IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 18, NO.

6, NOVEMBER 2003

1309

A New Active Common-Mode

EMI Filter for PWM Inverter
Yo-Chan Son, Student Member, IEEE, and Seung-Ki Sul, Fellow, IEEE

AbstractThis paper presents a new active common-mode electromagnetic interference (EMI) filter for the pulse-width modulation (PWM) inverter application. The proposed filter is based on
the current sensing and compensation circuit and it utilizes a fast
transistor amplifier for the current compensation. The amplifier
utilizes an isolated low-voltage dc power supply for its biasing and
it is possible to construct the active filter independent of the source
voltage of the equipment. Thus the proposed active filter can be
used in any application regardless of its working voltage. The effectiveness of the proposed circuit has been verified by experimental
results.
Index TermsCommon-mode, EMI, inverter, PWM, transistor
amplifier.

I. INTRODUCTION

HE PWM inverter as shown in Fig. 1, is the inherent noise

)
source that makes abrupt voltage transitions (high
accompanied by switching actions [1], [2]. Coupled with the
stray capacitance of the load machine the high frequency current
is generated, which can affect the operation of nearby equipments due to the conducted and radiated electromagnetic interference (EMI) [3], [4]. As the semiconductor technology progresses power devices are getting faster to reduce the switching
loss and to increase the controllability of the system, but the inis accompanied by the increase of the EMI level.
crease of
Especially the leakage current by the common-mode voltage of
the PWM inverter is the primary concern of the conducted and
radiated EMI. Because the high frequency leakage current at the
motor returns through the earth ground, its circulation loop is
relatively large compared with that of the normal-mode current.
This large circulation loop plays as a role of the antenna for the
radiated EMI. Recently, many countries limit the EMI emission
under their regulations [3][6]. To meet such regulations many
efforts to EMI mitigation have been concentrated and most of
them have been related to the design of input/output EMI filters
[1][3], [7]. Most of conventional filters were passive filters. In
designing passive filters the obtainable insertion loss is limited
by the available LC product that should be increased to meet the
requirement. In case of the common-mode EMI, one cannot increase the total capacitance for the safety reason [3] and should
increase the inductance of the common-mode choke in order

to increase the insertion loss, which makes the common-mode

filter bulky. Many researchers have tried to develop some active filtering techniques in order to overcome such limitations
[8][12].
In this paper, a new active common-mode EMI filter is introduced in order to mitigate the conducted common-mode EMI.
It can provide the sufficient attenuation under the limited LC
product and there is a promising possibility of its application
regardless of the working voltage of the system. Its analysis and
experimental verification will be given in this paper.
II. ACTIVE COMMON-MODE EMI FILTER
Fig. 2 shows the basic concept of the proposed active
common-mode EMI filter (ACEF). The circuit is based on the
topology using the current sensing and compensation [8], [9],
[11], [12]. The noise source is the PWM inverter in Fig. 2 and
the input filter can be an additional passive filter, which gives
additional insertion loss. The series-connected common-mode
works as the common-mode current sensing elechoke
ment by the additional winding. The high frequency current that
generates the high frequency flux in the
passes through
common-mode choke, which makes the high frequency voltage
at the input terminal of the trans-conductance amplifier. The
output of the amplifier is connected to the output capacitor
that is used for the current injection to the earth ground. In this
filter circuit, the injected current cannot be circulated within
because
the system without using the coupling capacitor
is used to provide
the closed loop cannot be made. Thus
the low-impedance path of the high frequency common-mode
current for the internal circulation. The supply voltage of the
is used to give the bias voltage to the amplifier.
filter circuit
is large enough to
At low frequency the impedance of
isolate the bias voltage from the main voltage.
The required bias voltage of ACEF can be calculated as in (1).
For example, if 0.5 A of the high frequency current at 1 MHz
should be supplied, then the bias voltage should be larger than
, which is the
7.96 V in order to drive a 10 nF capacitor
and the coupling
series impedance of the output capacitor
capacitor
(1)

