Você está na página 1de 5

Modeling of Electric Motors for Electromagnetic Compatibility

Analysis
Firuz Zare
School of Engineering Systems
Queensland University of Technology
Brisbane, QLD 4001,
f.zare@qut.edu.au
Abstract

This paper presents a method to model electric motors at


high frequency which can help designers to predict
leakage current and to design passive and active EMI
filters. A simple model of an electric motor at high
frequency used in high speed switching applications is
discussed in this paper as a key point for electromagnetic
compatibility analysis.

1.
Introduction
Modern electronic drives are large complex systems
consisting of a filter, a rectifier, dc-link capacitors, an
inverter and a motor as shown in Fig.1. In power
converters, high frequency leakage currents generated by
high dv/dt of hard and fast switches flow to the ground
wire through the stray capacitors of the motor and the
power module. Electromagnetic Compatibility (EMC) is
now an important issue in power electronics. Since more
than 60% of the world's energy is used to drive motors, it
is important to characterize and predict the
electromagnetic emission behavior of motor drive
systems. As switching speeds of power devices used in
power modules are increased, Electromagnetic
Interference (EMI) becomes a major problem for power
electronic circuits.

Fig. 1: A block diagram of a motor drive system


A passive EMI filter consisting of inductors and
capacitors, are normally used in power electronic circuits
in order to suppress common-mode and differentialmode noises. For high power applications, when the
magnitude of the common mode current increases, the
filter size and cost are significantly increased. Fig.2
shows a passive EMI filter for suppression of high
frequency noises. Common mode inductors in the filter
are made out of toroid cores with high impedance against
common mode current, but low impedance against
differential mode current. Main problems in use of
passive filters are:

Size of common mode inductors and X


capacitors
Power losses in common mode inductors
Cost of inductors and capacitors

Fig. 2: General circuit of a passive EMI filter


Active EMI filters based on current injection is another
solution to cancel the common mode high frequency
currents. Fig.3 shows a block diagram of an active EMI
filter and a common mode transducer (CT). In this
circuit, the common mode current is sampled at the input
of rectifier and the active EMI filter generates a similar
current in order to cancel the leakage current. The idea is
to create a low impedance path for the leakage current of
the electric motor instead of the ground path. Several
authors have reported different issues regarding EMI
suppressions. Isao Takahashi [5] and Seung-ki SuI [6, 7]
proposed active EMI filters with a feedback control
system. The filters can suppress low frequency noises
(150 kHz - 3 MHz) but it needs a passive filter to remove
all high frequency noises [4]. In [10], a topology has
been developed to apply active filtering to the utility
interface of a switching power supply and demonstrated
improvement of EMI filter for switching frequencies of a
few hundred kHz, although with the cost of a complex
compensation. In [9], an improved inverter output filter
is proposed for pulse width-modulated drive-systems.
The proposed filler is shown to effectively reduce both
the differential and common modes dv/dt at the motor
terminals, even in the presence of long motor leads.
Reducing differential mode dv/dt reduces over voltages
at the motor terminals and lowers the stress on the motor
insulation. Lowering common mode dv/dt significantly
reduces high-frequency leakage currents and shaft
voltage in motors.
The leakage current shows in Fig.3. (b) is created by a
high voltage stress during a short switching time (dv/dt).
As the switching time of modern power electronic
switches are decreased, the high frequency components
of the leakage current become significant and its

magnitude and waveform depend on the capacitive


coupling of the motor. In order to simulate and design an
active EMI filter, an accurate model of electric motor is
required when high dv/dt is applied to the electric
machine. This paper addresses a method to extract high
frequency parameters of a motor based on the
measurement results. It is recommended to estimate the
behavior of the motor in a frequency range, when the
motor acts as a capacitive or an inductive load according
to the phase values of the impedance, Then, the
parameters can be calculated directly from the magnitude
and the phase values of the impedance in these frequency
ranges, In this simple method, other issues like skin
effects and the effect of stray capacitance between the
windings will be discussed. This paper also presents that
a motor can be simplified as a single capacitor when a
high speed switching is applied to a motor drive system.

Each motor has different high frequency parameters due


to its structure, size and materials.

