You are on page 1of 12

High-Frequency Equivalent Circuit of an Induction Motor Driven by a PWM

Inverter
TAICHI HIDAKA,
1
MUNEAKI ISHIDA,
1
TAKAMASA HORI,
1
and HIDEKI FUJITA
2
1
Mie University, Japan
2
Chubu Electric Power Co., Inc., Japan
SUMMARY
This paper deals with the high-frequency equiva-
lent circuits in an induction motor driven by a PWM
inverter. The leakage current flows through stray capaci-
tance among stator windings and iron core (frame) of the
induction motor at the switching instants of the inverter
transistors. The high-frequency equivalent circuits are
derived based on the frequency characteristics of the
high-frequency zero-phase impedance and line-to-line
impedance of the induction motor. The validity of the
derived equivalent circuits is confirmed by simulation
and experimental results. 2001 Scripta Technica,
Electr Eng Jpn, 135(1): 65-76, 2001
Key words: PWM inverter; stray capacitance; leak-
age current; common mode current; high-frequency oscil-
lating current; normal mode current.
1. Introduction
Variable-speed drive systems of induction motors by
inverters are widely used in the field of industrial applications.
The recent trend in this area is to use high-speed switching
devices such as IGBT and MOSFET in order to reduce noise
and to raise the carrier frequency of the PWM inverter to as
high as 10 to 20 kHz. Concurrently, the 'leakage current
flowing to the ground line through the stray capacitances of
the armature winding, and so on (called common mode cur-
rent) is increased, which may cause problems such as faulty
operation of the leakage current breaker.
In addition, an antisurge capacitor is often connected
to the ac input of the rectifier circuit as the power supply
for the inverter. Then, it may happen that the capacitance
between the inverter and the case is larger than the stray
capacitance between the induction motor and the ground.
In such a case, the leakage current does not reach the power
line, but flows directly between the inverter and the induc-
tion motor. This may produce faulty operation of the in-
verter control circuit or peripheral equipment.
On the other hand, a 'high-frequency oscillating cur-
rent (below called 'normal mode current) flows between the
motor terminals due to switching; this produces high-voltage
high-frequency oscillations between the motor terminals. The
resulting burden on the insulation system is a problem. There
has consequently been much research on means of suppress-
ing such high-frequency oscillations [1-6].
Hitherto, however, a simple RC circuit has been used
as the model for analysis of the induction motor. But this
model can be applied only in a limited frequency range, and
it has been impossible to simulate the high-frequency os-
cillations of the induction motor accurately.
This paper considers induction motors with various
ratings. Based on the high-frequency impedance charac-
teristics of induction motors, in their high-frequency range
(several kilohertz to several megahertz in this study), we
derive an equivalent circuit for the common mode current
(common mode equivalent circuit) and an equivalent circuit
for the normal mode current (normal mode equivalent
circuit) which are effective for a fairly wide range of fre-
quencies. We then present a method for simple estimation
of the circuit constants. The validity of the equivalent circuit
is shown by examination of the frequency characteristics,
simulation of the operating waveform of the proposed
circuit, and experimental results.
2. High-Frequency Oscillating Current
2.1 Experimental system for the inverter and
definition of oscillating current
The experimental system shown in Fig. 1 was used
to examine the common mode and normal mode current
2001 Scripta Technica
Electrical Engineering in Japan, Vol. 135, No. 1, 2001
Translated from Denki Gakkai Ronbunshi, Vol. 119-D, No. 12, December 1999, pp. 1485-1493
65
waveforms. By 'common mode current is meant the cur-
rent which flows through the stray capacitance between the
load and the ground on switching of the inverter. In the
experimental system, a 3.3-F capacitor (constituting a
sufficiently low impedance for the current fluctuations to
be observed) was connected in a Y-connection on the source
side in order to define the neutral point as the virtual ground,
so that the source impedance did not affect the measurement
results. That virtual ground point was then connected to the
ground (frame) of the motor, and the common mode current
is defined as the leakage current flowing through the
grounding wire.
By 'normal mode current is meant the line current
that flows when the above ground connection is removed
(that is, when switch S
0
is off). In particular, the high-
frequency oscillating current that flows at the timing of
inverter switching is considered as the normal mode
current. A current probe capable of measuring oscillating
currents at frequencies up to 50 MHz was used in the
measurements.
