Differential Overvoltages and Common-Mode

Currents in PWM Motor Drives: The Influence of

the Cable Arrangement on Their Characteristics
H. de Paula, M. V. C. Lisboa, J. F. R. Guilherme, W. P. de Almeida* and M. L. R. Chaves**
Laboratório de Aplicações Industriais – Departamento de Engenharia Elétrica – UFMG, Brazil
Universidade Federal de Minas Gerais, *CEFET-MG, **Universidade Federal de Uberlândia
Av. Antônio Carlos, 6627 – Pampulha / CEP 31.270-901 / Belo Horizonte, MG, Brazil
[email protected] / www.cpdee.ufmg.br/~helder

protection, generate overheat and destruction of snubber

Abstract – Transient overvoltages at motor terminals and high- capacitances and also degrade the current measurements for
frequency common-mode currents in PWM drive systems have feedback purposes [6 – 8].
been extensively investigated in the last fifteen years. Another issue concerning PWM drives is the presence of
Nevertheless, some issues concerning such quantities still remain
unclear. That is the case of the converter-to-motor cabling high-frequency earth-return currents. The common mode
influence on the intensity of these high-frequency voltages and voltage produced by the inverter excites the capacitive
currents. Filters of different types have been traditionally used couplings between conductors and motor parts to the ground,
to mitigate these undesired phenomena. However, due to their generating successive high-frequency oscillatory currents in
cost, volume and losses, they can’t be considered the ultimate every dv/dt of the corresponding voltage. As a consequence,
solution of the problem. In this paper, typical industrial cable
arrangements will be characterized in terms of the generated EMI problems in industrial plants can arise, as well as
transient overvoltages and common-mode currents. It will be misactuation of fault-to-ground protection relays and motor
shown that variations in the range of 10 to 40 % in the peak bearing failures [9]. Common-mode current peaks of 50 A
values of these quantitites can take place due to different cable have been reported [8] in a textile industry subsystem
disposition. Thus, it can be concluded that special care in the comprising 75 motors of 3 kW rated power.
cable installation can lead to significant reduction in these
undesired phenom-ena, which in turn downsizes the required The phenomena mentioned in the preceding paragraphs
filter. have been largely discussed in the literature [2 – 12] for over
a decade; however, many issues still require further
Keywords - Cable installation, cabling, common-mode currents,
high frequency models, transient overvoltages, transformer investigation. That is the case of the influence of the
modeling. converter-to-motor cabling characteristics on the generated
transient overvoltages and common-mode currents.
I. INTRODUCTION Considerations on the cable modal parameter variation and
their mutual interaction and dependency in relation to the
In its early days, the application of frequency converters for
frequency and distance between conductors have been
motor speed control brought on a series of concerns, such as
presented in [13]. Advancing in this topic, interesting results
[1]: overheating in low speeds; additional losses due to non-
have been presented in [14], where the influence of the
sinusoidal supply; excessive temperature rise, possible useful
cabling on the high-frequency quantities under investigation
life reduction and need of de-rating; torque oscillations and
could be preliminarily illustrated. Nevertheless, aside from
associated problems; among others.
the lack of experimental results, the simulated common-mode
As a consequence of the remarkable advances in the power
currents were not very accurate in the referred work, once the
electronics and microprocessing fields in the last decades,
drive system modeling was not completed. In the present
such problems were overcomed, while others came upon [2,
paper, an appropriate representation of the power transformer
3]. One of them is the generation of transient voltage
was included, able to represent its common-mode impedance
oscillations at motor terminals, due to the PWM pulses
in a wide frequency range. Since the transformer plays a very
propagation and reflection at cable endings. The resulting
important role in the common-mode circuit, much more
voltage peaks can reach up to three times the converter D.C.
realistic results could now be obtained, contributing to the
link voltage (VDC) [4] and, in critical cases, can reduce the
subject under study.
motor insulation life to a few weeks [5]. Besides, in every
voltage transition from zero to VDC or vice-versa, all the II. COMPUTATIONAL REPRESENTATION OF THE
distributed cable capacitance is successively charged or INDUSTRIAL MOTOR DRIVE SYSTEM
discharged, resulting in a high-frequency transient current
component overlapped to the fundamental output inverter As it is well known, the simulation of the above-mentioned
current. This so-called “cable charging current” can be high problem demands sophisticated models, suitable for high-
enough to cause the inverter trip due to overcurrent frequency and transient studies [10, 12]. Regarding the cable,

