EEE - Electric Drives

Topic 13: Vector Control of AC Induction

Motors

Spring 2004

Introduction
Scalar control of ac drives produces good steady state
performance but poor dynamic response. This
manifests itself in the deviation of air gap flux linkages
from their set values. This variation occurs in both
magnitude and phase.
Vector control (or field oriented control) offers more
precise control of ac motors compared to scalar
control. They are therefore used in high performance
drives where oscillations in air gap flux linkages are
intolerable, e.g. robotic actuators, centrifuges, servos,
etc.

Introduction (contd)
Why does vector control provide superior
dynamic performance of ac motors compared
to scalar control ?
In scalar control there is an inherent coupling
effect because both torque and flux are
functions of voltage or current and frequency.
This results in sluggish response and is prone
to instability because of 5th order harmonics.
Vector control decouples these effects.

Torque Control of DC Motors

There is a close parallel between torque
control of a dc motor and vector control
of an ac motor. It is therefore useful to
review torque control of a dc motor
before studying vector control of an ac
motor.

Torque Control of DC Motors (contd)

A dc motor has a stationary field structure
(windings or permanent magnets) and a
rotating armature winding supplied by a
commutator and brushes. The basic
structure and field flux and armature MMF
are shown below:

Torque Control of DC Motors (contd)

The field flux f (f) produced by field current If
is orthogonal to the armature flux a (a)
produced by the armature current Ia. The
developed torque Te can be written as:
'
T

K
I
I
e
t
a
Because the vectors are forthogonal, they are

decoupled, i.e. the field current only controls

the field flux and the armature current only
controls the armature flux.

Torque Control of DC Motors (contd)

DC motor-like performance can be achieved
with an induction motor if the motor control
is considered in the synchronously rotating
reference frame (de-qe) where the sinusoidal
variables appear as dc quantities in steady
state.
Two control inputs ids and iqs can be used for
a vector controlled inverter as shown on the
next slide.

Torque Control of DC Motors (contd)

With vector control:

ids (induction motor) If (dc motor)
iqs (induction motor) Ia (dc motor)
Thus torque is given by:
where

is peak value of sinusoidal space vector.

'

Te K t r iqs K t ids iqs

r r

Torque Control of DC Motors (contd)

This dc motor-like performance is only
possible if iqs* only controls iqs and does
not affect the flux r , i.e. iqs and ids are
orthogonal under all operating conditions
of the vector-controlled drive.
Thus, vector control should ensure the
correct orientation and equality of the
command and actual currents.

Equivalent Circuit of Induction Motor

The complex de-qe equivalent circuit of an
induction motor is shown in the below figure
(neglecting rotor leakage inductance).

Equivalent Circuit of Induction

Motor (contd)
Since the rotor leakage inductance has been
air gapflux.
neglected, the rotor flux
=
, the
r

The stator current vector Is is the sum of the ids

and iqs vectors. Thus, the stator current magnitude,
is related to ids and iqs by:

I$s

I$s ids2 iqs2

Phasor Diagrams for Induction Motor

The steady state phasor (or vector)
diagrams for an induction motor in the
deqe (synchronously rotating) reference frame
are shown below:

Phasor Diagrams for Induction Motor

(contd)
The rotor flux vector r ( m ) is aligned
m
with the de axis and the air gap voltage V
is aligned with the qe axis. The terminal
voltage Vs slightly leads the air gap voltage
because of the voltage drop across the
stator impedance. iqs contributes real
power across the air gap but ids only
contributes reactive power across the air
gap.

Phasor Diagrams for Induction

Motor (contd)
The first figure shows an increase in the torque
component of current iqs and the second figure shows
an increase in the flux component of current, i ds.
Because of the orthogonal orientation of these
components, the torque and flux can be controlled
independently. However, it is necessary to maintain
these vector orientations under all operating
conditions.
How can we control the iqs and ids components of the
stator current Is independently with the desired
orientation ?

Principles of Vector Control

The basic conceptual implementation of vector control
is illustrated in the below block diagram:

Note: The inverter is omitted from this diagram.

