simulation results of the dynamic response under no-load
is shown in Fig. 9(b) and (d). It can be seen that the stator and
rotor currents reaching the steady state values in 0.4 s. The
phase difference of 90 is observed between the two stator

(f)

(g)

Fig. 9 (a-g): Simulation results of 4-leg inverter fed 2 I.M.

(vmain = 110 Vrms and vaux. = a*110 Vrms)

577

The experimental setup of the proposed control system is

done using digital signal controller and 4 leg VSI. It consists
of a TMS320F2812 Digital Signal controller board with on
chip ADC and PWM facility. A four-leg VSI is connected to
dc-link voltage supply at 200 V, with the output frequency
command, the switching frequency and the dead time were
set to 50 Hz, 14 to 16 kHz, and 4 s, respectively. The control
logic is written in C language.
Fig. 10 shows the measured waveforms for output
voltages (vsd and vsq) of the two-phase four-leg VSI when the
proposed method is implemented using digital signal
controller. It can be seen that, the output voltages are identical
in amplitude at 50 Hz, while the voltages are orthogonal.

REFERENCES
[1] A. M. Hava, R. J. Kerkman, and T. A. Lipo, Carrier- Based PWM-VSI Over
modulation Strategies: Analysis, Comparison, and Design, IEEE Transactions on
Power Electronics, vol. 13, no. 4, pp. 674-689, July 1998.
[2] F. Blaabjerg, F. Lungeanu, K. Skaug, and M. Tonnes, Two-Phase Induction Motor
Drives, IEEE Industry Applications Magazine, vol. 10, pp. 24-32, July/August2004.
[3] E. Muljadi, Y. Zhao, T. H. Liu, and T. A. Lipo,Adjustable ac Capacitor for a
Single-Phase Induction Motor, IEEE Transactions on Industrial Electronics, vol.29, no.
3, pp. 479-485, 1993.
[4] Zhang, R.; Prasad, V.H.; Boroyevich, D.; Lee, F.C., "Three-dimensional space
vector modulation for four-leg voltage-source converters", Power Electronics, IEEE
Transactions on, Volume 17, Issue 3, May 2002 Page(s):314- 326.
[5] C. M. Young, C. C. Liu, C. H. Liu. New Inverter- Driven Design and Control
Method for Two-Phase Induction Motor Drives, IEE Proceedings-Electric Power
Applications, volume 143, no. 6, pp. 458-466, (1996).
[6] M. B. de R. Correa C. B. Jacobina, A. M. N. Lima, E. R. C. da Silva. A Three-Leg
Voltage Source Inverter for Two-Phase AC Motor Drive Systems, IEEE Transaction
Power Electronics, volume 17, no. 4, pp. 517-523, (2002).
[8] V. Kinnares, Ch. Charumit. Modulating Functions of Space Vector PWM for
Three-Leg VSI-Fed Unbalanced Two-Phase Induction Motors, IEEE Transaction
Power Electronics, volume 24, no. 4, pp. 1135-1139, (2009).
[9] Do-Hyun. PWM Methods for Two-Phase Inverters, IEEE Industry Applications
Magazine, volume 13, pp. 50-61, (2007).
[10] Ch. Charumit, V. Kinnares, Carrier-Based Unbalanced Phase Voltage Space
Vector PWM Strategy for Asymmetrical Parameter Type Two-Phase Induction Motor
Drives, Electric Power System Research, volume 79, pp. 1127-1135, (2009).

Fig. 10: Measured waveforms of two-phase output voltages for 4 leg VSI
with proposed technique

VI.

CONCLUSION

In this paper the modified 3D-SVPWM method for two-phase fourleg VSI fed asymmetrical 2 I.M. is proposed. The control strategy
for two-phase operation has been developed to synthesize
unbalanced two-phase output voltage waveforms with open-loop
V/f control. The proposed SVPWM method can generate two phase
output voltages with independently adjustable magnitudes in two
output phases. The proposed PWM scheme allows the four-leg VSI
to feed sinusoidal output orthogonal voltages for asymmetrical 2
I.M. which is unbalanced two-phase load. The simulation and
experimental results confirm the feasibility of the developed control
scheme for the two-phase four-leg VSI drive system.

578