© 2013 IEEE

IEEE Transactions on Power Electronics, Vol. 28, No. 7, pp. 3114-3125, July 2013

Optimal Design of LCL Harmonic Filters for Three-Phase PFC Rectifiers

J. Mühlethaler,
M. Schweizer,
R. Blattmann,
J. W. Kolar,
A. Ecklebe

This material is published in order to provide access to research results of the Power Electronic Systems Laboratory / D-ITET /
ETH Zurich. Internal or personal use of this material is permitted. However, permission to reprint/republish this material for
advertising or promotional purposes or for creating new collective works for resale or redistribution must be obtained from
the copyright holder. By choosing to view this document, you agree to all provisions of the copyright laws protecting it.
3114 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 28, NO. 7, JULY 2013

Optimal Design of LCL Harmonic Filters

for Three-Phase PFC Rectifiers
Jonas Mühlethaler, Student Member, IEEE, Mario Schweizer, Student Member, IEEE, Robert Blattmann,
Johann W. Kolar, Fellow, IEEE, and Andreas Ecklebe, Member, IEEE

Abstract—Inductive components such as line filter inductors or optimization of inductive components employed in key power
transformers occupy a significant amount of space in today’s power electronics applications.
electronic systems, and furthermore, considerable losses occur in Numerous papers have been written about how to find opti-
these components. A main application of inductive components is
EMI filters, as, e.g., employed for the attenuation of switching fre- mal designs of power electronic systems, whereas the goal of
quency harmonics of power factor correction (PFC) rectifier sys- the according optimization task can be minimum costs, volume,
tems. In this paper, a design procedure for the mains side LCL filter losses, etc. Some papers investigated parts of a system, some
of an active three-phase rectifier is introduced. The procedure is others tried to model and optimize the overall system. For in-
based on a generic optimization approach, which guarantees a low stance, in [2], an overall optimization of a two-FET forward
volume and/or low losses. Different designs are calculated to show
the tradeoff between filter volume and filter losses. The design pro- converter is presented. For it, a two-stage optimization method
cedure is verified by experimental measurements. Furthermore, an to optimize the efficiency from light load to full load is proposed.
overall system optimization, i.e., an optimization of the complete Another approach to conduct an overall optimization is given
three-phase PFC rectifier, is given. in [3], where a flyback converter has been optimized. In [4], an
Index Terms—LCL filter, magnetic components, optimization. overall optimization of a boost power factor correction (PFC)
converter including the input electromagnetic interference filter
is conducted. Examples of papers in which the focus is put only
I. INTRODUCTION on parts of a system are, e.g., given in [5] and [6]. In [5], it
HE trend in power electronics research is toward is shown on the example of a buck converter, how the capaci-
T higher efficiency and higher power density of converter
systems. This trend is driven by cost considerations (e.g., mate-
tors of a dc-link capacitor tank are selected optimally. In [6], a
design methodology that finds the volumetric optimized induc-
rial economies), space limitations, (e.g., for automotive applica- tors employed in single-phase PFC boost converters is given,
tions), and increasing efficiency requirements (e.g., for telecom whereas the corresponding optimum switching frequency and
applications). The increase of the power density often affects input current ripple is found.
the efficiency, i.e., a tradeoff between these two quality indices The aim of this paper is to use the recently established com-
exists [1]. prehensive models of inductive power components to optimally
Inductive components occupy a significant amount of space in design LCL input filters for a three-phase PFC rectifier. The fo-
today’s power electronic systems, and furthermore, considerable cus of the paper at hand is put on the optimal design of the LCL
losses occur in these components. In order to increase the power harmonic filters. LCL input filters are an attractive solution to
density and/or efficiency of power electronic systems, losses attenuate switching frequency current harmonics of active volt-
in inductive components must be reduced, and/or new cooling age source rectifiers [7], [8]. In this paper, a design procedure
concepts need to be investigated. for LCL filters based on a generic optimization approach is in-
At the Power Electronic Systems Laboratory at ETH Zurich, troduced guaranteeing low volume and/or low losses. Different
a project has been initiated with the goal to establish compre- designs are calculated showing the tradeoff between filter vol-
hensive models of inductive power components which can be ume and filter losses. Furthermore, it is shown at the end of this
adapted to various geometric properties, operating conditions, paper how an overall system optimization, i.e., an optimization
and cooling conditions. These models will form the basis for the of the complete three-phase PFC rectifier (not only the filter),
could be conducted. It is important to consider the complete
Manuscript received April 17, 2012; revised June 11, 2012; accepted July 17, system, since there are parameters that bring advantages for one
2012. Date of current version December 24, 2012. Recommended for publica- subsystem but bring disadvantages for another. For instance, in
tion by Associate Editor J. Antenor Pomilio. the selected example of an LCL input filter, a higher switching
J. Mühlethaler, M. Schweizer, and J. W. Kolar are with the Power Elec-
tronic Systems Laboratory, ETH Zurich, 8006 Zurich, Switzerland (e-mail: frequency leads to lower volume, or, when keeping the volume
[email protected]; [email protected]; [email protected]. constant, to lower losses of the inductive components. However,
ch). higher semiconductor switching losses are expected in the case
R. Blattmann is with the ABB Switzerland, Ltd., CH-5400 Baden,
Switzerland (e-mail: [email protected]). of higher switching frequencies.
A. Ecklebe is with the Corporate Research Center, ABB Switzerland, Ltd., In Section II, the three-phase PFC rectifier is introduced; in
CH-5405 Baden-Dättwil, Switzerland (e-mail: [email protected]). Section III, the applied models of the LCL filter components
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org. are discussed; and in Section IV, the optimization algorithm for
Digital Object Identifier 10.1109/TPEL.2012.2225641 the LCL filter is described. Simulation and experimental results

0885-8993/$31.00 © 2012 IEEE

MÜHLETHALER et al.: OPTIMAL DESIGN OF LCL HARMONIC FILTERS FOR THREE-PHASE PFC RECTIFIERS 3115

Fig. 1. (a) Three-phase PFC rectifier with LCL input filter. (b) Cross sections of inductors employed in the input filter.

TABLE I is taken for the conductors. The windings are divided into two
SPECIFICATION OF THE THREE-PHASE PFC RECITIFIER
halves arranged on the two legs which lead to a more distributed
winding structure. A more distributed winding structure has
advantages such as better heat dissipation capabilities, lower
inductor volume, etc.

