Escolar Documentos
Profissional Documentos
Cultura Documentos
&
Control System Stability
Frequency Response
In practice the performance of a control system is
more realistically and directly measured by its timedomain response characteristics due to the
performance of most control systems is judged based
on the time response due to certain signal.
This is in contrast to the analysis and design of
communication systems for which the frequency
response is more importance, since it is the case,
most of signals to be processed are either sinusoidal
or composed of sinusoidal components.
Slide 2
Frequency Response
In design problems, the difficulties lie in the fact that
there are no unified methods of arriving at a designed
system given the time-domain performance
specifications, such as maximum over-shoot, rise time,
delay time, setting time and so on.
On the other hand, there is a wealth of graphical
methods available in the frequency domain, all
suitable for the analysis and design of the linear
control system.
Slide 3
Frequency response
The frequency response is a representation of the system's
open loop response to sinusoidal inputs at varying
frequencies.
The output of a linear system to a sinusoidal input is a
sinusoid of the same frequency but with a different
amplitude and phase.
The frequency response is defined as the amplitude and
phase differences between the input and output sinusoids.
The open-loop frequency response of a system can be used
to predict the behaviour of the closed-loop system .
Slide 4
Slide 5
o
Ae i
i
School of Engineering (TAFE)
Slide 6
Slide 7
Slide 8
Frequency response
The frequency response method may be less intuitive
than other methods.
However, it has certain advantages, especially in reallife situations such as modelling transfer functions from
physical data.
The frequency response of a system can be viewed two
different ways: via the Bode plot or via the Nyquist
diagram.
Both methods display the same information; the
difference lies in the way the information is presented.
Slide 9
Frequency response
To plot the frequency response, it is necessary to
create a vector of frequencies (varying between zero
(DC) and infinity) and compute the value of the system
transfer function at those frequencies.
If G(s) is the open loop transfer function of a system
and is the frequency vector, we then plot G(j)
versus .
Slide 10
Frequency response
Since G(j) is a complex number, we can plot both its
magnitude and phase (the Bode plot) or its position in
the complex plane (the Nyquist plot).
Phasor
In discussing sinusoidal signals, a convenient way of
representing such signals is by phasors.
Consider a sinusoidal signal y = Y sin t, i.e amplitude Y
and angular frequency , being produced by a radial
line of length Y rotating with a constant angular velocity
and taking the vertical projection y of the line at any
instant of time to represent the value of the sinusoidal
signal.
Information Technology Services
Slide 11
Frequency response
If we have another sinusoidal signal of different
amplitude then the radial line will be of a different length.
If we have a sinusoidal signal with a different phase
then it will start with a different value at time t = 0 and so
the radial line will start at t = 0 at some angle, termed
the phase angle, to the reference axis.
The reference axis is usually taken as the horizontal
axis.
Such lines are termed phasors and the representation is
said to be in the frequency domain.
Slide 12
Frequency response
Slide 13
Frequency response
Slide 14
Frequency response
The magnitude lzl of a complex number z and its angle
is thus given by: Z a b2 and tan 1 ba
We can describe a phasor used to represent a
sinusoidal quantity by a complex number.
If we have y = Y sin t then this is described by a
phasor (Figure a) consisting of just a real number.
Thus, a unit magnitude phasor with phase angle 0 is
represented by 1 + 0j.
2
Slide 15
Frequency response
Slide 16
Frequency response
If we have a phasor of length 1 and phase angle 0 then it will have a complex
representation of 1 + 0j.
The same length phasor with a phase angle of 90 will have a complex
representation of 0 + 1j; rotation of a phasor anticlockwise by 90 corresponds
to multiplication of the phasor by j since j(1 +0j) = 0 + j 1.
If we now rotate this phasor by a further 90, then as j(0+1j) = 0 + 1j we have
the original phasor multiplied by j.
As this phasor is just the original phasor in the opposite direction, it is just the
original phasor multiplied by-1 and so j =-1 and hence j = (-1).
Slide 17
Frequency response
Rotation of the original phasor through a total of 270i.e. 3 x 90 is equivalent to multiplying the original phasor by j = j(j) = -j.
Example 1:
What magnitude and phase is given the phasors described as (a) 3j, (b) 1 + 2j?
Slide 18
Frequency response
(a) The magnitude is 3 units and, since we only have
an imaginary component, the phase is 90.
(b) The magnitude is (a + b ) = (1 + 4) = 2.2 units
and the phase is given by tan = b/a = 2/1 and =
63.4 .
Slide 19
Frequency response
The frequency response of a system is the steady
state response of the system to a sinusoidal input
signal.
The steady state output is a sinusoidal signal of the
same frequency as the input signal differing only in
amplitude and phase angle.
In order to arrive at the principle of the frequency
response function we will consider a simple system
with a sinusoidal input and steady-state sinusoidal
output and recognise that our conclusions can be
applied in a more general way to all systems.