2.2 Measurement results
Table 1 shows the specifications of test machines A,
B, and C that were used in the measurements. Figure 2(a)
shows the measurement results when PWM control was
applied to the inverter and the test machine operated with
no load at 20 Hz and 70 V
rms
. It is evident that the current
waveform depends on the test machine, but that in essence
the high-frequency current flows immediately after the
application of a step voltage, followed by a relatively slow
decaying oscillation current.
Figure 2(b) shows the results of measurement of the
normal mode current. In this case too, the current waveform
depends on the test machine, but it is seen that the high-fre-
quency current flows immediately after the step voltage is
applied, followed by a slowly rising current. Thus, the
oscillating current differs considerably depending on the
kind of motor, indicating that the equivalent circuit cannot
be represented by a simple RLC circuit.
3. High-Frequency Impedance
The frequency characteristics of the zero-phase im-
pedance Z
tf
were measured in order to derive a common
mode equivalent circuit of the induction motor capable of
simulating the common mode current shown in Fig. 2(a).
The frequency characteristics of the impedance Z
uv
between
two terminals of the three-phase test machine were meas-
ured in order to derive a normal mode equivalent circuit of
the induction motor capable of simulating the normal mode
current shown in Fig. 2(b). The motor was loaded. A power
Fig. 1. Experimental system.
Table 1. Test machines (50 Hz/60 Hz)
Test
machine A
Test
machine B
Test
machine C
Number of poles 4 4 4
Rated output (kW) 2.2 0.75 2.2
Rated voltage (V) 200/220 200/200 200/200
Rated current (A) 9.6/8.6 3.8/3.4 9.3/8.8
Rated speed (rpm) 1430/1720 1410/1700 1420/1710
Slot shape fully closed fully closed half closed
66
operational amplifier was used as a high-frequency source
to apply a sinusoidal voltage with a constant amplitude of
10 V and a frequency of 10 kHz to 10 MHz across the
terminals. The impedance was calculated from the magni-
tudes of the voltage and current.
Figure 3 shows the results of measurement of the
zero-phase impedance for each test machine. In each ma-
chine, it is seen that the impedance decreases with the
frequency, accompanied by series and parallel resonances,
then declines slowly with small series and parallel reso-
nances. Figure 4 shows the results of measurement of the
line-to-line impedance. In each machine, a parallel reso-
nance appears and the impedance gradually decreases, ac-
companied by relatively small series and parallel
resonances. The line-to-line impedance is approximately 6
times the zero-phase impedance in the frequency range
from 5 to 10 MHz.
Fig. 2. Experimental waveforms of high-frequency
current.
Fig. 3. Frequency characteristics of common mode
impedance.
67
4. High-Frequency Equivalent Circuit of Induction
Motor
The windings of the induction motor include stray
capacitances between themselves and the core and between
various parts of the windings, in addition to the leakage
inductances and the resistances. From the viewpoint of its
essential structure, the equivalent circuit should be repre-
sented as a distributed-constant circuit. In this study, how-
ever, the model is represented by a lumped-constant circuit
[4]. Various equivalent circuits can be considered in this
approach, but we propose below an equivalent circuit that
provides for an extension from the common mode equiva-
lent circuit to the normal equivalent circuit, the symmetry
of the equivalent circuit between the connected three-phase
terminals and the frame, as well as between the neutral point
and the frame, the frequency characteristics of the imped-
ance, and so on.
4.1 Common mode equivalent circuit
In the common mode circuit, only the zero-phase
impedance is considered, and the stray capacitances be-
tween the three-phase windings are ignored, assuming that
their effect is small. The following properties are noted. (1)
The impedance is capacitive at low frequencies. (2) The
major resonances among the various resonances appearing
in the frequency characteristics of Fig. 3 are the two series
resonances and two parallel resonances (in the lower fre-
quency region). (3) The major resonances in the frequency
characteristics of the line-to-line impedance shown in Fig.
4 are two parallel resonances and one series resonance (in
the lower frequency region), taking an extension to the
normal mode equivalent circuit into consideration.
