Professional Documents
Culture Documents
Sc)
Mekelle University
2005
G(s)
+
_
Course Objectives
• To provide a general understanding of the
characteristics of dynamic systems and feedback
control.
3
Systems
Disturbance Inputs
System Outputs
Subsystem
Engineering systems
Biological systems
Information systems
Control Inputs
4
System Variables
• The system’s boundary depends upon the defined
objective function of the system.
• Process:
– (plant) is the main physical component of a control system
– Dynamic system whose output is to be controlled
6
Car and Driver Example
• Objective function: to control the direction and
speed of the car.
• Outputs: actual direction and speed of the car
• Control inputs: road markings and speed signs
• Disturbances: road surface and grade, wind,
obstacles.
• Possible subsystems: the car alone, power steering
system, braking system, . . .
7
Antenna Positioning Control System
• Original system: the antenna with
electric motor drive systems.
• Control objective: to point the
antenna in a desired reference direction.
• Control inputs: drive motor voltages.
• Outputs: the elevation and azimuth of the
antenna.
• Disturbances: wind, rain, snow.
8
Antenna Control System
Functional Block Diagram
Wind force
Antenna System
volts torque Angular
Ref. + volts power position
input _ Diff. Power
amp amp Motor Antenna
Error
volts
Angle
Feedback Path sensor
9
Control System Components
• Control system is interconnection of components that provide a desired
response
• The input and output relationship represents the cause and effect of the process.
• Control systems have similar structure & consists of same basic components.
System or process
– The main physical component of a control system
– Dynamic system whose output is to be controlled
Actuators
– Used to physically influence the process
– Converts the control signal to a power signal
Sensors
– Device provides measurement of the actual system output
– Not perfect and affected by errors
10
…cont’d
Reference input
– represents the desired value of the output
– Which is the objective of the control system
Comparator
Forms the control error signal
Speed sensor
Ks
Feedback Path
12
Open Loop System Characteristics
Assume that each component may be represented by a simple
gain, then
w o Kl (Tm Td )
K a K m Kl u KlTd
The accuracy of the open loop system depends upon the
calibration of the gains and prior knowledge of the
disturbance (choose the control u to give the desired wo ).
Problems:
– nonlinear or time varying gains
– unknown and varying disturbances
13
Closed Loop Characteristics
Now consider the case with feedback.
w o K l (Tm Td )
K a K m K l (w r K sw o ) K lTd
or
Ka Km Kl Kl
w o wr Td
1 Ka Km Kl K s 1 Ka K m Kl K s
14
Closed Loop Characteristics
rad/s volts 0
Ks is the sensor gain in units of volts per rad/s.
The input/output relationship is not very
sensitive to disturbances or changes in the
system gains 15
Closed Loop Characteristics
System Error
The control error is
e (w r K sw o )
Ka Km Kl K s Kl K s
1 w r Td
1 Ka Km Kl K s 1 Ka Km Kl K s
1 Kl K s
wr Td
1 Ka Km Kl K s 1 Ka Km Kl K s
forward gain: Ka Km Kl
feedback gain: Ks
loop gain: Ka Km Kl Ks
closed loop gain: forward gain
1 + loop gain
17
Advantages of closed loop system
• Closed loop system has the following advantages
– Faster response to an input signal
– Effective disturbance rejection
– Low sensitivity to system parameter errors
– Low sensitivity to change in calibration errors
– More accurate control of plant under disturbance and
internal variations
– Used to stabilize systems that are inherently unstable
in the open loop control
18
Disadvantage of closed loop systems
• The following are some of the disadvantages of
closed loop system
– Require the use of sensors which increase the system
costs, volume and weight
– Involve more components which leads to more costs
and complexity
– The power costs due to high gains are high
– More complex to design and harder to build
– Unnecessary when system inputs and the plant model
are known 19
System Dynamics
• Consider a sudden change in the speed reference, wr .
• The output speed, wo will not respond instantaneously
due to the inertial characteristics of the motor and
load, i.e. their dynamic characteristics.
• The motor and load need to be represented by
dynamic equations rather than simple gains.
• The output response will generally lag the input and
may be oscillatory.
20
System Dynamics
Step Responses
Ka = 2 Ka = 20
Step Response, Ka=2 Step Response, Ka=20
1 2
wo
0.9
wr
wo
0.8 1.5
0.7
0.6 1
wr
0.5
0.4 0.5
0.3
Tm
0.2
Tm 0
0.1
0 -0.5
0 2 4 6 8 10 0 2 4 6 8 10
Assume Ks = 1.0
21
Control System Design
Objectives
• Primary Objectives:
1. Dynamic stability
2. Accuracy
3. Speed of response
• Addition Considerations:
4. Robustness (insensitivity to parameter variation)
5. Cost of control
6. System reliability
22
Control System Design Steps
• Define the control system objectives.
• Identify the system boundaries.
– define the input, output and disturbance variables
• Determine a mathematical model for the
components and subsystems.
• Combine the subsystems to form a model for
the whole system.
23
Control System Design Steps
• Apply analysis and design techniques to determine
the control system structure and parameter values of
the control components, to meet the design
objectives.
Section 2.
System Modeling
Tw(s)
+ p(s)
Sa(s) 1 + 1 1
Gc(s) Kh G (s)
Bm p s J m
Ra s La s
+
-
Ks
X2W Helicopter Pitch Control Model
System Modeling
• Purpose of models in control systems:
1. The mathematical model of a system is the
basis for all control system analysis and
design methods.
26
Types of Models
• Physical Models
A model for a given system
– scale models depends upon:
– analogue models • defined system boundaries
• objective of the study
• Mathematical Models • level of approximation
– analytically based required
– experimentally based
27
Types of Models
• A design model will often have many
assumptions and simplifications made to allow
the use of analytical methods (we will normally
require linear, time-invariant models).
30
The Modeling Process
5. Make simplifications to create an
approximate model suitable for control design.
– Linearization of model equations
– Reduce the order of the model by eliminating
unimportant dynamics
– Use lumped parameter approximations for distributed
parameter system.
trade-off
Model Complexity Model Accuracy
31
Electrical Component Models
i
+ voltage/current voltage/charge
v
_
32
Mechanical Translation Models
x
force/velocity force/position
f
M
Mass f = M dv/dt f = M dx2/dt2
x
f
Viscous f=Bv f = B dx/dt
B friction
Spring f = k v dt f=kx
f x
33
Mechanical Rotational Models
J torque/velocity torque/position
T,
Inertia T = J dw/dt T = J d2/dt2
T,
Viscous T=Bw T = B d/dt
B friction
s
Stiffness T = s w dt T=s
T,
34
Transformation Models
i1 i2
v1 v2 v1 N1 i1 N2
Transformer = =
N1 N2 v2 N2 i2 N1
f2 , x2
f1 , x1 f1 L2 x1 L1
L2 Lever = =
L1
f2 L1 x2 L2
T2 , 2
N1 T1 N1 1 N2
Gears = =
T2 N2 2 N1
N2
T1 , 1
35
A Simple Example: R
An RC Circuit vi i vo
Find the transfer function Vo(s) / Vi(s) .
