Development of a conducted EMI model for a industrial power factor corrector BREHAUT STEPHANE

DEVELOPMENT OF A CONDUCTED EMI MODEL FOR AN

INDUSTRIAL POWER FACTOR CORRECTOR.

Stéphane BREHAUT(1), Jean-Charles LE BUNETEL(1), Ambroise SCHELLMANNS(1) ,

Didier MAGNON(1) , Antoine PUZO (2).
(1)
Laboratoire de Microélectronique de Puissance
7 avenue Marcel Dassault BP 407 - 37204 TOURS Cedex 3,
(2)
SAFT Power Systems Group
ZI n°1- 10 rue Jean Perrin 37173 Chambray les Tours Cedex
E-Mail : [email protected]

Keywords
«EMC/EMI, Power factor correction, Modelling, Simulation».

Abstract
The need to provide a good power factor correction is required in a growing number of applications, as
in telecommunications. The boost PFC circuit, widely used to fulfill this requirement, is developed at
low and medium power. However, PFC generates electromagnetic interferences in the power
converters. We must put an expensive and bulky EMI filter. The purpose of this paper is to present a
new methodology by calculating the EMI with a developed software. Experiments are performed to
verify the consistency with the result of the simulation tool. An industrial PFC for a 600W unit with
230Vac 50Hz input and 380V output has been used for this study.

I. Introduction
In power electronics, filters are commonly used for two purposes : to reduce the emission and
propagation of electromagnetic interferences (EMI) and to reduce the susceptibility of the converters
with respect to external EMI. The presence of filters is dictated by the proper operation of the
converters under the worst EMI constraints. In our case, the power supply studied is constituted of
both cascaded converters and it is used as a battery charger in telecommunication. The first one is a
power factor corrector (PFC) and the second one is a DC-DC insulated converter. Previous studies [1]
showed that the worst pollution came from PFC in the power supply. To reduce the EMI pollution, we
use expensive and bulky filters. All these facts demonstrate the importance of research in the
development of the EMI analysis of power converters working as power factor conversion.

There have been recent efforts to provide a more methodical design process by implementing
optimisation techniques [2][3]. The method proposed in this paper is based on frequency model for
EMI prediction and can thus be easily implemented in a common mathematical tool as MATLAB.
Using this software, it is possible to determinate with accuracy the conducted EMI before building an
industrial PFC without filter. This information can contribute to acquire a better understanding of the
system behaviour in high frequency. This software enables the simulation of the Line Impedance
Stabilization Network (LISN) and the EMI receiver. We can determine full EMI spectrum in dBµV in
accordance with the CEI CISPR 22 standard [4] in function of all parasitic elements.

The paper is presented as follows. In the first part, we present the system under study and its
specifications. Then, in the second part, the methodology for modelling of the EMI approach is
described and developed in a schematic method. Our purpose is a general application on other
converters. At least, the comparison between simulation and measurement results are showed and
analysed.

EPE 2003 - Toulouse ISBN : 90-75815-07-7 P.1

Development of a conducted EMI model for a industrial power factor corrector BREHAUT STEPHANE

II. System studied, specifications and fixed design variables

A. Interest for a follow-up EMI study of a PFC.
The power supply studied is a basic structure of a typical two-stages front end converter for DC
distributed power systems [5]. The first structure is constituted of a PFC boost converter by the need
for a good power factor correction. It is in series with a full-bridge DC/DC converter, a transformer
and a rectifier 48VDC 12A (figure 1).

High
Voltage Vbus
AC line PFCBoost Full Bridge
DC/DC

Fig.1. Two stage front end converter.

If we do a spectral analysis, we can say that the spectrum of the global converter has a similar form
with the spectrum of the PFC (figure 2). In fact, the two peaks observed for the full power supply at
12MHz and at 18MHz are at the same resonance than the peaks of the PFC. There is a difference of 10
to 15dBµV between the two spectra. The surface of the global converter’s layout is more important
than that of the layout of the PFC. So, the parasitic capacitors between the ground board and the layers
are more important. These parasitic elements produce common mode pollution [6]. We can conclude
that the PFC generates the predominant EMI. We concentrate our research on the modelisation of the
PFC.