Manuscript received July 20, 2002; revised February 1, 2003. Recommended

by Associate Editor S. Y. Hui.
Y.-C. Son is with the System Group, Discrete Team, Fairchild Semiconductor,
Gyeonggi-do 420-711, Korea (e-mail: yochan@fairchildsemi.co.kr).
S.-K. Sul is with the School of Electrical Engineering and Computer Science, Seoul National University, Seoul 151-744, Korea (e-mail:
sulsk@plaza.snu.ac.kr).
Digital Object Identifier 10.1109/TPEL.2003.818829

Fig. 3 shows the application of the proposed ACEF. In this

example the single-phase, three-phase, and dc applications are
shown. As introduced in [12], a push-pull amplifier is used and
is connected to the dc-bus of the PWM inverter or any dc load
system. In case of the PWM inverter supplied by the commercial

0885-8993/03$17.00 2003 IEEE

1310

IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 18, NO. 6, NOVEMBER 2003

Fig. 1. PWM inverter system using single-phase ac input.

impedance at the frequency band of interest. If the inductance of

the common-mode choke seen at the primary winding is
without the consideration of the secondary winding, then the relation between the input common-mode current and the base
current of the push-pull amplifier can be found as
(2)
Fig. 2. Basic concept of proposed ACEF.

60 Hz/220
utility line, the dc-bus voltage is about 300
.
If the push-pull amplifier should be placed across the dc-bus capacitor as suggested in [11], then each transistor should be able
to handle the full dc-bus voltage. The high-voltage/high-current
pnp-transistor is hardly available; hence that limits its application. Moreover the required voltage of the push-pull amplifier is
very small compared to the dc-bus voltage as shown in (1) and it
can make the inefficient voltage usage. However in the proposed
circuit the additional low voltage supply can be used to drive the
push-pull amplifier. This enables the use of low voltage devices
and extends the application of the proposed ACEF. Fig. 3(a)
shows its application when coupling capacitors are placed beand the filter supply voltage .
tween the dc-bus voltage
While coupling capacitors in Fig. 3(a) isolate the dc-bus and the
ACEF at low frequency, they make the low impedance path at
high frequency, which allows the internal circulation of the high
frequency leakage current between the system and the ACEF as
introduced in [12]. The filter supply voltage can be easily obtained by using a dc power supply fed to the control electronics
such as a gate drive circuit. Unlike an ideal voltage source, there
can be some voltage ripple in the filter supply if a switched
mode power supply (SMPS) is used for this dc power supply. In
this case, the noise produced by the dc power supply may flow
in Fig. 3(a),
into the dc bus through the coupling capacitors
which may affect the total conducted EMI including the PWM
inverter and the proposed ACEF. Besides the high-frequency
noise produced by the PWM inverter can be transmitted to the
with the same
control electronics via the coupling capacitors
should provide sufficient low
manner. Thus the filter supply
impedance to decouple such high-frequency noises. Coupling
capacitors also can be connected to ac input lines of the system
as shown in Fig. 3(b) and it is possible to construct a separate
input filter stage. The same idea can be extended to the 3-phase
applications, which is shown in Fig. 3(c) and (d).
The proposed circuit works as follows. In this analysis there
is an assumption that coupling capacitors have sufficiently low

is the turn ratio between the primary winding and

where
is the input impedance of the
the secondary winding, and
push-pull transistor amplifier including the additional resistor. If
the leakage inductance of the common-mode choke can be negcan be represented
ligible, the common-mode inductance
as
(3)
is the effective area, and
where is the relative permeability,
is the effective length of the magnetic core. Thus the cutoff
frequency can be represented as (4) regardless of the number
of turns
(4)
and the amIf the bandwidth of the transistor amplifier is
plifier is modeled as a first-order system, then the injection current of the amplifier can be derived as
(5)
is the ac current gain of the transistor.
where
Because the input common-mode current is the sum of the
, then the
injection current and the load leakage current
can be found as
relation between and
(6)
If the bandwidth of the transistor amplifier and
then (6) can be simplified as

are high,

(7)
where
higher then

and
, then (7) is approximated as

. If

is much

(8)

SON AND SUL: NEW ACTIVE COMMON-MODE EMI FILTER

1311

(a)

(b)

(c)

(d)

Fig. 3. Configurations of proposed ACEF: (a) ACEF using dc-bus coupling, (b) ACEF using ac line coupling for single-phase application, (c) ACEF using dc-bus
coupling, and (d) ACEF using ac line coupling for three-phase application.

