XIX International Conference on Electrical Machines - ICEM 2010, Rome

Braking torque analysis of the single phase line-

start permanent magnet synchronous motor
F. J. H. Kalluf, C. Pompermaier, M. V. Ferreira da Luz and N. Sadowski

Φ
Abstract — One of the most difficult procedures when conventional induction machine. Motor designers know
designing a single phase line-start permanent magnet that the LSPM is far different in terms of designing, if
synchronous motor is to predict the braking torque, mainly compared to induction motor, due to the effect of the
because it has multiple sources, which take place both in the magnets. The magnetic flux generated by the magnets
stator and in the rotor. This paper presents a way to explain the
braking torque sources and compares the torques obtained by must be high enough to provide a good synchronous
simulation with experimental results. torque, low enough to provide a good starting, and
correctly dimensioned to generate high efficiency.
Index Terms — Synchronous motors, induction motors,
permanent magnet, magnetic losses, squirrel cage motors, rare Torque
Induction
earth metals, neodymium, starting, eddy currents, rotating
machines. LSPM

I. INTRODUCTION

T HE Line-start Permanent Magnet (LSPM) motor for

years was not considered as an option for general-
purpose applications, mainly household appliances and
industrial motors due to a number of reasons. Some are
listed below:
• Cost of the NdFeB magnets: In the last 10 years, the Speed
rare-earth magnet price trend was really unpredictable,
Fig. 2. Induction versus LSPM asynchronous torque curve
which makes extremely dangerous to invest in
production lines dedicated to new permanent magnet
motors. Fig. 1 shows the 3-years price chart for • Manufacturing Issues: The main differences in the
Neodymium metal, were we can see the big difference production of an induction motor and a LSPM are
between the minimum and maximum price values. mostly in the rotor construction process. While the
Sometimes, increasing the copper volume has a lower induction rotor do not need any procedure other than
impact in the motor cost, and in these cases is difficult stamping and die casting, the LSPM rotor must be
to justify the use of permanent magnets. stamped, die casted, machined, the magnets should be
inserted and fixed, and then magnetized. These
procedures cost time and investment and can increase
the cycle time, reducing productivity.

Most problems that came up with LSPM cannot be solved

easily by a different design, but can be minimized if we
know how the constructive characteristics and the magnet
volume affect the performance (and cost).

Fig. 1. Neodymium price graph

II. INFLUENCE OF THE MAGNET VOLUME IN THE STARTING
(Source: metal-pages.com)
CURVE
• Starting: The effect of the braking torque in the motor As previously said, the magnet volume highly affects the
acceleration can be dangerously high, decreasing the motor starting performance. In fact, the higher is the flux
torque and generating torque oscillation. Fig. 2 shows generated by the magnets, the lower are the accelerating
an example of a LSPM torque curve, compared to a torques, as can be seen in Fig. 3.
Interesting to note that the shape and volume of the
This work was supported by Whirlpool SA Embraco Unit in Brazil.
F. J. H. Kalluf is with Department of Technological Development at Whirl- magnets and the way the magnet slots are distributed in the
pool S.A. - Embraco Unit, Joinville, SC, Brazil. rotor lamination have almost no influence over the starting
(e-mail: [email protected]). curve shape. Even changing the relation between Xd and Xq
C. Pompermaier is with Department of Technological Development at
Whirlpool S.A. - Embraco Unit, Joinville, SC, Brazil. (saliency ratio) makes no difference in the acceleration
(e-mail: [email protected]). torques, because of the high currents in the rotor cage, which
M. V. Ferreira da Luz is with Grupo de Concepção e Análise de creates a flux barrier and makes the inner rotor “invisible”
Dispositivos Eletromagnéticos at UFSC, Florianópolis, SC, Brazil.
(e-mail: [email protected]).
from the stator point of view. The shape of the rotor slots,
N. Sadowski is with Grupo de Concepção e Análise de Dispositivos however, has a considerable influence in the acceleration
Eletromagnéticos at UFSC, Florianópolis, SC, Brazil. curve, since it can limit the magnet flux, change the rotor
(e-mail: [email protected]).

978-1-4244-4175-4/10/$25.00 ©2010 IEEE

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY BOMBAY. Downloaded on April 06,2022 at 13:36:40 UTC from IEEE Xplore. Restrictions apply.
 cage resistance and change the amount of magnet flux First, we should note that, while tracing the braking
leakage. torque in the dynamometer (no voltage sources and short-
circuited windings), all the active power generated in the
motor as a whole will come from the magnet. This way, if
Low Magnet Volume we can calculate the current values in the stator and rotor at
Medium Magnet Volume this condition, it is possible to predict how the braking
Torque
High Magnet Volume
torque is divided according to the motor sources which
generate each torque component.