The ladder circuit shown in Fig. 5(a) is considered as
the common mode equivalent circuit. In the figure, the
circuit composed of r
4
, C
4
, r
3
, and C
3
is the circuit used to
simulate the behavior in the high-frequency range above 1
MHz. The inductive component is ignored, assuming that
it is smaller than the inductance of the cable connecting the
induction motor. In other words, the series and parallel
resonances of the third and higher modes are ignored,
assuming that their effect is small. In order to indicate
clearly the estimation of the circuit constants, a simplified
circuit is also used, in which the part of Fig. 5(a) enclosed
by the dotted line is denoted by r
2
and C
2
, as shown in Fig.
5(b).
4.2 Calculation of circuit constants in
common mode equivalent circuit
In the following, the circuit constants of the equiva-
lent circuit are determined from characteristic values such
as the resonant frequency and the impedance near the
resonances in the frequency characteristics for common
mode impedance (Fig. 3). Table 2 shows the items to be
noted in calculating the circuit constants. Since the equiva-
lent circuit itself is considerably approximated, approxi-
mate calculations are also used in order to simplify the
calculation of the circuit constants.
In the following, the series resonant frequency is
denoted by e
CS
(= 2tf
CS
) and the parallel resonant frequency
by e
CP
(= 2tf
CP
). The reactive components of C
i
(i = 1 ~ 4)
and L
0
are denoted by X
Ci
and X
L0
, respectively.
Fig. 4. Frequency characteristics of normal mode
impedance.
68
[1] Calculation of C
1
, C
2
, and L
0
In the circuits of Figs. 5(a) and 5(b), the zero-phase
impedance Z
.
C
is calculated by considering the relation at
resonance and ignoring the resistive component. The result
is as follows:
where
Applying the conditions for the series and parallel reso-
nances to the above equation, we obtain
Solving Eqs. (3) to (5), the relation between each
resonant frequency and C
2
/C
1
is obtained as shown in Fig.
6. Using this curve and the frequency dependence of the
zero-phase impedance shown in Fig. 3, C
2
/C
1
and the
resonant frequency are determined.
At frequencies that are 1/10 or less of the resonant
frequency (e
CL
= 2tf
CL
), the impedance can be approxi-
mately considered as a capacitance. The following relation
is then obtained:
C
1
and C
2
are determined from Fig. 6 and Eq. (6). L
0
is determined from the following relation:
Fig. 5. Common mode equivalent circuit.
Table 2. Constants of common mode equivalent circuit
Test machine
A
Test machine
B
Test machine
C
f
CL
[kHz] 10 10 10
Z
CL
[O] 5200 6700 4100
f
CP1
[kHz] 250 380 390
Z
CP1
[O] 425 1300 390
f
CP2
[kHz] 775 540 780
Z
CP2
[O] 190 900 208
f
CM
[kHz] 1400 1200 1500
Z
CM
[O] 98.4 355 152
ZZ
.
CM
[deg] 59.5 55.3 55.1
Z
CH
[O] 15 30 15
R
C
[O] 228 13.1 18.4
L
0
[H] 296 293 114
r
1
[O] 456 166 163
C
1
[nF] 0.32 1.18 0.97
r
3
[O] 44.5 1286 215
C
3
[nF] 1.22 0.26 0.97
r
4
[O] 15 30 15
C
4
[nF] 0.15 0.34 0.49
(1)
(2)
(3)
(4)
(5)
Fig. 6. Relationship between C
2
/C
1
and frequency.
(6)
69
[2] Calculation of C
3
, C
4
, r
3
, and r
4
In the high-frequency range beginning at several tens
of megahertz, the major current path is provided by r
4

C
4
. Applying further the approximation X
C4
<< r
4
, r
4
is
obtained as follows, where Z
CH
is the asymptotic line of the
impedance for f :
At frequencies at least twice as high as f
CP2
, the impedance
of the motor is well approximated by the input part to the
left of r
3
and C
3
in Fig. 5. Then, Z
.
C
is represented as follows:
where r
2
and X
C2
are given as follows:
At a frequency (e
CM
) that is about twice e
CP2
, the
effects of the resonances at lower frequencies as well as the
resonances of the ignored higher modes should be smaller.