Fig.4: Simple model of an electric motor


2.1.
Analysis of an electric motor at high
frequency
A low frequency model of an electric motor cannot be
used in high frequency analysis due to the stray
capacitances and frequency-dependency of losses. The
first step is to measure the magnitude and the phase
values of the impedance in terms of frequency based on
two different connections of the windings.
2.1.1
Calculation of Cstray
The first measurement is based on Fig.5.(a) where all
three phases are connected to each other at the terminal
side with assuming that the windings of the motor are
connected in a star form. Thus, the impedance between
the phases-stator (phase and magnitude values) is
measured in terms of the frequency using an impedance
analyzer.
Based on this configuration, the impedance of the stray
capacitance, Cstray, is much higher than the impedance of
Ls and Rloss at low frequencies [Fig.5.(c)] and the model
can be simplified as two parallel stray capacitances as
shown in Fig.5.(d).

Fig.3: (a) An active EMI filter (b) leakage current due to


dv/dt of a power module
2.
Model of electric motors at high frequency
An increase in the carrier frequency of voltage-source
PWM inverters based on high-speed switching devices
has improved operating characteristics of the inverters.
High speed switching can generate the following serious
problems due to high dv/dt:
Ground current escaping to earth through stray
capacitors inside a motor
Conducted and radiated noises
Shaft voltage and bearing current
The stray capacitance between the windings and the
stator frame is the most significant parasitic component
compare to the other stray capacitances. At high
frequency, an electric motor can be modeled as
distributed capacitors, inductors and resistors and the
maximum frequency can be determined using the
standing wave's equation. In this paper, we consider one
section to model the motor at high frequency as shown
in Fig.4 and only the stray capacitance between the
windings and the stator has been considered. The
proposed model can be used to analyze the effect of the
bearing and the common mode currents facilitate the
determination of solutions to suppress these currents.

2.1.2
Calculation of Ls and Rloss
The second measurement is based on Fig.5.(b) where
three phases are connected to each other and the
impedance between the phases-star point is measured.
Based on this configuration, the impedance of the stray
capacitance is much higher than the impedance of Ls and
Rloss at low frequencies [Fig.5.(e)] and it can be
simplified as an inductive and a resistive load as shown
in Fig.5.(f).
Fig.6 (a) & (b) show measurement results of the
impedance between the phases-stator of a 5.5KW motor.
The phase value is around 90 at <40KHz which
addresses that the two capacitors are connected to each
other in parallel [refer to Fig.5. (c)]. The stray
capacitance of the electric motor has been calculated
using the magnitude and the phase values from 10KHz
up to 40KHz and the result shows that Cstray is almost
around 0.9 nF.
Fig.6 (c) & (d) show measurement results of the
impedance between the phases-star point of the 5.5kW
motor. The inductance value can be calculated in two
different ways:
The inductance value can be calculated using
the measurement results and based on the circuit
diagram shown in Fig.5.(e). The inductance
value is decreased from 5.77mH down to

5.56mH when the frequency is increased from


10kHz up to 40kHz.
The inductance value can also be calculated at a
2
resonance frequency Ls =
=6.28mH. The
( 2f ) 2 C stray
error between this method and the previous one is due to
a stray capacitance between the windings, which has not
been taken into the account in this equation. Considering
the stray capacitance between the windings as 10% of
Cstray, the above equation gives Ls=5.7 mH.

1 .10

3.1 10

1 .10

Z_exp1
100

Zsimul( f )

10

2.79

1
4
1 .10

1 .10

1 10

1 .10

1 .10

Z_exp0 , f

1 10

(a)
78.3

100

50

Phase_exp 1
0

180

arg( Zsimul( f ) )

50

91.1

100
4
1 .10
1 104

1 .10

1 .10

Phase_exp 0 , f

1 .10
1 107

(b)
1 .10
741.102

100

Z_exp 1
Zsimul( f )
10

Fig.5: Two different connections to measure the


impedance between (a) phases and the stator (b) phases
and the star point; measurement between the phasesstator (c) a high frequency model of a motor (d) a simple
model of a motor; measurement between the phases-star
points (e) a high frequency model of a motor (f) a simple
model of a motor

3.378

1 .10

1 .10

1 .10
7
1 10

Z_exp 0 , f

(c)
79.087

100

50
Phase_exp

The same method is applied to extract Rloss and the result


shows that the resistance value is increased from 1129
Ohms up to 2215 Ohms when the frequency is increased
from 10 kHz up to 100 kHz. From the measurement
results, it can be seen that Rloss varies in terms of
frequency and it changes the characteristic of the motor
(between phases-star points). It is important to choose a
correct value for Rloss in order to have an accurate model
with a good performance. According to Fig.6, Rloss and Ls
influence the magnitude and the phase of the motor from
40 kHz up to 300 kHz. Thus, it is important to choose
Rloss within this range of the frequency and it is
recommended to calculate Rloss at the first resonance
frequency (100 kHz).