Fig. 4 shows the frequency response of the proposed ACEF

without including the effect of coupling capacitors. Parameters
used in Fig. 4 are listed in Table I. As can be seen from the
figure, the approximation (8) holds in the conducted EMI frequency band, 0.1530 MHz. But the attenuation performance
is degraded in the radiated EMI frequency band, 30300 MHz,
due to the limited bandwidth of the transistor amplifier. The
is achieved between
maximum attenuation of
and
. Simple passive filters can be inserted at the input or
output of the transistor amplifier to reduce the effect of the limited performance in the radiated EMI frequency band. Also as
concerned with the inflow of the low-frequency leakage current,
it cannot be attenuated because of the nature of coupling capacitors and the sensing circuit of (2).
The equivalent inductance of the common-mode choke due
to the ACEF can be calculated using the relation shown in (2)
and it is given as (9). Combined with (7) and (9), the input
common-mode current can be effectively suppressed without increasing the low frequency leakage current
(9)
III. EXPERIMENTS
In this section, several experimental results are shown. First,
Figs. 59 show the evaluation of the proposed ACEF using ac
line coupling for a single-phase application and Fig. 10 shows

Fig. 4. Frequency response of proposed ACEF.

the evaluation of the proposed ACEF for a three-phase application using ac line coupling.
Fig. 5 shows the configuration of the experimental system.
Detailed parameters of the experimental system are listed in
Table I. A 3.7 kW induction motor is used as a load machine of
the PWM inverter and a dc smoothing reactor is used in order
to reduce the input harmonic current. A printed circuit board

1312

IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 18, NO. 6, NOVEMBER 2003

TABLE I
SYSTEM PARAMETERS FOR FIG. 5

(a)

(b)
Fig. 6. No EMI filter installed: (a) Leakage currents (200 mA/div, 1 s/div)
and (b) conducted EMI spectrum.

Fig. 5.

Configuration of experimental system.

(PCB) for an input filter is installed separately from the PWM

inverter PCB. The input filter is composed of the ACEF and additional passive filtering elements. A single-phase LISN is used
to provide the stable source impedance at the high frequency as
shown in Fig. 1 and the peak detector is used in the measurement of the conducted EMI spectrum [3][5].
Waveforms of leakage currents and the conducted EMI spectrum of the system without any EMI filter are shown in Fig. 6.
Because there is no Y-capacitor, the motor leakage current generated by the switching of the PWM inverter is directly reflected
on the input common-mode current as shown in Fig. 6(a). The
conducted EMI spectrum includes both of the common- and
normal-mode EMI. Although they should be separately considered, the normal-mode EMI will not be discussed in this paper
with the assumption that some appropriate normal-mode filand
in Fig. 5, are installed
tering elements, such as
for each design stage. After sufficient normal-mode filtering,
the common-mode EMI plays as the bottleneck of the total conducted EMI [12], [13].
Fig. 7 shows leakage current waveforms and the conducted
EMI spectrum when the proposed ACEF is added in the system
of Fig. 6. Low-voltage complementary transistors (

(a)

(b)
Fig. 7. Proposed ACEF installed: (a) leakage currents (200 mA/div, 1 s/div)
and (b) conducted EMI spectrum.