Measuring the magnitude Z
CM
and the phase ZZ
.
CM
at e
CM
,
Z
CM
is decomposed into the series resistive component
(R
CM
) and the reactance component (X
CM
). Using Eqs. (9)
to (11), the following two expressions are obtained:
It is assumed that r
2
<< X
C2
, and that Eq. (2) is valid
even if there exists a resistive component. Then, C
3
, C
4
, and
r
3
are determined from Eqs. (2), (12), and (13).
[3] Calculation of R
C
and r
1
It follows from Fig. 5(a) that Z
.
C
is expressed as
follows:
where X
C1
= 1/ eC
1
, X
C2
= 1/ eC
2
, X
L0
= eL
0
. r
2
and X
C2
are given by Eqs. (10) and (11), respectively.
The resistive component of Eq. (14) is considered as
follows. When Q of the resonant circuit is high, it is in
general true that (resistive component)
2
<< (reactive com-
ponent)
2
. The approximation is also applicable to the case
where Q is low. Then, using the measured values of
Z
CP1
= Z
C
(e
CP1
) and Z
CP2
= Z
C
(e
CP2
), as well as C
3
, C
4
, r
3
,
r
4
, and L
0
, which have already been calculated, R
C
and r
1
are determined by the following relations (Appendix 1):
where r
2
is given by Eq. (10).
Based on the experimental results shown in Fig. 3,
the circuit constants for the common mode equivalent cir-
cuit were determined by the above procedure. Table 2
shows the results. When the resonance is not clear, it
suffices to set a point which is apparently close to the
resonance. A slight shift of the frequency will result in a
slight deviation of the circuit constants, but it is seen not to
have a large effect on the transient waveform.
4.3 Normal mode equivalent circuit
By extending the common mode equivalent circuit of
the induction motor, corresponding to the zero-phase com-
ponent, Fig. 7 is obtained. The impedance for the zero-
phase component and that for a single phase is in a ratio of
1 to 3. The stray capacitances C
a
C
and C
b
C
shown by the
dotted lines in the figure represent the stray capacitances
between lines, which are ignored in the common mode
circuit. The resistive components are ignored explicitly.
(8)
(9)
(10)
(11)
(12)
(13)
(14)
(15)
(16)
(7)
Fig. 7. High-frequency equivalent circuit.
70
Then the normal mode equivalent circuit is obtained as in
Fig. 8.
The winding resistance Rc is now written as R
n
, not
as 3 times R
C
, considering the effect of the line-to-line stray
capacitances in the extension to the normal mode equivalent
circuit. The internal stray capacitance C
a
C
and its resistive
component are combined with r
1
and C
1
, respectively, and
the results are designated as r
a
and C
a
. The line-to-line stray
capacitance C
b
C
and its resistive component, which produce
an effect while the voltage is rising, are combined with r
3
and C
3
, respectively, and the results are designated as r
b
and
C
b
. The part around the neutral point is mostly related to
the low-frequency characteristic when seen from the input
terminal, due to R
n
and L
0
. Consequently, they are neglected
for simplicity, assuming that r
4
<< r
2
, X
C2
, X
Cb
.
4.4 Calculation of circuit constants in normal
mode equivalent circuit
In the following, the circuit constants are calculated
from characteristic values such as the resonant frequency
and the magnitude of the impedance at resonance in the
frequency characteristics of the line-to-line impedance be-
tween two terminals of the three-phase system. Table 3
shows the characteristic values. In the calculation of the
circuit constants, approximations similar to those in the
common mode equivalent circuit are applied. Below, the
parallel resonant frequency is written as e
NP
, and the series
resonant frequency is written as e
NS
. The same values as in
the common mode equivalent circuit are used for L
0
, r
4
, and
C
4
.
[1] Calculation of C
a
and C
b
Considering the resonant frequency, and calculating
the line-to-line impedance Z
.
N
between two terminals of the
three-phase system while ignoring the resistive component,
the following expression is obtained:
The resonant frequency is determined from Eq. (17).