III. MODELING OF INPUT FILTER COMPONENTS

The loss modeling is based on the framework introduced
are given in Sections V and VI, respectively. In Section VII, the in [10], where a high level of accuracy has been achieved by
converter volume and losses are taken into consideration, i.e., combining the best state-of-the-art approaches and by embed-
an overall system optimization is performed. ding newly developed approaches into a novel loss calculation
framework. In [10], the impact of the peak-to-peak flux density
ΔB, frequency f , dc premagnetization HDC , temperature T ,
II. THREE-PHASE PFC RECTIFIER WITH INPUT FILTER
core shape, minor and major BH-loops, flux waveform, and
The three-phase PFC rectifier investigated in this paper is material on the core loss calculation has been considered. In
shown in Fig. 1. The rectifier comprises in each phase a boost order to calculate the winding losses, formulas for round con-
inductor L2 , a damped LC filter L1 /C/Cd /Rd , and a pair of ductors and litz wires, each considering skin and proximity
switches with free-wheeling diodes. The load is assumed to effects and also considering the influence of an air gap fringing
be a dc current source. The considered operating point of the field have been included. In the following, a discussion about the
PFC rectifier is described in Table I. A space vector modula- implemented models for the employed inductors [cf., Fig. 1(b)]
tion (SVM) scheme with loss-optimal clamping has been imple- is given.
mented to have a fundamental displacement factor of cos φ = 1.
In the applied SVM scheme with loss-optimal clamping, a A. Calculation of the Inductance
switching of the phase with the highest current is omitted; hence,
the switching losses are reduced accordingly. The functionality The inductance of an inductive component with N winding
and the detailed control of the used SVM scheme are described turns and a total magnetic reluctance Rm ,tot is calculated as
in [9]. The three-phase PFC rectifier with input filter has been N2
simulated in MATLAB/Simulink, where the nonlinearity of the L= . (1)
Rm ,tot
core material of the inductors, i.e., the change of the inductance
value with changing current, is taken into account. Accordingly, the reluctance of each section of the flux path
The three inductors L1,a , L1,b , and L1,c , and the three capac- has to be derived first in order to calculate Rm ,tot . The total
itors Ca , Cb , and Cc in star arrangement, together with the three reluctance for a general inductor is calculated as a function
boost inductors L2,a , L2,b , and L2,c , result in a third-order LCL of the core reluctances and air gap reluctances. The core and
low-pass filter between the mains and the switching stage. The air gap reluctances can be determined applying the methods
capacitor/resistor branches Cd,a /Rd,a , Cd,b /Rd,b , and Cd,c /Rd,c described in [11]. The reluctances of the core depend on the
are necessary to damp the resonance of the LC input filter. All relative permeability μr which is defined by the nonlinear BH-
inductors are assumed to have the same geometry, which is il- relation of the core material; hence, the reluctance is described as
lustrated in Fig. 1(b). The cores are made of grain-oriented steel a function of the flux. Therefore, as the flux depends on the core
(M165-35S, lamination thickness 0.35 mm). Solid copper wire reluctance and the reluctance depends on the flux, the system
3116 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 28, NO. 7, JULY 2013

Fig. 3. Cross section of the round conductor considered with a current in

Fig. 2. Reluctance model of the inductor shown in Fig. 1(b). It consists of one z-direction. The conductor is infinitely large in z-direction.
voltage source (representing the two separated windings), one air gap reluctance
Rg (representing the two air gaps), and one core reluctance Rc .
Eddy currents due to an external alternating magnetic field,
e.g., the air gap fringing field or the magnetic field from other
can only be solved iteratively by using a numerical method. In
conductors, lead to the proximity effect.
the case at hand, the problem has been solved by applying the
The sum of the dc losses and the skin effect losses per unit
Newton approach.
length in round conductors [cf., Fig. 3(a)] can be calculated
The reluctance model of the inductor shown in Fig. 1(b) is
as [14]
illustrated in Fig. 2. It consists of one voltage source (repre-
senting the two separated windings), one air gap reluctance Rg PS = RDC · FR (f ) · Iˆ2 (2)
(representing the two air gaps), and one core reluctance Rc . The
and1
fact that the flux density in core parts very close to the air gap is 
(slightly) reduced as the flux lines already left the core has been ξ ber0 (ξ)bei1 (ξ) − ber0 (ξ)ber1 (ξ)
FR (f ) = √
neglected. 4 2 ber1 (ξ)2 + bei1 (ξ)2

bei0 (ξ)ber1 (ξ) + bei0 (ξ)bei1 (ξ)
B. Core Losses − (3)
ber1 (ξ)2 + bei1 (ξ)2
The applied core loss approach is described in [10] in detail
with δ = √π μ1 σ f ,ξ = √d2δ ,RDC = σ π4d 2 ; δ is commonly
and can be seen as a hybrid of the improved–improved General- 0

ized Steinmetz Equation i2 GSE [12] and a loss map approach: named the skin depth, f is the frequency, d is the conductor
a loss map is experimentally determined and the interpolation diameter, and σ is the electric conductivity of the conductor
and extrapolation for operating points in between the measured material.
values is then made with the i2 GSE. The proximity effect losses in round conductors [cf., Fig. 3(b)]
The flux density waveform for which the losses have to be per unit length can be calculated as [14]
calculated is, e.g., simulated in a circuit simulator, where the PP = RDC · GR (f ) · Ĥe2 (4)
magnetic part is modeled as a reluctance model. This simulated
waveform is then divided into its fundamental flux waveform and

and into piecewise linear flux waveform segments. The loss ξπ 2 d2 ber2 (ξ)ber1 (ξ) + ber2 (ξ)bei1 (ξ)
energy is then derived for the fundamental and all piecewise GR (f ) = − √
2 2 ber0 (ξ)2 + bei0 (ξ)2
linear segments, summed up and divided by the fundamental 
period length in order to determine the average core loss. The dc bei2 (ξ)bei1 (ξ) − bei2 (ξ)ber1 (ξ)
+ . (5)
flux level of each piecewise linear flux segment is considered, as ber0 (ξ)2 + bei0 (ξ)2
this influences the core losses [13]. Furthermore, the relaxation RDC is the resistance per unit length; hence, the losses PS and
term of the i2 GSE is evaluated for each transition from one PP are losses per unit length as well. The external magnetic
piecewise linear flux segment to another. field strength He of every conductor has to be known when
Another aspect to be considered in the core loss calculation calculating the proximity losses. In the case of an ungapped
is the effect of the core shape/size. By introducing a reluctance core and windings that are fully enclosed by core material,
model of the core, the flux density can be calculated. Subse- 1-D approximations to derive the magnetic field exist. The most
quently, for each core section with (approximately) homoge- popular method is the Dowell method [16]. However, in the case
nous flux density, the losses can be determined. In the case at of gapped cores, such 1-D approximations are not applicable as
hand, the core has been divided into four straight core sections the fringing field of the air gap cannot be described in a 1-D
and four corner sections. The core losses of the sections are then manner. For the employed gapped cores, another approach has
summed up to obtain the total core losses. to be selected. The applied approach is a 2-D approach and is