978-1-4244-4649-0/09/$25.00 ©2009 IEEE 1103

there are two basic requirements: (i) accurate representation converters. The common-mode voltage appears naturally
of the cable resistance and inductance variation (skin and through the combination of the phase voltages. In relation to
proximity effect) up to the MHz range and (ii) reproduction the motor, the one represented in the simulations has a rated
of the wave propagation and reflection phenomena. In power of 2 h.p.
relation to the motor, it is crucial, among other aspects, to Just for the sake of illustration, in the sequence are shown
correctly represent its intrinsic capacitances, which behave as some preliminary results obtained from the above-mentioned
low impedance paths for the receiving voltage pulses. models, along with others experimentally obtained. Figure 1
In the present work, the cable modeling followed the illustrates the voltages applied to the cable, produced by the
methodology proposed in [12], which fulfills the above- pulse voltage source implemented in the program. These
established conditions and allows the simultaneous phase voltages reproduce those ones measured between phase
determination of the differential and common-mode terminals and the middle point of the DC link in a real
quantities (transient overvoltages and earth-return currents, converter. The resulting common-mode voltage, applied to
respectively), in a very simple way. With respect to the the corresponding circuit, generates the common-mode
motor, the model that was employed is that presented in [10], current. Figure 2 shows the occurrence of transient
whose results showed very good agreement with overvoltages, measured between motor phases (fig. 2a), and
measurements. In both cases, the whole procedure for the the common-mode current that flows through the ground (fig.
model determination is clearly detailed in the corresponding 2b), measured at the grounding point of the pulse voltage
references. It should be remembered that the modeling of the sources. Figure 3 presents the phase-to-neutral impedance
ground return path is part of the cable model [12], being thus frequency response of the motor, in terms of its magnitude
naturally included in it. It is also worthy to emphasize that and angle. It can be seen that the main resonance occurs
these modeling techniques have been extensively tested and around 100 kHz, from where the motor begins to present a
applied to similar studies [10, 15], proving to be accurate and capacitive behaviour. Regarding its impedance to ground
reliable. In this work, all the modeling was implemented in (result not show here), it was noted that this one is always
the ATP (Alternative Transients Program), which is a
established simulator for transients program and,
additionally, has the advantage of free use (freeware).
The analyzed system is comprised of three single-phase 4
mm2 cables, each one consisted of the conductor (copper),
insulation and external cover. The insulation material can be
EPR or PVC, while the cover is always made of PVC,
following the characteristics of the commercial cables
researched. Regarding the dielectric permittivities of these
materials, different manufacturers were consulted, but none
of them had this information. Therefore, the average values
found in the literature were used [16, 17], which are: εr
(PVC) = 5,0 and εr (EPR) = 3,0.
The converter-to-motor cable system is installed in a
metallic tray and arranged in plane and triangular geometries.
In the case of the former, the distance between cables is
equivalent to 2,5 times their external diameter. This spacing Fig. 1. Phase voltages (Va, Vb and Vc) and the resulting voltage (Vn-t)
was chosen to make possible the elimination of the “grouping between the motor neutral point and the ground (common-mode voltage).
factor” when following the standardized procedure [18] for
dimensioning the cabling cross section. Besides, since this
distance is quite small (less than 5 mm in the case of the
cable under study), it can be normally found in real
installations even when there is no intention to set this
spacing; the cables are generally just “loosely” placed on the
trays. All this information (cable characteristics, geometric
disposition and form of installation) was used in the ATP
“Cable Constants” routine for the calculation of the required
parameters for the cable model elaboration, as detailed in
[12].
The inverter was modeled as pulse voltage sources,
reproducing the PWM waveform. Each pulse was represented
as being trapezoidal, with rise time values between 90 and Fig. 2. Top: inverter and motor terminal voltages (curves in black and blue,
respectively). Bottom: common-mode current measured at source ground
400 ns, similar to those normally found in commercial point.