The relation between the resonant frequency and C
a
/(C
b
+
C
4
/6) is obtained as in Fig. 9. As the first step, C
a
/(C
b
+
C
4
/6) and f
NS
are determined from the figure. Then, C
a
is
determined from
C
b
is determined from Fig. 9.
[2] Calculation of r
b
As in the common mode equivalent circuit, it is
assumed that in the range of frequencies that are at least
twice f
NP2
, the line-to-line impedance of the motor is almost
entirely determined by the input part enclosed by the dotted
line in Fig. 8. Thus, the high-frequency impedance of the
equivalent circuit is
where X
4
= 1/ eC
4
, X
b
= 1/ eC
b
, X
L0
= eL
0
. r
b
C
and X
b2
C
are
given by
Table 3. Constants of normal mode equivalent circuit
Test machine
A
Test machine
B
Test machine
C
f
NP1
[kHz] 140 140 163
Z
NP1
[O] 6500 9450 6100
f
NS
[kHz] 410 320 510
Z
NS
[O] 1350 2853 1200
f
NM
[kHz] 1400 1000 1500
Z
NM
[O] 580 1350 400
L
0
[H] 296 293 114
r
4
[O] 15 30 15
C
4
[nF] 0.15 0.34 0.49
R
n
[O] 281 14.7 1.73
r
a
[O] 11.2 609 253
C
a
[nF] 0.17 0.28 0.28
r
b
[O] 407 1502 335
C
b
[nF] 0.29 0.22 0.53
Fig. 8. Normal mode equivalent circuit.
(17)
(18)
(19)
(20)
(21)
71
As in the common mode equivalent circuit, the meas-
ured impedance Z
NM
= Z
N
(e
NM
) at e
NM
is used to solve the
following equation for r
b
:
[3] Calculation of R
n
and r
a
The impedance Z
NS
= Z
N
(e
NS
) at e
NS
and the imped-
ance Z
NP1
= Z
N
(e
NP1
) at e
NP
are determined with consid-
eration of the resistive components. For the resistive
component of the series circuit, the approximation (resis-
tive component)
2
<< (reactive component)
2
is applied, as in
the calculation of R
C
and r
1
in the common mode equivalent
circuit. The following expressions are then obtained (Ap-
pendix 2):
By solving Eqs. (23) and (24) simultaneously, R
n
and r
a
are
obtained. The circuit constants of the normal mode equiva-
lent circuit calculated by the above procedure are shown in
Table 3.
5. Simulation Results
5.1 Frequency characteristics of impedance
In order to examine the validity of the equivalent
circuits in Figs. 5(a) and 8, the frequency characteristics of
the zero-phase impedance Z
C
and the line-to-line imped-
ance Z
N
were calculated using the derived circuit constants
(Tables 2 and 3). The results are shown in Fig. 10. Compar-
ing the results to experimental measurements (Figs. 3 and
4), the tendencies as a whole agree for test machines A, B,
and C. There are slight differences in the resonances near
the second series and parallel resonant frequencies. This is
Fig. 9. Relationship between C
a
/(C
b
+ C
4
/6) and
frequency.
(22)
(23)
(24)
Fig. 10. Frequency characteristics of impedance
(theoretical values).
72
due to the approximation of the equivalent circuit by a
lumped circuit, as well as to the approximations used in
deriving the circuit constants.
The purpose of this paper is to derive the circuit
constants by a simple algorithm. From this viewpoint, the
two proposed equivalent circuits seem adequate as equiva-
lent circuits to represent the impedances of an induction
motor.
5.2 Simulation of high-frequency oscillating
current
(1) Common mode current
In order to examine the extent to which the derived
common mode equivalent circuit can simulate the actual
common mode current, the transient current was calculated
for the case in which a step voltage is applied on the
common circuit, and the results were compared to the
experimental results in Fig. 2. Figure 11(a) shows the circuit
used for the simulation. The circuit constants shown in
Table 2 were used. The resistive component and the wiring
inductance of the common mode loop of the connecting
cable and the ground wire were measured with an LCR
meter (0.1 O, 7.5 H). They were connected in series to the
equivalent circuit. A step voltage of 280/3 V with a rise time
of 0.45 s corresponding to the zero-phase voltage was
applied, and the transient current i
C
was calculated.