A. Source 1: Windings
As known, the magnet flux generates in the motor
windings an induced voltage proportional to the flux
magnitude and rotational frequency. This voltage produces a
current inversely proportional to the winding resistance and
reactance. This way, there will be a dissipated power in the
Speed winding resistance, which is proportional to the torque
Fig. 3. Influence of the magnet volume over the starting curve generated (Fig. 5). The torque generated in the winding (Tq’)
is transferred to the rotor, generating a negative torque (Tq).
III. SOURCES OF THE BRAKING TORQUE

The procedures described below to calculate the braking FMM

(magnets)
torque are related to single-phase motors which use both
start and run capacitors.
To measure the braking torque is relatively easy if you ωr

have an active dynamometer, where you can set the speed Tq’
manually. The usual way to do this is short-circuiting the
windings (note the auxiliary winding must be short-circuited
by the start capacitor). With this simple procedure is possible ωr
to trace a curve which represents the sum of all negative Tq

torques generated by the magnet flux (T-). If you can test the
same rotor with no magnets, in order to measure only the
positive torque (T+), is possible to achieve the resultant
torque, as can be seen in Fig. 4.

Positive Torque (alum. cage)

Negative Torque (magnets) Fig. 5. Torque generated by the windings
Torque Resultant Torque

The only problem regarding this approach is that you

cannot consider both windings separately, since this is a
single phase unbalanced motor. In other words, the main
winding and the auxiliary winding are coupled by the rotor
and the mutual inductances must be considered in the
equivalent circuit in order to calculate the winding currents.
Another important issue to care about is the calculation of
the magnetizing reactance, which is almost constant in the
Speed
conventional induction motor but has a considerable
Fig. 4. Positive, Negative and Resultant torques variation in the LSPM motor depending on the rotor
position, as can be seen in Fig. 6. In some LSPM motor
These 3 curves are the output of the standard starting designs the magnet slots are placed to avoid the high
curve test procedure for most LSPM manufacturers and are saliency ratio, and this is an effective way to reduce the
more than sufficient to evaluate the starting performance, at difference between the maximum and minimum magnetizing
least in terms of torque values. reactance values, but in terms of reducing the braking torque
But if we want to analyze the braking torque deeper, is it may be not the best solution.
possible to split it in at least 3 sources of negative torque, In fact, the correct calculation of the magnetizing
which can be studied separately as listed below: reactance is important only because the motor is not
a) Windings currents (main/auxiliary); balanced (either in terms of geometry or windings design)
b) Rotor cage currents; and the backward field generates rotor currents which couple
c) Eddy currents in the steel (stator and rotor). the two windings. It is interesting to mention that in 2-phase
motors with balanced windings the sensibility of the braking
Knowing the negative torque sources may help a design torque to the magnetizing reactance is much lower, since this
optimization, in order to increase the total resultant motor has no backward field at no-load condition.
acceleration torques without losing any other performance
features, as proposed in [7].

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY BOMBAY. Downloaded on April 06,2022 at 13:36:40 UTC from IEEE Xplore. Restrictions apply.
 main), but the resistance and reactance of the two phases are
LSPM Induction
not, and there will be a current flowing through the rotor
cage in order to balance this difference, as can be seen in the
equivalent circuit of Fig. 8.
Magnetizing Reactance

In the equivalent circuit, the magnet effect can be

simulated by a constant-amplitude, variable-frequency
current source (Imag), where the frequency is proportional to
the rotor speed.