978-1-4244-4649-0/09/$25.00 ©2009 IEEE 1104

capacitive, as expected, with progressively lower values as currents flow through the earth return path (in this case, a
the frequency increases. Figure 4 shows some experimental metallic cable tray) and returns to its source (the inverter) by
results obtained in a drive system similar to that of fig. 5, the stray capacitances of the power switches to the grounded
illustrating the differential transient overvoltages (fig. 4a) and heat sink and by the transformer phases. In this way, these
the common-mode current (fig. 4b), measured at the neutral elements must also be represented for an accurate zero-
point of the transformer secondary winding. sequence current calculation.
With respect to the capacitive coupling of the converter
A. Inclusion of the transformer in the system switches to the ground, the values found in the literature [19 -
A typical industrial motor drive installation consists, aside 24] are in the range of 20 – 200 μF. Thus, a 500 μF capacitor
from the motor, frequency converter and interconnecting was decided to be included in the circuit to account for the
cable, of a power transformer as well. For the determination capacitances of all the converter switches.
of the transient overvoltages, which is fundamentally a In relation to the transformer, it can be observed by fig. 5
differential phenomenon, being thus restricted to the inverter, that the common-mode current flows through its grounded
cable and motor phases, the transformer exerts no influence neutral-point towards the phase terminals. Since the
and does not need to be included. However, when it comes to
frequency of these currents can be in the range of few
the common-mode current calculation, the transformer plays
hundreds of kHz up to some MHz, the model to be
a crucial role, since its impedance is part of the common-
implemented must represent the skin effect of the windings in
mode circuit, as can be seen in fig. 5. The common-mode
voltage produced by the inverter excites all the cable and this entire frequency range as well as all the capacitive
motor capacitive couplings to the ground. The generated couplings that exist in the transformer, in order to provide an
accurate determination of the common-mode currents. A vast
literature research was conducted [25 – 31] and, as a result, a
model [27, 29 – 31] that fulfills the stated conditions could be
found, which is briefly described in the sequence. Further
details can be found in the corresponding references.
This model corresponds to a 30 kVA, Δ-Y, 13,8 kV / 220-
127 V transformer, which is based on the concept of
quadripoles. This technique was sucessfully used in a study
[27] of transferred voltage surges between transformer
windings due to lightning or switching events. It correctly
represents the transformer impedance within the 10 kHz – 1
MHz range. Since the common-mode currents under analysis
Fig. 3. Phase-to-neutral impedance frequency response of the motor:
are normally situated in these limits, this model shows to be
magnitude (top) and angle (bottom). quite appropriate for this study. Improvements on this
transformer modeling were suggested by the same author in
[29 – 31]. However, since it was considered here that the
benefits in terms of accuracy where not so noticeable to
justify the increased complexity of the modified model, it was
decided to use in this work the originally proposed one.
Reporting to fig. 5, it can be observed that the common-
mode currents “see” the impedance between the transformer
neutral-point and its secondary terminals. Thus, in the case of
the present study, the quadripole must be re-arranged in order
to represent the afore-mentioned impedance, resulting in the
equivalent circuit shown in fig. 6. This two-node model,
(a) which is connected in series in the common-mode circuit, has
the frequency response depicted by fig. 7. Finally, fig. 8
illustrates the final structure of the circuit implemented in
ATP, for the high-frequency phenomena study in PWM
motor drives.
Transformer Connecting
Cable
Frequency
Converter

(b)
Induction Motor
Fig. 4. Measured results for a 100 m converter-to-motor connecting
cable. (a) Voltage at motor terminals. (c) Generated common-mode
current, measured at source ground point (1 V ≈ 300 mA). Fig. 5. Common-mode current circulation in a PWM motor drive system.

