U.S. patent application number 14/533330 was filed with the patent office on 2016-05-05 for sensorless rotor angle detection circuit and method for a permanent magnet synchronous machine.
This patent application is currently assigned to STMicroelectronics S.r.l.. The applicant listed for this patent is STMicroelectronics S.r.l.. Invention is credited to Dino Costanzo, Giuseppe Scarcella, Giacomo Scelba.
Application Number | 20160126873 14/533330 |
Document ID | / |
Family ID | 55754785 |
Filed Date | 2016-05-05 |
United States Patent
Application |
20160126873 |
Kind Code |
A1 |
Costanzo; Dino ; et
al. |
May 5, 2016 |
SENSORLESS ROTOR ANGLE DETECTION CIRCUIT AND METHOD FOR A PERMANENT
MAGNET SYNCHRONOUS MACHINE
Abstract
An estimate of the initial position of a rotor is made by
monitoring sensed motor current signals which are amplitude and
phase modulated with the rotor flux position in response to a high
frequency voltage signal injection. The motor current signals are
envelope detected to determine zero crossing points. Samples are
taken of the motor current signals at positive and negative offsets
from the zero crossing point, with the samples processed to
identify a direction of the rotor flux axis. Further samples of at
least one motor current signal are taken with respect to a certain
phase reference, and the samples compared to resolve a polarity of
the rotor flux axis which is indicative of the angular position of
the rotor.
Inventors: |
Costanzo; Dino; (Catania,
IT) ; Scelba; Giacomo; (Caltagrione, IT) ;
Scarcella; Giuseppe; (Catania, IT) |
|
Applicant: |
Name |
City |
State |
Country |
Type |
STMicroelectronics S.r.l. |
Agrate Brianza |
|
IT |
|
|
Assignee: |
STMicroelectronics S.r.l.
Agrate Brianza
IT
|
Family ID: |
55754785 |
Appl. No.: |
14/533330 |
Filed: |
November 5, 2014 |
Current U.S.
Class: |
318/400.04 |
Current CPC
Class: |
H02P 6/183 20130101;
H02P 21/18 20160201; G01R 19/175 20130101 |
International
Class: |
H02P 6/18 20060101
H02P006/18 |
Claims
1. A sensorless rotor angle detection circuit for a motor having
multiple phase windings, comprising: a current sensing circuit
configured to sense current in a plurality of the motor phase
windings; a first phase conversion circuit configured to convert a
multi-phase current sense signal output from the current sensing
circuit to a two-phase current signal; and a position detection
circuit configured to process said two-phase current signal and
determine a rotor angle of said motor; wherein processing of the
two-phase current signal comprises: determining zero crossing
points of an envelope of a first signal of the two-phase current
signal, said zero crossings being indicative of rotor angle
direction; sampling the first signal and a second signal of the
two-phase current signal at positions positively and negatively
offset from one of the zero crossing points to determine a
direction of a rotor flux axis; sampling one of the first and
second signals in correspondence with a phase reference to
determine information indicative of the rotor angle.
2. The circuit of claim 1, wherein the determined information is a
polarity of the determined direction of the rotor flux axis,
wherein said polarity is indicative of the rotor angle.
3. The circuit of claim 1, further comprising: a control circuit
configured to generate said two-phase input signal; a second phase
conversion circuit configured to convert the two-phase input signal
to a multi-phase control signal; and a multi-phase inverter circuit
configured to receive said multi-phase control signal and output a
multi-phase motor drive signal for application to said multiple
phase windings.
4. The circuit of claim 3, wherein the two-phase input signal
generated by said control circuit is configured to inject a
roto-pulsating high frequency stator flux in said motor, and said
first and second signals of the two-phase current signal are
amplitude and phase modulated with a position of said rotor
flux.
5. The circuit of claim 3, wherein the phase reference is provided
by an alternating voltage signal generated by the control circuit
and applied to one phase of the two-phase input signal.
6. The circuit of claim 5, wherein the one phase of the two-phase
input signal which receives the alternating voltage signal
corresponds to a direction of rotor flux axis.
7. The circuit of claim 3, wherein the control circuit is
configured to apply a zero magnitude signal as a first signal of
the two-phase input signal and a non-zero magnitude signal as a
second signal of the two-phase input signal.
8. The circuit of claim 7, wherein the first and second phase
conversion circuits operate in a reference frame synchronous to an
arbitrary angle.