C. Winding Losses 1 The solution for F

R (and G R ) is based on a Bessel differential equation
The second source of losses in inductive components is the that has the form x2 y  + xy  + (k 2 x2 − v 2 )y = 0. With the general solution
y = C 1 Jv (kx) + C 2 Yv (kx), whereas Jv (kx) is known as Bessel function
ohmic losses in the windings. The resistance of a conductor of the first kind and order v and Yv (kx) is known as Bessel function of the
increases with increasing frequency due to eddy currents. Self- second kind and order v [15]. Furthermore, to resolve Jv (kx) into its real and
3
induced eddy currents inside a conductor lead to the skin effect. imaginary part, the Kelvin functions can be used: Jv (j 2 x) = berv x + j beiv x.
MÜHLETHALER et al.: OPTIMAL DESIGN OF LCL HARMONIC FILTERS FOR THREE-PHASE PFC RECTIFIERS 3117

temperature. The ambient temperature TA is assumed to be

constant at 25 ◦ C.
The heat transfer due to convection is described with

P = αA(TL − TA ) (7)

where P is the heat flow, A the surface area, TL the body surface
temperature (i.e., inductor temperature), and TA the fluid (i.e.,
ambient air) temperature. α is a coefficient that is determined
by a set of characteristic dimensionless numbers, the Nusselt,
Grashof, Prandtl, and Rayleigh numbers. Radiation has to be
considered as a second important heat transfer mechanism and is
described by the Stefan–Boltzmann law. Details about thermal
modeling are not given here; the interested reader is referred
to [18], from where the formulas used in this paper have been
taken.

Fig. 4. (a) Illustration of the method of images (mirroring). (b) Illustration of E. Capacitor Modeling
modeling an air gap as a fictitious conductor.
The filter and damping capacitors are selected from the EP-
COS X2 MKP film capacitors series, which have a rated voltage
of 305 V. The dissipation factor is specified as tan δ ≤ 1 W/kvar
Fig. 5. Thermal model with only one thermal resistance. (at 1 kHz) [19], which enables the capacitor loss calculation. The
capacitance density to calculate the capacitors volume can be
approximated with 0.18 μF/cm3 . The capacitance density has
described in detail in [10], where it has been implemented based
been approximated by dividing the capacitance value of several
on a previously presented work [17].
components by the according component volume.
The magnetic field at any position can be derived as the
superposition of the fields of each of the conductors. The impact
of a magnetic conducting material can further be modeled with F. Damping Branch
the method of images, where additional currents that are the An LC filter is added between the boost inductor and the
mirrored version of the original currents are added to replace mains to meet a total harmonic distortion (THD) constraint.
the magnetic material [17]. In case of windings that are fully The additional LC filter changes the dynamics of the converter
enclosed by magnetic material (i.e., in the core window), a new and may even increase the current ripple at the filter resonant
wall is created at each mirroring step as the walls have to be frequency. Therefore, a Cd /Rd damping branch has been added
mirrored as well. The mirroring can be continued to pushing for damping. In [20] and [21], it is described how to optimally
the walls away. This is illustrated in Fig. 4(a). For this study, choose Cd and Rd . Basically, there is a tradeoff between the
the mirroring has been stopped after the material was replaced size of damping capacitor Cd and the damping achieved. For
by conductors three times in each direction. The presence of this study, Cd = C has been selected as it showed to be a good
an air gap can be modeled as a fictitious conductor without compromise between additional volume needed and a reason-
eddy currents equal to the magneto-motive force across the air able damping achieved. Hence, the volume of the damping ca-
gap [17] as illustrated in Fig. 4(b). pacitor is the same as the volume of the filter capacitor. The
According to the previous discussion, the winding losses have value of the damping resistance that leads to optimal damping
to be calculated differently for the sections A and B illustrated is then [20], [21]
in Fig. 1(b), as the mirroring leads to different magnetic fields. 
L1
Rd = 2.1 . (8)
D. Thermal Modeling C
A thermal model is important when minimizing the filter vol- The Cd /Rd damping branch increases the reactive power con-
ume, as the maximum temperature allowed is the limiting factor sumption of the PFC rectifier system. Therefore, often other
when reducing volume. The model used in this paper consists damping structures, such as the Rf -Lb series damping structure,
of only one thermal resistance Rth and is illustrated in Fig. 5. are selected [21]. For this study, however, the Cd /Rd damping
The inductor temperature TL is assumed to be homogenous; it branch has been favored as its practical realization is easier
can be calculated as and lower losses are expected. Furthermore, as will be seen in
TL = TA + Ploss Rth (6) Section V, the reactive power consumption of the PFC rectifier
system including the damped LC input filter is in the case at hand
where Ploss are the total losses occurring in the inductor, con- rather small, and, if necessary, it could be actively compensated
sisting of core and winding losses, and TA is the ambient by the rectifier.
3118 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 28, NO. 7, JULY 2013

The losses in the damping branch, which occur mainly in

the resistors, are calculated and taken into consideration in the
optimization procedure as well.

IV. OPTIMIZATION OF THE INPUT FILTER

The aim is to optimally design a harmonic filter of the intro-
duced three-phase PFC rectifier. For the evaluation of different
filter structures, a cost function is defined that weights the filter
losses and filter volume according to the designer needs. In the
following, the steps toward an optimal design are described. All
steps are illustrated in Fig. 6. The optimization constraints are
discussed first.

A. Optimization Constraints and Conditions

The high-frequency ripple in the current i2,a/b/c is limited
to the value IHF,pp,m ax , which is important as a too high
IHF,pp,m ax , e.g., impairs controllability (for instance, an accu-
rate current measurement becomes more difficult). Furthermore,
the THD of the mains current is limited. In industry, a typical
value for the THD that is required at the rated operating point is
5 % [22]. In the design at hand, a THD limit of 4 % is selected in
order to have some safety margin. Two other design constraints
are the maximum temperature Tm ax and the maximum volume
Vm ax the filter is allowed to have. A fixed switching frequency
fsw is assumed. The dc-link voltage VDC and the load current
IL of the converter are also assumed to be given and constant.
All constraints/condition values for the current system are given
in Fig. 6.