Frequency global
system result.

Frequency PFC
result.

Fig.2. Influence of the PFC’s pollution on the global system (150KHz-30MHz).

We can notice that the peaks of EMI of the global converter are nearer. Indeed, the switching
frequency of the inverter is 100KHz whereas the switching frequency of the PFC is 40.5KHz.
B. Description of the PFC.
The system to be designed consists in a boost PFC converter with a part of an EMI filter as showed in
Figure 3, which gives an electrical schematic of the circuit. We keep a part of the EMI filter because
we need the capacitor above the bridge rectifier for the tripping of the voltage measured by the
controller. The capacitors below the bridge will permit to do many tests of robustness. The designed
specifications include : output power (Po), input voltage (Vin), line frequency, output voltage (Vout).
The controller components are not included in Figure 3. The constant-frequency average-current-mode
control for continuous-current-mode operation is the control strategy for the switch. The
characteristics of the converter used for this application are given in table I. We use a battery to supply
the controller in order to limit pollution.

EPE 2003 - Toulouse ISBN : 90-75815-07-7 P.2

Development of a conducted EMI model for a industrial power factor corrector BREHAUT STEPHANE

Inrush limiting
Input Voltage[VAC] 230 limiter circuitry

Output Voltage [VDC] 382

C 106
C 102 Load
Switching frequency 40.5 C 105 Capacitor
Control
Fs [KHz] filter

C 107
Input current [Arms] 3.2 Ac Capacitor
230Vac 50Hz filter
source

Table I. Specifications Fig.3. Boost PFC Stage Schematic

The snubber composed of a ferrite and a parallel diode with the free wheel diode of the commutation
cell allows to reduce the peak of the inverse overlay of the power diode. The diode in parallel allows
to discharge the accumulated energy in the ferrite.

III. EMI Tool Description

This section is a description of the proposed tool. The main parts of this procedure are presented in the
following steps. The principle is to work directly in the frequency domain. The software enables to
determine the conducted EMI in dB/µV for the PFC converter, in accordance to the CISPR 16-2
standard.
A. Boost modelling and assumptions.
The first step for conducted EMI prediction is to propose a complete electrical equivalent circuit
including the converter itself with all the parasites, the measurement equipment (Line Impedance
Stabilisation Network-LISN), the cabling impedance between LISN and converter and the
commutation cell. We do many hypotheses in order to limit the complexity of the modelisation. In
spite of the effect of saturation of core, we assume that the value of the boost inductance requires a
unique value in high frequency. The system configuration between the mains and the bridge rectifier is
assumed to be symmetrical with respect to ground. We consider that only the switching cell of Boost
converter with a snubber generates disturbances. We admit that the impedance of the network is
endless.
LISN Commutation cell

ground

Bridge Propagation paths : parasitic

rectifier inductances and capacitors

Fig.4. Equivalent scheme for EMI modelling, including parasitic components, LISN and converter.

EPE 2003 - Toulouse ISBN : 90-75815-07-7 P.3

Development of a conducted EMI model for a industrial power factor corrector BREHAUT STEPHANE

The propagation paths correspond to electrical connections between interference components and the
switching cell. The study of propagation paths requires a previous knowledge of the high frequency
behaviour of every components and interference elements. These elements are the parasitic capacitors
and inductances, the passive and active components, the circuit layout, the bridge rectifier and the line
impedance stabilization network (LISN). The components and layouts were carefully measured. This
electrical scheme is presented Figure 4.

Each element of the circuit respects the impedance variation along the frequency range (150KHz
30MHz). The switch and the diode are two sources of pollution. These sources produce the pollution
in the commutation cell. The layouts are assimilated to inductances and the ground board is
represented by capacitors.