9. The circuit of claim 1, wherein sampling the first signal and a
second signal of the two-phase current signal comprises: sampling
the first and second signals at a positive offset position;
calculating a first sign of a product of the samples at the
positive offset position; sampling the first and second signals at
a negative offset position; calculating a second sign of a product
of the samples at the negative offset position; and determining a
corrected angle as equaling: an initial estimated angle if the
first and second signs exhibit a first opposite relationship; and
the initial estimated angle plus 90.degree. if the first and second
signs exhibit a second opposite relationship.
10. The circuit of claim 9, further comprising determining a fault
if the first and second signs are the same.
11. The circuit of claim 1, wherein sampling one of the first and
second signals in correspondence with a phase reference comprises
sampling said one of the first and second signals at positions
corresponding to a defined angle.
12. The circuit of claim 11, further comprising: determining
whether a magnitude of the odd samplings exceeds a magnitude of the
even samplings; and if so, then outputting the rotor angle equal to
the direction of the rotor flux axis offset by 180.degree.; and
otherwise, outputting the rotor angel equal to the direction of the
rotor flux axis.
13. The circuit of claim 11, wherein the defined angle corresponds
to zero crossings of the phase reference.
14. The circuit of claim 1, wherein said processing of the
two-phase current signal occurs when a rotor of said motor is at
standstill.
15. A method for sensorless detection of rotor angle for a motor
having multiple phase windings, comprising: sensing current in a
plurality of motor phase windings to provide a multi-phase current
sense signal; converting said multi-phase current sense signal to a
two-phase current signal; and processing said two-phase current
signal to determine a rotor angle of said motor; wherein processing
comprises: determining zero crossing points of an envelope of a
first signal of the two-phase current signal, said zero crossings
being indicative of rotor angle direction; sampling the first
signal and a second signal of the two-phase current signal at
positions positively and negatively offset from one of the zero
crossing points to determine a direction of a rotor flux axis;
sampling one of the first and second signals in correspondence with
a phase reference to determine information indicative of rotor
angle.
16. The method of claim 15, wherein the information comprises a
polarity of the determined direction of the rotor flux axis,
wherein said polarity is indicative of the rotor angle.
17. The method of claim 15, further comprising: generating said
two-phase input signal; converting the two-phase input signal to a
multi-phase control signal; and generating a multi-phase motor
drive signal for application to said multiple phase windings from
said multi-phase control signal.
18. The method of claim 17, further comprising injecting a
roto-pulsating high frequency stator flux in said motor in response
to said two-phase input signal, wherein said first and second
signals of the two-phase current signal are amplitude and phase
modulated with a position of said rotor flux.
19. The method of claim 17, wherein the phase reference is provided
by an alternating voltage signal applied to one phase of the
two-phase input signal.
20. The method of claim 19, wherein the one phase of the two-phase
input signal which receives the alternating voltage signal
corresponds to a direction of rotor flux axis.
21. The method of claim 17, further comprising appling a zero
magnitude signal as a first signal of the two-phase input signal
and a non-zero magnitude signal as a second signal of the two-phase
input signal.
22. The method of claim 21, wherein the first and second phase
conversion circuits operate in a reference frame synchronous to an
arbitrary angle.
23. The method of claim 15, wherein sampling the first signal and a
second signal of the two-phase current signal comprises: sampling
the first and second signals at a positive offset position;
calculating a first sign of a product of the samples at the
positive offset position; sampling the first and second signals at
a negative offset position; calculating a second sign of a product
of the samples at the negative offset position; and determining a
corrected angle as equaling: an initial estimated angle if the
first and second signs have a first opposite relationship; and the
initial estimated angle plus 90.degree. if the first and second
signs have a second opposite relationship.
24. The method of claim 23, further comprising identifying a fault
if the first and second signs are the same.
25. The method of claim 15, wherein sampling one of the first and
second signals in correspondence with a phase reference comprises
sampling said one of the first and second signals at positions
corresponding a defined angle.
26. The method of claim 25, further comprising: determining whether
a magnitude of the odd samplings exceeds a magnitude of the even
samplings; and if so, then outputting the rotor angle equal to the
direction of the rotor flux axis offset by 180.degree.; and
otherwise, outputting the rotor angel equal to the direction of the
rotor flux axis.
27. The method of claim 25, wherein the defined angle corresponds
to zero crossings of the phase reference.
28. The method of claim 15, wherein said processing of the
two-phase current signal occurs when a rotor of said motor is at
standstill.
Description
TECHNICAL FIELD
[0001] This disclosure relates generally to synchronous machines
and, more particularly, to techniques for sensorless detection of
rotor position in a permanent magnet motor.