0 60 120 180 240 300 360

Rotor Angle [°]
Fig. 6. Magnetizing reactance – LSPM versus induction motor
Fig. 8. Equivalent circuit – magnet effect
The first problem faced to define the magnetizing
reactance for analytical calculations is that you must choose Note that all the reactance values have to be adjusted
a constant value, and the natural choice is to choose the depending on the rotor speed, since they are proportional to
medium value of the curves in Fig. 6. Unfortunately, this is the EMF frequency.
probably not the best value due to the distortion of the flux The generated braking torque has 3 components:
waveform in the air gap. In fact, harmonic distortion in the
• the positive torque generated by the current in the
air gap flux is much higher in the LSPM if compared to the
direct component of the rotor resistance (Trd);
conventional induction motor, which reduces even more the
fundamental component of the flux. Since most LSPM • the positive torque generated by the current in the
analytical models use only the fundamental component, the quadrature component of the rotor resistance (Trq);
equivalent reactance is much lower than the medium value
for this type of motor. Finite element simulations show that • the negative torque generated by the current in the
while the medium value of the LSPM magnetizing reactance -Rr/2 resistance (Trneg).
is near half of the induction motor, taking only the Thus, the total negative torque generated by the currents
fundamental component give us values as low as 1/6 of that. in the rotor cage is:
With the correct magnetizing reactance value already 2
calculated, the braking torque due to the windings can be Ird 2 Rr Irq 2 Rr I mag Rr / 2
Tcage = + − (1)
calculated. Fig. 7 shows the calculated curves for main and 2 π fr 2 π fr 2 π fr
auxiliary windings of a 220V, 50Hz, 50W, 2-poles single-
phase LSPM motor with start and run capacitor: The negative resistance is used as a mathematical shortcut
to make the proposed equivalent circuit applicable for
0.5
Braking Torque (main)
current and torques calculation, not only by the magnets
Braking Torque (auxiliary) point of view, but also while calculating the induction motor
0.4
torques, windings resistive torques and eddy currents
resistive torque.
0.3
As can be seen by the equivalent circuit, in a balanced
TORQUE

motor the rotor currents Ird and Irq are equal and the sum of
0.2
torques Trd and Trq are equal to the negative torque Trneg,
and the total resistive torque is zero. In unbalanced motors,
0.1
however, Ird and Irq are different, and the value of Tcage will
not be zero. According to the proposed equivalent circuit, the
0
0
3
1×10 2×10
3
3×10
3 higher the unbalance, the higher the value of Tcage, and this is
theoretically correct.
SPEED [RPM]
The correct calculation of the rotor resistance is a key
Fig. 7. Braking torques due to the motor windings factor for the precise resistant torque calculation. Several
methods can be used and probably the most common are
B. Source 2: Rotor Currents shown in [1] and [2]. As can be seen in these references, the
rotor rings and rotor bars dimensions affect differently the
As the single-phase induction motor is not balanced, there rotor resistance and a good approach to check the precision
will be always a current flowing in the rotor cage, even when of these methods is to perform several consecutive tests with
the motor is at synchronous speed, due to the backward field different rotor rings, by progressively machining both rings
and by the magnets point of view, it is not different. before each test. This will show how much the resistant
Considering only the effect of the magnet, the motor is torque is affected by the rotor ring transversal area and will
always at synchronous (no-load) speed, and the winding give a clue about the sensibility of the rotor resistance to the
terminals can be seen as short-circuits. Note that the rotor ring dimensions.
magnetizing reactance can be considered the same in both Fig. 9 shows a set of curves, where the influence of the
windings (considering the auxiliary winding referred to the rotor cage rings dimension (in this case, the transversal area

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY BOMBAY. Downloaded on April 06,2022 at 13:36:40 UTC from IEEE Xplore. Restrictions apply.
 of the ring) is explored via simulation. IV. EXPERIMENTAL RESULTS
After the calculation of each portion of the total braking
Influence of the rotor ring dimension torque is possible to compare it with the measurements and
in the Cage Braking Torque x Speed curve check the precision of the proposed model.
0.16
Fig. 11 shows each individual calculated portion and the
0.14 total calculated braking torque, and Fig. 12 shows a
0.12
comparison between the calculated curve and the real curve
measured in the dynamometer.
Braking Torque

0.1

0.08 0.7
Total Braking Torque
0.06 100mm² Main Winding Braking Torque
0.6 Auxiliary Winding Braking Torque
80mm²
0.04 Rotor Cage Braking Torque
60mm² Eddy Currents Braking Torque
0.02 40mm² 0.5

BRAKING TORQUE
20mm²
0
0.4
0 500 1000 1500 2000 2500 3000 3500

SPEED [RPM]
0.3
Fig. 9. Braking torque due to the rotor cage currents
0.2
C. Source 3: Steel Eddy Currents
0.1
Probably the most difficult task while modeling the
behavior of the electrical steel when submitted to the magnet 0
0 1000 2000 3000
rotating field is the fact that the rotor and the stator eddy
SPEED [RPM]
current losses do not vary the same way. While the stator
Fig. 11. Individual and total braking torque (calculated)
losses are proportional to the rotating speed, the rotor eddy
current losses are proportional to the unbalance of the stator
currents, which cause flux variations in the rotor. A way to 0.7
Braking Torque (Calculated)
verify the influence of each one is using experimental tests Braking Torque (Measured)
0.6
to separate stator and rotor portions of the braking torque, by
constructing special rotors, as can be seen in Fig. 10. 0.5
BRAKING TORQUE

