You are on page 1of 9

TRANSACTIONS ON ELECTRICAL AND ELECTRONIC ENGINEERING

IEEJ Trans 2007; 2: 109–117


Published online in Wiley InterScience (www.interscience.wiley.com). DOI:10.1002/tee.20117

Review

Parameter Measurements for Permanent Magnet Synchronous Machines

Minoru Kondo∗a , Member

The advances in inverter drive technology and permanent magnet materials have enabled the application of
permanent magnet synchronous machines (PMSMs) to various fields. PMSMs are rapidly spreading in their
applications with their advantage of high efficiency, which is highly needed in many applications in recent
years. Although there have been standardized measurement methods of the machine parameters for synchronous
machines, they are not intended to apply to PMSMs. Therefore, several new measurement methods suitable for
PMSMs have been proposed. This paper presents a review and analysis of these methods.  2007 Institute of
Electrical Engineers of Japan. Published by John Wiley & Sons, Inc.

Keywords: parameter measurement, permanent magnet synchronous machines

Received 11 September 2006; Revised 29 November 2006

1. Introduction circuit, which consider the winding loss R and iron-


loss resistance Rc . Although many types of equivalent
The advances in inverter drive technology and per- circuits have been presented in the papers introduced
manent magnet materials have enabled the application later, Figures 1 and 2 are sufficient to explain the sug-
of permanent magnet synchronous machines (PMSMs) gested method of determination of the parameters of
to several fields. The PMSM is rapidly spreading its PMSMs experimentally. In this equivalent circuit, the
application with its advantage of high efficiency which leakage reactance is lumped into the synchronous reac-
is highly needed in many applications in recent years. tance because it is impossible to isolate the leakage
Although the PMSM is a type of synchronous machine, reactance from the synchronous reactance. Besides, the
many of the conventional test methods [1] for syn- circuit elements that represent the reactance are divided
chronous machines are not applicable to PMSM, because into two elements, a resistance Rr , which represents the
the permanent magnetic field cannot be changed easily. reluctance torque output, and a reactance Xs which rep-
Therefore, some new parameter measurement methods resents the reactive power. Although these two elements
are proposed, and in some cases, different methods are are not presented in any paper introduced later, the phasor
used to measure the same characteristics. Therefore, it diagram is the same as that presented in those.
is useful to review the measurement methods proposed The parameters used in this paper are based on
and to demonstrate the differences between them. In this the classical two-reaction analysis and they are the
paper, the parameter measurement methods that have permanent magnet induced EMF E, d-axis reactance
been proposed are reviewed and analyzed, and further Xd , and q-axis reactance Xq . In addition, the iron-
possibilities are also demonstrated. loss resistance Ri is also considered. Therefore, this
paper deals with only steady-state characteristics, and
transient characteristics are not covered. A salient pole,
2. Phasor Diagram, Equivalent circuit, and star-connected, polyphase machines are assumed in the
Parameters of the PMSM analysis in this paper. The reason why a salient pole
machine is assumed is because many PMSMs are of the
Figures 1 and 2 show respectively the phasor dia- interior magnet type possessing saliency.
gram of the PMSM [2] and corresponding equivalent

3. Testing Sets
a Correspondence to: Minoru Kondo. E-mail: min@rtri.or.jp
∗ Railway Technical Research Institute, 2-8-38, Hikari-cho, Koku- Practical test methods depend on the available test
bunjishi, Tokyo 185-8540 equipment. For example, it is often difficult to perform a

 2007 Institute of Electrical Engineers of Japan. Published by John Wiley & Sons, Inc.
M. KONDO

q-axis
RIa
XqIq
PMSM PM
X d Id
RrI (a) Testing set for generator test
Xs I
Ia
E
V SG PM
Ei
Iq
d jd
PMSM L
j I

(b) Testing set for no-load test


Ic Id
d-axis VSI
Fig. 1 Phaseor diagram of a PMSM

Ia R I Rr Xs S PMSM L

Ic (c) Testing set for load test


PMSM: Permanent magnet synchronous machine (tested machine),
V Ei E PM: Prime mover, SG: Synchronous generator,
Rc
L: Load (dynamometer), S: Rotational position sensor

