Professional Documents
Culture Documents
© copyright FACULTY of ENGINEERING ‐ HUNEDOARA, ROMANIA 51
ACTA TECHNICA CORVINIENSIS – Bulletin of Engineering
parameters from physically measured data. and consequently the rotor flux will align with it.
Therefore, a two phase orthogonal dq model in Then the d-axis of the rotor is aligned with the
synchronous reference frame has been used to center of stator phase A and acquired inductance of
analyze the synchronous reluctances of the machine. the circuit will be equivalent to Ld inductance [4].
The mathematical model of PMSM motor is as Q-axis alignment requires a certain modification of
follows: the connection scheme according to Figure 2.
⎡u d ⎤ ⎡id ⎤ ⎡ Ld 0 ⎤ d ⎡id ⎤
⎢ u ⎥ = R S ⎢i ⎥ + ⎢ 0 Lq ⎥⎦ dt ⎢⎣iq ⎥⎦
+
⎣ q⎦ ⎣ q⎦ ⎣
⎡0 − Lq ⎤ ⎡id ⎤ ⎡0 ⎤
+ ωe ⎢ ⎥ ⎢i ⎥ + ω eψ PM ⎢ ⎥ (1)
⎣ Ld 0 ⎦⎣ q ⎦ ⎣1⎦
where RS is stator phase resistance, Ld, Lq are
synchronous inductances in dq axes and ψPM
represents a flux of permanent magnets.
Figure 2. Connection scheme performing q-axis
All tests were executed at standstill or very slow
alignment
rotary movement, therefore cross-coupling terms in
As it shown in Figure 2, the scheme considers only
both d- and q-axis as well as the back-EMF voltage
two motor phases energized and the phase A is
component in q-axis are canceled and the PMSM
disconnected. This approach causes generation of the
model is linearized as follows:
armature flux ψBC that is 90 degrees shifted against
⎡u d ⎤ ⎡id ⎤ ⎡ Ld 0 ⎤ d ⎡id ⎤ the resulting armature flux ψABC depicted in Figure 1.
⎢u ⎥ = RS ⎢i ⎥ + ⎢ 0 i
Lq ⎥⎦ dt ⎢⎣ q ⎥⎦
(2)
The rotor moves to the position of the armature flux
⎣ q⎦ ⎣ q⎦ ⎣ ψBC and aligns with it. It means that the q-axis of the
From motor control perspective, the equation (2) rotor is now aligned with the center of stator phase A
describes the model of the plant for d and q current and acquired inductance of the circuit will be
loop. Electrical torque in dq reference frame is equivalent to Lq inductance. Once the rotor is
developed by contribution of synchronous and aligned, the shaft of the motor must be mechanically
reluctance torque, and is derived as follows: locked. Having the rotor aligned and properly locked,
pp(ψ PM iq + (Ld − Lq )id iq )
3 the connection for measurement purposes has to be
Te = (3)
2 changed to the one shown in Figure 1.
To establish a dq frame model of a given PM MOTOR PARAMETER CHARACTERISTICS
synchronous motor, there are four motor parameters A. Stator resistance RS
that are to be determined. These are stator Generally, parameter RS defines a phase resistance
resistance RS, inductances Lq and Ld, and PM flux measured between line-to-neutral points. In the most
linkage ψPM. common PMSMs, the midpoint of Y-connected winding
A. Connection scheme definition is not accessible and therefore line-to-line resistance
Due to the mutual magnetic coupling between the has to be measured. The measurement can be
three phases, for stationary tests with single-phase provided either according to the scheme in Figure 1
excitation, the inductance should be measured with or scheme in Figure 2. In first case, the phase
all three phases excited as shown in Figure 1 [3]. In resistance will be two-thirds of measured line-to-line
order to investigate the behavior of inductances in resistance, in second case the phase resistance will
both d- and q-axis, the same scheme can be used, but be half of measured line-to-line resistance.
the rotor has to be locked at two different positions. During measurement an attention has to be paid
For d-axis measurement the rotor flux ψPM must be because winding resistance value is highly
aligned with the resulting armature flux ψABC and for temperature-dependent.
q-axis measurement the angle between rotor flux and B. Synchronous inductances Ld and Lq
resulting armature flux must be 90 degrees, Figure 1. Synchronous inductances measurement assumes the
connection scheme shown in Figure 1. As it has been
mentioned, the stator winding is Y connected without
any neutral line for external access and the only line-
to-line value can be measured. When the rotor’s d-
axis/q-axis are aligned with the stator phase A, the
synchronous inductance Ld and Lq can be derived from
equivalent line-to-line inductance LL-L as two-thirds
of LL-L, depending on the rotor angle θ. This
assumption can be accepted only in case of
simplification, where the mutual inductances are not
Figure 1. Connection scheme arrangement taken into account.
