U.S. patent application number 10/402107 was filed with the patent office on 2006-04-06 for motor current reconstruction via dc bus current measurement.
Invention is credited to Eddy Ying Yin Ho, Toshio Takahashi.
Application Number | 20060071627 10/402107 |
Document ID | / |
Family ID | 35500484 |
Filed Date | 2006-04-06 |
United States Patent
Application |
20060071627 |
Kind Code |
A1 |
Ho; Eddy Ying Yin ; et
al. |
April 6, 2006 |
Motor current reconstruction via DC bus current measurement
Abstract
A technique for reconstructing phase currents based on
measurements of a DC bus current. Non-observable regions of DC bus
current samples to reconstruct phase currents are reduced in size
by using two phase space vector modulation. The non-observable
regions can be further reduced by omitting deadtime insertions for
switch actuations when output current is higher than a given
threshold. Voltage command vectors in non-observable areas can be
formed by two additive vectors of differing phase and magnitude to
obtain an observable DC bus current reflecting phase current. The
additive vectors have the same combined time average value as that
of the voltage command vector.
Inventors: |
Ho; Eddy Ying Yin;
(Torrance, CA) ; Takahashi; Toshio; (Rancho Palos
Verdes, CA) |
Correspondence
Address: |
OSTROLENK FABER GERB & SOFFEN
1180 AVENUE OF THE AMERICAS
NEW YORK
NY
100368403
US
|
Family ID: |
35500484 |
Appl. No.: |
10/402107 |
Filed: |
March 27, 2003 |
Related U.S. Patent Documents
|
|
|
|
|
|
Application
Number |
Filing Date |
Patent Number |
|
|
60368860 |
Mar 28, 2002 |
|
|
|
Current U.S.
Class: |
318/807 |
Current CPC
Class: |
G01R 19/0092 20130101;
H02M 1/0009 20210501; H02M 7/5395 20130101; H02P 23/14 20130101;
H02P 2205/01 20130101; H02P 8/12 20130101 |
Class at
Publication: |
318/807 |
International
Class: |
H02P 27/04 20060101
H02P027/04 |
Claims
1. A method for reducing a non-observable region in a PWM inverter
drive system, comprising: forming a command voltage using two phase
space vector modulation; increasing a time spent on any given basic
vector to be above a specified threshold; and localizing an amount
of time spent on any given basic vector, such that an available
current measurement time is increased, whereby the non-observable
regions are reduced.
2. The method according to claim 1, further comprising using an
alternate zero vector when a voltage angle is larger than
30.degree. to maximize an available observation interval.
3. The method according to claim 2, wherein zero vector 111 is used
as the alternate zero vector.
4. The method according to claim 1, further comprising eliminating
a deadtime insertion when motor current is higher than a given
threshold, thereby reducing a time period of the non-observable
regions.
5. The method according to claim 1, further comprising sampling a
DC bus current at least three times in each PWM cycle.
6. The method according to claim 5, further comprising: defining a
time interval over which DC bus current samples are obtained during
a PWM cycle; and spacing the DC bus current sample within the
defined interval to permit output phase current reconstruction in
at least two phases.
7. A method for obtaining phase current in a power inverter by
measuring DC bus current, comprising: determining a non-observable
sector having a width; forming a first voltage vector with an
observable phase angle and a magnitude being greater than the
non-observable sector width when a command voltage vector is in the
non-observable sector; forming a second voltage vector having a
phase and magnitude such that said command voltage vector results
from summing said first and second voltage vectors, such that a
time average of the sum of the first and second voltage vectors is
approximately equal to a time average of the command voltage
vector.
8. The method according to claim 7, wherein the non-observable
sector width is related to a DC bus current sample time lag.
9. The method according to claim 8, wherein the non-observable
sector width A is given by the equation A = T min T pwm .times.
.pi. 3 ##EQU2## wherein T.sub.min is the minimum time required to
obtain a DC bus current sample and T.sub.pwm is a PWM cycle time
interval.
10. A processor programmable to produce representations of phase
output current based on DC bus current samples in accordance with
the method of claim 7.
11. A program code for execution by a processor to provide
representations of phase current reconstructed from measuring DC
bus current in an inverter, comprising instructions for operating
the processor in accordance with the method of claim 7.