1
4
1 .10
4
1 10

arg ( Zsimul( f ) )

180

50

85.5 100
4
1 .10

1 .10

1 10

1 .10
,f

Phase_exp

1 .10
7
1 10

(d)
Fig. 6: Experimental and simulation results of a 5.5 kW
motor; (a) magnitude (b) phase values between phasesstator (c) magnitude (d) phase values between phases-star
point
There is a small cable from the motor terminals to the
impedance analyzer, which can be modeled as a R_L in

series with the motor as shown in Fig. 7. The second


resonance frequency shown in Fig.6 (a) & (c) is
associated with this inductance and the first stray
capacitance, Cstray. Thus the inductance value of the cable
can be extracted as Lcable =378 nH. The resistance value
of the cable can be calculated at the second resonance
frequency. Based on this analysis, the parameters of a 5.5
kW motor have been extracted as follow:
Ls=5.56 mH; Cstray=O.9 nF; Rloss=2215 Ohms;
Lcable=378nH; Rcable=3 Ohms

Fig. 7: A high frequency model of a motor with a small


cable.
There are several issues, which have to be discussed:
 The main point is that Rloss is a function of
frequency and it can be modeled as a non-linear
load but it makes the model very complex.
Another alternative is to choose a correct value
for Rloss, with a performance close to
experimental results. It is recommended to
calculate Rloss at the first resonance frequency.
 The inductance value is not significantly
changed from low frequency up to the first
resonance frequency; and it is recommended to
calculate the inductance value using the phase
and the magnitude values from the measurement
results shown in Fig.6.
 In this paper, the stray capacitance between the
windings has not been taken into the account.
The simulation result shows that the stray
capacitance between the windings has less effect
on the impedance between the phases-stator, but
it can improve the performance of the motor
(only between the phases-star point) from
300kHz up to 3MHz.
2.2 A simple high frequency model of an electric
motor
A high frequency model of an electric motor has been
discussed in the last section with considering three main
parameters, the stray capacitance between the windings
and the stator, the losses and the inductance. The main
issue is to analyze a common mode current generated by
high speed switching devices such as insulated gate
bipolar transistors (IGBTs), which generate high dv/dt
across the terminals of the motor. Fig.8 shows a threephase motor based on the proposed model and it is
connected to an inverter. When a PWM inverter operates,
only the voltage of one of the phases is changed at each
instant. Thus, the leakage currents due to the switching
transients can be simulated using the high frequency
model of the motor and PWM pulses.

Fig.8: a high frequency model of a three-phase motor


connected to an inverter
An IGBT and its gate driver determine the switching
time and the dv/dt. Assuming that DC link voltage is
around 600 Volt, the dv/dt is determined by the
switching time of the device. Fig.9 shows a PWM pulse
with a rising time of tr. To analyze and to find out the
influence of the switching time on the leakage current,
harmonic contents of a PWM pulse are calculated based
on an approximate method [5]. The first corner
frequency, f = 1 , is determined by duty cycle of the
1

pulse and the second corner frequency, f = 1 , is


2

tr

inversely proportional to a rising time of the signal. For


PWM applications, f2 is normally much higher than f1
and it determines the maximum frequency of the PWM
pulse (refer to Table.I).