BACKGROUND
[0002] Motors in the form of a Permanent Magnet Synchronous Machine
(PMSM) are gradually replacing classic DC motor drives as well as
other AC motor drives in a large number of industrial and domestic
applications. This is due, at least in part, on increasingly
stringent specifications on energy consumption with a view to
sustainable development that tend to place other motor drives at a
disadvantage with respect to PMSM drives. Those skilled in the art
recognize a number of advantages with respect to the key features
of a permanent magnet motor such as compactness, efficiency, power
factor, robustness and reliability. Notwithstanding those
advantages, a known drawback with respect to controlling such a
motor is that a vector controlled PMSM drive needs to use an
encoder or a resolver for correctly aligning the stator current
vector (i.e., correctly detecting rotor position). Such an
electromechanical position transducer is not typically present in
low cost/low power DC motor drives or constant V/Hz induction motor
drives. The inclusion of such a transducer presents a significant
disadvantage to the adoption of a PMSM drive for use in low
cost/low power applications.
[0003] The performance of sensorless PMSM drives depends on torque
control capabilities as well as on position and speed estimation
accuracy and bandwidth. There is accordingly a need in the art for
a sensorless PMSM drive capable of operation at any speed
(including low and zero speed) which includes functionality for
making both initial and incremental rotor position detections.
SUMMARY
[0004] In an embodiment, a sensorless rotor angle detection circuit
for a motor having multi-phase windings comprises: a current
sensing circuit configured to sense current in plural motor phase
windings; a first phase conversion circuit configured to convert a
multi-phase current sense signal output from the current sensing
circuit to a two-phase current signal; and a position detection
circuit configured to process said two-phase current signal and
determine a rotor angle of said motor. The processing of the
two-phase current signal comprises: determining zero crossing
points of an envelope of a first signal of the two-phase current
signal, said zero crossings being indicative of rotor angle
direction; sampling the first signal and a second signal of the
two-phase current signal at positions positively and negatively
offset from one of the zero crossing points to determine a
direction of a rotor flux axis; sampling one of the first and
second signals in correspondence with a phase reference to
determine information indicative of the rotor angle.
[0005] In an embodiment, a method for sensorless detection of rotor
angle for a motor having multi-phase windings comprises: sensing
current in plural motor phase windings to provide a multi-phase
current sense signal; converting said multi-phase current sense
signal to a two-phase current signal; and processing said two-phase
current signal to determine a rotor angle of said motor. The
processing comprises: determining zero crossing points of an
envelope of a first signal of the two-phase current signal, said
zero crossings being indicative of rotor angle direction; sampling
the first signal and a second signal of the two-phase current
signal at positions positively and negatively offset from one of
the zero crossing points to determine a direction of a rotor flux
axis; sampling one of the first and second signals in
correspondence with a phase reference to determine information
indicative of the rotor angle.
[0006] The circuit and method may, for example, be configured to
implement a control algorithm in software executed on a
microcontroller or computer. Alternatively, the circuit and method
may implement the control algorithm using analog signal
circuitry.
BRIEF DESCRIPTION OF THE DRAWINGS
[0007] For a more complete understanding of the present disclosure,
and the advantages thereof, reference is now made to the following
descriptions taken in conjunction with the accompanying drawings,
in which:
[0008] FIG. 1 is a block diagram of a sensorless vector-control
drive for a Permanent Magnet Synchronous Machine (PMSM) motor;
[0009] FIG. 2 illustrates the amplitude and phase modulation of the
current components derived from sensed motor current through
reference frame translation to an arbitrary angle;
[0010] FIG. 3 is a block diagram of signal processing operations
performed by a position detection circuit;
[0011] FIG. 4 illustrates a sampling and envelope detection
operation;
[0012] FIG. 5 shows current waveforms obtained by injecting a
roto-pulsating high frequency magnetic field with respect to a
motor operating at standstill (or when the high frequency test
signal is injected on a reference frame that travels with a
different speed than the rotor);
[0013] FIG. 6 illustrates the high frequency current signals with
identified zero crossing and offset sampling locations;
[0014] FIG. 7 illustrates details of the high frequency current
signals at the sampling locations;
[0015] FIGS. 8A-8D illustrate the phase displacement between high
frequency magneto motive force and corresponding flux; and
[0016] FIG. 9 illustrates sampling in support of determining rotor
flux polarity.
DETAILED DESCRIPTION OF THE DRAWINGS
[0017] Significant effort has been expended towards the development
of a reliable, low cost control strategy for an AC motor drive that
does not require a position transducer. Such control strategies,
referred to in the art as `sensorless` techniques, generally
estimate the rotor position by processing electrical motor
variables, such as phase stator currents and voltages.
[0018] The simplest sensorless techniques, based on rotor flux
position estimation, integrate the back EMF of the motor. This
approach is simple and effective, but fails when the machine is
operating at low and zero speed because the motor back EMF is too
low to effectively measure and the rotor flux estimation result is
sensitive to stator resistance variations and measurement
noises.