1 1
vi R i i dt , vo i dt
C C
Taking the Laplace
1 1
Vi ( s ) R I ( s ) , Vo ( s ) I (s)
sC sC
then, eliminating I(s)
1 Vo ( s ) 1
Vo ( s ) Vi ( s) or
1 RCs Vi ( s ) 1 s
where, = RC
36
Bandwidth Calculation for the Lab:
Frequency Response:
1
vi Ri i , vo R f i i dt Assumptions: Op Amp
C input zero current
Take the Laplace and solve and voltage.
1
Vi ( s ) Ri I ( s ) , Vo ( s ) ( R f ) I ( s )
sC
1
(R f ) X f (s)
Vo ( s ) sC Vi ( s ) Vi ( s ) G ( s ) Vi ( s )
Ri X i ( s)
Mechanical Example k
M
u(t)
B
spring force f s k x
dx
sliding force f b B v B
dt
net force on mass u f s f b , then
d 2x dx
M 2 u f s f b u k x B , or
dt dt
d 2x dx
M 2 B k x u
dt dt
39
Mechanical Example x(t)
Continued k
M
u(t)
Ms 2 X ( s) Bs X ( s) k X ( s) U ( s)
and solving
1
X ( s) U ( s)
Ms Bs k
2
40
R L
Electrical Example vi i C vo
I (s)
s
Vi ( s ) G ( s )Vi ( s ) • L <=> M
1
Ls Rs
2
• R <=> B
C
• 1/C <=> k
Transfer Function , G(s)
41
Classes of Model Equations
• Continuous differential equations.
• Discrete difference equations.
• Algebraic equations
42
Differential Equations
Partial Ordinary
Linear Nonlinear
44
Linearization of Equations
g
slope
x x xo
yo y=g(x)
Consider the function
y g ( x)
Expand this function about x = xo ,
using a Taylor series expansion. xo
g 1 2g
y g ( xo ) ( x xo ) ( x xo ) 2 H .O.T
x x xo 2! x 2
x xo
Armature Circuit:
dia
eb back emf K bw m , and ea eb Ra ia La
dt
Mechanical Load:
d 2 m d m
Jm 2
the net torque Tm Td Bm
dt dt
51
The DC Servomotor:
Model Equations (variable count)
• There are 7 variables in the equations.
(m , wm , Tm , Td , ia , ea , eb )
K i Ea ( s ) Td ( s )
m(s)
s[ La J m s 2 ( Ra J m La Bm ) s ( Ra Bm K b K i )]
53
The DC Servomotor:
Transfer Function Model
m ( s) Ki
Ea ( s) s[ La J m s 2 ( Ra J m La Bm ) s ( Ra Bm K b K i )]
G(s)
Ea(s) m (s)
G(s)
54
The DC Servomotor:
Block Diagram
Td(s)
Ia(s) Tm(s)+ wm(s) m(s)
Ea(s) 1 1
+ 1
Ra s La
Ki Bm s J m
+ s
--
Eb(s) Kb
• Feedback rule:
G(s) G( s)
+ _
1 GH ( s)
H(s)
56
Block Diagram Reduction
• Feedback rule derivation:
R(s) E(s) C(s)
+ _ G(s) E ( s ) R( s ) H ( s )C ( s )
C ( s) G ( s) E ( s)
H(s)
then
C ( s ) G ( s )[ R( s ) H ( s )C ( s )]
and
This is a rule
C (s) G ( s)
that is used
extensively. R( s) 1 G ( s) H ( s)
57
Servomotor
Block Diagram Reduction
Td(s)
G1(s) G2(s) G3(s) G4(s)
Ea(s) 1
+
1 1
m(s)
+
Ra s La
Ki Bm s J m
+ s
--
H1(s)
Kb
H1(s)
58
Servomotor
Block Diagram Reduction
Next combine the feedback loop.
60
The Signal Flow Graph Method
• The main steps are as follows:
a) construct the signal flow graph either from a
block diagram or from the basic physical
connection of system components (the transfer
functions of the components must be known).
b) Identify and calculate the various paths and
loops in the signal flow graph.
c) With the results from b), apply a formula,
Mason’s formula, to determine the overall
transfer function.
61
Signal Flow Graph Elements
• Nodes, branches and transmission elements
x1 t12 x2 x1 x2
t12
node node
branch
x2 = t12 x1 t12 G12(s)
63
Signal Flow Graph Example:
Servomotor System
Td(s)
G1(s) G2(s) G3(s) G4(s)
Ea(s) x1 1 x2 + x3 1 x4 m(s)
+ 1
Ra s La
Ki Bm s J m
+ s x5
--
H1(s)
Kb
Td(s)
1
Ea(s) 1 x1 G x2 G
1 2
x3 G x4 G4
3
m(s)
-H1
64
Signal Flow Graphs
Definitions
• Source node: only has outgoing branches.
• Sink node: only has incoming branches.
• Path: continuous unidirectional succession of branches
(passes through no node more than once).
• Forward path: a path from input to output.
• Feedback path or loop: originates and terminates at the
same node.
• Non-touching paths: paths with no common nodes.
• Path gain or loop gain: product of branch gains or
transmission functions along the path. 65
Signal Flow Graphs
Mason’s Formula
1
T M k k k 1 p ; p the number of forward paths.
k
where
loop gain products
1 (all loop gains)
of non touching pairs
loop gain products of loop gain products of
non touching groups of three non touching groups of four
M k k th forward path gain
k defined only using loops not touching the k th forward path.
66
Mason’s Formula:
A Systematic Approach
1. Identify all forward paths and write the path gains Mk .
2. Identify all loops and write the loop gains.
3. Identify all non touching loop pairs and write down
the loop gain products.
4. Do the same for groups of 3, 4, … non touching loops.
5. Calculate as defined.
6. Identify all loops not touching forward path k , and
repeat steps 2 -> 5 to calculate k .
7. Apply Mason’s formula to calculate the overall
transfer function.
67
Mason’s Formula:
m (s)
Servomotor Example, E a (s)
Td(s)
Ea(s) 1 x1 G x2 G x3 G x4 G4 m(s)
1 2 3
-H1
• Forward paths:
M1 Ea x1 x2 x3 x4 m Gain = G1G2G3G4
• Feedback loops:
L1 x1 x2 x3 x4 x1 Loop gain = - G1G2G3H1
• Non touching loop pairs: none
68
Mason’s Formula:
m (s)
Servomotor Example, E a (s)
then, = 1 + G1G2G3H1
69
Mason’s Formula:
m (s)
Servomotor Example, Td ( s )
R’(s) 1 G1 x1 G2 x2 G x3 G4
3
x4 G5 C(s)
R(s)
-H1 -H2
• Forward Paths:
M1 R’R x3 x4 C Gain = G6G4G5
M2 R’R x1 x2 x3 x4 C Gain = G1G2G3G4G5
71
Mason’s Formula:
Example 2.8 in P&H
• Feedback loops:
L1 x1 x2 x1 Loop gain = – G2H1
L2 x3 x4 x3 Loop gain = – G4H2
• Non touching loop pairs:
L1 L2 Loop gain = G2G4H1H2
73
Eceg 402
w n2
G(s) 2
s 2Vw n s w n2
Section 3.