Fig.9: a PWM pulse with a rising time, tr in (a) time


domain (b) frequency domain
Table.I
Vdc=600 Volt
f2

tr=150 ns
2123 KHz

tr = 600
ns
530 KHz

tr = 1500
ns
212.3 KHz

Fig.10.(a) shows that a common-mode current generated


by a dv/dt consists of two parts. The first part is due to
the first stray capacitor Cstray1 and the second part is due
to Rloss, Ls and Ctray2. The impedance of the first leg
(Cstray1) is significantly lower than the impedance of the

second leg (Rloss, Ls and Ctray2) at high frequency, while


their impedances are almost same at low frequencies.
Thus, the common-mode currents strongly depend on the
harmonic contents (or switching times) of the input
pulses.
If a PWM pulse is applied to this electric motor, with f2 =
212 kHz, (tr=1500 ns) which is close to the first
resonance frequency of the motor, the second leg (Rloss,
Ls and Ctray2) carries a significant current. But if f2 (for
example f2 =2.1MHz) of a PWM pulse is higher than the
first resonance frequency of the motor (10 times higher),
then the significant parts of the high frequency contents
of the input pulse are influenced by the first stray
capacitor, Cstray1. The impedance of the second loop is
increased at high frequencies, while the impedance of the
Cstray1 is decreased. Thus, the leakage current generated
by Cstray1 is much more than the one generated by the
second loop when a high dv/dt is applied to the motor. In
this case, the high frequency model of the motor can be
simplified as one stray capacitor, Cstray1, (with
considering the high frequency model of a small cable)
as the leakage current in the first leg is much more than
the current in the second leg.

Fig.10: a high frequency model of an electric motor used


for (a) slow switching (b) fast switching
3. Conclusions
A high frequency model of an electric motor has been
analyzed in this paper. One of the advantages of the high
frequency model of the electric motor is to estimate and
to analyze the leakage and the bearing currents and to
design EMI filters using simulations. This paper
addresses a simple method to extract high frequency
parameters of an electric motor based on the
measurement results. It is recommended to estimate the
behavior of the motor in a frequency range, when the
motor acts as a capacitive or an inductive load according
to the phase values of the impedance. Then, the
parameters can be calculated directly from the magnitude
and the phase values of the impedance. An impedance of
a 5.5kW motor has been measured in terms of the
frequency using an impedance analyzer and the results
have been used to extract the high frequency parameters.
In this simple method, other issues like skin effects and
the effect of stray capacitance between the windings have
been discussed. This paper also presents that a motor can
be simplified as a single stray capacitor when a high
speed switching is applied to a motor drive system.
References
[1]. H.Zhu, Jih-Sheng
Lai.
Modeling-Based
Examination of Conducted EMI Emissions from Hard

and Soft Switching PWM Inverters, IEEE Transactions


on Industry Applications, Vol. 37, No.5, Sep/Oct 2001.
[2]. S.Chen. T.A.Lipo, D.Fitzgerald. Modeling of Motor
Bearing Currents in PWM Inverter Drives, IEEE; IEEE
Transactions on Industry Applications, Vol. 32, No.6,
and Nov/Dec 1996.
[3]. Michel Mardiguian, EMI trouble shooting
Techniques, the McGraw-Hill companies, Inc, 2000.
[4]. Satoshi Ogasawara, Hideki Ayano, Hirofumi Akagi.
An Active Circuit for Cancellation of Common Mode
Voltage Generated by a PWM Inverter, IEEE
Transactions on Power Electronics, Vol. 13, No.5, Sep
1998, pp.835-841.
[5]. Isao Takahashi, "Active EMI Filter for Switching
Noise of High Frequency Inverters, IEEE Power
Conversion Conference, Aug. 1997, pp.331-334.
[6]. Seung-Ki Sul, "A Novel Active Common-mode EMI
Filter for PWM Inverter APEC 2002. Seventeenth
Annual IEEE March 2002 Volume: 1
[7]. Yo-Chan Son, Seung-Ki Sul, "A New Active
Common-Mode EMI Filter for PWM Inverter", IEEE
Transactions on Power Electronics, Vol. 18, NO.6, NOV
2003
[8]. A.Boglietti, E.Carpaneto. An Accurate Induction
Motor High-Frequency Model for Electromagnetic
Compatibility Analysis, Electric Power Components
and Systems, 29:191-209, 2001
[9]. Dudi A. Rendusara, Prasad N. Enjeti, "An Improved
Inverter Output Filter Configuration Reduces Common
and Differential Modes at the Motor Terminals in PWM
Drive Systems", IEEE Transactions on Power
Electronics, Vol. 13, NO.6, Nov 1998
[10]. Thomas Farkas, Martin Schlechet, "Viability of
Active EMI Filters for Utility applications, IEEE
Transactions on Power Electronics, Vol.9, No.3, May
994, pp.328-337.

Você também pode gostar