SON AND SUL: NEW ACTIVE COMMON-MODE EMI FILTER

1313

(a)

(a)
(b)
Fig. 8. Additional passive EMI filter installed: (a) C
installed.

installed and (b) C

V) are used in this filter circuit and a 15 V dc power supply

is used as the bias voltage . This voltage source is available out of the control power supply of the inverter system of
and
in Fig. 5 are inFig. 5. In the experiment of Fig. 7,
stalled along with the ACEF. Using parameters listed in Table I,
and , can be calculated as 11.2 MHz
corner frequencies,
and
are assumed to be
and 14 kHz, respectively, when
and , the proposed
100 and 100, respectively. Between
ACEF attenuates the conducted EMI according to (7) as shown
in Fig. 4. Above the attenuation by the proposed ACEF is effective as shown in Fig. 7(b). Most of the high frequency leakage
from the motor is absorbed by the proposed ACEF
current
and only the low frequency ripple can be seen from the waveas shown in Fig. 7(a). Consequently the high freform of
quency input common-mode current is much reduced compared
with that of Fig. 6(a). Because its peak magnitude is considerably reduced and this helps preventing the magnetic core in
the additional filter from its saturation, which helps improving
the performance of the additional filter stage. Thus one can get
the enhanced attenuation performance of the additional filter, or
make the size of the additional magnetic core smaller.
In Fig. 8, an additional filter stage is added to the system of
is installed
Fig. 7. In Fig. 8(a) an additional Y-capacitor
at the input terminal along with an additional common-mode
. With the aid of the additional filter stage, the conchoke
ducted EMI is suppressed about 20 dB V until 2 MHz, but the
level of the EMI spectrum is increased especially above 2 MHz.
Although the entire level is placed under the given limit line, it
is inis not favorable to have the small margin. In Fig. 8(b),
. The entire EMI level is much attenuated
stalled instead of
and the conducted EMI above 5 MHz is not increased as in the
case of Fig. 8(a).
The proposed ACEF was adopted for the commercial air conditioner and Fig. 9 shows its evaluation result. The circuit of
Fig. 3(a) was used in this experiment. Fig. 9(a) shows the con-

(b)
Fig. 9. Application example of ACEF for commercial air conditioner:
(a) two-stage passive filters and (b) one passive filter with 1 ACEF.

ducted EMI spectrum of the system with two-stage cascaded

passive filters. An ACEF was implemented using the existing
common-mode choke of filter stages of Fig. 9(a) and the result
is shown in Fig. 9(b). As can be seen from Fig. 9, the EMI spectrum of 150 of 500 kHz is much attenuated so that it is possible
to meet the EN standard (EN 55014) for the quasipeak and average measurements marked as and , respectively, although the EMI spectrum was increased above 1 MHz.
Fig. 10 shows the evaluation result of the proposed ACEF for
a three-phase application. The circuit of Fig. 3(d) was used in
this experiment. Fig. 10(a) shows the configuration of the current experiment. Instead of using LISN, a power line interference probe (PLIP) was used for conducted EMI measurement
for Fig. 10(b) and (c). This voltage probe is used in place of
LISN when an EUT requires high current supply or when there
are difficulties in connecting a LISN into the circuit [5]. Main

1314

IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 18, NO. 6, NOVEMBER 2003

to various power electronics system regardless of the working

voltage. Also it helps improving the performance of additional
filter stages. Its effectiveness has been verified by experiments.
REFERENCES

(a)

(b)

(c)
Fig. 10. ACEF example for three-phase system: (a) configuration of
experimental system, (b) comparison with raw EUT, and (c) comparison with
the case of L
only.

power source is in corner-grounded delta-connection. Because

the PLIP attenuates the detected voltage signal by 30 dB V, it
was taken into account in the measurement. From the Fig. 10(b)
and (c), the proposed ACEF provides good attenuation performance until 10 MHz and the EMI spectrum above 10 MHz is
not much decreased due to the limited bandwidth of the amplifier circuit, which shows the validity of the proposed ACEF for
the three-phase application.
IV. CONCLUSION
In this paper, the concept of the new active common-mode
EMI filter (ACEF) has been proposed and its various implementation examples have been introduced. With the proposed ACEF
it is possible to use low-voltage transistors for the amplifier by
introducing coupling capacitors between the power line and the
amplifier circuit. Thus the proposed ACEF can be applicable