Performance Specifications
and Second-Order Systems
S tep Res pons e
1.8
1.6
1.4
1.2
Amplitude
1
0.8
0.6
0.4
0.2
0
0 10 20 30
Time (s ec.)
Performance Specifications
• Consider the time domain characteristics of
systems in terms of special test inputs.
– step input: r (t ) 0 t 0
r(t)
A t >0 A
A
R( s )
s 0 t
– ramp input: r (t ) 0 t 0 r(t)
At t > 0
A
R( s ) 2 slope A units/s
s
0 t
75
Performance Specifications
Test Inputs
r(t)
– parabolic input: r (t ) 0 t 0
At 2 t > 0 At2
A
R( s ) 2
s3
0 t
76
A Second-Order System
• Consider the general solution of the
following second-order system.
c(t)
d 2c(t ) d c(t ) r(t)
2
a b c(t ) r (t ) System
dt dt
R( s) ( s a ) c(0) d c(0) / dt
C (s)
s2 a s b s2 a s b s2 a s b
77
Second-Order System Response
Step Input (real roots)
1 1
C(t) L
s(s a s b)
2
78
Second-Order System Response
Step Input (real roots)
ss s (s s ) s (s s )
1 2 1 1 2 2 2 1
exponential responses
80
Second-Order System Step Response
s 2 s 1t s s 2 t
c (t )
1
1 e 1
e
s 1s 2 s 1 s 2
s 1 s 2
1
1
s 1s 2
C(t)
0.5
s2 s 1t Note:
s 1 s 2
e
0
The total response is the
weighted sum of the
responses associated
-0.5
s1 s 2t with each root of the
e characteristic equation.
-1
s 1 s 2
-1.5
0 1 2 3 4 5 6 7 8 9 10
seconds
81
Fundamental Responses or
System Modes
• The total response is made up of the sum of individual
exponential responses and a constant. These responses
are the system fundamental responses or modes,
e-s1 t and e-s2 t .
• These responses will decay to 37% of their initial value
when t = 1/s1 for the first mode and t = 1/s2 for the
second mode. These times correspond to the time
constants of the modes,
1 = 1/s1 and 2 = 1/s2 .
82
Fundamental Responses or
System Modes
• What would one expect if any of the roots were
positive?
• What governs the speed of the system response?
• Remember, the characteristic equation is define by
the denominator polynomial of the transfer
function, and s1 , s2 are the roots of this
characteristic equation.
83
Second-Order System Response
Step Input (complex roots)
• Use partial fraction expansion to obtain the inverse Laplace.
The roots of the characteristic equation are
a a2
s asb 0 s
2
b s jw
2 4
Then by partial fraction expansion
1 1
s( s a s b) s( s s jw )( s s jw )
2
k1 k2 k3
s s s jw s s jw
where k1 , k2 , k3 are the residues and are calculated as follows:
84
Second-Order System
partial fraction expansion residues
1 1
k1 2
( s s jw )(s s jw ) s 0 s w 2
k2
1
1
1
e j ( / 2)
s( s s jw ) s s jw (s jw )(2 jw ) s 2 w 2 2w
k3 k 2
1
e j ( / 2)
s 2 w 2 2w
w
where tan 1
s
Now the time response is found by applying the inverse
Laplace transform to each term as follows.
85
Second-Order System
Time Response (complex roots)
1 1 j ( / 2) s t jw t
c(t ) e e e
s w
2 2
2w s w
2 2
e j ( / 2)es te jw t
s t
2
1
e {cos(w t / 2)}
s w w s 2 w 2
2
This is a
1 s 2 w 2 s t damped
2 1 e sin(w t )
s w2 w sinusoidal
time response
exponential decay
86
Second-Order System
Time Response (complex roots)
Step Response
1
2 2
s w
1.6
s t sin(w t )
1.4 e A sinusoidal
1.2
waveform with
frequency w and a
Amplitude
0
0 10 20 30 40 50 60
Time (sec.)
87
Second-Order Systems
Natural Frequency and Damping Ratio
2
a a
s2 a s b 0 s b s jw
2 4
• When a = 0 , the damping is zero and the resulting
frequency is the natural frequency, w n b
• The ratio of the damping constant, s to the natural
s
frequency is the damping ratio, V
w n
88
Second-Order Systems
Natural Frequency and Damping Ratio
2
Vw
s 2 ns w n
2
1 1 V
2
where tan
V
89
Standard Second-Order
Step Responses
Step Response
wn = 1.0
1.8
1.6 __ V = 0.1
1.4
__ V = 0.2
1.2
__ V = 0.5
Amplitude
0.8 __ V = 1.0
0.6
w n2
0.4
0.2
0
0 10 20 30
2
Vw
s 2 ns w n
2
Time (sec.)
90
Roots Plotted in the
Complex “s-plane”
• The roots of the characteristic equation, also
known as the system poles, may be plotted
and interpreted in the complex s-plane.
• Consider the complex root pair
s s 1 jw 1
s1
with w n1 w 1 s 1 and V 1
2 2
w 12 s 1 2
91
Roots Plotted in the
Complex “s-plane”
Step Response
1.8
the angle, by
Amplitude
0.8
0.6
s1
0.4
w1 cos( ) V
wn1 w 21 s 2
0.2
0
0 10 20 30
Time (sec.)
s
The natural
-s1 frequency is the
distance from the
-w1
origin to the poles.
w n 1 w 12s1 2
92
Effects of Root Locations
in the Complex “s-plane”
• Moving the roots to the right
reduced damping, similar natural frequency.
• Moving the roots away from the real axis
reduced damping, higher frequency.
• Moving the roots along a constant line
constant damping, higher natural frequency.
• Moving the roots to lie on the imaginary axis
an oscillator, zero damping.
• Moving the roots to the right of the imaginary axis
growing oscillations, an unstable system.
93
Effects of Root Locations
in the Complex “s-plane”
jw Step Response
w1 1.2
Amplitude
0.8
-s1 0.6
0.4
-w1 0.2
0
0 2 4 6 8 10 12
Time (sec.)
94
Performance Specifications
• Stability: jw
– Specified damping ratio or s
a region of the s-plane with
a minimum damping ratio.
– Specified maximum
Step Response
1.6
Mp
M p css
1.4
Amplitude
1
css 0.8
V /
0.6
1V 2
100e
0.4
0.2
0
0 10 20 30
95
Performance Specifications
• Speed of Response:
– Specified bandwidth ( wb corresponding to the
point where the gain is -3db from the low
frequency gain). For a second-order system,
wb = wn when, V = 1/2.
jw
min.
wn
– Specified region in
s
the s-plane with a
minimum wn .
96
Performance Specifications
• Speed of Response (continued):
– Specified rise time, Tr (time from 10% to 90%
of steady-state). 1.8
Step Response
1.6
1.4
Tp
– Specified time to 1.2
0.8
Tp
0.6
w n 1 V
0.4
2
10% 0.2
Tr Time (sec.)
97
Performance Specifications
• Combined stability and speed max.
jw
-s
of response: s
– Specified maximum of the real
part of the roots. Roots to lie to
the left of a boundary value. 1.8
Step Response
1.4
1
s Ts
For 2% , e 0 . 02
0.6
0.4
Ts
4 4
Ts
0.2
then
s Vw
0
0 10 20 30
n 98
Time (sec.)