B. D- and Q- axis rotor alignment Since inductance in the motor is also subject of
In order to locate the d- and q-axis in PMSM the saturation phenomena, the way to correctly monitor
easiest way is to apply DC voltage to the terminal of the inductances is to measure them at various
the motor. Using the connection shown in Figure 1, current levels. The nonlinearity ψ=f(i), L=f(i) are
the rotor will follow the resulting armature flux ψABC caused by the saturation of iron parts of the stator
52 2013. Fascicule 1 [January–March]
ACTA TECHNICA CORVINIENSIS – Bulletin of Engineering
and has the same shape as magnetizing curve B=f(H). MOTOR PARAMETER MEASUREMENT
This dependency is described in Figure 3, where the A. Superposition of DC and AC excitation signal
relations ψ=f(i), L=f(i) in both axes are presented. The method utilizes compounded DC and AC voltages
as an excitation signal. The major signal is DC voltage
that will be adjustable during the test. The
magnitude and the frequency of AC signal are set by
a voltage source and will be fixed during the test.
Such excitation signal is connected to the terminal of
the motor according to the scheme in Figure 1. The
idea of proposed approach is clearly visible in Figure
4.
to have a capability to cover also saturation region, Figure 5. Excitation voltage and the current response
several methods can be used. The rough division of The DC values of voltage and current can be used for
suitable methods depends on the type of voltage the resistance measurement.
exciting signal and subsequently on the signal 2 2 V DC
processing. There are three different approaches RS = RL − L = (4)
3 3 I DC
presented in this paper.
2013. Fascicule 1 [January–March] 53
ACTA TECHNICA CORVINIENSIS – Bulletin of Engineering
The peak values of the voltage AC component and
current AC component and the phase shift between
them is used for inductance calculation as follows:
2 1
Ld (θ = 0 ) = Lq (θ =90 ) = R S tg (ϕ ) (5)
3 2πf
2 1 V ACm
Ld (θ = 0 ) = Lq (θ = 90 ) = sin (ϕ ) (6)
3 2πf I ACm
Figure 7. DC decay test connection scheme
Equations (5) and (6) offer two possibilities to
The default state of the circuit shown in Figure 7
calculate the line-to-neutral inductances. The
assumes the switch SW in the ON state and the
important information for both calculations is the
current i flows through the winding. By switching off
phase shift ϕ. In order to extract this value from the
the switch SW, new current iR will flow through the
waveforms shown in Figure 5 a proper modulation
circuit. An additional adjustable resistance RADD
technique has to be used.
allows setting the current iR according to the
B. Instantaneous Flux Linkage
required current step change Δi as it is described in
This method utilizes only AC voltage as an excitation
Figure 8.
signal for measuring the inductance saturation
curves. The same test arrangement as shown in
Figure 1 is used. Due to the fact that rotor is
mechanically locked the voltage equations (2) is
taken into account and used for linkage flux
calculation. By applying mathematical rules to (2),
a new set of equations can be derived:
ψ d = L d i d = ∫ (U d − R S i d )dt (7) Figure 8. The current transient during DC decay test
∫ (U − R S i q )dt
ψ q = Lq iq = (8) The principles and effect of a partial current
q
excitation is graphically described in Figure 9.
Estimated linkage fluxes are shown in Figure 6. Then
the inductances are calculated as follows:
ψd
Ld = − avg
(9)
id
ψq
Lq = − avg
(10)
iq
The current and voltage waveforms were measured
by an oscilloscope and processed in Matlab. In order Figure 9. Different levels of inductance saturation
to increase the accuracy of calculation, several curves excited by DC current step change
periods of current/voltage waveforms were stored It can be observed from Figure 9, that the particular
and averaged values were used for calculation. -3
regions of saturation curves relate to the particular
-3
x 10 x 10
2
hysteresis loop ψd=f(id)
2.5
2
hysteresis loop ψq=f(i q) time electrical constants due to changed inductance.
This phenomenon is valid for both axes and can be
1.5 averaged ψd=f(id) averaged ψq=f(i q)
1.5
1
0.5
1
0.5
used for synchronous inductances calculation. The
0 0 current transient has to be plotted by the
-0.5
-0.5
-1
oscilloscope and by using cursors the time electric
-1
-1.5 constant has be found for every step change. The
-1.5 -2
data would be triggered on switch SW change from
ON to OFF state.
-2 -2.5
-10 -8 -6 -4 -2 0 2 4 6 8 10 -10 -8 -6 -4 -2 0 2 4 6 8 10
i [A] i [A]
d q
54 2013. Fascicule 1 [January–March]
ACTA TECHNICA CORVINIENSIS – Bulletin of Engineering
360
A. Compounded excitation signal
Equations (5, 6) were used for inductances and 340
300
L [μH]
Ld = f(id) 280
q
280 260
Inductance [µH]
260 240
220
240
200
-6 -4 -2 0 2 4 6
220 Iq [A]
200
Equation (5) Figure 13. Measured q-axis saturation curve
Equation (6) The method is sensitive for correctly estimated
180 fluxes. An issue can arise during the calculation, by
‐10 ‐8 ‐6 ‐4 ‐2 0 2 4 6 8 10 using (7),(8), because of a DC offset that can appear
Phase current [A] in the current or voltage waveforms. This can be a
Figure 10. Measured d-axis saturation curves reason why the saturation curves are slightly
300 misaligned in comparison with the other methods.