[1] G. Skibinski, R. Kerkman, and D. Schlegel, EMI emissions of modern

PWM ac drives, IEEE Ind. Applicat. Mag., pp. 4781, Nov./Dec. 1999.
[2] S. Chen, Generation and suppression of conducted EMI from
inverter-fed motor drives, in Proc. IEEE IAS99 Conf., 1999, pp.
15831589.
[3] L. Tihanyi, Electromagnetic Compatibility in Power Electronics. New
York: IEEE Press, 1995.
[4] T. Williams, EMC for Product Designers, 2nd ed. New York: Newnes,
1996.
[5] Limits and methods of measurement of radio disturbance characteristics of electrical motor-operated and thermal appliances for household
and similar purposes, electric tools and electric apparatus, Standard EN
55014, 1993.
[6] L. Rossetto, P. Tenti, and A. Zuccato, Electromagnetic compatibility
issues in industrial equipment, IEEE Ind. Applicat. Mag., pp. 3446,
Nov./Dec. 1999.
[7] M. J. Nave, Power Line Filter Design for Switched-Mode Power Supplies. New York: Van Nostrand Reinhold, 1991.
[8] L. Lawhite and M. F. Schlecht, Design of active ripple filters for power
circuits operating in the 110 MHz range, IEEE Trans. Power Electron.,
vol. 3, pp. 310317, July 1988.
[9] N. K. Poon, J. C. P. Liu, C. K. Tse, and M. H. Pong, Techniques for
input ripple current cancellation: Classification and implementation,
IEEE Trans. Power Electron., vol. 15, pp. 11441152, Nov. 2000.
[10] S. Ogasawara and H. Akagi, Circuit configurations and performance
of the active common-noise canceler for reduction of common-mode
voltage generated by voltage-source PWM inverter, in Proc. IEEE
IAS00 Conf., 2000, pp. 14821488.
[11] I. Takahashi, A. Ogata, and H. Kanazawa, Active EMI filter for
switching noise of high frequency inverters, in Proc. IEEE PCC-Nagaoka97 Conf., 1997, pp. 331334.
[12] Y.-C. Son and S.-K. Sul, Conducted EMI in PWM inverter for
household electric appliance, in Proc. IEEE IAS01 Conf., 2001, pp.
24412447.
[13] M. J. Nave, A novel differential mode rejection network for conducted
emissions diagnostics, in Proc. IEEE Nat. Symp. Electromagn.
Compat., 1989.

Yo-Chan Son (S99) was born in Korea, in 1974.

He received the B.S., M.S., and Ph.D. degrees from
the School of Electrical Engineering and Computer
Science, Seoul National University, Seoul, Korea, in
1997, 1999, and 2003, respectively.
He is now working for the Fairchild Semiconductor, Gyeonggi-do, Korea, where he is involved
with the development of the smart power module
for the motor control applications. His research
interests are power semiconductor module design,
high-performance ac drives, and mitigation of
electromagnetic interferences for drive systems.

Seung-Ki Sul (S78M87SM98F00) was born

in Korea in 1958. He received the B.S., M.S., and
Ph.D. degrees in electrical engineering from Seoul
National University, Seoul, Korea, in 1980, 1983, and
1986, respectively.
He was with the Department of Electrical and
Computer Engineering, University of Wisconsin,
Madison, as an Associate Researcher from 1986 to
1988. He then was with Gold-Star Industrial Systems
Company as a Principal Research Engineer from
1988 to 1990. Since 1991, he has been a member
of the faculty of the School of Electrical Engineering and Computer Science,
Seoul National University, where he is currently a Full Professor. His current
research interests are electric machines, electric vehicles, custom powers, and
power converter circuits.