[0019] A closed loop rotor position observer has been also
developed in order to decrease the sensitivity to parameter
variations, while automatically performing the correct system
initialization. However, this approach, as well as position
estimation techniques based on stator phase voltage third harmonic
or spatial phenomena inherent within the machine (such as
rotor/stator eccentricities and winding asymmetries) cannot work at
zero speed for the lack of useful signals to process.
[0020] Model Reference Adaptive Controller (MRAC), Kalman Filters,
Luenberger Observers and other sensorless control schemes based on
the application of sophisticated identification procedures allow,
under some limitations, low and zero speed operations, but are too
complex and expensive to be used in practical low cost systems.
[0021] Sensorless vector controlled alternating current (AC) motor
drives are available in the market, with these products
demonstrating the practical use of sensorless drive technology.
Commercially available sensorless AC motor drives are mainly Direct
Field Oriented Controlled (DFOC) induction motor drives, requiring
only speed (not position as in the case of PMSM drives) estimation,
or Direct Torque Control (DTC) drives that require only coarse
position estimation. Such systems work in a speed range between 1%
and 100% of the rated speed, featuring speed control accuracy as
high as 5% of the base speed. Theses motors can favorably compete
with constant Volt/Hertz configurations at low level and with
standard Indirect Field Oriented Controlled (IFOC) induction motor
drives at high level, although only at sufficiently high speed in
the last case.
[0022] These sensorless AC drives show dynamic performance fairly
approaching those of standard vector controlled drives, at least at
speeds higher than 3-5% of the rated value. Below such a speed,
however, the dynamic stiffness quickly decreases, making it
impossible to perform speed control. The flux/rotor position is
incrementally estimated with a typical accuracy of 8-12 bits, and
this is sufficient to perform the vector control. Speed estimation
features depend on the specific algorithm used but, again, the
typical accuracy is 8-12 bits, allowing for the performance of
speed control. Thus, these AC sensorless drives are not suitable to
control shaft position, which requires an absolute shaft position
estimation with 12-16 bits accuracy and a high bandwidth, as well
as full torque control capability at zero speed.
[0023] Some known application fields of sensorless AC motor drives
include: offset printing machines, manufacturing machines in the
pulp and paper industry, iron industry and textile industry,
movement of cranes, conveyors and trolleys, air conditioning
systems, sealed compressors and submerged pumps.
[0024] Further development of sensorless AC motor drives will
permit such drives to compete with standard vector controlled AC
drives, thus showing full torque control at zero speed and at least
12 bits of absolute position estimation accuracy. Expected
application fields include the same fields as noted above, although
with improved dynamic performance and shaft position control
capability.
[0025] Among sensorless AC drives, PMSM drives feature the maximum
power density and are well suited for those applications where the
weight and the shape of the motor are of concern. All of the
aforementioned sensorless methods able to operate even at zero
speed will also need to estimate the initial rotor position. A
classification can be made according to the detection method.
Techniques belonging to the indirect speed estimation category
exploit the stator equations to estimate the amplitude, the angular
position and/or the frequency of the rotor flux and the rotor
voltage equations to calculate the speed. Indirect approaches
generally show a good level of technological maturity and have been
already introduced onto the market, being particularly suitable to
equip medium-low performance drives, where compensation of
parametric variations and low speed sustained operations are not
required. When compared with traditional systems based on the
constant Volt per Hertz (VbyF) techniques, that are naturally
sensorless featuring a predictive speed control, induction motor
drives based on indirect speed estimation feature superior dynamic
performance and an higher speed control robustness.
[0026] The predominant indirect speed estimation techniques in the
art can be grouped as follows:
[0027] Back EMF (BEMF) techniques: compute speed from the
fundamental Back EMF and stator currents. These techniques feature
a good dynamic response at speeds exceeding 10% of the rated speed.
Below this value, a correct Back EMF estimation cannot be
accurately performed.
[0028] Flux estimation techniques: compute speed from the rotor
voltage equations. These techniques use full order observers,
reduced order rotor flux observers or simple rotor flux
calculators, as well as Kalman filters and adaptive observers to
obtain flux and speed estimation in open or closed loop modes.
Although such approaches lead to different performance, not always
directly correlated to the complexity of the algorithm or to the
required computation effort, the performance is satisfactory in the
high-medium-low speed range. However, these approaches are not able
to consistently estimate rotor speed at zero or very low stator
frequency.