12. A method for obtaining phase current in a power inverter from
DC bus current, comprising: forming a command voltage vector using
two phase space vector modulation; arranging a combination of basic
voltage vectors to produce the command voltage vector with a
non-observable region having a reduced width; and when the command
voltage vector is in the non-observable region: forming a first
voltage vector in an observable region to have a magnitude greater
than the width; forming a second voltage vector to add to the first
voltage vector; adding the first and second voltage vectors to form
a third voltage vector having approximately a same phase and
magnitude as the command voltage, such that a time average of the
sum of the third voltage vector is approximately the same as a time
average of the command voltage vector.
Description
RELATED APPLICATION
[0001] The application is based on and claims benefit of U.S.
Provisional Application No. 60/368,860, filed on Mar. 28, 2002,
entitled Motor Current Reconstruction Via DC Bus Current
Measurement, to which a claim of priority is hereby made.
BACKGROUND OF THE INVENTION
[0002] 1. Field of the Invention
[0003] The present invention relates generally to motor current
feedback measurements, and relates more specifically to a
computational reconstruction of motor current obtained through
measurement of a DC bus current.
[0004] 2. Description of Related Art
[0005] Inverters for three phase motor drives are well known in the
industry. Typically, a DC bus supplies switched power to different
phases of an AC motor. A design approach used to supply switching
commands and sequences to the inverter involves the use of space
vector modulation. For example, a switch vector plane is
illustrated in FIG. 1 with specific switch states noted at the
vertices of the hexagon.
[0006] With this type of motor control, it is desirable to
accurately measure motor phase current to provide a high
performance control. However, it is often difficult to accurately
measure motor phase current over wide current and temperature
ranges. For example, Hall effect sensors can be used, but are
inherently bulky and costly. In a pulse width modulated (PWM)
inverter drive system, motor phase current can be determined from
measurement of the DC bus current when non-zero basic vectors are
used. Each basic vector is assigned a specific time in a PWM cycle
to generate the command voltage vector. However, if a basic vector
is used only for a very short period of time, motor phase current
cannot be directly determined from the DC bus current. This lack of
observability of motor phase current is due to practical
considerations in the implementation of the PWM inverter drive
system. For example, time delays caused by A/D converter sample and
hold times, slewing of voltage during turn on, and other delay
factors prevent the effects of basic vectors used for a very short
time from being observed.
[0007] In the space vector plane shown in FIG. 1A, the
non-observable regions are illustrated as being located along the
borders of the sections of the space vector plane. Without being
able to observe motor phase currents during these control periods,
it is difficult to achieve a robust and high performance motor
drive.
SUMMARY OF THE INVENTION
[0008] The present invention provides an algorithm for the
reconstruction of motor currents from measurement of DC bus
current. The non-observable operation of motor phase current is
restricted to a much smaller domain with the use of the
reconstructing algorithm. 2-phase space vector modulation permits
the minimum time for an observable effect of a basic vector to be
decreased. In practical application, the time constraint related to
non-observability is cut in half. By reducing this minimum time,
the available time for measurement of phase current according to
this technique is doubled. When the voltage vector angle is large
than 30.degree., the zero vector 111 is used instead of 000. By
using the different zero vector, a switched phase pulse time is
maximized.
[0009] When 2-phase space vector modulation is used, and motor
current exceeds a certain threshold level, dead time need not be
inserted and the time constraint can be further reduced.
[0010] When command voltage falls inside the non-observable
domains, the command voltage vector is formed from two vectors
generated in two PWM periods. One generated vector is a voltage
vector having a phase equal to 30.degree. and a magnitude equal to
two times the width of the non-observable section. Use of this
vector insures the observability of two of the three motor phase
currents. The second vector is added to form the resulting command
voltage vector that falls inside the non-observable domain. The
time average of the two combined vectors are equal to the time
average of the command voltage vector. Using the combination of the
two vectors permits the controller execution cycle to be reduced by
half.
BRIEF DESCRIPTION OF THE DRAWINGS
[0011] The present invention is described in detail below, with
reference to the accompanying drawings having appropriate reference
numeral designators, in which:
[0012] FIG. 1A illustrates a voltage space vector plane with
conventional non-observable zones;
[0013] FIG. 1B illustrates a modified voltage space vector plane
according to the present invention;
[0014] FIG. 2 illustrates a reduced non-observable zone through
2-phase commutation;
[0015] FIG. 3 illustrates a current sampling technique according to
the present invention; and
[0016] FIG. 4 illustrates the insertion of a known voltage vector
to obtain a reference command voltage vector.
DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
[0017] The present invention provides an algorithm to reconstruct
3-phase motor current information from measurement of a DC bus
current supply. A voltage space vector plane 10, as illustrated in
FIG. 1B, contains non-observable regions near sector borders.
According to the present invention, a voltage space vector plane 11
is produced with reduced non-observable zones.
[0018] Referring to FIG. 2(a), a conventional 3-phase inverter
provides a 3-phase voltage space vector PWM modulation. In FIG.
2(b), a same command voltage can be formed using a 2-phase voltage
space vector PWM, as illustrated in commutation diagram 21.
According to the 2-phase vector modulation, a minimum time
constraint for non-observability is cut in half.
[0019] In a two level PWM inverter drive system, eight possible
basic voltage vectors can be produced, and any desired command
voltage vector can be formed by the eight basic voltage vectors.
The desired command voltage vector is limited by the maximum output
voltage of the inverter, as determined by the DC bus voltage level.
In a PWM inverter drive system, motor phase current information can
be determined from the DC bus current when non-zero basic vectors
are used. Each basic vector is assigned a specific time in a PWM
cycle to generate a command voltage vector. If the command voltage
vector is used only for a very short period of time, the motor
current cannot be observed from the DC bus current. The shortness
of this time constraint results from time delays associated with
A/D conversion, including sample and hold times, in addition to
voltage slewing resulting from device turn on. It is this time
constraint that forms the non-observable regions in the voltage
space vector plane illustrated in FIG. 1.
[0020] Referring again to FIG. 2(a), a command voltage vector is
shown in non-observable region of sector 1. In this instance, all
three phases of the motor are PWM driven in a PWM cycle Tpwm. The
total time T2 occurs in two different spots in PWM cycle Tpwm.
[0021] Referring now to FIG. 2(b), a non-observable region is
reduced through the use of a 2-phase voltage space vector
modulation. The same command voltage can be formed by the 2-phase
PWM in FIG. 2(b), as is formed in the system of FIG. 2(a). However,
in the 2-phase voltage space vector modulation, the non-observable
time constraint period is cut in half. The total T2 time is
localized in one spot. Accordingly, the available time for
measurement of current is doubled. When the voltage angle is larger
than 30.degree., the zero vector 111 is used instead of the zero
vector 000 to maximize T1 time.
[0022] With the use of a 2-phase voltage space vector modulation, a
further reduction in the time constraint can be achieved. When
motor current is higher than a given threshold, the need to insert
dead time is eliminated. Typically, the time constraint can be
written as a minimum time Tmin as follows: T .times. .times. min =
Td + T .function. ( d v / d t ) + T .function. ( A / D ) ##EQU1##
where .times. [ Td = [ Tdeadtime abs .function. ( i ) >
Threshold 0 abs .function. ( i ) < Threshold T .function. ( A /
D ) .times. .times. A / D .times. .times. converter .times. .times.
sample / hold .times. .times. time T .function. ( d v / d t )
.times. .times. Power .times. .times. switching .times. .times.
device .times. .times. slew .times. .times. time ##EQU1.2##
[0023] Accordingly, when Td is equal to zero, Tmin is reduced
accordingly.
[0024] Referring now to FIG. 3, the DC bus current is sampled three
times every PWM cycle. In FIG. 3, samples idc1 and idc3 are taken
from the same motor phase, but at different time instance. Current
samples idc1 and idc3 are computed based on the equations shown in
FIG. 3 to provide a timing synchronization with sample idc2.
[0025] Referring now to FIG. 4, the insertion of a known voltage
vector is illustrated in diagram 40. The insertion of the voltage
vector is used to form a command voltage vector Vref. When the
command voltage is inside the non-observable sector bands, the
command voltage vector is formed by two vectors generated in two
PWM periods. Voltage vector V1 has a phase equal to 30.degree. and
a magnitude equal to two times the non-observable sector width A.
By forming vector V1 according to these constraints, observation of
two motor phase currents is insured. Vector V2 is added to vector
V1 to form the resulting command voltage vector Vref. The time
average of V1 plus V2 is equal to the time average of Vref, as
illustrated in FIG. 4. By forming the command voltage vector with
vectors V1 and V2, controller execution cycle is reduced by half
when the command voltage enters the non-observable sector.
[0026] Although the present invention has been described in
relation to particular embodiments thereof, many other variations
and modifications and other uses will become apparent to those
skilled in the art. It is preferred, therefore, that the present
invention be limited not by the specific disclosure herein, but
only by the appended claims.
* * * * *