Figure 12(a) shows the calculation results for test
machines A, B, and C. It is seen in each machine that a steep
current flows immediately after the step voltage is applied,
followed by a very high frequency oscillation and then a
relatively slow oscillating current. Compared to Fig. 2(a),
it is seen that the general tendencies, such as the peak value
of the oscillating current and the frequency of the immedi-
ately following oscillation, agree well.
Based on the above observations, it is concluded that
the proposed common mode equivalent circuit is fairly
adequate as an equivalent circuit to simulate the common
mode current.
(2) Normal mode current
In order to examine the extent to which the derived
normal mode equivalent circuit can simulate the actual
normal mode current, the transient current was calculated
for the case in which a step voltage was impressed on the
normal mode equivalent circuit, and the result was com-
pared to the measurements. Figure 11(b) shows the circuit
Fig. 11. Simulation circuit. Fig. 12. Waveforms calculated by equivalent circuit.
73
used for simulation. The circuit constants of Table 3 were
used. The resistive component and the inductance of the
normal mode loop of the connecting cable were measured
with an LCR meter (0.1 O, 20 H). They were connected
in series to the equivalent circuit. A step voltage of 280 V
with a rise time of 0.45 s corresponding to the line-to-line
voltage was impressed, and the transient current i
N
was
calculated. Figure 12(b) shows the calculation results for
test machines A, B, and C.
It is seen that in each machine a steep current flows
immediately after the step voltage is applied, followed by
a very high frequency oscillation and then a slowly rising
current with oscillation. Compared to the observed wave-
form of Fig. 2(b), it is seen that the general tendencies, such
as the peak value, the frequency of the high-frequency
oscillation, and the decay of the oscillation, agree well. The
slow rise of the current, however, cannot be simulated on
every test machine. This is because the equivalent circuit
constants are derived for high resonant frequencies (10 kHz
to 10 MHz), and it is somewhat difficult to apply the
equivalent circuit directly to lower frequencies.
It is thus concluded that the proposed normal mode
equivalent circuit is fairly adequate as an equivalent circuit
that simulates the normal mode current immediately after
the rise of the inverter output voltage. It will also be possible
to simplify the equivalent circuit as an RC circuit based on
the proposed equivalent circuit by restricting the frequency
range. It should be noted, however, that the proposed
equivalent circuit must be extended or replaced by another
circuit for the low-frequency range. Furthermore, the actual
oscillating current waveform is likely to be affected by the
operating point of the motor, the temperature of the motor,
the phase currents immediately before switching, the stray
capacitance of the switching element, and the impedance
of the wiring. These factors are left for future studies.
6. Conclusions
The frequency characteristics of the high-frequency
impedance of the locked induction motor (zero-phase im-
pedance Z
C
between three phases combined and the ground,
and line-to-line impedance Z
N
between two terminals of the
three-phase system) are measured. Based on the measure-
ment results, the common mode and the normal mode
equivalent circuits are derived. The procedure for determin-
ing the circuit constants is discussed. By comparing the
results of simulations using the equivalent circuit and the
results of experiments, it is concluded that the derived
equivalent circuit is valid in the frequency range from 10
kHz to 10 MHz.
This paper has principally discussed the validity of
the equivalent circuits. In the future the authors plan to
investigate the effectiveness and validity of the simplified
equivalent circuit in the restricted frequency range.
Acknowledgments
The authors are grateful for the advice of Mr. T. Goto,
a student in the first course of the doctoral program, and for
the assistance of Mr. N. Ito, an undergraduate student, in
the experiments.
REFERENCES
1. Murai Y, Kubota T, Kawase Y. Leakage current reduc-
tion for a high-frequency carrier inverter feeding an
induction motor. IEEE/IAS Annual Meeting Conf
Rec, p 344-348, 1987.
2. Ogasawara, Fujita, Akagi. Modeling and theoretical
analysis for high-frequency leakage current gener-
ated in voltage type PWM inverter. Trans IEE Japan
1995;115-D:77-83.
3. Goto, Ishida, Hori. A high-frequency zero-phase
equivalent circuit for inverter-driven induction motor.
Tech Rep Semiconductor Power Devices, IEE Japan,
SPC-95-37, 1995.