[0029] Model Reference Adaptive System (MRAS) technique: exploits
the redundancy of two different models computing the same motor
variables to estimate the speed. In its simplest form a MRAS scheme
is composed of two rotor flux calculators. The first exploits a
voltage and a current based flux model, while the second is based
on a current model using the estimated speed as the input. The
difference between the outputs of the two flux calculators is used
to correct the estimated speed. Compared with previous approaches,
this technique provides for improved speed estimation in the low
speed range. However, the technique is not able to determine speed
at low and zero speeds.
[0030] One type of direct sensorless speed estimation is based on
the analysis of the harmonic content of some motor variables. This
technique is almost motor parameter independent, but generally
requires a heavier computational effort (for example, in comparison
to indirect techniques). Direct estimation of the rotor speed is
made possible by the presence in the harmonic spectrum of motor
currents and voltages of components featuring an angular frequency
multiple of the rotor speed. Such components are commonly generated
on standard machines by parasitic motor saliencies, caused by rotor
asymmetries, rotor eccentricity, rotor slotting and magnetic
saturation.
[0031] A further distinction can be made among direct techniques
using standard machine excitation and those introducing suitable
test signals on standard current and voltages.
[0032] Estimation techniques using a standard excitation are based
on the detection of stator voltage rotor slot harmonics. The speed
can be determined by measuring the amplitude or the angular
frequency of slot harmonics. However, there is a problem with
extracting the harmonic components of the speed information from
the whole harmonic content of the machine. Such components, in
fact, feature a low amplitude and disappear at very low and zero
speed. Moreover, their frequency ranges in a wide frequency
interval, according to speed variations, making mandatory the
introduction of adaptive PLL systems. In practice, application of
such methods at low speed is a quite complex task, while at high
speed indirect techniques give similar results, with a much lower
amount of computations.
[0033] The direct speed estimation technique that exploits suitable
HF test signals presents a solution for use at the low speed range.
The technique based on HF signal injection, in theory, allows for
making a precise estimate of the motor speed at any frequency and
with a good bandwidth. However, some presented techniques, to be
practical, require special rotor designs in order to introduce a
kind of saliency on standard machines This limits the applicability
of these speed estimation techniques on standard AC motor
drives.
[0034] For a PMSM, estimation of the initial rotor position at zero
speed and at any load is critical. In particular, embodiments
herein present an apparatus and process for estimating an initial
rotor position for a PMSM sensorless vector-control drive. The
proposed solution overcomes the drawbacks mentioned above and, in
particular, provides an accurate estimation of the rotor position
at standstill, thus avoiding any mechanical movement of the rotor.
The technique is further applicable, without necessitating rotor
standstill, in situations where a reference frame for measurement
travels with a different speed than the rotor.
[0035] Reference is now made to FIG. 1 showing a block diagram of a
sensorless vector-control drive 10 for a Permanent Magnet
Synchronous Machine (PMSM) motor 12. The motor 12 includes three
motor phases (referred to as the "as" phase, the "bs" phase the
"cs" phase (wherein each motor phase includes a motor winding
connected to a common node in a "Y" configuration, as known in the
art). The drive 10 includes a three-phase inverter (VSI) 14
configured to apply three corresponding voltage signals (Vas, Vbs
and Vcs) to the three motor phases in response to three
corresponding drive control signals 16. The three-phase inverter 14
may, for example, include three half-bridge drive circuits, wherein
each half-bridge drive circuit receives one of the drive control
signals 16 and generates one of the drive voltage signals (Vas, Vbs
and Vcs). A phase conversion circuit 18 is provided to convert an
input signal 20 from a two-phase (qd) frame of reference to a three
phase (abc) frame of reference for said drive control signals 16.
Such a phase conversion circuit 18 is known to those skilled in the
art. The phase conversion circuit 18 operates with reference to an
arbitrarily selected electrical angle (.theta.arb) whose frequency
is selected based on the magnetic structure of the motor. The
frequency may, for example, comprise a frequency selected between 1
and 50 Hz.
[0036] Although the description herein is made with reference to a
three-phase motor, it will be understood that the techniques
disclosed herein are more generally applicable to multi-phase
motors.
[0037] A control circuit 30 controls the configuration of the input
signal 20 applied on the q and d axes. For example, the control
circuit 30 may apply a zero magnitude signal to one of the q or d
axes and a non-zero magnitude signal (DC or AC) on the other one of
the q or d axes. The phase conversion circuit 18 and three-phase
inverter 14 accordingly function in response to the input signal 20
to inject roto-pulsating high frequency flux into the motor 12.