B. Calculation of L2,m in
The minimum value of the inductance L2,m in can be calcu-
lated based on the constraint IHF,pp,m ax as
√
3 VDC −
2
2Vm ains
L2,m in = δ(100) · (9)
IHF,pp,m ax · fsw
√
with δ(100) = 3M 2 cos(π/6) the relative turn-on time of the
space vector (100) when the current of phase a peaks. Equa-
tion (9) is based on the fact that, in case of a fundamental
displacement factor of cos φ = 1, the maximum current rip-
ple IHF,pp,m ax occurs when the current reaches the peak value
IˆLF of the fundamental (at this instant, the voltage 23 VDC −
√ √
2Vm ains is across the inductor). With M = 2 2V m a in s
VD C , (9) Fig. 6. Design procedure for three-phase LCL filters based on a generic opti-
becomes mization approach.
√ √ √
3 2|Vm ains | 3 VDC −
2
2Vm ains
L2,m in = cos(π/6) · .
VDC IHF,pp,m ax · fsw
(10)

C. Loss Calculation of Filter Components

In the foregoing sections, it has been shown how an accurate
loss modeling based on simulated current and voltage wave-
forms is possible. However, such a calculation based on simu-
lated waveforms is time consuming and therefore, for an efficient
optimization, simplifications have to be made. In Fig. 7, ideal-
ized current waveforms for each filter component of a phase are Fig. 7. Idealized current waveforms for each filter component.
MÜHLETHALER et al.: OPTIMAL DESIGN OF LCL HARMONIC FILTERS FOR THREE-PHASE PFC RECTIFIERS 3119

illustrated. The current in L1 is approximated as purely sinu- qLoss · P and qVolum e · V are in the same range.2 In the case
soidal with a peak value of at hand, this was achieved with qLoss = 1/W and qVolum e =
3 · 104 /m3 (these values differ a lot from case to case and,
2 IL VDC
Iˆ = √ (11) accordingly, are valid for the case at hand only).
3 2Vm ains In order to calculate the volumes VL 1 and VL 2 of the inductors,
their boxed volumes are calculated based on the parameters in
where Vm ains is the RMS value of the mains-phase voltage. A X.
possible reactive current is rather small and has been neglected. The following steps are conducted to calculate the filter losses
With the mains frequency fm ains = 50 Hz, losses, volume, and P and the filter volume V of (13) (cf., Fig. 6).
temperature of L1 can be calculated. 1) Calculate the inductance value L2 as a function of the
The current in L2 has a fundamental (sinusoidal) component, current. Furthermore, the losses PL 2 , the volume VL 2 ,
with an amplitude as calculated in (11) and a fundamental fre- and temperature TL 2 of the boost inductors L2 are calcu-
quency of fm ains = 50 Hz, and a superimposed ripple current. lated. In case a constraint cannot be met, the calculation is
The ripple current is, for the purpose of simplification, in a aborted and the design is discarded and a new design will
first step considered to be sinusoidal with constant amplitude be evaluated.
IHF,pp,m ax over the mains period. The losses for the fundamen- 2) Calculate the inductance value L1 as a function of the
tal and the high-frequency ripple are calculated independently, current. Furthermore, the losses PL 1 , the volume VL 1 , and
and then summed up. By doing this, it is neglected that core temperature TL 1 of the filter inductors L1 are calculated. In
losses depend on the dc bias condition. case a constraint cannot be met, the calculation is aborted
The ripple current is assumed to be fully absorbed by the filter and the design is discarded and a new design will be
capacitor C, hence, with the given dissipation factor tan δ, the evaluated.
losses in the capacitor can be calculated as well. 3) For the purpose of simplification, the THD without filter
How the aforementioned simplifications affect the modeling is approximated with THD = IHF,pp,m ax /I(1),pp , where
accuracy will be discussed in Section V. I(1),pp is the fundamental peak-to-peak value of the mains
current.3 Furthermore, it is assumed that the dominant har-
D. Optimization Procedure monic content appears at fsw (this assumption is motivated
by simulation results). The LC filter has then to attenuate
After the optimization constraints and the simplifications cho-
the ripple current by
sen for the loss calculation have been described, the optimization
 
procedure itself will be explained next. A = 20 log10 I(1),pp THDm ax /IHF,pp,m ax (14)
A filter design is defined by
(in dB) at a frequency of fsw . Therewith, C is calculated
⎛ ⎞
aL 1 aL 2 as
⎜ L1
w wL 2 ⎟
⎜ ⎟ 1 1
⎜ NL 1 NL 2 ⎟ C= = (15)
⎜ ⎟ L1 ω02 A

⎜ do doL 2 ⎟ L1 (2πfsw · 10 4 0 d B )2
X = ⎜ L1 ⎟ (12)
⎜ hL 1 hL 2 ⎟ where ω0 is the filter cutoff frequency. The losses and
⎜ ⎟
⎜ tL 1 tL 2 ⎟ the volume of the capacitors and damping resistors can
⎝ ⎠
wwL 1 wwL 2 then be calculated. In case a constraint cannot be met, the
dL 1 dL 2 calculation is aborted and the design is discarded and a
new design will be evaluated.
where all inductor parameters are defined as in Fig. 1(b). The 4) The volume, temperature, and losses are now known and
subscripts L1 and L2 describe to which inductor each param- the cost function (13) can be evaluated.
eter corresponds. The capacitance value of the filter capacitor The optimal matrix X is found by varying the matrix param-
is calculated based on the L1 value to guarantee that the THD eters, evaluating these matrixes by repeating previous steps, and
constraint is met. The parameters in X are varied by an optimiza- minimizing the cost function (13). After the optimal design is
tion algorithm to obtain the optimal design. The optimization is found, the algorithm quits the loop of Fig. 6.
based on the MATLAB function fminsearch() that applies the
downhill simplex approach of Nelder and Mead [23].
The optimization algorithm determines the optimal parameter 2 The filter losses are in the range of approximately 100 W . . . 250 W, whereas
values in X. A design is optimal when the cost function the filter volumes are in the range of approximately 0.001 m3 . . . 0.01 m3 .
With qL o ss = 1/W and qVo lu m e = 3 · 10 4 /m3 , the volume range is lifted to
F = kLoss · qLoss · P + kVolum e · qVolum e · V (13)
30 . . . 300 and therewith becomes comparable to the losses.
I 22 + I 32 + I 42 + . . . + I n2
3 The THD is defined as THD = , where In is the RMS
I1
is minimized. kLoss and kVolum e are weighting factors, qLoss value of the nth harmonic and I1 is the RMS value of the fundamental current.
and qVolum e are proportionality factors, and P and V are the Under the assumption that only the dominant harmonic content IH F , p p , m a x
(which, for the purpose of simplification, is assumed to appear at one fre-
filter losses and filter volume, respectively. The proportionality quency) leads to a harmonic distortion, the THD becomes IH F , p p , m a x /I(1 ), p p ,
factors are chosen such that, for a “comparable performance,” consequently.
3120 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 28, NO. 7, JULY 2013

Fig. 8. P –V Pareto front showing filter volumes V and filter losses P of

different optimal designs.