Principles of Vector Control (contd)

The motor phase currents, ia, ib and ic are
converted to idss and iqss in the stationary
reference frame. These are then converted to
the synchronously rotating reference frame d-q
currents, ids and iqs.
In the controller two inverse transforms are
performed:
1) From the synchronous d-q to the
stationary d-q reference frame;
2) From d*-q* to a*, b*, c*.

Principles of Vector Control (contd)

There are two approaches to vector control:
1) Direct field oriented current control
- here the rotation angle of the iqse vector with
respect to the stator flux qrs is being directly
determined (e.g. by measuring air gap flux)
2) Indirect field oriented current control
- here the rotor angle is being measured
indirectly, such as by measuring slip speed.

Direct Vector Control

In direct vector control the field angle is
calculated by using terminal voltages and
current or Hall sensors or flux sense
windings.
A block diagram of a direct vector control
method using a PWM voltage-fed inverter
is shown on the next slide.

Direct Vector Control (contd)

Direct Vector Control (contd)

The principal vector control parameters, ids*
and iqs*, which are dc values in the
synchronously rotating reference frame, are
converted to the stationary reference frame
(using the vector rotation (VR) block) by
using the unit vector cose and sine. These
stationary reference frame control parameters
idss* and iqss* are then changed to the phase
current command signals, ia*, ib*, and ic* which
are fed to the PWM inverter.

Direct Vector Control (contd)

A flux control loop is used to precisely control
the flux. Torque control is achieved through
the current iqs* which is generated from the
speed control loop (which includes a bipolar
limiter that is not shown). The torque can be
negative which will result in a negative phase
orientation for iqs in the phasor diagram.
How do we maintain idsand iqs orthogonality?
This is explained in the next slide.

Direct Vector Control (contd)

Direct Vector Control (contd)

Here the de-qe frame is rotating at
synchronous speed e with respect to the
stationary reference frame ds-qs, and at
any point in time, the angular position of
the de axis with respect to the ds axis is
e (=et).
From this phasor diagram we can write:
s

r cos e and qr r sin e

s
dr

Direct Vector Control (contd)

Thus,

cos e
r

s
dr

qrs
, sin e , and r

s 2
dr

The cose and sine signals in correct

phase position are shown below:

s 2
qr

Direct Vector Control (contd)

These unit vector signals, when used in the
vector rotation block, cause ids to maintain
orientation along the de-axis and the iqs
orientation along the qe-axis.

Summary of Salient Features of

Vector Control

A few of the salient features of vector

control are:
The frequency e of the drive is not
controlled (as in scalar control). The motor
is self-controlled by using the unit vector
to help control the frequency and phase.
There is no concern about instability because
limiting within the safe limit automatically
limits operation to$the stable region.

Is

Summary of Salient Features of

Vector Control (contd)

Transient response will be fast because

torque control by iqs does not affect
flux.
Vector control allows for speed control
in all four quadrants (without
additional control elements) since
negative torque is directly taken care
of in vector control.

Flux Vector Estimation

The air gap flux can be directly measured in a
machine using specially fitted search coils or Hall
effect sensors. However, the drift in the integrator
with a search coil is problematic at very low
frequencies. Hall effect sensors tend to be
temperature-sensitive and fragile.
An alternative approach is to measure the terminal
voltage and phase currents of the machine and use
these to estimate the flux. These techniques are
discussed on pp. 363-368 of the Bose text.

Indirect Vector Control

Indirect vector control is similar to direct
vector control except the unit vector
signals (cose and sine) are generated in
a feedforward manner.
The phasor diagram on the next slide can
be used to explain the basic concept of
indirect vector control.

Indirect Vector Control (contd)

Indirect Vector Control (contd)

The ds-qs axes are fixed on the stator and
the dr-qr axes are fixed on the rotor. The
de-qe axes are rotating at synchronous
speed and so there is a slip difference
between the rotor speed and the
synchronous speed given by:

e r sl

Since, e e dt , we can write:

e r sl

Indirect Vector Control (contd)

In order to ensure decoupling between the
stator flux and the torque, the torque
component of the current, iqs, should be
aligned with the qe axis and the stator flux
component of current, ids, should be aligned
with the de axis.
We can use the de-axis and qe-axis
equivalent circuits of the motor (shown on
the next slide) to derive control expressions.