978-1-4244-4649-0/09/$25.00 ©2009 IEEE 1105

 of cable is reached, named “critical length”, a full reflection
occurs and the voltage at the cable ending is twice the
receiving pulse. This behavior can be observed in fig. 9. The
pulse propagation, reflection and combination mechanisms
and the resulting overvoltage at motor terminals are fully
detailed in [15]. Simple and accurate equations for the
frequency and amplitude determination of the transient
overvoltages as a function of the cable parameters and pulse
rise time are presented in [14]. The faster the pulse rise times
Fig. 6: Equivalent circuit configuration for the transformer neutral-to-short- are, the lower the generated voltage peaks become, as
circuited phases impedance representation. observed in fig. 9. Different conductor geometrical
disposition within a cable system can lead to changes in the
1.5K
positive and zero-sequence capacitance and inductances [13],
1.0K
which in turn modifies propagation speed and surge
impedance for the differential and common modes, resulting
0.5K in different transient overvoltage and return currents
intensity, respectively. In the case of the voltage peaks at
0

100d
M(V(V1:+)/I(R5)) motor terminals, considering a specific rise time, the faster
the pulse travels along the cable, the lower are the generated
overvoltages. Since a higher propagation speed was obtained
0d in the triangular configuration, the resulting voltage peaks at
the motor were then lower, as illustrated by fig. 10. This
SEL>> becomes more noticeable as the rise time increases. It should
-100d
100Hz 300Hz
P(V(V1:+)/I(R5))
1.0KHz 3.0KHz 10KHz 30KHz 100KHz 300KHz 1.0MHz 3.0MHz 10MHz
be stressed that a reduction in the order of 5 – 10 % in the
Frequency
expected transient overvoltage can significantly increase the
Fig. 7: Frequency response of the transformer common-mode impedance. life expectancy of standard-insulated motors [32].
Top: impedance magnitude (kΩ). Bottom: impedance phase angle (degrees). Proceeding with the analysis, figs. 11 – 13 present the
results regarding the common-mode currents. Since the model
Cable Model
(featuring skin proposed in this work for the converter high-frequency
Pulses Voltage effect and
wave representation is quite simple, the converter was put into
Source propagation) operation initially ungrounded, in order to eliminate the
Induction possible inaccuracy in the results associated to its modeling.
Common Inverter
Mode Switches-to- Phases
Motor High- Latter, the converter was grounded and the results were
Frequency
Transformer Grounded Model compared, in fig. 14.
Equivalent Heat Sink
Circuit Capacitance Return Path During the analysis, it was observed that both higher
Motor
(included in the
Ground propagation speed and surge impedance in the zero-sequence
cable model)
circuit were also obtained with the triangular cable
Fig. 8: Diagram illustrating the system implemented in ATP. disposition. As far as the earth-returning currents are
concerned, this is highly desirable: the amplitude of the
III. SIMULATED RESULTS common-mode currents is proportional to the relation
From the computational platform presented in the previous between the common-mode voltage distribution over the
section, a large number of cases were simulated to cable and the common-mode surge impedance. Lower over -
characterize the intensity of the transient overvoltages and
common-mode currents as a function of the cable arrange-
ment. Besides, further studies were also conducted in order to
analyze the influence of the pulse rise time and the condition
of not grounding the motor. Having in mind that in each of all
these cases various cable lengths had to be used, a huge
number of simulations was then necessary; due to the extreme
time effort involved, not all cases were performed with all
rise times and cable lengths, as it will be seen in the sequence.
Figures 9 to 14 present the obtained results.
Figure 9 shows the overvoltage peak in relation to cable
length for several inverter pulse rise times. As it is well
known, the amplitude of the transient voltage oscillations at
motor terminals is a function of the pulse rise time and the Fig. 9. Transient peak voltage at motor terminals as a function of cable
length, for different pulse rise times (triangular cable configuration).
pulse propagation speed in the cable. When a certain length