Fig. 3 Examples of the testing sets

R: winding resistance, Rc: iron loss resistance, PMSM as a motor, an inverter drive is needed to start
E: permanent magnet induced EMF,
the run because the starting frequency is zero. However,
Xd – Xq Xd + Xq Xd – Xq once the machine under test has started to rotate, it
Rr ≡ sin 2jd, Xs ≡ – cos 2jd
2 2 2 can be driven by a sinusoidal alternating voltage source.
Xd: d-axis reactance, Xq: q-axis reactance Therefore, when performing a test with this set, the test
procedure is as follows: At first, the tested machine is
Fig. 2 Equivalent circuit per phase of PMSM started with an inverter drive system. Then, the source is
switched to a sinusoidal source. Another starting method
load test for a large machine, because the test requires a is manually rotating it with a belt wound on the shaft end.
larger power source and load. Therefore, it is important to When the PMSM has started to rotate, it is connected to
review the testing sets before discussing the test method. sinusoidal voltage source with a frequency low enough
Figure 3 show the list of testing sets in the test methods to synchronize. After that, the frequency is increased to
presented in this paper. the test frequency.
In the testing set presented as Fig. 3(a), tested machine In testing set (c), the tested machine is driven by an
is driven by a prime mover and run as a generator. No inverter as a motor with shaft load. Although a loaded
load is prepared in this set. The prime mover output test can be performed in the testing set (b) with some
power needed is approximately the total loss of the tested shaft load added, the available range of current vectors
machine. Therefore, this testing set is suitable for larger is limited with a sinusoidal voltage source that does not
machines. use feedback control. Therefore, the most meaningful test
In testing set (b), the tested machine is run as a motor results are expected in the testing set (c) because PMSMs
without shaft load from a voltage source. One example of are controlled by inverters in the most applications.
the voltage source is a synchronous generator driven by On the other hand, it is sometimes difficult to prepare
a prime mover, which can generate sinusoidal alternating an inverter and a shaft load for the test of larger machines.
voltage with adjustable frequency. When running the Besides, in the case of PMSM sold without any inverter,

110 IEEJ Trans 2: 109–117 (2007)


PARAMETER MEASUREMENTS FOR PMSM

since they can be driven by different inverters, it is Id


E
very important to measure the characteristics of rotating Id =
machine itself without any influence of the inverter Xq – Xd
I = Imin
control. Therefore, the tests in testing set (a) or (b) are
also useful and needed. Iq
Current vector
A C
in no-load test
D
4. Parameter Measurements
Ei = E
4.1. Induced EMF E In the testing set (a), the
induced EMF E can be measured as the open circuit
voltage of the armature winding when the tested machine
is driven as a generator by a prime mover (at A in Fig. 5). E
B Id = –
A Fourier analysis has to be carried out to extract the Xd
cosj = 1
fundamental component when the harmonic waves in the
voltage waves are not negligible. Ei = 0
In testing set (b), the tested machine is run without Fig. 5 Current vector diagram of PMSM
any shaft load at a fixed rotational speed. In addition,
the phase current is measured as the varying applied
In addition, if the mechanical loss is negligible compared
voltage. Plotting the measured current versus the voltage,
to the iron loss, Mellor claims that the core loss resistance
a curve such as the one shown in Fig. 4 can be
Rc also can be determined as
obtained. Yamamura [3] called this the ‘New V-curve’
and proposed to determine the reactance of the PMSM
Rc = E/Ia (2)
from this curve as introduced later. Nee [4] proposed to
read the terminal voltage corresponding to the minimum
In testing set (c), the PMSM is run under a constant
current point (D in Fig. 5) as the induced EMF, assuming
speed, and the terminal voltage is measured while
that the induced EMF equals the measured terminal
varying the q-axis current and keeping the d-axis current
voltage (and the winding resistance is negligible). Mellor
zero. By plotting the terminal voltage versus the current,
[2] proposed almost the same method. In this method,
a characteristics curve can be obtained. The induced EMF
the PMSM runs at unity power factor (at C in Fig. 5, the
can be determined from the intercept [5].
current expected to be almost minimum), the induced
In the latter two methods, the induced EMF is
EMF is determined using (1)
determined from the terminal voltage when the PMSM is
run as a motor at nearly zero current with a power factor
E = V − R Ia (1) of unity. Therefore, these methods are also based on the
same idea as the conventional measurement method [1].
In addition, the measurement of open circuit voltage
cosj = 1 can be performed in testing sets (b) and (c) also. It can
be performed during the retardation of the PMSM.
V = Xq I When performing these tests, it is very important
to clarify the temperature of the permanent magnets,
because their characteristics change with the temperature.
Current I Power factor cosj