[0038] A current sensing circuit 22 is provided for a plurality
(but not necessarily all) of the motor phases to sense drive
current and output corresponding drive current signals (ias, ibs,
ics). The current sensing circuits 22 are illustrated, for
convenience only, as nodes on the lines carrying the drive voltage
signals (Vas, Vbs and Vcs) to the corresponding three terminals of
the motor 12. Circuitry for performing such current sensing
operations with respect to motor drive are well known to those
skilled in the art. A phase conversion circuit 24 is provided to
convert the drive current signals (ias, ibs, ics) from a three
phase (abc) frame of reference to a two-phase (qd) frame of
reference to output high frequency current component signals (iqshf
and idshf) 26. Such a phase conversion circuit 24 is known to those
skilled in the art. The phase conversion circuit 24 operates, like
that of the circuit 18, with reference to said arbitrarily selected
electrical angle (.theta.arb).
[0039] In an alternative embodiment, a single current sensing
circuit 22 may be implemented to measure the individual motor
currents using a technique known to those skilled in the art as a
single common DC link current sensor.
[0040] The high frequency current components iqshf and idshf,
calculated in a reference frame synchronous to the arbitrarily
selected electrical angle .theta.arb, are thus amplitude and phase
modulated with the rotor flux position, as a consequence of the
physical phenomenon of the flux deviation occurring in anisotropic
motors. An example of such modulation is shown in FIG. 2. In this
example, the d axis current component is modulated at twice the
electrical angular position, and the orthogonal q axis current
component is modulated at four times the electrical angular
position. See, also, FIG. 5.
[0041] The drive 10 further includes a position detection circuit
28 that receives and processes the output current signals (iqshf
and idshf) 26 and determines the position (.theta.final) of the
rotor for the motor 12. FIG. 3 illustrates a block diagram of the
general signal processing operations performed by the position
detection circuit 28.
[0042] In a first signal processing operation (reference 100), the
control circuit generates signal 20 which comprises a first (for
example, one of zero or non-zero) magnitude signal on the q axis
and a second (for example, the other of zero or non-zero) magnitude
signal (Vdshf) on the d axis. A roto-pulsating high frequency flux
(synchronous to the arbitrarily selected electrical angle
(.theta.arb)) is injected into the motor 12 as a result of the
operations performed by the phase conversion circuit 18 and
three-phase inverter 14. One or more of the high frequency current
signals, iqshf or idshf, are then demodulated by circuit 28 to
extract information about flux deviation. Due to the flux deviation
phenomenon, the high frequency current signals, iqshf or idshf,
hold information about rotor angle displacement with respect to the
injection angle .theta.arb (and thus consequently about the rotor
angle .theta.final of the motor 12 itself). Taking the high
frequency current signal iqshf, for example, the first signal
processing operation 100 samples the current signal in
correspondence of the zero crossing or a defined angle of the input
non-zero voltage signal Vdshf (according to the resistor-inductor
(RL) characteristic at the injection frequency of the motor 12).
This signal Vdshf, as discussed above, is generated by the control
circuit 30 and thus is known. A sign of the sampled current is
alternately inverted (without any constraint in the initial
polarity selection) so as to generate a low frequency current
envelope signal iqshf_env (which is a function of rotor position)
as shown in FIG. 4. This operation arises as a consequence of an
assumption that the high frequency voltage Vdshf is 90.degree. (or
more precisely according to the RL characteristic of the motor at
the injection frequency) phase shifted in advance with respect to
the high frequency current iqshf.
[0043] It will be noted that the computational load required for
performing the demodulation process (FIG. 3, reference 100) with
respect to one or more of the high frequency current signals, iqshf
or idshf, is considerably reduced by combining this sampling
synchronization between the high frequency voltages and currents.
Thus, the demodulation process permits implementation of sensorless
control solutions, based on the injection of an additional high
frequency magnetic field, with processing by a relatively lower
performing microcontroller.
[0044] Reference is now made to FIG. 5 showing current waveforms
with respect to the motor 12 operating, for example, at standstill
(i.e., no rotation) during the operation 100. It will be understood
that the waveforms of FIG. 5 are applicable as well to the
situation where the high frequency test signal is injected on a
reference frame that travels with a different speed from the rotor.
The first waveform in FIG. 5 illustrates the sensed drive current
signal for one phase of the motor 12 (in this case, for phase as
showing the current ias). The second waveform in FIG. 5 illustrates
the high frequency current signal iqshf. The third waveform in FIG.
5 illustrates the high frequency current signal idshf. It will be
noted that the waveforms are illustrated for at least one entire
electrical cycle (360.degree.), and that within such one electrical
cycle there are four zero crossing points 102 on the high frequency
current signal iqshf. Thus, for an injection of the roto-pulsating
high frequency flux into the motor 12 for a duration of at least
one-fourth of the electrical period, the high frequency current
signal iqshf will exhibit at least one minimum point (zero
crossing) that is orthogonal to the injected current. The zero
crossing point(s) provide(s) information indicative of the actual
rotor flux position (0final) with a potential error of +90.degree.,
180.degree. or -90.degree.. The next processing two steps operate
to resolve the position ambiguity.