Slide 20
Frequency response
Consider a system where the input x is related to the
output by y = kx.
If we have an input of a sinusoidal signal x = sin t then
the output is y = k sin t and so a sinusoidal signal with
the same frequency but a different amplitude.
Thus, if we represent the sinusoidal signals by
phasors: Output phasor Y k
Input phasor X
Now consider a system where the input x is related to
the output y by the differential equation:
dy
y kx
dt
Information Technology Services
Slide 21
Frequency response
Input phasor X 1 j
This is the definition of a frequency response function as
the output phasor divided by the input phasor.
Information Technology Services
Slide 22
Frequency response
We can compare this with the different equation written
in the s-domain as: sY(s) + Y(s)= kX(s)
Y ( s)
k
Slide 23
Frequency response
Example 2:
Determine the frequency response function for a
system having a transfer function of G(s) = 5/(2 + s).
Replacing s by j gives the frequency response
function of G(j) = 5/(2 + j).
Slide 24
1 j 1 j 1 j 2 2 2
1
But j 2 1, G ( j )
j
1 2 2 1 2 2
G ( j )
Slide 25
2 2
2 2
1
1
1
1 2 2
G ( j ) is referred to as the gain of the system.
G ( j )
Slide 26
Slide 27
Slide 28
Slide 29
2 2 n j n2 n2 2 2 j n
1
2 j
n
n
Slide 30
1
n
1
2
1
2 j
1
n
n
n
gives G(j )
1
n
2
1
n
2 j
2 j
2 j
2
2 j
n
1
n
2 j
n
Slide 31
2
the output is given by : tan -
1-
Slide 32
Graphical Representations of
the Frequency Response of a
System
The characteristics of a system, or of an element of a
system, are determined by the way in which the
amplification or gain A and the phase lag vary with
the frequency of the input signal.
i= A sin t and o= A sin (t+)
There are a number of ways of presenting this
information, each of the various methods having its
own advantages and disadvantages.
Information Technology Services
Slide 33
Graphical Representations of
the Frequency Response of a
System
If the loop or open-loop response of a system is known,
the corresponding closed-loop response can easily be
determined.
For the reason, most of the graphical presentations for
closed-loop system plot the open-loop characteristics
and not the actual closed-loop ones.
In this way, it is possible to check the stability
characteristic of a proposed closed-loop system, before
it is actually connected up as a closed-loop system.
Information Technology Services
Slide 34
Frequency Diagrams
One of the most straightforward methods of presenting
the necessary information is a graphical presentation
of the values of A and plotted against the frequency .
This representation is unsatisfactory, and however,
and the same information can be plotted more usefully
in the form of log10A against log10 and against
log10.
One advantage of these logarithmic plots is that the
information corresponding to very low frequencies (a
few cycles per second) and to very high frequencies
(hundreds of thousands of cycles per second) can all
be conveniently presented on the one graph.
Information Technology Services
Slide 35
Frequency Diagrams
In this logarithmic presentation, the vertical scale of
log10A (i.e. log10 o/i) is usually calibrated in terms of a
unit called a decibel.
The gain in decibels is defined as:
Example:
1. Find the gain for o/i=100
2. When the gain=60 decible, what is the value of o/i.
Slide 36
Slide 37
Nyquist Diagram
The ordinate corresponding to any point on the
diagram are the real and imaginary parts, respectively,
of the particular function being represented.
Since
Slide 38
Nyquist Diagram
As the frequency of the input is varied from = 0 to
= , the values of A (i.e. the length of the vector)
and (i.e. the angle of the vector to the real axis)
vary and the end of the vector sweeps out a curve or
a locus plot.
This curve is the Nyquist diagram of the system.
Slide 39
Response of First-Order
Systems to Sinusoidal Inputs
When a sinusoidal input is fed to a first-order system,
a sinusoidal output is produced.
At low frequencies the signals are transmitted
with little change of amplitude (unless the system has
a steady-state gain represented by K, in which case
the amplitude is approximately K times the input) or
phase.
At the higher frequencies the amplitude is reduced and
the output lags behind the input and as follow
,
A
0,
90.
Information Technology Services
Slide 40
Response of Second-Order
Systems to Sinusoidal Inputs
The response of a second-order system to sinusoidal
oscillations of varying frequency is shown in the
frequency response curves.
Slide 41
Response of Second-Order
Systems to Sinusoidal Inputs
Slide 42
Response of Second-Order
Systems to Sinusoidal Inputs
The corresponding phase lag increases with
increasing frequency, being higher for the higher
values of up to = n, and smaller at the higher
values of > n.
Irrespective of the amount of damping ip the system,
the phase difference is always -90 when = n
(when the applied frequency is equal to the natural
frequency of the system).
When = 0 the gain of the system is K (i.e. the
steady-state gain) and the phase is zero.