Power factor cosj


If possible, it is desirable to measure the temperature
with a thermometer (or a thermocouple). The maximum
value of the open circuit voltage is important from the
Current I
viewpoint of inverter protection, and the minimum value
of induced EMF is important from the viewpoint of
performance guarantee. Therefore, it is recommended to
DI
measure the open circuit voltage at the lowest and the
DV highest permanent magnet temperatures. For example,
Minimum current
the measurement at the lowest temperature is performed
when the rotor temperature is assumed to be the same
as the ambient temperature and the measurement at the
Voltage V
highest is performed right after the temperature rise test.
Fig. 4 New V-curve of PMSM under no load The measurement of the permanent magnet temperature

111 IEEJ Trans 2: 109–117 (2007)


M. KONDO

is also important in the following tests. When the and


direct measurement is not available, an alternative is the 
measurement of the open circuit voltage before and after Id = Ia2 − Ei2 /Rc2 (10)
the test.
If the effect of the losses are neglected, the d-axis
4.2. The d-axis reactance Xd In testing set (a),
reactance can be written simply [4] as:
the d-axis reactance Xd is determined using the result
of the short circuit test. In the short circuit test, the V −E
PMSM is driven as a generator by a prime mover with the Xd = (11)
Ia
terminals of armature winding connected to each other
(at B in Fig. 5). The d-axis reactance is determined from or
the short circuit current Is measured during the test and
the induced EMF E. V
Xd = (12)
I
Xd = E/Is (3)
The trajectory of current vector during the test is shown
in Fig. 5. If the mechanical loss is negligible, the
This equation is deduced as follows:
trajectory agrees with the Id -axis around the minimum
During the short circuit test, the voltage is kept zero.
current point. In that case, the q-axis component of the
Therefore, the voltage equation of PMSM can be written
current and flux linkage can be neglected. Therefore, the
as:
d-axis reactance can be determined by (11) or (12).
In testing set (c), the d- and q-axis reactance is
R Id − Xq Iq = 0 (4)
determined from the current, voltage, power factor, and
R Iq + E + Xd Id = 0 (5) torque angle δ during the loaded test [7].
From the phasor diagram the d- and q-axis reactances
Solving these equations yields [6] (see Appendix) are:
 V cos δ − E − R Iq
E R 2 + Xq2 Xd = (13)
Id
Is = (6)
R 2 + Xd Xq V sin δ + R Id
Xq = (14)
Iq
Here, assuming that the voltage drop due to the winding
resistance is negligible, (3) can be obtained from (6) If core loss is neglected, the d- and q-axis currents are
In testing set (b), the d-axis reactance is determined
from the no-load test result. Id = Ia sin(ϕ − δ) (15)
Yamamura [3] proposed to determine the d-axis
impedance from inclination of linear portion in New Iq = Ia cos(ϕ − δ) (16)
V-curve shown in Fig. 4 as:
Although V , Ia , and ϕ can be measured using a digital
V power meter, the torque angle δ should be measured
Zd = |Rd + j Xd | = (7)
I separately. If a rotational position sensor is available,
the angle can be measured.
However, the d-axis reactance cannot be determined with However, some special treatment should be carried
this equation, because it was not shown how to determine out to eliminate the influence of the harmonic compo-
Rd . nent in the measured current and voltage, because the
Mellor [2] proposed to determine the d-axis reactance PMSM is usually driven from an inverter in this testing
Xd from the no-load test result considering the winding set. Morimoto [8] proposed to measure the fundamental
resistance and core loss resistance: component of current and voltage using a fast Fourier
transform (FFT) analyzer and calculate the d-axis reac-
|Ei − E|
Xd = (8) tance using the equations above. The permanent magnet
Id flux linkage and the winding resistance should have been
already obtained from other tests. Obviously, if the volt-
where age source driving the PMSM is not an inverter but a
sinusoidal voltage source, the d-axis reactance can be
Ei = V − Ia R cos ϕ (9) determined without an FFT analyzer.