V. OPTIMIZATION OUTCOMES
The optimization procedure leads to different filter designs
depending on the chosen weighting factors k1 and k2 in (13),
i.e., depending whether the aim of the optimization is more
on reducing the volume V or more on reducing the losses P .
Limiting factors are the maximum temperature Tm ax (limits
the volume from being too low) and a maximum volume Vm ax
(limits the efficiency from being too high). Different designs are
shown by a P –V plot, i.e., a P –V Pareto front in Fig. 8; the
tradeoff between losses and volume can be clearly identified.
One design of Fig. 8 has been selected for further investi- Fig. 9. Simulations and calculations to one selected design (cf., Fig. 8).
gations. Particularly, a comparison between the (for the opti- (a) Filter inductor L 1 , a . (b) Boost inductor L 2 , a .
mization procedure) simplified and the more elaborate calcula-
tion based on voltage/current waveforms from a circuit sim-
ulator has been made. The filter parameters of the selected VI. EXPERIMENTAL RESULTS
design are detailed in Fig. 8. The circuit of the three-phase Experimental measurements have been conducted to show
PFC rectifier with the selected input filter has been simulated that the previous introduced calculations are valid. The filter
in MATLAB/Simulink, and the simulated current and voltage prototype built in the course of this study has been assembled
waveforms have been taken to calculate the losses according of laminated sheets and coil formers of standard sizes, in order
to Section III. The results are given in Fig. 9. The THD con- to keep the cost low. This avoids an exact implementation of an
straint is met and the current IHF,pp,m ax is only insignificantly optimum as can be seen in Fig. 8 (the prototype built is not on the
higher. The simplified loss calculation used for the optimization 8 kHz-line). However, this does not impair the significance of the
leads to an overestimation of the boost inductor losses. This is measurement results. Specifications, dimensions, and photos of
due to the fact that the maximum ripple current has been as- the LCL filter built are given in Fig. 10. All measurements have
sumed to be constant over the mains period. The losses in the been carried out with a Yokogawa WT3000 Precision Power
filter inductors, on the other hand, have been underestimated as Analyzer.
any high-frequency ripple in the current through L1 has been The measurements have been conducted with the T-type con-
neglected in the simplified calculations. One could try to im- verter introduced in [24]. In Fig. 11, a photo of the converter is
prove/change the simplifications made for the optimization and given. The T-type converter is a three-level converter; however,
therewith improve the simplified loss calculation. However, the a two-level operation is possible as well. A two-level opera-
difference between the two calculation approaches has been tion with the same modulation scheme (optimal-loss clamping
considered as acceptable for this study. modulation scheme) as in the MATLAB simulation has been
Another important design criteria is the achieved power fac- implemented.
tor. The reactive power consumption of the PFC rectifier system, The results of the comparative measurements and simulations
including the damped LC input filter, is in the case at hand rather are given in Figs. 12 and 13. As can be seen, the calculated and
small (power factor = [real power]/[apparent power] = 0.998). measured loss values are very close to each other. The maxi-
So far, all results are based on simulations and calculations. mum current ripple in the actual system is (slightly) higher than
The models have to be verified experimentally to prove the in the simulation. This can be explained by the fact that in the
validity of the optimization procedure. In the following section, simulation the inductance of the boost inductor is assumed to be
experimental results are shown. constant over the full frequency range. However, in reality, the
MÜHLETHALER et al.: OPTIMAL DESIGN OF LCL HARMONIC FILTERS FOR THREE-PHASE PFC RECTIFIERS 3121

Fig. 12. (a) Simulations and (b) measurements on one of the filter inductors
L 1 of the implemented design.

VII. OVERALL RECTIFIER OPTIMIZATION

As can be seen in Fig. 8, an increase in switching frequency
leads to lower filter losses and lower filter volumes. However, in
return, an increase in switching frequency leads to higher switch-
Fig. 10. Specifications, dimensions and photos of (a) one filter inductor L 1 , ing losses in the converter semiconductors. In other words, it is
(b) all filter and damping capacitors/resistors C , C d , and R d , and (c) one boost important to consider the system as a whole in order to achieve
inductor L 2 .
a truly optimal design. In the following, first, a model for the
converter is derived, which allows to determine a P –V Pareto
front of the converter. Later, the tradeoff in the switching fre-
quency is illustrated and the optimal frequency for the overall
system is determined.

A. Overall Optimized Designs

The Infineon Trench and Field Stop 1200 V IGBT4 series
has been selected to determine the P –V pareto front of the con-
verter. These semiconductors are the same as the ones employed
in the converter of Fig. 11. A converter model has been set up,
which determines the junction temperature of the transistors
Fig. 11. Photo of the PFC converter. Tj,T , the junction temperature of the diodes Tj,D , the volume
of the converter V , and the losses of the converter P . As input
effective inductance decreases with increasing frequency due variables, it needs the transistor chip area AT , the diode chip
to inductor losses and parasitic capacitances. The higher THD area AD , the cooling system volume VCS , and the operating
value can also be explained with the same effect since the filter point defined by the switching frequency fsw , and peak-to-peak
inductance decreases with increasing frequency as well. The fre- fundamental current I(1),pp . The high-frequency ripple current
quency behavior could be modeled analytically by representing IHF,pp has, as long as IHF,pp  I(1),pp , only a negligible im-
the inductors as RLC networks. pact on the losses and is therewith not taken into consideration.
3122 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 28, NO. 7, JULY 2013