Performance Specifications
• Steady-state accuracy:
– Specification of the maximum error in steady-
state (after all dynamics have settled).
1.8
1.6
reference
1.4
cref css
1.2
0.6
0.4
0.2
0
0 10 Time (sec.) 20 30
99
Step Responses of
Higher-Order Systems
• The step response of higher-order systems
may be determined through partial fraction
expansion as before.
k0 k1 k2 k2 k3
C ( s)
s s s 1 s s 2 jw 2 s s 2 jw 2 s s 3
where k0 , k1 , … are the residues, then
s t s t s t
c(t ) k 0 k1e 1 k2e 2 w
sin( 2t
) k3e 3
2
Characteristics of
Closed-Loop Systems
feedforward control
Gf(s)
+
R(s) E(s) + C(s)
+
Gc(s) Gp(s)
– cascade compensator
H(s)
Control System Design Steps
• Define the control objectives.
• Define a control system structure
• Define the desired performance specifications.
• Design a controller or compensator to meet the
specifications. plant
error plant
output
signal input Physical
Reference Controller
representing the + Plant
desired measured –
output measured output
Sensor
104
Control System Structures
• Consider the standard control feedback.
R(s) E(s) C(s)
Gc(s) Gp(s)
+
– cascade compensator
H(s)
Then
Gc G p ( s )
C (s) R( s)
1 Gc G p H ( s )
105
Control System Structures
Cascade Compensation
• Ideally, we would like the measured output to
be equal to the reference input, H(s)C(s)=R(s) .
• If Gc(s)Gp(s) = Kc , where Kc ,
then
Kc 1
C ( s) R( s) C (s) R( s)
1 K c H (s) H (s)
108
Control System Structures
Feedback Compensation
• Consider the inner block.
Modified plant
C ( s) G p ( s)
Gp ( s ) dynamic
U ( s) 1 GcG p ( s ) characteristics
then
C ( s) K c G p ( s) K cG p (s)
R( s) 1 K c G p H (s) 1 Gc G p ( s ) K c G p H ( s )
H(s)
110
Eceg 402
Section 5.
Stability
complex GH-plane
complex s-plane mapping s GH(s) Im{GH(jw)}
jw
s Re{GH(jw)}
-1
RHS
Stability Definitions
• Bounded Input Bounded Output Stability:
A system is BIBO stable if, for every bounded
input, the output remains bounded with
increasing time (all system poles must lie in the
left half of the s-plane).
• Marginal Stability:
A system is marginally stable if some of the poles
lie on the imaginary axis, while all others are in
the LHS of the s-plane. Some inputs may result
in the output becoming unbounded with time.
112
Stability Analysis
• To test the stability of a LTI system we need only
examine the poles of the system, i.e. the roots of
the characteristic equation.
• Methods are available for testing for roots with
positive real parts, which do not require the
actual solution of the characteristic equation.
• Also, methods are available for testing the
stability of a closed-loop system based only on
the loop transfer function characteristics.
113
Routh-Hurwitz Stability Criterion
• A quick method for checking BIBO stability.
• Assume the characteristic polynomial is
n 1
Q ( s ) an s an 1s
n
a1s a0
where a0 0 .
• A necessary (but not sufficient) condition for
all roots to have non-positive real parts is that
all coefficients have the same sign.
• For the necessary and sufficient conditions,
first form the Routh array. 114
Routh-Hurwitz Stability Criterion
The Routh Array
n 1
Q( s ) an s an 1s
n
a1s a0
n
s an an-2 an-4 an-6 …
n-1
s an-1 an-3 an-5 an-7 … where
n-2
s b1 b2 b3 b4 … an1 an 2 an an 3
sn-3 c1 c2 c3 c4 … b1
an1
: : : :
2 an 1 an 4 an an5
s k1 k2 b2
1 an 1
s l1
0 etc.
s m1
115
Routh-Hurwitz Stability Criterion
The Routh Array
• In a similar manner, elements in the 4th
row, c1, c2 , … are calculated based on the
two previous rows.
b1 an 3 an1 b2
c1
b1
b1 an 5 an 1 b3
c2
b1
• The elements in all subsequent rows are
calculated in the same manner.
116
Routh-Hurwitz Stability Criterion
Necessary and sufficient conditions:
• If all elements in the first column of the Routh
array have the same sign, then all roots of the
characteristic equation have negative real parts.
• If there are sign changes in these elements, then
the number of roots with non-negative real parts
is equal to the number of sign changes.
• Elements in the first column which are zero
define a special case.
117
Routh-Hurwitz Stability Criterion
Example 1:
Q ( s ) 2 s s 3s 5s 10
4
3
2
4 3 10 10 0
s 2 3 10 0 b1 7 b2 10
3 1 1
s 1 5 0 0 35 10
2 b1 b2 0 c1 6 . 43
s 7
1 c1 0
s 10 ( 6 . 43 ) 0
0 d1 10
s d1 6 . 43
The characteristic equation has two roots with positive real parts
since the elements of the first column have two sign changes.
(2,1,-7,6.43,10)
118
Routh-Hurwitz Stability Criterion
Special Case 1:
• A zero in the first column:
• Remedy: substitute for the zero element, finish
the Routh array, and then let 0 .
Q( s) s 3 3s 2 3 2 2
b1 (negative)
3 1 -3 0 0
s b1 2
c1 2
0() 2
2 0
s b1
1 b1 0
s There are two roots with positive
0
s c1 real parts (1, , -2/ , 2)
119
Routh-Hurwitz Stability Criterion
Special Case 2:
• An all zero row in the Routh array which
corresponds to pairs of roots with opposite signs.
• Remedy:
– form an auxiliary polynomial from the coefficients in
the row above.
– Replace the zero coefficients from the coefficients of
the differentiated auxiliary polynomial.
– If there is not a sign change, the roots of the
auxiliary equation define the roots of the system on
the imaginary axis.
120
Routh-Hurwitz Stability Criterion
Special Case 2 (example):
Q( s) s s s 1
4
3
Auxiliary polynomial
4
s 1 0 -1 0 s2 1
3 1 -1 0 0
s then
s2 1 -1 0 d ( s 2 1)
2s
1 02 0 ds
s
0 d1 d1 = –1
s
The system has one root with a positive real part ( 1, 1, 1, 2, -1).
The root is found from the auxiliary eq. s2 – 1 = 0 , s = ± 1
121
Routh-Hurwitz Stability Criterion
Parameter Range Test
• The Routh-Hurwitz stability criterion may
be used to find the range of a parameter for
which the closed-loop systems is stable.
• Leave the parameter as an unknown
coefficient in the characteristic polynomial,
form the Routh array, check the range of the
parameter such that the first column does
not change sign.
122
Routh-Hurwitz Stability Criterion
Parameter Range Example
Q( s) s 6s 11s 6s K
4
3
2
60 6 K
s4 1 11 K 0 c1 d1 K
3 10
s 6 6 0 0
2 10 K 0 Then for stability,
s
1 c1 0 K >0
s
0 d1 60 6 K > 0 K < 10
s
\ 0 < K < 10
123
Relative Stability Analysis
The Nyquist Criterion
• The Routh-Hurwitz Criterion provides a check
of absolute stability based on the closed-loop
characteristic equation.