We suppose that the rectifier does not generate disturbances but the impedances of diode rectifier are
included in the simulation. The duty cycle frequency’s of PFC is 10ms, so, the cycle of pollution is
10ms too. During this time equals to the half period of mains, diodes are always in the same position,
open or closed, so the position of the diodes is static in the modelisation. For each frequency, electrical
elements can be modelled by their impedance. The resolution of the equivalent electrical circuit
estimates the EMI level in the LISN resistors. We use a matrix method [7] designed with Kirchoff law.
The models presented with the equation 1 and figure 5 are considered.

U=Z.I (eq.1)
with U : sources of pollution produced by the commutation cell.
Z : The converter is converted into an impedance matrix.
I : Currents of common and differential mode.

Z59

Z48
I33 I25 Z4 7
I26 I39
I40
Z66 Z72 Z73
Z67
Vd2 Z63
7

I34 I35 Z4 Z8 Z37 Z41

Z12 Z51 Z17 Z21 Z24 Z27 Z30 Z33 Z43 Z57 VD Z62 Z68 Z69
I2 I4 I6 I3 I8 I10 I12 I14 I16 I18 I20 I22
Z56 I32
Z50 Z2 Z5 Z9 Z52 0
Z13 ZZ
18 Z22 Z25 Z28 Z31 Z34 Z38 Z45 I36
1
I27
I2 VK
I29 Z60 Z60 Z16 4 Z44 Z64//Z65
Z61

I28 Z3 Z6 Z10 Z53 Z14 ZZ

19 Z23 Z26 Z29 Z32 Z35 Z39 Z46 Z55
1
I3 I5 I7 I31 I9 I15 I17 I19 I21 I23
9 I11 I13
Z7 Z11 Z15 Z54 Z20 Z36 Z40 Z71 Z7 0
Z42
I37 I38
Z74 Z75

Fig.5. Impedance model.

B. One sort of Matrix for a frequency range.
We develop a simulation, between 150KHz and 30MHz, which takes the impedance evolution
of elements in function of frequency into account. The impedance models such as the boost
inductance, figure 6, are very complicated. To obtain a more simplified model, we decide to
work on many frequency ranges. We use accurate models on some frequency ranges limited.

EPE 2003 - Toulouse ISBN : 90-75815-07-7 P.4

Development of a conducted EMI model for a industrial power factor corrector BREHAUT STEPHANE

So, the impedance models of the component are more accessible and the time of calculation is
better. An example of an equivalent model of the boost inductance is given in table II.

Module of the boost inductance.

Phase of the boost inductance.

Fig.6. measurement of impedance of boost PFC between 150KHz and 30MHz.

Frequency 150KHz- 1MHz- 10MHz- 12MHz- 18MHz-

range 1MHz 10MHz 12MHz 18MHz 30MHz

Model
used

Table.II. Evolution of impedance model function of frequency.

C. Use of dynamics matrices for the model of semiconductor.
To account for the effects of the commutation cell with regard to the rest of the system placed between
this cell and the mains, the commutation cell can be replaced by equivalent voltage sources with
impedances in series. The first equivalent voltage source VK (t) (figure 7) is located between the
collector and the source of the switch. The second voltage source Vd (t) is situated between the anode
and the cathode of the diode. The sources of pollution of the semi conductor are represented by a
trapezoidal waveform with the ringing. The rise and fall times of the simple waveform have been
adjusted in function of real commutations.

Impedance VB(t)
Vk switch closed

t t t
VC(t)
Impedance
switch open

Fig.7. Description of the dynamic comportment of the switch.

We must use a dynamic matrix because the impedance of both switch and diode changes depending on
whether they are open or closed. The source of disturbances of the switch Vk (t) is realised by the

EPE 2003 - Toulouse ISBN : 90-75815-07-7 P.5

Development of a conducted EMI model for a industrial power factor corrector BREHAUT STEPHANE

superposition of two sources : the first one for the opening and closing and the second one for the
ringing at the conducting and the opening state. Then, we use the theorem of superposition to make the
summation of harmonics created by the rise, the fall and the ringing of the switch.