where CSPI = 15 W/(K · L) can be approximated to be constant

for a given cooling concept. The value 15 W/(K · L) has been
calculated for the cooling concept of the converter shown in
Fig. 11 according to [27]. According to [25], for the selected
insulated-gate bipolar transistor (IGBT) series, the junction to
sink surface thermal resistance Rth,js can be approximated as a
function of the chip area as
K
Rth,js = 23.94 · A−0.88 . (17)
W · mm2
The switching and conduction losses per chip area have been
extracted from data sheets. Different IGBTs with different cur-
rent ratings, but from the same IGBT series have been analyzed.
To determine the losses as a function of the chip area, the IGBT
and diode chip area as a function of the nominal chip current IN
has to be known. This chip area-current dependency has been
taken from [25] and is for the transistor
mm2
AT = 0.95 · IN + 3.2 mm2 (18)
A
and for the diode
mm2
AD = 0.47 · IN + 3.6 mm2 . (19)
A
The switching losses scaled to the same current do not vary much
with the chip size in the considered range, as a comparison of
different data sheets of the selected IGBT series has shown;
therefore, for this study, the switching loss energies have been
considered as independent of the chip area. The switching loss
Fig. 13. (a) Simulations and (b) measurements on one of the boost inductors
L 2 of the implemented design. energies have been extracted at two junction temperatures (25 ◦ C
and 150 ◦ C) and, for intermediate temperatures, a linear interpo-
lation has been made. These, from the data sheet extracted and
interpolated, switching loss energies are then assigned to each
switching instant in order to determine the switching losses.
In return, the conduction losses depend substantially on the
Fig. 14. Illustration of a converter model. chip size area. This has been modeled as in [26], but with taking
the impact of the temperature into consideration. The following
TABLE II equation describes the conduction losses
CONSTRAINTS AND CONDITIONS FOR CONVERTER OPTIMIZATION
Ron,N (Tj,T /D ) · AN 2
Pcond (A, i, Tj,T /D ) = Vf · i + ·i
A
(20)
where Vf is the forward voltage drop, Ron,N (Tj,T /D ) is the
nominal on-resistance, AN is the nominal chip size area (to
which Ron,N (Tj,T /D ) is taken from the data sheet), i is the cur-
rent through the transistor, and A is the actual chip size area. The
on-resistance has been extracted at two junction temperatures
(25 ◦ C and 150 ◦ C) and, for intermediate on-resistance values,
The inputs and outputs of the converter model are illustrated in a linear interpolation has been made. The current i has been
Fig. 14 and all optimization constraints and conditions are listed calculated for the chosen modulation scheme. Equation (20) is
in Table II. This optimization based on the semiconductor chip evaluated with different values for Vf and Ron,N , depending
area is motivated by previously presented works [25], [26]. whether the conduction losses of the transistor or the diode are
The thermal resistance of the cooling system, i.e., the heat sink calculated.
surface to ambient thermal resistance Rth,sa , has been modeled The converter volume is the sum of the cooling system vol-
with the cooling system performance index (CSPI) [27]; the ume, the dc-link capacitor volume, and the volume of the switch-
Rth,sa is then calculated as ing devices. The switching device volume has been calculated
by multiplying the chip areas by a depth of 1 cm in order to
1 approximate the transistor volume. The cooling system volume
Rth,sa = (16)
CSPI · VCS is known as it is a model input parameter. The dc-link capacitor
MÜHLETHALER et al.: OPTIMAL DESIGN OF LCL HARMONIC FILTERS FOR THREE-PHASE PFC RECTIFIERS 3123

Fig. 15. P –V Pareto front showing converter volumes V and converter losses
P of different optimal designs.

CDC has been selected such that the dc-link voltage in case of
an abrupt load drop from nominal load to zero load does not ex-
ceed a predefined value. The maximum dc-link voltage increase
ΔVDC after the load drop can then be approximated as
√  
1 3 2Vm ains Iˆ2 1ˆ
ΔVDC = · + I2 tm ax
CDC 2 VDC fsw 2

where Iˆ2 is the peak value of the phase current through the Fig. 16. (a) Losses of loss optimized designs for different frequencies.
boost inductor L2 and tm ax is the time difference between the (b) Losses of volumetric optimized designs for different frequencies.
moment where the system sampled the load drop (i.e., latest
1/fsw after the load drop occurred, since the sampling interval
is 1/fsw ) and the moment where the dc-link voltage peaks (i.e.,
VDC + ΔVDC is reached). The chosen value for the maximum
voltage overshoot is ΔVDC = 50 V. The dc-link capacitors have
been selected from the EPCOS MKP dc-link film capacitors
series, which have a rated voltage of 800 V. The capacitor losses
are low and, therewith, have been neglected. The capacitance
density to calculate the capacitors volume can be approximated
with 0.6 μF/cm3 .
The losses and volumes of other system parts, such as the DSP,
auxiliary supply, gate driver, etc., have not been considered.
An optimization algorithm has been set up similar to that of Fig. 17. Volume of volumetric optimized designs for different frequencies.
the filter. A design is characterized by the parameters AT , AD ,
and VCS . The optimization algorithm varies these parameter
in order to find optimal designs. The optimization procedure In case of loss-optimized designs, the major volume part
leads to different designs depending whether the aim of the comes from the filter volume. This becomes clear when Fig. 8
optimization is more on reducing the volume V or more on is compared with Fig. 15: the maximum filter volume is much
reducing the losses P . Different designs are shown by a P –V higher than the maximum converter volume. This has to do
plot, i.e., a P –V Pareto front in Fig. 15; the tradeoff between with the selected constraints; however, a further increase of
losses and volume can be clearly identified. It becomes clear that the converter volume would not have a big impact in further
a higher switching frequency leads to higher losses; therefore, decreasing the losses; therefore, it can be concluded that the
there must exist an overall optimal switching frequency at which constraints have been selected well.
the system losses are minimized. In Fig. 17, the volume of volumetric optimized designs for
Basically, one can combine the results from the filter and different frequencies is plotted. As both, the filter size and the
from the converter and, therewith, determine the overall system converter size, decrease with increasing frequencies, a high fre-
performance. The losses of loss optimized designs for different quency is favorable with respect to system size. The converter
frequencies have been calculated and are shown in Fig. 16(a). size decreases with increasing frequency because the dc-link
In the case at hand, the optimal switching frequency is at ap- capacitance reduces with increasing frequency. The fact that the
proximately 5–6 kHz. In Fig. 16(b), the losses of a volumetric volume decreases with increasing switching frequency is only
optimized designs are given for different frequencies; it can be true in a limited frequency range; above this frequency range,
seen that the optimal switching frequency is in the same range. the losses become very high and the size of the components
3124 IEEE TRANSACTIONS ON POWER ELECTRONICS, VOL. 28, NO. 7, JULY 2013