• The Nyquist Criterion may be used to analyse
the relative stability of the closed-loop system
based on the loop characteristics.
• Relative stability refers to how close the system
the system is to the absolute stability boundary.
124
Relative Stability Analysis
The Nyquist Criterion
• Consider the general feedback system
R(s) C(s) The closed-loop
+
G(s)
transfer function is
–
C ( s) G ( s)
H(s)
R ( s ) 1 GH ( s )
• The characteristic equation is
1 GH ( s ) 0
• Define F(s) as
F ( s ) 1 GH ( s)
125
The Nyquist Criterion
• F(s) is a rational polynomial in s and can be
written generally as
( s z1 )( s z2 )( )
F (s)
( s p1 )( s p2 )( )
where -zi are the zeroes of F(s) and -pi are the
poles of F(s) .
132
The Nyquist Criterion
• Map F(s) for the Nyquist path enclosing the RHS.
• Then the number of clockwise encirclements of
the origin of the F(s) plane is N = Z – P , where
Z is the number of zeroes of F(s) in the RHS,
P is the number of poles of F(s) in the RHS.
Z, corresponds to the number of poles of the closed- loop
system in the RHS.
P, corresponds to the number poles of the loop transfer
function in the RHS.
133
The Nyquist Criterion
• Normally, P is known. If the loop transfer
function is stable, then P = 0 and N must be zero
for a stable closed-loop system.
• If P is non zero, then there must be P
counterclockwise encirclements of the origin.
• A slight modification of the process is to map the
GH(s) function rather than F(s) and then check
for encirclements of the –1 point in the complex
GH(s) plane. This works since F(s) = 1 + GH(s) .
134
The Nyquist Criterion:
A Simple Example
• Consider a system with the loop TF
K
s-plane GH ( s) GH-plane
s(s a)
jw Im{GH}
j
j mapping
Pole of
GH
s Re{GH}
-1
-j -j
135
The Nyquist Criterion:
A Simple Example
• The number of encirclements of the –1 point in
the GH-plane is zero, N = 0.
• The number poles of the loop transfer function
GH(s) in the RHS, is zero (P = 0). Note: the
Nyquist path excludes the pole at the origin.
• Therefore, the number of poles of the closed-
loop system = the zeroes of 1 + GH(s), in the
RHS, is Z = N + P = 0 + 0 = 0 .
• The closed-loop system is stable.
136
The Nyquist Criterion:
Sketching the GH(s) Function
• In most cases only an approximate sketch of the
GH(s) mapping is required.
• Consider various segments of the Nyquist path.
jw Section I : s = e j ; 0
j
= –90 +90
II III
Section II : s = j0+ j
s
I Section III : s = R e j ; R
IV = +90 –90
-j Section IV : s = – j j0–
137
The Nyquist Criterion:
Sketching the GH(s) Function
• Consider a general loop transfer function
K( s z1 )( s z2 )( )
GH ( s ) k
s ( s p1 )( s p2 )( )
• Section I: ( s = e j , 0 , = – 90º + 90º )
K ( e j
z1 )( )
GH ( e ) k j k
j
e ( e j p1 )( )
as 0 As varies from
– 90º +90º (CCW)
Kz1z 2
GH / k /GH varies from
e p1p2
k j k
k 90º – k 90º (CW)
138
The Nyquist Criterion:
Sketching the GH(s) Function
• Section II: (s = jw , w = 0+ )
w 0 w
GH ( j )
/ k 90
w w
GH ( j ) 0 / (n m)90
139
The Nyquist Criterion:
Sketching the GH(s) Function
• Section IV:
This section is the mirror image of Section II.
• Section III: ( s = Re j , R , = +90º –90º )
j K ( R e j
z1 )( ) K
GH ( R e ) k j k j
( n m ) j ( n m )
R e ( R e p1 )( ) R e
0 / (n m)
II
Sec. III: no effect.
III
s Sec. IV: mirror image of II.
I Sec. I: (k = 2) Then GH(j0)
IV rotates 360º clockwise from
-j GH(j0–) to GH(j0+) with a
magnitude of .
141
The Nyquist Criterion:
Example 1 GH (s) K
s 2 (s a)
Sec. II:
jw s = j 0+ , GH(jw) /-90º
j
s = j , GH(jw) 0 /-270º
II III Sec. III: no effect.
s
I
Sec. IV: mirror image of II.
IV Sec. I: (k = 1) Then GH(j0)
-j
rotates 180º clockwise from
GH(j0–) to GH(j0+) with a
magnitude of .
143
The Nyquist Criterion:
Example 2 GH ( s )
K
s ( s a )( s b)
j Im{GH}
• Where is the “–1” point?
GH(jwc)
• If the “–1” point is inside
the GH path, then
Re{GH}
N=2 , P= 0 and Z=2
\ the system is unstable.
-1
K K
Then, GH ( jw c )
( a b)w c ( a b)ab
2
145
Relative Stability
Gain and Phase Margins
• Gain and phase margins are a measure of
how close the system is to instability.
Im{GH}
Gain Margin: the additional gain
in db, that will make the system
|GH(jwc)|
critically stable.
1
20 log10
Re{GH} GH ( jw c )
-1 fm
Phase Margin: the additional phase
lag in degrees, that will make the
system critically stable.
fm 180
146
The Nyquist Criterion:
Special Example GH (s) K
s2 2
e j
Sec. II:
j 2 s = j 0+ , GH(jw) K/2
s = j , GH(jw) 0 /-180º
Sec. I: Let s j 2 e j
jw
j where 0 ; 90o 90o
II III then K
GH ( s )
I s (j 2 j 2
e ) 2
IV
K
2 j 2 2 e j 2e j 2 2
-j 147
The Nyquist Criterion:
Special Example GH (s) K
s2 2
j Im{GH}
For 0
K
GH ( s )
2 2 e j ( 90 )
Re{GH}
/ ( 90)
-1 K/2
• Section III: no effect
• Section IV: mirror image
-j of II.
There are no net rotations about the “–1” point, therefore, N = 0.
Since P = 0 ( no roots of GH in the RHS), Z = 0 and the closed
loop system is stable for all values of K.
148
r(t) = A sin(wt)
Eceg 402
G(s)
0
Phase (deg)
-50
-100
-150
-200
-250
10-1 100 101 102
Frequency (rad/sec)
Frequency Response Methods
Stability Analysis
• The Nyquist plot provides absolute and
relative stability information about the
closed-loop system, based on a plot of the
loop characteristics of the system.
• The most important section of the Nyquist
plot corresponds to section II of the Nyquist
path, s = jw , with w = 0+ .
• This corresponds to the frequency response
characteristics of the loop transfer function.
150
Frequency Response Methods
Stability Analysis
• In the Nyquist plot, this information is plotted in
polar form, magnitude and phase angle as
frequency is varied.
• This same information can be plotted in two
rectangular plots of magnitude (in dB) vs.
frequency and phase angle vs. frequency (Bode
plots).