978-1-4244-4649-0/09/$25.00 ©2009 IEEE 1106

 Figure 13 illustrates the common-mode currents obtained
for both grounded and ungrounded motor conditions. In the
latter case, since there is no zero-sequence path provided by
the motor, only the amount of current the flows through the
cable distributed capacitances is measured at the transformer
neutral point. Therefore, without the contribution of the
current associated to the motor, the total current that returns
through the ground towards the transformer is lower, as
depicted by fig. 13.
Figure 14 shows the importance of the whole system
modeling. The common-mode determination without the
presence of the transformer had been already realized in [14],
Fig. 10. Transient peak voltage at motor terminals for plane and triangular
cable configurations and distinct pulse rise times. where the computational platform was not completed.
Nevertheless, as can be seen in fig. 14, the obtained result
when the transformer is disregarded (curve in blue) is way far
from that originated with its (accurate) representation (curve
in black). In relation to the converter high-frequency
modeling employed in this work, the result is depicted by the
curve in green (as mentioned before, in figs. 11-13 the
converter was ungrounded, in order not to influence the
results). Analyzing this curve and the corresponding results
for other rise times (not shown here), it was concluded that
the importance of the inclusion of the inverter switches-to-
ground capacitance reduces as the rise time and the cable
length increase. However, since this is an aproximate
Fig. 11. Common-mode current peak for several pulse rise times, in the representation, conclusions concerning the converter model
case of triangular cable configuration. must be verified by the experimental results, which will be
shown in next section.

Fig. 12. Common-mode current peak for different cable arrangements

and motor grounding conditions. Fig. 13. Common-mode current peak for distinct pulse rise times and motor
grounding conditions (triangular configuration).
voltages associated to higher impedance in the zero-sequence
mode led to more reduced common-mode current amplitudes
in the case of the triangular configuration. If the common-
mode circuit was comprised of the cable only, the behavior of
the currents intensity in relation to the pulse rise time could
be fully predictable, being able to be determined with
appropriate formulae, as demonstrated in [14]. In the present
case, having been included the transformer and the motor, the
analysis becomes quite complex, since it must be performed
under the light of surge impedances and wave propagation
and reflection theory. Figure 11 shows that the dependency of
the common-mode currents on the pulse rise time is restricted
to shorter cables and faster rise times, while fig. 12 indicates Fig. 14. Common-mode current results obtained with different modeling of
the zero-sequence circuit: (1) only interconnecting cable and motor
that the plane geometry resulted in 25 – 35 % higher representation, (2) inclusion of the transformer and (3) also inclusion of the
magnitude for such currents. capacitive coupling of the converter to the ground.