0.4

0.3

0.2

0.1

0
0 1000 2000 3000

SPEED [RPM]
Fig. 10. Rotors for eddy current tests
Fig. 12. Comparison between calculated and measured braking torques
The rotor shown in the left side of Fig. 10, constructed
without the aluminum cage, will generate the same magnetic V. CONCLUSIONS
flux as the rotor with the full cage (right side), but there will The proposed model is a valid methodology to simulate
be no coupling between the two phases and the braking the LSPM motor braking torques. The precision of the
torque of both phases will be independent. Using the calculation is highly sensible to the magnetizing reactances
difference in the resistive torque for the two tested rotors, the and rotor cage equivalent resistance values. In order to
braking torque generated by the stator eddy currents and the generate good results, the method requires that these values
predicted results for the cage torque (from the equivalent are correctly calculated.
circuit), is possible to calculate the value of the braking The use of a dynamometer can be an important addition
torque generated only by the eddy currents in the rotor steel. to increase the precision of the calculation, since some of the
Once separated this portion of torque, the equivalent circuit model parameters are not dependant only on the rotor
can be completed using a resistance in parallel with the geometrical characteristics, but also on the interaction
magnetization reactance, which should be equal for both between the rotor materials (electrical contact resistance
phases. This procedure should be made once for each rotor between rotor cage and rotor steel).
geometry to extract the steel loss equivalent resistance.
VI. ACKNOWLEDGMENT
It is important to add that the rotor steel eddy currents are
also affected by the contact resistance between the aluminum This work was supported by the Department of Technolo-
cage and the rotor steel. Good quality rotors have a high gical Development at Whirlpool S.A. - Embraco Unit, Bra-
contact resistance, which can generate lower losses in the zil.
rotor, contributing for a low braking torque and a better
efficiency at normal operation.

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY BOMBAY. Downloaded on April 06,2022 at 13:36:40 UTC from IEEE Xplore. Restrictions apply.
 VII. REFERENCES group. In 1989 he went left to France to prepare his Doctoral thesis in the
Laboratoire d´Electrotechnique et d´Electronique Industrielle (LEEI) in
Books: Toulouse. He finished his thesis in 1993, his work receiving an honor
[1] Stepina, J., Single-Phase Induction Motors: Construction, Theory and award. Returning to Brazil he joined again the GRUCAD group where he
Calculation, Magna Physics Publishing, 2006. was the advisor for several M.Sc. and Doctoral theses, coordinating gradu-
[2] Veinott, C. G. and Martin, J. E., Fractional and Subfractional ate courses and participating at several works with Brazilian and foreign
Horsepower Electric Motors, New York: McGraw-Hill Inc., fourth companies. In 2002 he received his “Habiliation à Direction des Re-
edition, 1986. cherches” diploma from the Institut National Polytechnique de Toulouse
(France). In 2003 the published with his colleague Prof. João Pedro As-
Papers from Conference Proceedings (Published): sumpção Bastos the book entitled “Electromagnetic Modeling by Finite
Element Methods” by Marcel Dekker Inc., New York. He worked on sever-
[3] Knight, A. M. and McClay, C. I., The Design of High-Efficiency al research projects financial supported by Brazilian federal research coun-
Line-Start Motors, IEEE Transaction on Industry Applications, Vol. cils (FINEP, CNPq) as well by companies. He participated of the organiza-
36, No. 6, Nov/Dec 2000, pp. 1555-1562. tion of CBMAG-1987, CBMAG-1995, COMPUMAG-1997 conferences
[4] Williamson, S. and Knight, A. M., Performance of Skewed Single- and was, with Prof. José Ricardo Bergmann (CETUC PUC-Rio), the general
Phase Line-Start Permanent Magnet Motors, IEEE Transaction on In- chairman of MOMAG 2008 (Microwaves and Magnetics) conference held
dustry Applications, Vol. 35, No. 3, May/June 1999, pp. 516-525. in Florianópolis. He was the vice-chairman of COMPUMAG-2009 confe-
[5] Zhou, J. and Tseng, K., Performance Analysis of Single-Phase Line- rence which held in Florianópolis at November 2009. Topics of interest:
Start Permanent-Magnet Synchronous Motor, IEEE Transaction on numerical methods in electromagnetics, 2D and 3D finite element method,
Energy Conversion, Vol. 17, No. 4, Dec 2002, pp. 453-462. electromagnetic devices (design, analysis and applications), electrical ma-
[6] Popescu, M., et al. Asynchronous Performance Analysis of a Single- chines, and magnetic material losses.
Phase Capacitor-Start, Capacitor-Run Permanent Magnet Motor,
IEEE Transaction on Industry Applications, Vol. 20, No.1, March
2005, pp. 142-150.