112 IEEJ Trans 2: 109–117 (2007)


PARAMETER MEASUREMENTS FOR PMSM

Although, this method needs a rotational position normally operated with negative d-axis current and the
sensor and an FFT analyzer, this is the most admirable d-axis flux path is not saturated.
measurement of the reactance, because the reactance of There are measurement methods for the reactance that
the loaded machine can be measured directly. are achieved by the locked rotor test. A conventional
method is a locked rotor test with a low-frequency
4.3. The q-axis reactance Xq It is desirable that alternating current. Miller [7] proposed a locked rotor test
the q-axis reactance Xq is measured in a load test, using an inductance bridge to prevent the influence of the
because it varies as the current due to saturation. In voltage induced by the alternating current. However, this
testing set (c), the q-axis reactance can be determined method requires a relatively complex testing set.
in the same manner as d-axis reactance using (14). It is required to pass the q-axis current to measure the
Methods have been proposed with some special operating q-axis reactance, and the PMSM generates a torque when
conditions, such as keeping Id = 0 [2], [9] and keeping the q-axis current is passed. Therefore, it seems difficult
δ = 90◦ (Id = E/Xd ) [9]. to measure the q-axis reactance in the testing set (a) and
In the case that δ = 90◦ (b). However, theoretically it may be possible to measure
the q-axis reactance by the following methods.
V + R Ip V − R Id
Xq = = (17) In testing set (a), the q-axis reactance can be calculated
Ir Iq from the results of short circuit test.
As mentioned above, the short circuit current is
where, Ip is the input power component and Ir is the expressed with (6). If the rotational speed is high enough
reactive component of input current I . Resolving Id and to neglect the voltage drop due to the winding resistance,
Iq onto the p- and r-axes, the short circuit current is expressed as E/Xd and it does
not vary with the rotational speed but becomes constant.
Ip = Iq cos δ − Id sin δ (18) However, when operating at a small rotational speed,
Ir = Iq sin δ + Id cos δ (19) it does not become so. Therefore, after determining the
d-axis reactance with the short circuit test, the q-axis
In the case that Id = 0 reactance can be determined using (6) from the result of
the lower-speed short circuit test. Substituting nE, nX d
V sin δ V sin ϕ and nX q E, for Xd and Xq and solving (6) for nX q yields
Xq = = (20)
Iq Ia
λXq
However, these methods are not attractive since the 
λXd (RIs )2 + λER (RIs )2 + (λXd Is )2 − (λE)2
q-axis reactance measured under actual operating con- =
dition is the most important. (λE)2 − (λXd Is )2
(24)
While these methods require the measurement of the
rotational position, Nee [4] suggested a method that does
Here, λ is the ratio of low-speed test frequency to the
not require a rotational position sensor to determine
rated frequency.
q-axis reactance. He suggests that the angle δ be
In testing set (b), the q-axis reactance can be deter-
calculated from the load test result using following
mined from the no-load test result [10]. In the no-load
equations. These equations can be obtained by solving
test, as the voltage increases, the d-axis current comes
the equation concerning the q-axis components.
close to a fixed value (Fig. 4 and Fig. 5)
1   
cos δ = E B − B 2C 2 − C 2E2 + C 4 E
B2 +C 2 Id = (25)
Xq − Xd
(21)
B = V − Xd I sin ϕ − R I cos ϕ (22) In this condition, the q-axis reactance can be determined
from the voltage and the current:
C = Xd I cos ϕ − R I sin ϕ (23)
Ei
After the angle is determined, the q-axis reactance can be Xq = (26)
I
calculated using (14). In this calculation, Xd should be
determined by another test prior to the determination of and if the losses are disregarded,
Xq , and it is assumed that the Xd does not vary between
these tests. This assumption is reasonable in the case V
Xq = (27)
of the interior permanent magnet machine, because it is Ia

113 IEEJ Trans 2: 109–117 (2007)


M. KONDO

Table I. Comparison of measurement methods

Test Ref. Eqn. Approximation error Testability

E Open circuit generator test (1) — None Easy


No-load test (2) (1) Mechanical loss is neglected Easy
(4) — Losses are neglected Easy
Xd Short circuit test (1) (3) Xd is assumed to be constant Easy
No-load test (2) (8) Mechanical loss is neglected Easy
(3) (12) Losses are neglected Easy
(4) (11) Hysteresis losses are neglected Easy
Locked rotor test (7) — Effects affects the result Special testing set is needed
Load test (7) (13) None Rotor position measurement is needed
Xq Short circuit test (6) (24) Parameters are assumed to be constant Easy
No-load test (10) (27) Parameters are assumed to be constant Easy
Losses are neglected
Locked rotor test (7) — Histeresis effects affects the result Special testing set is needed
Load test (4) (21) Xd is assumed to be constant Rotor position measurement is not needed
(7) (14) None Rotor position measurement is needed