• The same Nyquist relative stability criteria can
be interpreted in terms of these rectangular plots
151
Frequency Response of a System
• Consider the response of a system, G(s) to a
sinusoidal input, u(t) = A sin(wt) . Then the
output is C ( s) G ( s ) R( s) G ( s ) Aw
s2 w 2
• Solve for c(t) through partial fraction expansion.
b1 b1 k1 k2
C (s)
s jw s jw s p1 s p2
Aw G ( s ) A G ( jw ) w
A G( j ) e j
b1 ( s jw ) 2
s w s jw
2
2j 2j
152
Frequency Response of a System
w
A G(s) w w
AG( j ) A G( j ) e
j
w
b1 ( s j ) 2
s w 2 s jw 2j 2j
where
/ G ( jw )
G(s)
154
Frequency Response Analysis
Bode Plots
• Bode plots are a pair of rectangular plots of the
magnitude and phase of a transfer function vs.
the log of frequency.
• Plot the magnitude or gain of the system in dB,
20 log10 G( jw ) vs. frequency on a semi-log scale.
• Plot the angle, /G ( jw ) vs. frequency on a semi-
log scale.
• This represents the information from Section II
of the Nyquist plot.
155
Bode Plots: 5 100
G ( s) 2
Example s 1 s 2 s 100
Bode Diagram Constructing Bode Plots
Magnitude (dB)
jw (1
k
)(1 2V )( )
p1 wn wn
157
Constructing Bode Plots:
Asymptotic Approximations
K z1 z2
where the Bode Gain is K B
w
p1 n
2
jw 2V jw w
2
( jw )k
1 1
dB
p1 dB wn wn dB
158
Constructing Bode Plots:
Asymptotic Approximations
• The angle of G(jw) may be written as
jw jw
/ G( jw ) K B 1 1
z1 z2
w 2V jw w
2
( jw ) 1 j
k
1
p1 wn wn
• Thus it is clear that for both magnitude in
dB and the angle, the total transfer function
may be written in terms of the sum of its
components 159
Individual Components:
Poles and Zeros at the Origin
• Poles at the origin: 1
Bode Diagrams ( jw ) k
Plot
1
vs. log 10 w
Magnitude (dB)
40
jw
20 log 10
20 ( jw )k
50
0
The angle is constant at – k90
-50
-90 • Zeroes at the origin: ( jw )k
10
-1
10
0
10
1
10
2 the same, but with the
Frequency (rad/sec)
opposite sign
160
Individual Components:
1
First-Order Poles 1 jw / p
w w
2
1
angle tan
1
gain dB 20 log10 1
1 jw / p dB p p
– for w << p
gain dB 20 log10 (1) 0 , angle 0o
– for w = p
gain dB 20 log10 ( 2 ) 3dB , angle tan -1 (1) 45 o
– for w >> p w
gain dB 20 log10 , angle 90o
p
161
Individual Components:
First-Order Poles and Zeros
First-Order Pole and Zero
• Plot these on a scaled
frequency axis w/p.
Magnitude (dB)
20
1 jw / z
10 • Zeroes are the same but
0 Slope =
-3dB 1 -20 dB/dec.
with the opposite sign.
-10
-20
1 jw / p • Asymptotic
approximations fit the high
Phase (deg)
50
frequency and low
0
-45
-50
frequency characteristics.
10 -1 10 0 10 1
• w = p is the corner
Frequency (rad/sec) w/p frequency of the pole/zero.
162
Individual Components:
1
Second-Order Poles w w
2
1 2V j
• for w << wn wn wn
1
gain dB 20 log10 0 dB , angle 0
1 0 0
• for w = wn
1 1
gain dB 20 log10 , angle 90
1 2V j 1 2V dB
• for w >> wn
1
w
gain dB 20 log10 w w 40 log10 w
2
2V j
w w
n
A slope of 40dB n n
per decade. 163
Individual Components:
1
Second-Order Poles
w w
2
1 2V j
2V wn wn
angle - tan
-1
180
w / wn
40dB/dec.
-20 z = 1.0
6dB
-40
0
Phase (deg)
-50
-100
90
-150
w
-1 0 1
10 10 10
Frequency (rad/sec) wn
165
Individual Components:
Non-Minimum Phase
• Non-minimum phase refers to a system with a
final phase more negative than expected from
the order of denominator and numerator.
• A zero in the RHS, (s + z) has the same
magnitude characteristics as the zero in the LHS,
but has the opposite angle. The final angle is
90 rather than +90 .
• The final phase of a system with a RHS zero will
be 180 more negative than with a LHS zero
166
Individual Components:
s
Non-Minimum Phase e
• A second non-minimum phase component is a
time delay or transportation lag which is
represented by es , where is the delay
time in seconds.
• The frequency characteristic is ejw . The
magnitude is 1.0 (0dB) for all frequencies and so
it does not affect the magnitude plot.
• The phase angle is w. When plotted against
log10w, this results in an exponential decrease in
phase. 167
Bode Plot Construction:
Example 1 10( s 10)
G (s)
s( s 5)(s 2)
• Bode Form:
10 ( j w 10) 10 (1 jw / 10)
G ( jw )
jw ( jw 5)( jw 2) jw (1 jw / 5)(1 jw / 2)
40
1 jw / 10
Magnitude (dB)
20
0 1
-20 1 1 jw / 5
jw
-40
1
-60 1 jw / 2
-90
-100
Phase (deg)
-120
-140
-160
-180
10
-2 -1
10 .2 .5 100 2 5
1
10 10
2
Frequency (rad/sec)
169
10 (–s 10)
Non-Minimum Phase: Example 1A G( s)
s ( s 5)(s 2)
Non-Minimum Phase Example
60
40
Magnitude (dB)
20
-20
-40
-60
-100
–45/dec. –90/dec.
-150
Phase (deg)
-200
–135/dec.
-250
-300
–90/dec. –45/dec.
-350
10-2 10-1 100 101 102
Frequency (rad/sec)
170
Bode Plot Construction:
Example 2 G(s) 5 100
s 1 s 2 2 s 100
• Bode Form:
5 100
G ( jw )
j w 1 j( w )2
2 jw 100
5 1
1 jw / 1 1 0.2 jw / 10 (w / 10 )2
0
-20dB/dec. 14dB
Magnitude (dB)
-20
-40 -60dB/dec.
-60
-50
-45/dec.
Phase (deg)
-100
-135/dec.
-150
-200
-90/dec.
-250
-270
10 -1 10 0 10 1 10 2
p=1 wn = 10
Frequency (rad/sec)
172
Transfer Function Identification
• Frequency response characteristics can be
obtained experimentally by applying
sinusoidal inputs of various frequencies, and
measuring the gain and phase relationships
between input and output.
• By fitting asymptotic approximations to the
frequency response characteristics obtained
from experimental measurements, an
approximate transfer function model of the
system can be obtained. 173
Transfer Function Identification:
Major Steps
1. Determine the initial slope order of poles at
the origin.
2. Determine the final slope difference in order
between the denominator and numerator (nm).
3. Determine the initial and final angle confirm
the results from above or detect the presence of
a non-minimum phase system (delays or zeroes
in the RHS).
4. Determine the low frequency gain (Bode gain).
174
Transfer Function Identification:
Major Steps
5. Detect the number and approximate location
of corner frequencies and fit asymptotes.
– possibly subtract well defined components.