Dissertations:
[7] Kalluf, F. J. H., Análise do funcionamento assíncrono de um motor
síncrono de ímãs permanentes com partida direta (Portuguese).
Master’s Degree Dissertation at UFSC, Florianópolis, SC, Brazil,
2010.

VIII. BIOGRAPHIES

Flavio Jorge Haddad Kalluf was born in C. Procopio, Paraná, Brazil in

1971 and graduated as Electrical Engineer at Federal University of Santa
Catarina (UFSC) in 1995 and obtained his M.Sc. at UFSC on March 2010
with an analytical modeling of the asynchronous operation of a line-start
permanent magnet synchronous motor (LSPM). Since 1995 he works at
R&D department of Whirlpool/Embraco in Joinville, Brazil, as a motor de-
velopment engineer. His fields of interest are the design of induction ma-
chines, line-start synchronous motors and brushless permanent magnet mo-
tors. Have several applied patents in the last years on induction, LSPM and
brushless motor design. IEEE and ICS member since 2008.

Cristófaro Pompermaier was born in Foz do Iguaçu, Paraná, Brazil, on

September 29, 1979. He graduated as Bachelor of Electrical Engineering
from the Santa Catarina State University, Joinville, 2005 and obtained the
Master’s Degree Diploma in linear motors at Federal University of Santa
Catarina, Florianópolis, 2009, Brazil. He works as an electrical motor re-
searcher engineer at Whirlpool SA Embraco Unit in South Brazil, since
2004. The main activities are related to Brushless DC motors, Linear Mo-
tors and Single-Phase Induction motors applied to Hermetic Compressors.
IEEE and ICS member since 2008.
Mauricio Valencia Ferreira da Luz was born in Jaguariaíva, Paraná,
Brazil in 1973. His electrical Engineering diploma was obtained at the State
University of Santa Catarina (UDESC) in 1997. He finished his master de-
gree in 1999 at the Federal University of Santa Catarina (UFSC), in the field
of energy systems. The subject was related to the studies and development
of a method to stray load losses calculation in induction machines. In 1999
he started his doctorate thesis at UFSC. In between April 2001 to October
2002 he studied at University of Liège (Belgium). His doctorate finished in
GRUCAD in the field of energy systems in 2003. Its main issue was the de-
velopment of a software for 3D electromagnetic field calculation using edge
elements, taking into account the movement and the feeding circuit. Be-
tween 2003 to 2005, he was a professor at the University Center of Jaraguá
do Sul (Santa Catarina, Brazil). In December 2005, he became a permanent
member at the Department of Electrical Engineering at UFSC. He has also
been working in some research projects financially supported by the Brazil-
ian Federal Research Councils (FINEP - CNPq) as well as by private com-
panies. He helped on the organization of MOMAG 2008 (Microwaves and
Magnetics), a conference held in Brazil. He helped on the organization of
COMPUMAG 2009, a conference held in Florianópolis at November 2009.
Topics of interest: analytical and numerical methods in electromagnetics,
coupled problems, material modeling, analysis and design of electrical ma-
chines and power transformers.
Nelson Sadowski was born in São Bento do Sul (27/03/1959). He obtained
his Electrical Engineer degree at Universidade Federal de Santa Catarina
(UFSC) in 1982 and his M.Sc. degree in 1985 also at UFSC. In November
1985 he obtained a teaching position at the same university as an Assistant
Professor and started his research activities in the newly created GRUCAD

Authorized licensed use limited to: INDIAN INSTITUTE OF TECHNOLOGY BOMBAY. Downloaded on April 06,2022 at 13:36:40 UTC from IEEE Xplore. Restrictions apply.