This method can determine the q-axis reactance roughly the semi-input and the second power of the internal EMF,
but easily. and according to this method, it becomes possible to
The comparison between the test methods are shown separate the iron loss from no-load loss. The second
in Table I. There is a trade-off between the accuracy and power of internal the EMF can be calculated from
testability. following equation:

4.4. Iron-loss resistance Ri The iron-loss char-


acteristic is determined from the no-load examination Ei2 = V 2 − 2 R Pin /m + R 2 Ia2 (29)
result performed using test equipment (b).
At the time of a no-load test, the input electric power The example of a no-load examination result is shown
becomes equal to the generating loss in an electric motor. in Fig. 6. The tested motor is a 140 kW totally enclosed
Among these losses, the stator copper loss can be PMSM [12].
determined from winding resistance and the current According to the figure, the semi-input is increasing
measurement result, and when this is deducted from the linearly to the 2nd power of internal electromotive power,
input, electric power Pin becomes the sum of other losses and iron-loss resistance can be determined from the
and it is hereafter called semi-input Psi slope.

Psi = Pin − m R Ia2 (28) (Ei2 )


Ri = (30)
Psi /m
where, m is the number of phases.
The remaining losses can be considered to be an iron
loss and a mechanical loss, and when the mechanical loss 2.5
can be disregarded, the semi-input can consider them to
be the value of an iron loss. 2
Mellor [2] proposed to determine iron-loss resistance
by this approach, and the formula used in the method 1.5
Psi (kW)

have already been shown in (2).


Yamamura [3] also proposed to determine iron-loss 1
resistance by the same approach.
However, in an actual electric motor, a mechanical loss 0.5
cannot be ignored in many cases with respect to the iron
loss, and one needs to separate each loss from the sum 0
0 50000 100000 150000 200000
of the iron loss and the mechanical loss. Ei2 (V2)
Urasaki [11] has proposed the method of determining
iron-loss resistance from inclination of a graph plotting Fig. 6 Semi-input power versus square of Ei

114 IEEJ Trans 2: 109–117 (2007)


PARAMETER MEASUREMENTS FOR PMSM

Moreover, since the iron loss becomes zero when the 6 No-load loss (no-load test)
internal EMF is zero, the value of the section of the semi-
No-load loss (retardation test)
input becomes the mechanical loss. Thus, the mechanical 5
Mechanical loss (no-load test)
loss can be determined.
4 Mechanical loss (retardation test)
This method of determination is fundamentally the

Loss (kW)
same as that for induction motors.
3
However, since a PMSM will become unstable if the
voltage is made small in a no-load test, it is difficult to 2
conduct the test on a voltage small enough to determine
the mechanical loss with sufficient accuracy. 1
Urasaki’s iron-loss resistance measuring method is
suitable for the permanent magnet synchronous machine, 0
0 1000 2000 3000 4000 5000
since it determines the iron-loss resistance from the slope Rotational speed (/min)
of the semi-input and the test with small voltage is
Fig. 7 Comparison of no-load loss measurement
unnecessary.
There are also iron-loss measurement methods that do
not pass current in stator winding. estimated from the interpolation of the loss characteristics
In the testing set (a), if the torque T when rotating a curve using an exponential function as the interpolating
PMSM by a prime mover can be measured, the no-load function.
loss Pnl can be determined by the following formula from
a mechanical input: [13] 5. Conclusion