[0045] In a second signal processing operation (reference 104, for
example carried out on data collected during the first phase), the
control circuit generates signal 20 which comprises a zero
magnitude signal on the q axis and a non-zero magnitude signal
(Vdshf) on the d axis. A roto-pulsating high frequency flux
(synchronous to the arbitrarily selected electrical angle
(.theta..sub.fictious) which can rotate clockwise or
counter-clockwise with respect to the motor) is injected into the
motor 12 as a result of the operations performed by the phase
conversion circuit 18 and three-phase inverter 14. The angular
position of the axis q and the axis d of the motor is discriminated
by analyzing at least two samplings of both the high frequency
current signals iqshf and idshf that are offset on either side of
one zero crossing of the high frequency current signal iqshf
identified in operation 100.
[0046] In a preferred embodiment, the two samplings are taken
simultaneously (or nearly simultaneously) offset from the
identified zero crossing location by a sampling phase angle
.alpha.. In a preferred implementation, .alpha.<90.degree., and
in an example implementation, .alpha.=45.degree.. The two samplings
are accordingly made at the zero crossing angle minus .alpha. and
at the zero crossing angle plus .alpha.. Reference is made to FIG.
6 showing the high frequency current signals iqshf and idshf, the
identified zero crossing 102 of the high frequency current signal
iqshf, and the locations 106 and 108 of the two samplings at the
example implementation where .alpha.=45.degree.. The sampling of
the high frequency current signals iqshf and idshf can be
performed, in a manner similar to that described above with respect
to operation 100, synchronously to the applied high frequency
voltage Vdshf.
[0047] At each sampling location, the process calculates a value c
which is equal to the sign of the product of the sampled high
frequency current signals iqshf and idshf in accordance with the
following formula:
c=sign(iqshfidshf)
[0048] The process then compares the value of c for sampling
location 106 (c106) to the value of c for the sampling location 108
(c108). It is assumed that the identified zero crossing location
102 is an initial estimated rotor position (.theta.est). In the
case where the arbitrarily selected electrical angle
(.theta..sub.fictious) rotates counterclockwise with respect to the
rotor, if c106 and c108 are respectively positive and negative,
then the corrected rotor position (.theta.cor) is given by the
initial estimated rotor position (i.e., .theta.cor=.theta.est).
Conversely, if c106 and c108 are respectively negative and
positive, then the corrected rotor position is given by the initial
estimated rotor position plus 90.degree. (i.e.,
.theta.cor=.theta.est +90.degree.). On the other hand, in the case
that the arbitrarily selected electrical angle
(.theta..sub.fictious) rotates clockwise with respect to the rotor,
if c106 and c108 are respectively negative and positive, then the
corrected rotor position is given by the initial estimated rotor
position. Conversely, if c106 and c108 are respectively positive
and negative, then a corrected rotor position (.theta.cor) is given
by the initial estimated rotor position plus 90.degree. (i.e.,
.theta.cor=.theta.est+90.degree.). Other results, i.e. c106 and
c108 both positive or both negative, can be considered as errors in
method application or signal measurement and decoding chain.
[0049] It will be noted that, for the process, when a zero crossing
point has been detected in operation 100, the value of .theta.arb
in that instant is indicated .theta.est.
[0050] Reference is now made to FIG. 7 showing details of an
example for the high frequency current signals iqshf and idshf at
the sampling locations 106 and 108. In this example, assuming that
the .theta.arb is rotating counterclockwise, the sign of the
product of the high frequency current signals iqshf and idshf at
location 106 is "negative" and the sign of the product of the high
frequency current signals iqshf and idshf at location 108 is
"positive". So, in this example, the corrected rotor position
.theta.cor is given by the initial estimated rotor position plus
90.degree. (i.e., .theta.cor=.theta.est+90.degree.). If, on the
other hand, both samples were like that illustrated on the right
hand side of FIG. 7 (or both like that illustrated on the left hand
side), then this would be an indication of errors in the method
application or signal measurement and decoding chain.
[0051] The analysis performed with c106 and c108 allows for
determining whether the identified zero crossing of the high
frequency current signal iqshf occurs along the d or q axis. From
this, one can know exactly the direction (but not polarity) of the
rotor flux vector (.theta.final). The foregoing arises from the
following physical phenomenon: the different reluctance along the d
(rotor flux direction) and q axes produces a phase displacement
between the high frequency magneto motive force Fh and the
corresponding flux .lamda.hf. In particular, Fh lags and leads
alternatively the flux .lamda.hf, as can be seen in FIGS. 8A-8D.