Indirect Vector Control (contd)

Indirect Vector Control (contd)

The rotor circuit equations may be written as:
d dr
Rr idr ( e r ) qr 0
dt

d qr
dt

Rr iqr ( e r ) dr 0

Indirect Vector Control (contd)

The rotor flux linkage equations may be
written as:

dr Lr idr Lmids
qr Lr iqr Lmiqs
These equations may be rewritten as:
Lm
1
idr dr
ids
Lr
Lr
iqr

L
1
qr m iqs
Lr
Lr

Indirect Vector Control (contd)

Combining these with the earlier equations allows
us to eliminate the rotor currents which cannot
be directly obtained. The resulting equations are:

d dr Rr
Lm
dr
Rr ids sl qr 0
dt
Lr
Lr

d qr

Lm
Rr

Rr iqs sl dr 0
qr
where
.
dt
Lr
Lr

sl e r

Indirect Vector Control (contd)

For decoupling control the total rotor flux needs to be

r
aligned with the de-axis and so we want: qr=0 =>
dqr/dt =0
If we now substitute into the previous equations, we
get:

Lr d r
r Lmids
Rr dt

and
where

Lm Rr
has
for dr .
slbeen
substituted
iqs
L
r

Indirect Vector Control (contd)

For implementing the indirect vector control strategy, we
need to take these equations into consideration as well as
the equation:

e r sl

Note:
A constant rotor flux results in the equation:

so that the rotor flux is directly proportional to i ds in steady

L i
state.
m ds
r

Indirect Vector Control (contd)

An implementation of indirect vector control
for 4-quadrant operation is shown below:

Indirect Vector Control (contd)

Features of this implementation:
Diode rectifier front-end with a PWM inverter with
a dynamic brake in the dc link.
Hysteresis-band current control.
Speed control loop generates the torque
component of current, iqs*.

Constant rotor flux is maintained by using the

desired ids*.

The slip frequency sl* is generated from the

desired iqs*.

Indirect Vector Control (contd)

Slip gain Ks is given by:

sl* Lm Rr
Ks *
iqs Lr r

e and e are given by:

and

e sl* r

e dt

The incremental encoder is necessary for

indirect vector control because the slip signal
locates the rotor pole position with respect to
the dr axis in a feedforward manner.

Indirect Vector Control (contd)

If iqs*<0 for negative torque, phasor iqs is

reversed and sl (and sl) will be negative.

The speed control range can be extended

into the field weakening region by
incorporating the dotted line part of the
implementation (see figure below). Note:
Closed loop flux control is now required.

Indirect Vector Control (contd)

Harmonic content of hysteresis-band
current control is not optimum. Also,
at higher speeds the current controller
will saturate in part of the cycle
because of the high back emf.
Synchronous current control can be
used to overcome these problems. See
Bose text, pp. 372-374 for details.

Indirect Vector Control (contd)

A dc motor-like electromechanical model
can be derived for an ideal vectorcontrolled drive using the following
equations:
Lr d r
r Lmids
Rr dt
3 P Lm
Te
r iqs
2 2 Lr

d r
2
Te TL J
dt
P

Indirect Vector Control (contd)

A transfer function block diagram is shown below:

Note: The torque Te responds instantly but the

flux has first order delay (with time constant
=Lr/Rr).

Indirect Vector Control (contd)

The physical principle of vector control can
be explained more clearly with the help of
the below de-qe equivalent circuits:

Indirect Vector Control (contd)

Since ids and iqs are being controlled, we can
ideally ignore the stator-side parameters.
With qr=0 under all conditions, the emf
source on the rotor side de-circuit slqr=0.
This means that in steady state ids flows only
through the magnetizing inductance, Lm, but
in the transient case, is shared by the rotor
circuit whose time constant = Llr/Rr.