increases again so that the heat can be dissipated. This effect is [11] J. Mühlethaler, J. W. Kolar, and A. Ecklebe, “A novel approach for 3D
not visible here. air gap reluctance calculations,” in Proc. 8th Int. Conf. Power Electron.,
2011, pp. 446–452.
[12] J. Mühlethaler, J. Biela, J. W. Kolar, and A. Ecklebe, “Improved core loss
calculation for magnetic components employed in power electronic sys-
VIII. CONCLUSION AND FUTURE WORK tems,” in Proc. Appl. Power Electron. Conf. Expo., Mar. 2011, pp. 1729–
A design procedure for three-phase LCL filters based on a 1736.
[13] J. Mühlethaler, J. Biela, J. W. Kolar, and A. Ecklebe, “Core losses under
generic optimization approach is introduced guaranteeing low DC bias condition based on Steinmetz parameters,” in Proc. IEEE/IEEJ
volume and/or low losses. The cost function, which character- Int. Power Electron. Conf., Jun. 2010, pp. 2430–2437.
izes a given filter design, allows a weighting of the filter losses [14] J. Lammeraner and M. Stafl, Eddy Currents. London, U.K.: Iliffe Books,
1966.
and of the filter volume according to the designer’s need. Differ- [15] L. C. Andrews, Elementary Partial Differential Equations With Boundary
ent designs have been calculated to show the tradeoff between Value Problems. New York: Academic Press, 1986.
filter volume and filter losses. Experimental results have shown [16] P. Dowell, “Effects of eddy currents in transformer windings,” Proc. Inst.
Electr. Eng., vol. 113, no. 8, pp. 1387–1394, 1966.
that a very high loss accuracy has been achieved. To improve [17] A. V. den Bossche and V. C. Valchev, Inductors and Transformers for
the THD and current ripple accuracy, the frequency behavior of Power Electronics. Boca Raton, FL: CRC Press/Taylor Francis Group,
the inductors could be modeled as well. 2005.
[18] VDI Gesellschaft, VDI Heat Atlas. Heidelberg/Berlin, Germany:
As can be seen in Fig. 8, a higher switching frequency leads Springer-Verlag, 2010.
to lower filter volume, or, when keeping the volume constant, [19] Film Capacitors, EPCOS AG, Munich, Germany, 2009.
to lower filter losses. However, higher switching losses are ex- [20] R. D. Middlebrook, “Design techniques for preventing input filter os-
cillations in switched-mode regulators,” in Proc. Nat. Solid-State Power
pected in case of higher switching frequencies. Therefore, an Convers. Conf., 1978, pp. A3.1–A3.16.
overall system optimization, i.e., an optimization of the com- [21] R. W. Erickson and D. Maksimovic, Fundamentals of Power Electronics.
plete three-phase PFC rectifier including the filter, has been New York: Springer, 2004.
[22] J. W. Kolar and T. Friedli, “The essence of three-phase PFC rectifier
performed. Generally, it is important to consider the system systems,” in Proc. 33rd IEEE Int. Telecommun. Energy Conf., Amsterdam,
to be optimized as a whole, since there are parameters that Netherlands, Oct. 2011, pp. 1–27.
bring advantages for one subsystem while deteriorating another [23] J. A. Nelder and R. Mead, “A simplex method for function minimization,”
Comput. J., vol. 4, pp. 308–313, 1965.
subsystem. [24] M. Schweizer and J. W. Kolar, “High efficiency drive system with
3-level t-type inverter,” in Proc. 14th Eur. Conf. Power Electron. Appl.,
Birmingham, U.K., Aug. 30–Sep. 1, 2011.
ACKNOWLEDGMENT [25] T. Friedli and J. W. Kolar, “A semiconductor area based assessment of ac
motor drive converter topologies,” in Proc. 24th Annu. IEEE Appl. Power
The authors would like to thank ABB Switzerland, Ltd., for Electron. Conf. Expo., Feb. 2009, pp. 336–342.
giving them the opportunity to work on this very interesting [26] M. Schweizer, I. Lizama, T. Friedli, and J. W. Kolar, “Comparison of the
project. chip area usage of 2-level and 3-level voltage source converter topologies,”
in Proc. 36th Annu. Conf. IEEE Ind. Electron. Soc., Nov. 2010, pp. 391–
396.
[27] U. Drofenik, G. Laimer, and J. W. Kolar, “Theoretical converter power
REFERENCES density limits for forced convection cooling,” in Proc. Int. PCIM Eur.
Conf., Nuremberg, Germany, 2005, pp. 608–619.
[1] J. W. Kolar, J. Biela, S. Waffler, T. Friedli, and U. Badstuebner, “Perfor-
mance trends and limitations of power electronic systems,” in Proc. 6th
Int. Conf. Integr. Power Electron. Syst., 2010, pp. 1–20. Jonas Mühlethaler (S’09) received the M.Sc. and
[2] R. Yu, B. Pong, B. Ling, and J. Lam, “Two-stage optimization method Ph.D. degrees from the Swiss Federal Institute of
for efficient power converter design including light load operation,” IEEE Technology Zurich (ETHZ), Zurich, Switzerland, in
Trans. Power Electron., vol. 27, no. 3, pp. 1327–1337, Mar. 2012. 2008 and 2012, respectively, both in electrical engi-
[3] C. Larouci, M. Boukhnifer, and A. Chaibet, “Design of power converters neering. During his master studies, he focused on
by optimization under multiphysic constraints: Application to a two-time- power electronics and electrical machines and in-
scale AC/DC-DC converter,” IEEE Trans. Ind. Electron., vol. 57, no. 11, volved on compensating torque pulsation in perma-
pp. 3746–3753, Nov. 2010. nent magnet motors. During the Ph.D. studies, he
[4] S. Busquets-Monge, J.-C. Crebier, S. Ragon, E. Hertz, D. Boroyevich, was involved on modeling and multiobjective opti-
Z. Gurdal, M. Arpilliere, and D. Lindner, “Design of a boost power factor mization of inductive power components.
correction converter using optimization techniques,” IEEE Trans. Power He is currently a Postdoctoral Fellow at Power
Electron., vol. 19, no. 6, pp. 1388–1396, Nov. 2004. Electronic Systems Laboratory, ETHZ.
[5] P. Pelletier, J. Guichon, and J. Schanen, “Optimization of a dc capacitor
tank,” in Proc. Conf. Rec. 40th IAS Annu. Meet. Ind. Appl. Conf., Oct.
2005, vol. 2, pp. 1470–1474.
[6] J. Zientarski, R. Beltrame, D. Candido, M. Martins, and H. Hey, “De-
sign methodology for universal line input boost power factor correction Mario Schweizer (S’09) was born in Switzerland on
magnetics,” IET Power Electron., vol. 4, pp. 715–724, Jul. 2011. February 20, 1983. He studied electrical engineering
[7] M. Liserre, F. Blaabjerg, and S. Hansen, “Design and control of an LCL- at ETH Zurich, Zurich, Switzerland. During his stud-
filter-based three-phase active rectifier,” IEEE Trans. Ind. Appl., vol. 41, ies, he focussed on power electronics, mechatronics
no. 5, pp. 1281–1291, Sep./Oct. 2005. and energy systems. He received the M.Sc. degree
[8] S. V. Araujo, A. Engler, B. Sahan, and F. Antunes, “LCL filter design for in May 2008. His Master thesis research involved
grid-connected NPC inverters in offshore wind turbines,” in Proc. 7th Int. implementation of a new control hardware and dif-
Conf. Power Electron., 2007, pp. 1133–1138. ferent control strategies for an ultra sparse matrix
[9] T. Friedli, “Comparative evaluation of three-phase Si and SiC AC-AC converter. He has been working toward the Ph.D.
converter systems” Ph.D. dissertation, ETH Zurich, Switzerland, 2010. degree at the Power Electronic Systems Laboratory
[10] J. Mühlethaler, J. W. Kolar, and A. Ecklebe, “Loss modeling of inductive since July 2008.
components employed in power electronic systems,” in Proc. 8th Int. Conf. His current research interests include eco’intelligent drive systems and grid-
Power Electron., 2011, pp. 945–952. friendly active rectifier interfaces.
MÜHLETHALER et al.: OPTIMAL DESIGN OF LCL HARMONIC FILTERS FOR THREE-PHASE PFC RECTIFIERS 3125