4. Goto, Ishida, Hori. A high-frequency zero-phase
equivalent circuit for inverter-driven induction motor.
Trans IEE Japan 1996;116-D:1018-1026.
5. Ogasawara, Ayano, Akagi. Active cancellation of
common mode voltage generated in voltage type
PWM inverter. Trans IEE Japan 1997;117-D:565-
571.
6. Hidaka, Ishida, Hori, Fujita, Fuwa. A high-frequency
equivalent circuit (ladder circuit) PWM inverter-
driven induction motor. Trans Res Group Power
Electron 1998;23:56-62.
APPENDIX 1
Derivation of Eqs. (15) and (16)
When e = e
CP1
in Eq. (14), we have
Then, Z
CP1
.
is obtained as follows:
Applying the approximations r
2
2
<< X
C2
2
, (X
L0
X
C1
)
2
>> { R
C
X
C1
+ X
L0
(R
C
+ r
1
+ r
2
) X
C1
(R
C
+ r
2
)
2
, the fol-
lowing relation is obtained:
(A.1)
(A.2)
74
As the next step, let e = e
CP2
. Then,
Z
CP2
.
is written as follows:
Applying the approximation r
2
2
<< X
C2
2
the following relation is obtained:
Equations (15) and (16) are derived from Eqs. (A.3)
and (A.6).
APPENDIX 2
Derivation of Eqs. (23) and (24)
Applying the approximations in the main text, Z
.
N
is
derived as follows:
When e = e
NS
, it follows from Eq. (17) that
Substituting this into Eq. (A.7) and rearranging,
is obtained. Applying the approximation r
b
C2
<<
X
b
C2
, {3X
L0
(4R
n
+ 12r
4
+ 4r
a
+ r
b
C
) X
b
C
(2R
n
+ 6r
4
+ ra)
2
<< (3X
b
C
X
L0
)
2
, Eq. (23) of the main text is obtained.
Next, let e = e
NP
. It is assumed that the following
relation applies by virtue of the resonance relation in Eq.
(17), even if there exists a resistive component:
By virtue of these relations, Eq. (A.7) is written as follows:
Appl yi ng t he approxi mat i on r
b
C2
<< X
b
C2
,
{6X
L0
(4R
n
+ 12r
4
+ 2r
a
) - X
a
(4R
n
+ 12r
4
)
2
<<
{12X
L0
(X
a
3X
L0
)
2
, Eq. (24) of the main text is derived.
(A.3)
(A.4)
(A.5)
(A.6)
(A.7)
(A.8)
(A.9)
(A.10)
(A.11)
75
AUTHORS (from left to right)
Taichi Hidaka (student member) received his B.S. degree from the Department of Electrical and Electronic Engineering
of Mie University in 1996 and then entered the first half of his doctoral program.
Muneaki Ishida (member) received his D.Eng. degree in electrical and electronic engineering from Nagoya University in
1980. He has been a professor in the Department of Electrical Engineering of Mie University since 1996. He received an IEEJ
Paper Award in 1993 and an IEEJ Publication Award in 1999. He is a member of IEEE, the Research Group on Power Electronics,
the Society of Instrumentation and Control Engineers, and the Japan Society of Mechanical Engineers.
Takamasa Hori (member) received his B.S. degree in electrical engineering from Osaka University in 1961 and then joined
Hitachi Ltd. (Hitachi Central Research Laboratory). He later received a D.Eng. degree. Since 1987, he has been a professor in
the Department of Electrical Engineering of Mie University. He was the recipient of an IEEJ Paper Award in 1993, an IEEJ
Control Award in 1994, and an IEEJ Publication Award in 1999. He is a Fellow of IEEE and of the Society of Instrumentation
and Control Engineers, and a member of the Research Group on Power Electronics.
Hideki Fujita (member) received his B.S. degree in electrical engineering from Nagoya University in 1973 and then joined
Chubu Electric Power Co., where he engaged in business activity and research related to power systems. He received his D.Eng.
degree from Nagoya University in 1995. Currently, he is Head of Research, Chubu Electric Power Co., Power Technology
Research Laboratory. He received an IEEJ Progress Award in 1990. He is a member of IEEE.
76

You might also like