Pnl = ωm T (31) The parameter measurement methods for PMSM are


reviewed in this paper. There are various measuring
where ω m is the rotational speed. methods that have merits and demerits in respect of
Moreover, also by the test equipment (b), the no-load accuracy or the ease of measurement. Since the ease of
loss can be determined from the moment of inertia of the carrying out each test or the loss that can be disregarded
rotor J and rotation deceleration in a no-load retardation differs, the optimal test method for different designs of
test result [6]. PMSMs will be different. It was apparent that the no-
load test is very important in the parameter measurement
d ωm
Pnl = ωm J (32) of PMSM, because every parameter can be determined
dt from the no-load test result. The author hopes this paper
will help selecting the test method.
By either of the methods, what can be determined is the
no-load loss, and the iron loss and the mechanical loss
are not separated. References
However, if it is an interior permanent magnet type
motor, the mechanical loss can be determined from (1) IEC 34-4 Rotating electrical machines Part 4: Methods for
the same test as for a machine without the magnets. determining synchronous machine quantities from tests 1995.
Therefore, the iron loss can be calculated by subtracting (2) Mellor PH, Chaaban FB, Binns KJ. Estimation of Parameters
the mechanical loss from the no-load loss measured. and performance of rare-earth permanent-magnet motors
avoiding measurement of load angle. IEE Proceedings-B 1991;
Furthermore, the iron-loss resistance can also be
138(6):322–330.
computed by dividing the second power of the induced (3) Yamamura S, Takano S. New V-curve of the Permanent-Magnet-
EMF by the iron loss. Excited Synchronous Motor, IEE Japan Annual Conference 2001,
2001; 1809–1910 (in Japanese).
E2 (4) Nee H-P, Lefevre L, Thelin P, Soulard J. Determination of d
Ri = (33) and q Reactances of Permanent-Magnet Synchronous Motors
Pnl /m Without Measurements of the Rotor Position. IEEE Transactions
on Industry Applications 2000; 36(5):1330–1335.
The no-load test result compared with the no-load (5) Senjyu T, Kuwae Y, Urasaki N, Uezato K. Loss and Parameter
retardation test result the author performed is shown Measurement Method for Low power Permanent magnet
in Fig. 7. As shown in this figure, both results are Synchronous Motors, IEE Japan Conference Record, RM-00-161,
consistent. The value of an iron-loss measurement results 2000; 19–24 (in Japanese).
(6) Kondo M, Kawamura J, Terauchi N. Energy Consumption
in the no-load test has been determined by subtracting Calculation of Permanent Magnet Synchronous Motor for
the mechanical loss from the semi-input at the minimum Railway Vehicle Traction Using Equivalent Circuit. IEE Japan
current point. In addition, the mechanical loss has been 2005; 125(4):313–320 (in Japanese).

115 IEEJ Trans 2: 109–117 (2007)


M. KONDO

(7) Miller TJE. Methods for testing permanent magnet polyphase AC Vr = Xd Id sin ϕδ + Xq Iq cos ϕδ
motors. IEEE, IAS 1981; 23D:494–499.
(8) Morimoto S, Takeda Y, Hirasa T. Parameter Measurement of = Xd I sin2 ϕδ + Xq I cos2 ϕδ
PM Motor in dq Equivalent Circuit. IEE Japan 1993; 113-
D(11):1330–1331 (in Japanese). 1 − cos 2ϕδ 1 + cos 2ϕδ
(9) Chalmers BJ, Hamed SA, Baines GD. Parameters and perfor-
= Xd I + Xq I (36)
2 2
mance of a high-field permanent-magnet synchronous motor  
for variable-frequency operation. IEE Proceedings-B 1985; Xd + Xq Xd − Xq
= − cos 2ϕδ I
132(3):117–124. 2 2
(10) King TL, Novotny DW. Discussion contribution. IEEE Transac-
tions on Power Apparatus and Systems 1980; PAS-99(4):1517. In a equivalent circuit, a resistor and a reactance can
(11) Urasaki N, Senjyu T, Uezato K. A Novel Calculation Method for
represent the active and reactive power, respectively.
Iron Loss Resistance Suitable in Modeling Permanent-Magnet
Synchronous Motors. IEEE Transactions on Energy Conversion Therefore, the author introduced a resistor Rr and a
2003; 18(1):41–47. reactor Xs defined in Fig. 2, and (35) and (36) can be
(12) Matsuoka K, Kondo M, Shimizu Y. A Totally Enclosed Traction rewritten as
Motor Using a Permanent Magnet Synchronous Motor. Electrical
Engineering in Japan 2005; 151(3):71–80. Va = Rr I (37)
(13) Slemon GR, Liu X. Core Losses in Permanent Magnet Motors.
IEEE Transactions on Magnetics 1990; 26(5):1653–1655. Vr = Xs I (38)