Lq = f(iq )
C. Partial current step changes- decay test
280
For testing purpose, a current range from -7 up to
260 7Amps has been chosen with current step change Δi
Inductance [µH]
-10 -8 -6 -4 -2 0 2 4 6 8 10 240
Phase current [A]
220
Figure 11. Measured q-axis saturation curves
The results plotted in Figure 10 and 11 are slightly 200
deferent in depends on the approach (5) or (6) of 180 Ld
final estimation. The reason of these differences is Lq
most likely caused by incorrect calculation, because 160
-10 -8 -6 -4 -2 0 2 4 6 8 10
approach with (5) utilized the same RS value for Phase current[A]
whole current range calculation. On the contrary, the Figure 14. Measured d-and q-axis saturation curves
approach with (6) utilized instantaneous values of The shape of the curves has an expected behavior.
VACm and IACm for every current step. So the The temperature of the winding plays significant role
calculation with (6) can be considered as more during the test, where the value of resistance can be
correct and also the shape of saturation curve is changed. This can have a negative impact to the
nearer to the theoretical assumption shown in Figure results, as it can be seen in Figure 14, where a
3. certain misalignment can be observed in comparison
B. Instantaneous Flux Linkage with inductance curves depicted in Figure 13 and
The fluxes ψd, ψq estimated by (7, 8) are shown in Figure 14.
Figure 6. A simple mathematical rule must be used to CONCLUSIONS
get averaged values ψd_avg, ψq_avg. Then the saturation The aim was to propose and verify the methods for
curves, depicted in Figure 12 for d-axis and in Figure PMSM motor parameters investigation that can be
13 for q-axis, are calculated by applying (9, 10). later used for microprocessor implementation. Three
235
different approaches for inductance saturation curves
230
measurement are presented in the paper. Comparing
225 the results it can be seen, that the inductance curves
220 taken from all methods are roughly similar. The
215
small differences are caused by signal measurement
L [μH]
and processing.
d
210
The method utilizing compounded AC/DC signals
205
offers a possibility to estimate the resistance as well
200
as both inductances in dq coordinates. A huge
195 advantage of this approach is a variability that allows
190 measuring and verifying the motor parameters and
-6 -4 -2 0 2 4 6
Id [A] their behavior within a wide operational range of the
Figure 12. Measured d-axis saturation curve motor. On the contrary, requirement of flexible AC-
DC source and consequently complicated
2013. Fascicule 1 [January–March] 55
ACTA TECHNICA CORVINIENSIS – Bulletin of Engineering
demodulation method for phase shit estimation can according to the winding connection during the test.
be considered as a certain drawback. Such approach is enough precise for resistance
Instantaneous flux method requires a very precise measurement, but certain differences can be
flux estimator. An issue can arise during the flux expected in inductances behavior, because the
estimation process caused by integration. The mutual inductances are not taken into account.
acquired waveforms of motor quantities like current The same PMSM motor is planning to be used for
and voltage are affected by a small DC signal that is further work. The proposed solutions for parameters
introduced by a measuring process (offset of ADC estimation will be implemented into the
channels). Then the integrated offset generates the microcontroller and used for parameters estimation
varying offset in the time which makes drift of the of overall electric drive. The aim is to verify the
resulting fluxes. To avoid any drift in flux influence of the voltage inverter to the system
waveforms, the input signals of the estimator have to parameters and consequently to the quality of motor
be pure sinusoidal waveforms without any DC offset. control.
Due to this fact this method is quite laborious. REFERENCES
The last method, shown in the paper, is a simple [1] P. Vas: “Parameter Estimation, Condition
current step change – decay test. The success and the Monitoring, and Diagnosis of Electrical
final precision of this method is linked with the time Machines”, Clarendon Press – Oxford, 1993, ISBN
electric constant. The more accurate the time 0-19-859375-9
constant is deducted from the acquired exponential [2] V. Hrabovcová; P. Rafajdus, P. Hudák, M. Franko:
waveform the more accurate parameters are “Meranie a modelovanie elektrických
reached. The modification of origin decay test to the strojov, EDIS – Žilinská univerzita v Žiline,, Žilina,
partial step current changes allows investigation of 2004, ACB, str.: 335, ISBN 80-8070-229-2
the motor parameters in the wide operational range. [3] W. L. Soong: “Inductance Measurements for
The PMSM motor has been chosen because of its Synchronous Machines”, Power Engineering
popularity in servo-driven application. Respecting the Briefing Note Series #2, 8 May 2008
trend that only few applications require neutral [4] Dal Y. Ohm: “Dynamic Model of PM Synchronous
point to be accessible; the standard Y-connected Motors”, Drivetech, Virginia
PMSM motor without any access to the neutral point [5] R. Filka: “Sensorless Control of IPM Synchronous
has been chosen and used during all tests. In such Motors for Entire Speed Range Including Start-
case, the only line-to-line inductances are measured Up”, PhD Thesis, 2008
and then converted to the phase inductance value
56 2013. Fascicule 1 [January–March]