The processing performed in operation 104 exploits this phenomenon:
the projection of the flux .lamda.hf along the estimated q' axis
generates a high frequency current iqshf which is approximately in
phase or in opposition to the current idshf, depending on the
operating quadrant. From this analysis, a correction is made (if
necessary) to the rotor position from the initial estimated rotor
position at the zero crossing location.
[0052] More specifically: FIG. 8A illustrates the condition where
iqshf is 180.degree. shifted with respect to idshf; FIG. 8B
illustrates the condition where iqshf is 0.degree. shifted with
respect to idshf; FIG. 8C illustrates the condition where iqshf is
180.degree. shifted with respect to idshf; and FIG. 8D illustrates
the condition where iqshf is 0.degree. shifted with respect to
idshf;.
[0053] The processing operation 104 advantageously requires only
two samplings of high frequency current signals iqshf and idshf and
the performance of only two tests in order to obtain the exact
direction of the rotor flux vector. The process can be performed
relatively quickly using simple computational techniques.
[0054] In a third signal processing operation (reference 110), the
control circuit generates signal 20 which comprises an alternating
non-zero magnitude signal (Vdshf) applied to the axis along the
direction of rotor flux (as determined in operation 104) and a zero
magnitude signal applied to the other axis. The polarity of the
previously identified rotor flux vector (.theta.cor) is then
determined by monitoring the high frequency current signal for the
corresponding axis. Thus, assume that the alternating non-zero
magnitude signal (Vdshf) is applied to the d axis, then the high
frequency current signal idshf is monitored.
[0055] Sampling of the high frequency current signal idshf can be
performed as noted above synchronously to the applied high
frequency voltage. More particularly, the high frequency current
signal idshf is sampled at each zero crossing of the alternating
non-zero magnitude signal (Vdshf). It is important to sample the
high frequency current signal idshf in correspondence with a
specific phase of applied high frequency voltage. The sampled
values of signal idshf at odd zero crossings (first half phase) of
the signal Vdshf are summed to produce a first summed value
.SIGMA.odd and the sampled values of signal idshf at even zero
crossings (second half phase) of the signal Vdshf are summed to
produce a second summed value .SIGMA.even. The first and second
summed values (.SIGMA.odd and .SIGMA.even) are then compared. If
.SIGMA.odd>.SIGMA.even, then a final rotor position is given by
the corrected rotor position (from operation 104) minus
180.degree.. Conversely, if .SIGMA.even>.SIGMA.odd, then the
first rotor position is given by the corrected rotor position.
[0056] Reference is now made to FIG. 9 showing details of an
example for the phase relationship between the alternating non-zero
magnitude signal Vdshf and the high frequency current signal idshf.
The zero crossing points for the signal Vdshf are marked by a
".smallcircle." symbol. Samples of the signal idshf made at odd
zero crossing points are marked by a ".diamond." symbol. Samples of
the signal idshf made at even zero crossing points are marked by a
".quadrature." symbol. The first summed value .SIGMA.odd is equal
to the summation of the .diamond. samples (i.e., .SIGMA..diamond.),
while the second summed value .SIGMA.even is equal to the summation
of the .quadrature. samples (i.e., .SIGMA..quadrature.). The final
rotor position is given by the following equations:
If .SIGMA..diamond.>.SIGMA..quadrature.; then
.theta.final=.theta.cor-180.degree.
If .SIGMA..diamond.<.SIGMA..quadrature.; then
.theta.final=.theta.cor
[0057] The illustration in FIG. 9 shows an example where the
.SIGMA..diamond. is greater than the .SIGMA..quadrature., and so
the corrected rotor position must be switched by 180.degree. in
order to produce the final rotor position. The final rotor position
.theta.final is output from the position detect circuit 28 (FIG. 1)
for further processing in support of controlling operation of the
motor 12. Conversely, if the signal idshf instead had the opposite
phase from that illustrated in FIG. 9, the .SIGMA..diamond. would
be less than the .SIGMA..quadrature., and so the final rotor
position would equal the corrected rotor position.
[0058] It will be readily understood by those skilled in the art
that materials and methods may be varied while remaining within the
scope of the present disclosure. It is also appreciated that the
present disclosure provides many applicable inventive concepts
other than the specific contexts used to illustrate embodiments.
Accordingly, the appended claims are intended to include within
their scope such processes, machines, manufacturing, compositions
of matter, means, methods, or steps.
* * * * *