Information Technology Services
Slide 43
Response of Second-Order
Systems to Sinusoidal Inputs
As the frequency increases the gain and phase vary,
until at = n, = -90 and the gain = l/2.
With further increase in , the phase increases from
-90 to - 180 , and the gain decreases steadily until,
when = -180, the gain is zero.
Slide 44
Stability Characteristics of
Closed-loop Control Systems
Ideally, the output of a control system should follow
exactly any variations of the input signal.
In practice this is not possible, but, it is desirable that
the system should respond quickly, that steady-state
errors should be eliminated, and that the system
should be a stable one.
The various errors involved depend upon the type of
system and the type of input being considered.
Slide 45
Stability Characteristics of
Closed-loop Control Systems
Various feedback signals can be used to improve the
performance of a particular system, e.g. signals
proportional to the error, to the derivative of the output
(velocity feedback), to the derivative of the error, to
the integral of the output, to the integral of the error,
etc., or any desired combination of these.
A compromise is always necessary in the design of any
system; a high gain produces good response, but may
lead to instability, whereas a large amount of damping
improves the stability but increases the response time
and the magnitude of the steady-state errors.
Information Technology Services
Slide 46
Stability Characteristics of
Closed-loop Control Systems
Slide 47
Stability Criteria
The closed-loop transfer function is given by equation
o
G
i 1 GH
Slide 48
Nyquist Criterion
The critical can be represented on the Nyqist diagram
by the (-1, 0) point and if the locus of the loop transfer
function passes through, or to the left of this point, the
corresponding closed-loop system is unstable one.
Slide 49
Nyquist Criterion
More generally, the Nyquist criterion can be stated by
saying that the locus, when travelled in the direction
from = 0 to = , must pass the (-1, 0) point in
such a way that the point lies to the left of the locus.
The Nyquist diagram of the loop transfer function can
also be used to determine the amount of stability that
would be possessed by the corresponding closed loop
system.
Slide 50
Nyquist Criterion
This stability is expressed in terms of
(a) the gain margin, which is the amount by which the
gain differs from unity when = 180,
(b) the phase margin, which is the amount by which
the phase angle differs from 180 when the gain is
unity.
Slide 51
Routh Criterion
Another criterion which can be used to determine
whether a particular system is stable or not is the
Routh criterion.
This criterion is based on the signs of the coefficients
of the characteristic equation of the system and the
sign of the Routhian Discriminant and, for stability, the
signs must all be positive.
Although it indicates whether a system is a stable or
an unstable one, it does not, unfortunately, give an
exact measure of the amount of stability possessed by
the system.
Slide 52
Routh Criterion
The characteristic equation of a system is obtained from
the expression for the closed loop transfer function
by equating the denominator (i.e. I + GH) to zero.
The solution of the equation
(1 + GH)o = 0
And represents the transient (i.e. free) response of the
system and this is indicative of the overall stability of the
system.
In order that this equation should have no solutions
involving real roots (i.e. no divergent motions), the
coefficients of the equation must satisfy the Routh
criterion.
Information Technology Services
Slide 53
Root-Locus Plots
Slide 54
Root-Locus Plots
The characteristic equation of the closed-loop system is
given by 1 + GH = 0.
As the transfer function can be represented in its most
general form as KF(p) where K is the steady- state gain and
F(p) represents the frequency-dependent part of the gain.
Replacing the loop gain GH in 1+GH=0 by KF(p) we have
the characteristic equation expressed in the form 1+ KF(p)
=0.
Slide 55
Root-Locus Plots
The frequency-dependent part of the gain usually consists of a
numerator and a denominator, both of which are, in general,
functions of p. Thus 1+ KF(p) =0 can be written as:
FN ( P )
FN ( P )
1 K
0
(OR)
K
1
The startingFDpoint
of
the
root
locus
plot
can
be
( P)
FD ( P ) considered to be
Slide 56
Root-Locus Plots
The values of p which satisfy this equation represent the basic
dynamic stability characteristics of the system; they are usually
referred to as the poles of the root locus plot.
When feedback is introduced the values of the roots of the
characteristic equation are changed and as the value of K is
increased the roots sweep out the root locus plot. The ends of the
root locus plot correspond to K =.
In order to satisfy, the equation
, FN(p) = 0 which is the ends
of the root locus plot correspond to the solution of the equation with
the numerator of the loop transfer function equated to zero.
F ( P)
K N
1
FD ( P )
Slide 57
Root-Locus Plots
These points are referred to as the zeros of the root locus plot.
The system is a stable one if the roots lie in the left hand half
of the Argand diagram (i.e. if the real parts of the roots have
negative values), and any points in the right hand half indicate
instability.
The root locus plot technique is particularly useful for
representing the effects on the dynamic characteristics of an
aircraft of moving the C.G (i.e. changing the static stability), or
of changing the gain in an autostabilization system.
Slide 58
Slide 59
Slide 60
Slide 61