978-1-4244-4649-0/09/$25.00 ©2009 IEEE 1107

 IV. EXPERIMENTAL RESULTS

An experimental setup similar to the system modeled in the

previous section was used in the tests. Unfortunately, a
transformer with the constructive characteristics and rated
power of that represented in the simulations (a 30 kVA
distribution transformer) was not available, which had to be
replaced in the tests by a 5 kVA, 220/380 V unity. Its
secondary winding was connected in star with its neutral
point connected to the cable tray, completing the common-
mode circuit, as illustrated by fig. 5.
Experimental results for the voltage and current oscillations
under study have been already presented in fig. 4. The pulse (a)
rise time of the converter used in the tests (WEG CFW-08)
was also measured and showed to be around 80 ns. Due to
practical difficulties and time involved, in the experiments
only cable lenghts up to 50 meters were analyzed, which were
tested in plane and triangular configurations and for following
conditions: (i) ungrounded motor and converter, (ii) grounded
motor/ungrounded converter and (iii) grounded motor and
converter. Figures 15 shows the obtained results.
A large discrepance in the current peaks was observed
when comparing the simulated with the experimental results,
which shall be attributed to the the mismatch between the
characteristics of the transformer used in the testes and those (b)
of the modeled one. However, the percentual variation in the Fig. 15. Common-mode current peak as a function of the cable length.
amplitude of such currents in relation to the cable geometries In blue: ungrounded motor and converter; In black: grounded motor only; In
green: grounded motor and inverter.
was, in average, similar to that observed in the simulations. A Results for (a) triangular and (b) plane cable geometries.
reasonable agreement in the behaviour of the current peaks in
relation to the cable length was obtained, as well. Comparing successfully used in past works; however, in this paper, the
the results for the cases when the converter was also system modeling was completed by the inclusion of an
grounded (figs. 14 and 15), a discrepance was observed, appropriate common-mode representation of the transformer,
indicating that the (simplified) model used here is not which plays an important role in transient phenomena
reasonable for the converter high-frequency representation. involving the zero-sequence circuit. Analyzing the impedance
With regard to the transient overvoltages (not shown), a very frequency response of the system components (cable, motor,
good proximity between measurements and simulations was transformer and inverter capacitance to ground) up to the
observed, since the differential circuit, which involves only MHz range, it can be observed that, for higher cable lengths
the cable and the motor, was very well represented by the (as the ones used here), the cable impedance is dominant, in
models previously developed [10, 12]. such a way that the resulting frequency response of the
complete system becomes similar to that of the cable. At a
V. CONCLUSIONS first sight, this could wrongly suggest that the transformer
could be disregarded in the analysis; however, keeping in
The main goal of this work was to investigate the influence mind that transient overvoltages and common-mode currents
of typical industrial cable configurations on the generated are originated by pulse propagation and reflection in the
level of transient overvoltages and common-mode currents in positive and zero-sequence circuits, respectively, the
PWM motor drive systems. The study comprised cable impedances involved are the surge ones, which do not depend
systems installed in metallic trays in triangular and plane on length and are different from those obtained in the
geometries; in the case of the latter, the distance between frequency response, even for very high frequencies. In this
cables was equal to 2,5 times their diameter. The chosen context, the following conclusions could be drawn from the
distance avoids the use of the “grouping factor” in the cable research:
cross section dimensioning, which can result in thinner
(and cheaper) cables. Besides, since this distance is quite (a) The triangular cable configuration led to lower transient
small, it can be normally found in practice even when there is voltage peaks at motor terminals, being the obtained
no intention to set this spacing, once the cables are generally overvoltage reduction dependent of the pulse rise time and
just “loosely” placed on the trays. cable length;
For the present investigation, sophisticated models were (b) Simulations indicated that the common-mode current
used for cable and motor representation, which have been magnitude was between 25 – 35 % higher in the case of the