– examine expected -3dB points.
– try to separate second-order terms and use the
standard responses to estimate damping.
6. Sketch the phase plot for the identified
transfer function as a check of accuracy.
175
Transfer Function Identification:
Major Steps
7. Use the phase plot to check for non-minimum
phase terms and to calculate the time delay
value if one is present.
8. Calculate the frequency response for the
identified model and check against the
experimental data (MatLab or a few points by
hand calculation).
9. Iterate and refine the pole/zero locations and
damping of second-order terms.
176
Transfer Function Identification Example 1
40
20
Magnitude (dB) 0
-20
-40
-60
-80
-100
-120
-80
-100
Phase (deg)
-120
-140
-160
-180
-200
-220
10 -1 10 0 10 1 10 2 10 3
Frequency (rad/sec)
177
Transfer Function Model
Identification: Example 1
• Initial slope = 20dB/dec. a 1/jw term.
• Final slope = 40dB/dec. (n-m) = 2 .
• Initial angle = 90 and the final angle is
180 which checks with the gain curve.
• Low frequency gain is found to be |KB/w|dB
= 35dB, where w = 0.1 KB = 5.62 .
• Through asymptotic fitting there are two
poles found at wc = 4 and 25, and one zero
at wc = 70 . 178
Transfer Function Identification Example 1
40
20
Magnitude (dB)
-20dB/dec.
0
-20
-40dB/dec.
-40
-60
-40dB/dec.
-80
-60dB/dec.
-100
-120
4 25 70
-80
-100
Phase (deg)
-120
-140
-90/dec.
-160
-180 -45/dec.
-200 +45/dec.
-220
10 -1 10 0 10 1 10 2 10 3
Frequency (rad/sec)
179
Transfer Function Model
Identification: Example 1
• The estimated transfer function in Bode
form is 5.62(1 jw / 70)
w
G( j )
jw (1 jw / 4)(1 jw / 25)
Magnitude (dB)
20
-20
-40
-60
0
Phase (deg)
-50
-100
-150
-200
10 -2 10 -1 10 0 10 1 10 2
Frequency (rad/sec)
181
Transfer Function Model
Identification: Example 2
• Initial slope = 0dB/dec. no pole at the origin.
• Final slope = 40dB/dec. (n-m) = 2 .
• Initial angle = 0 and the final angle is 180
which checks with the gain curve.
• Low frequency gain is found to be
|KB|dB = 20dB KB = 10 .
• Through asymptotic fitting a simple pole is
found at wc = 0.2, a simple zero at wc = 1.0 and a
complex pole at wn = 5.0 .
182
Transfer Function Identification Example 2
40
Magnitude (dB)
20 -20dB/dec.
15dB.
0
-20 -40dB/dec.
-40
-60
0.2 1.0 5.0
0
-45/dec.
Phase (deg)
-50
-90/dec.
-100
-90/dec.
-150
-200
10 -2 10 -1 10 0 10 1 10 2
Frequency (rad/sec)
183
Transfer Function Model
Identification: Example 2
• The peak at wn = 5.0 is 15dB which
corresponds to a damping ratio of z 0.1 .
• The estimated transfer function in Bode form is
10 (1 jw / 1)
G ( jw )
(1 jw / 0.2)(1 0.2 jw / 5 w 2 /25)
184
Delay Example
20
10
Magnitude (dB)
0
-10
-20
-30
-40
-100
Phase (deg)
-200
-300
-400
185
Transfer Function Model
Identification: Example 2A
• The magnitude plot is the same as the previous
example.
• The angle plot continues to go more negative
and does not asymptotically approach 180 as
expected.
• This is a non-minimum phase system with a
time delay, e–s.
• How do we determine the time delay, ?
186
Transfer Function Model
Identification: Example 2A
• Plot the phase of the transfer function
identified from the magnitude plot and find the
phase difference between this plot and the
phase data.
• This difference represents the e–jwterm.
• At w = 20 rad/sec, –w –/3 0.052 s
(– 60)
• At w = 50 rad/sec, –w – 5/6 0.052 s
(– 150)
187
50 e–0.052s ( s 1)
Transfer Function with Delay G ( s )
( s 0.2)( s 2 s 25)
Phase Data and Identified Bode Plot Without Delay
20
10
Magnitude (dB)
0
-10
-20
-30
-40
0
Phase (deg)
-100
-200 60
150
-300
-400
10 -2 10 -1 10 0 10 1 10 2
Frequency (rad/sec)
188
Bode Plots and Stability Analysis
• In the Nyquist analysis, it became clear that
Section II of the plot was the most critical in
determining the stability of the closed-loop
system.
• The Bode plot of the loop transfer function,
GH(jw) provides the same magnitude and angle
information as Section II of the Nyquist plot.
• Therefore, the Bode plot of GH(jw) can be used
to evaluate the stability of the closed-loop
system. 189
Bode Plots and Stability Analysis
• Consider the definitions of the gain and phase
margins in relation to the Bode plot of GH(jw) .
– Gain Margin: the additional gain required to make
| GH(jw) | = 1 when /GH(jw) = 180 . On the Bode
plot this is the distance, in dB, from the magnitude
curve up to 0dB when the angle curve crosses 180 .
– Phase Margin: the additional phase lag required to
make /GH(jw) = 180 when | GH(jw) | = 1 . On the
Bode plot this is the distance in degrees from the
phase curve to 180 when the gain curve crosses
0dB. 190
8( s 70)
Gain and Phase Margin Example G ( s)
Magnitude (dB) 40 s ( s 4)( s 25)
20
0
-20
GM = 19dB
-40
-60
-80
-100
-120
-80
-100
Phase (deg)
-120
-140
-160 fm = 40
-180
-200
-220
10 -1 10 0 10 1 10 2 10 3
Frequency (rad/sec)
191
Closed-loop Frequency Response
• Assuming the closed-loop system is stable, the
frequency response of the closed-loop system will
directly give the bandwidth of the system, i.e. 3dB
down from the steady-state gain.
• This will be related to the zero crossing frequency
of the loop transfer function plot.
• Information about the overshoot of the step
response can also be obtained from the peak of
the magnitude curve in the closed-loop Bode plot.
192
Closed-Loop Bode Plot
0
3dB
Magnitude (dB)
-10
-20
-30
-40
-50
-60
wB
G(s)
0
1 G ( s)
Phase (deg)
-50
-100 8( s 70)
-150 s 3 29 s 2 108s 560
-200
10-1 100 101 102
Frequency (rad/sec)
193
Eceg 402
R(s) E(s)
1 s
C(s)
+
Kc
1 s
Gp(s)
– cascade compensator
H(s)
Design Criteria and Specifications
• Design Criteria: • Frequency domain:
accuracy
– bandwidth
stability (absolute and relative)
– phase margin
speed of response
– gain margin
• Time domain: • S-plane:
– steady-state error
– damping ratio, z
– % overshoot
– natural frequency, wn
– rise time, Tr
– damping constant, s
– time to 1st peak, Tp
– settling time, Ts
195
Effects of Cascade a Controller
R(s) E(s) C(s)
Kc Gp(s)
+
– cascade compensator
H(s)
-80
increasing the bandwidth.