Minoru Kondo (Member) received the Masters’s degree Finally (33) can be written as
in mechanical engineering, and the
Ph.D. degree in electrical engineer- j Xd I˙d + j Xq I˙q = V̇a + j V̇r = (Rr + j Xs )I˙ (39)
ing, both from the University of
Derivation of equation (6) Inserting Iq = Is
Tokyo. He is a researcher at the
cos ϕδ and Id = Is sin ϕδ into (4) and (5) yields
Railway Technical Research Insti-
tute, Tokyo, Japan, with responsibil-
Is (R sin ϕδ − Xq cos ϕδ ) = 0 (40)
ity in the area of research and devel-
opment of traction motors for railway E + Is (R cos ϕδ + Xd sin ϕδ ) = 0 (41)
vehicles. From 2005 to 2006, he was
an electrical engineer with Toshiba It is obvious that Id should be negative. Therefore,
Corp., Japan. He was responsible for designing traction solving (40) for sin ϕδ and selecting the negative solution,
motors for railway vehicles. His main interests are the
design and testing of electric motors, including permanent Xq
sin ϕδ = −  (42)
magnet synchronous motors and synchronous reluctance R 2 + Xq2
motors.
r
cos ϕδ = −  (43)
R 2 + Xq2
Appendix

Derivation of PMSM equivalent circuit In the Inserting (42) and (43) into (41) yields
two-reaction method, the synchronous reactance voltage
drop is expressed with two components (d- and q-axis R 2 + Xd Xq
E + Is  =0 (44)
components). R 2 + Xq2

j Xd I˙d + j Xq I˙q (34) Solving (44) for Is yields (6)

If this voltage drop is divided into an active power Derivation of equation (21) If the iron loss is
component Va and a reactive power component Vr , it is neglected, it is obvious from the phasor diagram that
obvious from the phasor diagram that they can be written the components in the q-axis direction can be written as:
as
E + Xd Id + R Iq = V cos δ. (45)
Va = Xd Id cos ϕδ − Xq Iq sin ϕδ
Inserting (15) and (16) into (45) and neglecting the
= Xd I sin ϕδ cos ϕδ − Xq I cos ϕδ sin ϕδ iron loss yields:
Xd − Xq
= I sin 2ϕδ (35) E + Xd I sin(φ − δ) + R I cos(φ − δ) = V cos δ (46)
2

116 IEEJ Trans 2: 109–117 (2007)


PARAMETER MEASUREMENTS FOR PMSM

60 Examples of measurement results The examples


Phase current of measurement results of the parameters are shown
Phase current (A), power factor (%)

50 below and an example of the no-load test result is shown


Power factor
in Fig. 8. The tested motor is a 4-pole 140 kW totally
40 enclosed PMSM [12] (rated speed: 2550/min)
Winding resistance R
30
0.0420 (at 115 ◦ C)
20 Permanent magnet induced EMF E
306 V (open circuit voltage)
10 318 V (no-load test, voltage at minimum current point)

0
d-axis reactance Xd
0 200 400 600
Phase voltage (V)
2.13 (short circuit test, (3), E = 306 V, Is = 144 A)
Rotational speed = 2550/min 2.21 (no-load test, (11), V = 376, E = 318, I =
26.3 A)
Fig. 8 Example of the no-load test result
2.51 (no-load test, (12), V = 376V-318V, I =
26.3 A–3.2A)
Since 2.23 (no-load test, (8), Ei = 376, E = 318, Id =
26.0 A)
sin(φ − δ) = sin φ cos δ − cos φ sin δ
q-Axis reactance Xq
cos(φ − δ) = cos φ cos δ − sin φ sin δ (47)
11.6 (short circuit test, (24), λ= 0.00910, Is = 67.2A,
(46) can be rewritten as E = 306 V, Xd = 2.13)
8.69 (no-load test, (27), V = 490, I = 56.4 A)
E = B cos δ + C sin δ (48) 7.58 (load test, (14)–(16) and (21)–(23), V = 433
V, I = 118 A, Xd = 2.21, cosφ = 0.963, E =
where B and C is defined as (22) and (23), respectively. 318, B = 358, C = 250, cosδ = 0.206, Id =
Rearranging and squaring both sides of (48) yields: −105 A, Iq = −54.5 A)

 2
E − B cos δ
= 1 − cos2 δ (49)
C

Solving (49) for cosδ yields (21).

117 IEEJ Trans 2: 109–117 (2007)

You might also like