978-1-4244-4649-0/09/$25.00 ©2009 IEEE 1108

cable plane disposition, while, in the experimental results, a [11] A. C. S. Lima, R. M. Stephan and A. L. Costa, “Comparison of
Overvoltage Mitigation Methods in Industrial Drives with Long
variation range of 25 – 50 % was found; Cables”, IEEE Internat. Symposium on Industrial Electronics, 2003.
(c) From (i) and (ii) it is thus recommended that the [12] H. de Paula, D. A. Andrade, M. L. Chaves, J. L. Domingos, M. A. A.
triangular arrangement is employed and special care is taken Freitas, “Methodology for Cable Modeling and Simulation for High-
Frequency Phenomena Studies in PWM Motor Drives”, IEEE
during the cable installation, preserving the intended Transactions on Power Electronics, pp. 744 – 752, March 2008.
configuration over the whole cabling length; [13] H. de Paula, M. L. Chaves, A. F. Moreira, T. F. Milagres, M. V. C.
(d) The common-mode results obtained without the Lisboa e J. F. R. Guilherme, “Experimental Investigation of Conveter-
presence of the transformer are way far from those originated to-Motor Cabling Influence on Transient Overvoltages and Common
Mode currents”, COBEP (Brazilian Conference on Power Electronics),
with its (accurate) inclusion; pp. 59-64, 2007.
(e) The dependency of the common-mode currents on the [14] H. de Paula et. al., “Novas Contribuições ao Estudo das Sobretensões
rise time is restricted to shorter cables and faster rise times; Transitórias e Correntes de Modo-Comum em Sistemas de
Acionamento de Motores por Inversores”, VIII Induscon (Brazilian
(f) Being the motor ungrounded, the total current that Conference on Industry Applications), 2008.
returns through the ground towards the transformer is lower; [15] H. de Paula, “Uma Nova Metodologia para a determinação dos
(g) The overvoltage and common-mode reduction Fenômenos de Alta Freqüência em Sistemas de Acionamento de
Motores de Indução Através de Inversores PWM”. Ph. D. Thesis, UFU,
associated with the cabling is much lower than that obtained Brazil, 2005.
with filters; however, there are no concerns involving costs, [16] Braskem Petroquímica, www2.braskem.com.br (produtos → boletins
weight, space or power losses. In this way, many applications técnicos → Propriedades de ref. dos compostos PVC).
will still require further mitigation of these high-frequency [17] G. Skibinski, D. Leggate and R. Kerkman, “Cable Characteristics and
Their Influence on Motor Overvoltages”, IEEE 12th APEC, vol. 1, pp.
phenomena, thus demanding the use of filters. In such 114-121, February 1997.
situations, special care in the cable installation can downsize [18] ABNT NBR 5410: Instalações Elétricas de Baixa Tensão (Brazilian
the required filter; standard for low voltage electrical installations – cable dimensioning),
2004.
(h) It was observed that the cable positive-sequence [19] L. Arnedo and K. Venkatesan, “High Frequency Modeling of Induction
inductance is lowered when the triangular geometry is used; Motor Drives for EMI and Overvoltage Mitigation Studies”, IEEE
therefore, aside from the afore-mentioned benefits, a lower IEMDC, pp. 24 – 29, Nov. 2003.
[20] Y. Weens, N. Idir, J-J. Franchaud and R. Bausière, “High Frequency
voltage drop over the cable is obtained as well. Modeling of an Adjustable Speed Drive”, 12th International EPE-
PEMC, pp. 456 – 461, August 2006.
ACKNOWLEDGEMENT [21] A. Kempski, R. Smolenski and R. Strzelecki, “Common Mode Current
Paths and Their Modeling in PWM Inverter-Fed Drives”, 33rd IEEE
The authors would like to thank FAPEMIG, for the PESC, pp. 1551 – 1556, June 2002.
finnancial support. [22] J. He, J. Jiang, J. Huang and W. Chen, “Model of EMI Coupling Paths
for and Off-Line Power Converter”, 19th IEEE APEC, pp. 708 – 713,
REFERENCES 2004.
[23] E. G. Villabona, P. S. Gúrpide, O. A. Sádaba and A. L. Azanza,