Robert Blattmann received the M.Sc. degree in elec- Andreas Ecklebe (M’07) received the Dipl.-Ing. and
trical engineering from the Swiss Federal Institute Dr.-Ing. degrees from the Otto-von-Guericke Univer-
of Technology Zurich, Zurich, Switzerland, in 2011. sity, Magdeburg, Germany, in 2002 and 2009, respec-
During his master studies, he focused on power elec- tively, both in electrical engineering.
tronics and energy systems and was involved on a From 2002 to 2004, he was with SMS Demag
new concept for a magnetic bearing system. AG and Alstom Power Conversion, where he was in-
In 2012, he joined the Trainee Program of ABB volved in the design of technological control systems
Switzerland, Ltd., Baden, Switzerland. for large-scale industrial automation solutions. Since
2008, he has been with ABB Corporate Research,
Baden-Daettwil, Switzerland, where he is currently
leading the Power Electronics Integration Research
Group.

Johann W. Kolar (S’89–M’91–SM’04–F’10) re-

ceived the M.Sc. and Ph.D. degrees (summa cum
laude/promotio sub auspiciis praesidentis rei pub-
licae) from the University of Technology Vienna,
Vienna, Austria.
Since 1984, he has been an Independent Interna-
tional Consultant in close collaboration with the Uni-
versity of Technology Vienna, in the fields of power
electronics, industrial electronics and high perfor-
mance drives. He has proposed numerous novel con-
verter topologies and modulation/control concepts,
e.g., the VIENNA Rectifier, the Swiss Rectifier, and the three-phase ac–ac
sparse matrix converter. He has published more than 450 scientific papers in
international journals and conference proceedings and has filed more than 85
patents. He became a Professor and Head of the Power Electronic Systems
Laboratory, Swiss Federal Institute of Technology (ETH) Zurich, on Febru-
ary 1, 2001. His current research interests include ac–ac and ac–dc converter
topologies with low effects on the mains, e.g., for data centers, more-electric-
aircraft and distributed renewable energy systems, and solid-state transformers
for smart microgrid systems. Further main research interests include the real-
ization of ultracompact and ultraefficient converter modules employing latest
power semiconductor technology (SiC and GaN), micropower electronics and/or
power supplies on chip, multidomain/scale modeling/simulation and multiob-
jective optimization, physical model-based lifetime prediction, pulsed power,
and ultrahigh speed and bearingless motors.
Dr. Kolar has been appointed the IEEE Distinguished Lecturer by the IEEE
Power Electronics Society (PELS) in 2011. He received the IEEE Transactions
Prize Paper Award of the IEEE Illuminating Engineering Society in 2005, of
the IEEE Institute of Aeronautical Sciences in 2009 and 2010, the IEEE/ASME
in 2010, and of the IEEE PELS in 2009, 2010, and 2011. He also received the
Best Paper Award at the International Conference on Performance Engineering
2007, the IAS 2008, the IECON 2009, the International Symposium on Param-
eterized and Exact Computation 2010 (for two papers), the Energy Conversion
Congress and Exposition (ECCE) Asia 2011, and the ECCE USA 2011. Fur-
thermore, he received the ETH Zurich Golden Owl Award 2011 for Excellence
in Teaching and an Erskine Fellowship from the University of Canterbury, New
Zealand, in 2003. He initiated and/or is the founder/cofounder of four spin-
off companies targeting ultrahigh speed drives, multidomain/level simulation,
ultracompact/efficient converter systems, and pulsed power/electronic energy
processing. In 2006, the European Power Supplies Manufacturers Association
awarded the Power Electronics Systems Laboratory of ETH Zurich as the Lead-
ing Academic Research Institution in Power Electronics in Europe. He is a
member of the IEEJ and of International Steering Committees and Technical
Program Committees of numerous international conferences in the field (e.g.,
Director of the Power Quality Branch of the International Conference on Power
Conversion and Intelligent Motion). He is the founding Chairman of the IEEE
PELS Austria and Switzerland Chapter and Chairman of the Education Chapter
of the EPE Association. From 1997 to 2000, he has been serving as an As-
sociate Editor of the IEEE TRANSACTIONS ON INDUSTRIAL ELECTRONICS and
since 2001 as an Associate Editor of the IEEE TRANSACTIONS ON POWER ELEC-
TRONICS. Since 2002, he has been an Associate Editor of the Journal of Power
Electronics of the Korean Institute of Power Electronics and a member of the
Editorial Advisory Board of the IEEJ Transactions on Electrical and Electronic
Engineering.