First we can characterize these voltage sources in the frequency domain by means of the Laplace
transform and then by applying the appropriate conversion to the Fourier representation. The previous
voltage waveform is represented by an addition of sinusoids, each of them being a multiple of the
fundamental frequency (in our case, the line frequency).
D. Theoretical and experimental EMI results and their comparisons.
We make several high frequency (HF) schemes of the PFC with an increasing precision. We show the
evolution of the accuracy of the simulation. The first scheme, figure 8, is a simplified model defined
by a square matrix (20-20). The second model, figure 4, is defined by a square matrix (40-40). In the
simplified model, we keep the source of disturbances of the MOSFET, the parasitic capacitors of the
bridge rectifier, of the power switch and of the bus capacitors. We have a simulation similar to the
measurement between 150KHz et 500KHz (figure 9).

Fig.8. Equivalent simplified scheme for EMI modelling.

Simula tion

Measurement
Standart CEI
CISPR 22

Fig.9. Simulation and measurement for the simplified HF scheme (150kHz-30MHz).

For the complete HF scheme, the validity of the EMI model is expected and was experimentally
verified in high frequency range (150kHz 30MHz). Figure 10 shows the EMI envelope of the PFC.
Firstly, the comparison between disturbance levels and standard limits [6] clearly shows that the

EPE 2003 - Toulouse ISBN : 90-75815-07-7 P.6

Development of a conducted EMI model for a industrial power factor corrector BREHAUT STEPHANE

studied converter without the filter does not respect the standard. Secondly, the simulation is nearer to
the measurement.

Simulation

Measurement
Standart CEI
CISPR 22

Fig.10. Simulation and measurement for the complete HF scheme (150kHz-30MHz).

The three peaks at 8MHz, 12MHz and 18MHz are due to some resonance loops of disturbance around
the commutation cell. In simulation, we find the three main peaks of resonance, so we can use this tool
to determinate the passive elements which produce the resonance. The oscillating circuit is excited by
the gradient of voltage of the power switch of the converter.

Two solutions can be used to eliminate these peaks of resonance. The first one is to modify the
damping factor, with the modification of one of the elements R,L,C of the loop [8][9]. The bigger the
coefficient, the less important the peak of resonance is. The second solution is to move the resonance
frequency above 30MHz because the filtering is theoretically easier [10]. To reduce the effects of the
parasitic loops which may cause EMI noise, the loops area needs to be as small as possible.

IV. Validation of the simulation tool.

We keep a part of the PFC filter to know the viability of the model studied. We had changed each
value of each capacitor of the filter figure 3 and we have compared the measured and simulated
spectrum. We had done the EMI measurements of robustness with a other prototype which is lightly
different by her layout. So, the comparaison is qualitative.

In the following example, figures 11 and 12,we have change the value of the differential capacitor of
220nF (C105 ) by a value of 440nF.

EMI with the C105

capacitor of 220nF

EMI with the C105

capacitor of 440nF

Fig.11. Measurement of EMI for two different values of the differential capacitor C105 .

EPE 2003 - Toulouse ISBN : 90-75815-07-7 P.7

Development of a conducted EMI model for a industrial power factor corrector BREHAUT STEPHANE

EMI with the C105

capacitor of 220nF

EMI with the C105

capacitor of 440nF

Fig.12. Simulation of EMI for two different values of the differential capacitor C105 .

We can see that in the two case, measurement and simulation, the EMI disturbance is less important
from 12 MHz on. In the second example, figures 13 and 14, we replace two capacitors (C106 , C107 ) , of
common filter, of a value of 4.7nF by two capacitors of 1nF.

EMI with the C 106 and

C107 capacitors of 4.7nF EMI with the C 106 and
C107 capacitors of 1nF

Fig.13. Measurement of EMI for different values of the common capacitors C106 and C107 .