[1] J. M. D. Murphy, F. G. Turnbull, Power Electronic Control of AC “Simplified High-Frequency Model for AC Drives”, 28th IEEE IECON,
Motors, Pergamom Press, Second Edition, 1989. pp. 1144 – 1149, Nov. 2002.
[2] R. J. Kerkman, “Twenty Years of PWM AC Drives: When Secondary [24] B. Revol, J. Roudet, J. L. Schanen and P. Loizelet, “EMI Study of a
Issues Become Primary Concerns”, IEEE 22o IECON, vol. 1, pp. LVII- Three Phase Inverter-Fed Motor Drive”, IEEE 39th IAS Annual
LXIII, August 1996. Meeting, pp. 2657 – 2664, 2004.
[3] L. A. Saunders, G. L. Skibinski, S. T. Evon and D. L. Kempkes, [25] P. Sarma-Maruvada and N. Hylten-Cavallius, “Capacitance calculations
“Riding the Reflected Wave – IGBT Drive Technology Demands New for some basics high voltage electrode configurations”, IEEE
Motor and Cable Considerations”, IEEE 43rd IAS Annual Meeting, pp. Transactions on Power App. Syst., vol. PAS-94, no. 5, pp. 1708-1713,
75-84, September 1996. Sep./Oct. 1975.
[4] R. J. Kerkman, D. Leggate, G. L. Skibinsky, “Interaction of drive [26] F. de Leon and A. Semlyen, “Efficient calculation of elementary
modulation and cable parameters on AC motor transients”, IEEE parameters of transformers”, IEEE Transactions on Power Delivery,
Transactions on Industry Applications, vol. 33, n. 3, pp. 722-731, 1997. vol. 7, no. 1, pp. 376-383, Jan. 1992.
[5] M. Fenger, S. R. Campbell and J. Pedersen, “Dealing with Motor [27] A. R. De Conti, “Proteção de Redes Elétricas de Baixa Tensão:
Winding Problems Caused by Inverter Drives”, IEEE 44th IAS Annual Sobretensões Transferidas em Transformadores de Distribuição”. M.
Meeting, pp. 65-76, May 2002. Sc. Dissertation, Universidade Federal de Minas Gerais, Brazil, 2001.
[6] D. Leggate, J. Pankau, D. Schlegel, R. J. Kerkman, G. L. Skibinski, [28] J. A. Martinez, R. Wallking, B. A. Mork, J. Martin-Arnedo, and D.
“Reflected Waves and Their Associated Current”, IEEE Transactions Durbak, “Parameter Determination for Modeling System Transients –
on Industry Applications, vol. 33, n. 6, pp. 1383-1392, 1999. Part III: Transformers”, IEEE Transactions on Power Delivery, vol. 20,
[7] M. F. Rahman et al., “Modelling and Experimental Studies of Effects no.3, July 2005.
of Steep Fronted Inverter Waveforms on Motor and Supply Cabling [29] Bassi, W. et. al., “A simple transformer model for analisys of
and Their Remedies”, IEEE International Conference on Power transferred lightning surges from MV to LV lines”, International
Electronics and Drive Systems, vol. 1, pp. 519-524, 1999. Conference on Electricity Distribution, Nice, France, 1999.
[8] S. Suwankawin et al., “Destruction by Charge Current and Its Solution [30] Kanashiro A. G., Piantini A., Burani, G. F., “A methodology for
for PWM Inverter Feeding Multiple Motors Through Long Cables”, transformer modeling concerning high frequency surges”, Proceedings
IEEE 40o IAS Annual Meeting, pp. 1382-1389, 2005. of the International Symposium on Lightning Protection (VI SIPDA),
[9] A. V. Jouanne et al., “An Evaluation of Mitigation Techniques for pp. 275-280, São Paulo, Brazil, 2001.
Bearing Currents, EMI and Overvoltages in ASD Applications”, IEEE [31] Piantini A., Kanashiro A. G., “A distribution transformer model for
Trans. on Industry Applications, vol. 34, n. 5, pp. 1113-1122, 1998. calculating transferred voltages”, 26th International Conference on
[10] A. F. Moreira, T. A. Lipo, G. Venkataranaman, S. Bernet, “High Lightning Protection, pp. 429 – 434, Cracow, Poland, 2002.
Frequency Modeling for Cable and Induction Motor Over-Voltage [32] E. J. Bulington, S. Abney, G. L. Skibinsky, “Cable Alternatives for
Studies in Long Cable Drives”, IEEE Transactions on Industry PWM AC Drives Applications”, IEEE 46th Industry Applications
Applications, vol. 38, pp. 1297-1306, 2002. Society Conference, pp. 247 – 259, September 1999.

978-1-4244-4649-0/09/$25.00 ©2009 IEEE 1109