Indirect Vector Control (contd)

In the qe-circuit when torque is controlled by
iqs the emf sldr changes instantaneously
(because sl dr Lm Rr iqs / Lr ). Since qr=0, this
emf causes a current (Lm/Lr)iqs to flow
through the rotor resistor Rr. If Llr is

neglected and flux r is constant, ids is seen

to only flow through Lm and iqs only flows
through the rotor side, as desired.

Indirect Vector Control (contd)

A serious issue with respect to indirect
vector control is that of slip gain
detuning. This is due primarily to
variation in rotor resistance. This effect is
illustrated below where Rr=actual rotor
r
resistance
and
R
= estimated rotor resistance.

Indirect Vector Control (contd)

Continuous on-line tuning of Ks is very
complex and computationally intensive.
However, two methods, one based on
extended Kalman filtering (EKF) for
parameter estimation and a second one
based on a model referencing adaptive
controller (MRAC) approach are good options.
The EKF method will be considered later when
studying sensorless vector control but the
MRAC method is described next.

Indirect Vector Control (contd)

In the MRAC approach a reference model
output signal X* that satisfies the tuned vector
control condition is usually a function of i ds* and
iqs*, motor inductances, and operating
frequency. The adaptive model X is estimated
based on motor feedback voltages and currents
as shown in the next slide. X is compared to X *
an the resulting error used to estimate the slip
gain
through a P-I compensator. Slip gain
tuning is achieved when X=X*.

s
K

Indirect Vector Control (contd)

Indirect Vector Control (contd)

Suppose we decide to use torque as the model
parameter X. Thus,
3 P Lm *
X T
r iqs
2 2 Lr
*

*
e

Substituting Lmids* for

gives:
r

2
L
3
P

m * *
X * Te*
ids iqs
2 2 Lr

The actual torque can be estimated from the

stator frame variables using the equation:
3 P
X Te dss iqss qss idss
2 2

Indirect Vector Control (contd)

Note: Lm and Lr parameter variations
affect the estimation accuracy of X*
and at low speeds, the stator
resistance Rs affects the estimation
accuracy of X.

Stator Flux-Oriented Vector Control

Until now we have only considered
rotor flux-oriented vector control.
Airgap flux or stator flux-oriented
vector control is also possible but at a
cost of a coupling effect that requires
decoupling compensation. See Bose
text pp. 381-384 for details.

Vector Control of Current-Fed

Inverter Drive
Vector control can also be extended to
current-fed drives as illustrated below:

Vector Control of Current-Fed

Inverter Drive (contd)
Drive operates with regulated rotor flux and the
speed control loop is the outer loop. The speed loop
generates the torque command Te* which is then
divided by K to generate iqs*. The flux loop

r
generates ids*.
is used to control the firing angle of the phase
$
Is
controlled
rectifier through a feedback loop. The

inverter frequency is controlled by a phase-locked

loop (PLL) so that the stator current, , is
maintained at the desired torque angle with respect
to the rotor flux.
$

Is

Vector Control of Cycloconverter

Drive
Vector control can also be used with a
Scherbius drive with cycloconverter as shown:

Vector Control of Cycloconverter

Drive (contd)
Recall in the Scherbius drive, +sPg is sent to
the line in subsynchronous motoring and -sPg
is sent to the line in supersynchronous
motoring, where sPg is the slip energy.
Currents Ip and IQ are the in-phase and
quadrature current components, with respect
to the slip voltage, Vr. The error from the
speed control loop generates the desired
current, Ip* and IQ* may be set to zero (as
shown).

Vector Control of Cycloconverter

Drive (contd)
The unit vector signals are obtained from
the following equations:
cos sl cos e r cos e cos r sin e sin r
sin sl sin e r sin e cos r cos e sin r
s
s
v
vds
qs
where cos e
and sin e ,
Vs
V s

s
V
and
is the amplitude of the line voltage.

Vector Control of Cycloconverter

Drive (contd)
To illustrate how the drive works, consider a
drive that is accelerating from a
subsynchronous speed with a command
supersynchronous speed.
At subsynchronous speed:
IP >0, sl >0, and sPg>0.
At synchronous speed, sl=0 and IP is dc.
At supersynchronous speed:
IP <0, => sl <0, and sPg<0.