EMI with the C 106 and

C107 capacitors of 4.7nF EMI with the C 106 and
C107 capacitors of 1nF

Fig.14. Simulation of EMI for different values of the common capacitors C106 and C107 .

We notice an increase of the EMI pollution between 150KHz and 8MHz as the disappearance of the
peak at 18MHz. On the contrary, there is an increase of the spectrum of pollution between 26MHz and
30MHz for the simulation. For the capacitors of filtering, the putting up is sensible. If we change the
position of the capacitor, figure 15, we modify the length of the capacitor legs and its parasitic
inductance. The parasitic inductance, varying from 1 to 2nH, has not effect at low and medium
frequency (150KHz-10MHz) but changes the result of EMI at high frequency (10MHz-30MHz).

Parasitic inductance of the

common capacitor.

Fig.15. HF representation of the common capacitor.

EPE 2003 - Toulouse ISBN : 90-75815-07-7 P.8

Development of a conducted EMI model for a industrial power factor corrector BREHAUT STEPHANE

V. Conclusion
The results obtained by the EMI tool are very closed with the experimental results and tests of
robustness which demonstrate the validity of this tool on the range 150kHz-30MHz. With this new
model which includes active and passive components, we can make a detailed and quantitative
pollution analysis of the PFC. Now, it is possible to change or add new passive or active components
easily. We can change the commutation frequency [11] to show the repercussion on the EMI
comportment. If we want to add an additional branch, we must develop a new matrix computation.
The purpose of this work is to investigate how and under which conditions, EMI constraints can be
lowered. We want to achieve a model of the global system of the battery charger with the same
methodology. Thus, we will be able to identify the causes of pollution of this circuit and to reduce the
EMI before putting in a less expensive filter.

REFERENCES
[1]. D. BARILLET-PORTAL, «Rapport de stage», DESS micro-électronique de L’Université de Bordeaux,
2000.
[2]. J-C CREBIER, «Contribution à l’étude des perturbations conduites dans les redresseurs commandés, Thèse
de L’Institut National Polytechnique de Grenoble, 1999.
[3]. S. BUSQUETS-MONGE, J. C. CREBIER, S. RAGON, E. M. HERTZ, J. WEI, J. ZHANG, D.
BOROYEVICH, Z. GURDA, P. K. LINDNER, A. ARPILLIERE, «Optimization Techniques Applied to the
Design of a Boost Power Factor Correction Converter», PESC 2001 IEEE 32nd Annual, Volume: 2, 2001 pp.
920-925.
[4]. CEI CISPR 22, «Radio disturbance characteristics – Limits and methods of measurement (edition 3), 1997.
[5] R. WATSON, «New Techniques in The Design of Distributed Power System», Thèse de l’Institut
polytechnique de Virginie, 1998.
[6]. F.COSTA, «Contribution à l’étude des perturbations conduites par les convertisseurs H.F.» Thèse de
doctorat 3ème cycle, Avril 1992, Université d’Orsay.
[7]. J-C CREBIER, M. BRUNELLO, J. P. FERRIEUX,«PFC full bridge rectifier EMI forecast analysis», EPE
99.
[8]. A. PONS, «Optimisation de la fonction de filtrage dans les convertisseurs de traction», Alcatel, 1998.
[9]. A.PUZO, «C.E.M Chargeur, Méthode de dépollution à la source », Alcatel Alsthom recherche, 1997.
[10]. E. M. HERTZ, «Thermal and EMI modelling and analysis of a boost PFC circuit designed using a genetic-
based optimisation algorithm, Thèse de l’Institut polytechnique de Virginie, 2001.
[11]. E. M. HERTZ, S. BUSQUETS-MONGE, D. BOROYEVICH, «Analysis of the Tradeoffs between Thermal
Behavior and EMI Noise Levels in a Boost PFC Circuit», Proceeding of the IEEE industry Applications
Conference, 2001.

EPE 2003 - Toulouse ISBN : 90-75815-07-7 P.9