0
• For Kc = 1, the phase
-50 margin is about 55, but
Phase (deg)
-100
-150 fm
with Kc = 3, the phase
-180
-200 margin becomes -10 ,
-250
-2 -1 0 1 2
resulting in an unstable
10 10 10 10 10
Frequency (rad/sec) closed-loop system.
198
Effects of Increasing Kc
Nyquist Diagrams
• On the Nyquist plot,
1.5
1
increasing the gain Kc from 1
Imaginary Axis
0.5
Kc = 3 to 3 results in encirclements of
0 the –1 point and the system
–1
-0.5 becomes unstable. Note with
-1 Kc = 1, the gain margin is
-1.5
about 2.5 or about 8 dB.
-1.5 -1 -0.5
Real Axis
0.5
Kc = 3 the system in the frequency
0
–1
range where |GH(jw)| = 1.
-0.5 fm
-1
• This may be done through
-1.5 cascade phase-lead
-1.5 -1 -0.5 compensation.
Real Axis
200
Phase-Lead Compensation
fm = 16 at wc = 9.9 rad/s
Gm=8.1303 dB (at 15.324 rad/sec)
Pm=15.772 deg. (at 9.9897 rad/sec)
20
• Add phase-lead
compensation to the
Magnitude (dB)
0
-20
-40 controller.
1 0 .6 s
Gc ( s ) 3
-60
1 0.017 s
-80
-100
50
compensator • This introduces phase
0 advance of 70 and a
Phase (deg)
-50
-100
uncompensated
gain increase of 15 dB
-150
-200
compensated at the new zero-crossing
-250
-2 -1 0 1 2 3
frequency, wc = 9.9
10 10 10 10 10 10
0
compensator
-20 controller.
-40
1 1.8s
-60
Gc ( s ) 3
-80
compensated 1 3.3s
0 • This introduces gain
-50 reduction at and a
Phase (deg)
-100
-150
small phase lag at the
-200 new zero-crossing
-250
-2 -1 0 1 2
frequency, wc = 3.8
10 10 10 10 10
203
Phase-Lead Compensator
Maximum Phase Advance, m
w w
• The phase lead is tan 1
tan 1
z p
• Find the maximum.
d 1 1 1 1 z ( p 2 w 2 ) p( z 2 w 2 )
0
dw w z
2
w p
2
( z w )( p w )
2 2 2 2
1 1
z p
• Factoring the numerator
( z p)( m pz ) 0 w m
w Design
2
pz
equation
• The maximum phase advance is
p z
m tan 1 tan 1
z p 204
Phase-Lead Compensator
Maximum Phase Advance, m
p 1/ 1
• Since, , tan m
z 2 2
1
then sin m Design equation
1
-5
-10
advance occurs at
-15 ½ | |dB wm = zp.
m
z p • The gain at wm is
40 ½ | |dB above the low-
Phase (deg)
20
frequency gain, | 1/ |dB.
0
• Maximum phase advance is
10-2 10-1
zp
101 102
defined by 1
Frequency (rad/sec) sin m
206 1
Phase-Lead Compensator
Implementation
R2
Passive circuit: (1 s R1C1 )
R R2
C1 Eo ( s ) 1
Ei ( s )
R1R2C1
1 s
R1 R2
then
ei R1
R2 eo
Eo ( s )
1 1 s
1 s
Ei ( s )
R2 where
Eo ( s ) Ei ( s )
R1 R1 R2
R2
1 s R 1C1 RRC 1 2 1
,
R1 R2 R2
207
Phase-Lead Compensator
Implementation
• The full controller • The relationship to the
includes the passive pole-zero form is
circuit and an amplifier.
1 1 s
p1 , 1
, p
Gc ( s ) K c z z
1 s • The maximum phase
• The low frequency gain is advance occurs at
1
Kc w m pz 1
208
Bode Plot Design of Phase-Lead
1 1 s
Compensators G (s) K 1 s
c c
wc , fm m ½| |dB
Bode specs.
sin m
1 h.f. gain
plot 1 increase
211
Bode Plot Design of Phase-Lead
Compensators: Design Steps
4. Required :
– Select such that, w m 1
is at the new
predicted zero-crossing frequency.
5. Required Kc :
– Select Kc to give the required gain increase in
the loop transfer function. Remember, the
compensator low-frequency gain is Kc/.
6.Implement: G (s) K 1 1 s
c c
1 s
212
Phase-Lead Design Example 1
• Steady-state error:
• The plant
1 1
1 E ( s) R( s )
G p ( s) 1 KGp (s) sK
s(s 2)
2
s
s2
1
• Specifications:
ess lim s E( s) < 0.05
s0 K /2
– steady-state error <5%
for a ramp input. K > 40
– Phase margin >45° • Draw the Bode plot of
40
G p ( s)
s(s 2)
213
Phase-Lead Design Example 1
Phase-Lead Design Example
• Initial fm = 18° , so
20
choose m = 45° – 18°
Magnitude (dB)
0 4.8dB
+ 3° = 30° .
-20 • Then sin(30o ) 0.50
1
-40
3.0
8.5 1
-120
• The gain increase is
½| |dB =4.8 dB .
-135
Phase (deg)
-140
27°
-160 • The new predicted
-180
18° 13° wc = 8.5 rad/s and
100 101
Frequency (rad/sec)
102
fm(uncomp) = 13°
214
Phase-Lead Design Example 1
• Second iteration: • Calculate
1
choose m = 45° – 13° w m w c 8 .7
+ 3° = 35° .
1
0.06
• Then sin(35o ) 0.57 8.7 3.7
1 • Calculate the gain Kc
3.7
1 1
Kc 40 K c 40 148
• The gain increase is
½| |dB =5.7 dB . • The final compenstor
• The new predicted 1 1 0.22s
wc = 8.7 rad/s and Gc ( s ) 148
3.7 1 0.06s
fm(uncomp) = 12° amplifier
OK 215
Phase-Lead Design Example 1
Final Bode Plot
Phase-Lead Compensation Design
40
Magnitude (dB)
20
-20
-40
8.7rad/s
-100
Phase (deg)
-120
-140
compensated
uncompensated 47°
-160
-180
10--1 10 0 10 1 10 2
Frequency (rad/sec)
216
Phase-Lead Design Example 1
Closed-Loop Step Response
Closed-Loop Step Response
1.6
compensated
1.4
uncompensated, K=40
1.2
Amplitude
0.8
0.6
uncompensated, K=1
0.4
0.2
0
0 1 2 3 4 5
Time (sec.)
217
Phase-Lag Compensator
Characteristics
• Time constant form. • Pole-zero form.
1 s Kc s z
Gc ( s) Kc Gc ( s )
1 ' s ' s p
amp. passive 1 , 1
z
gain network where p
'
where ’ > 1 Note: | p | < | z |
218
Phase-Lag Compensator
Characteristics
• Consider the phase lag, one decade above
the zero of the compensator, w = 10 z.
lag tan
10 z
1
tan 1 10 z ’ lag
p z 2 2.8
tan 1 10 ' tan 1 10
5 4.5
tan 1 10 ' 84.3o
50 5.6
219
Phase-Lag Compensator
Bode Plot
Lag Compensator z
0
• Note : '
p
Magnitude (dB)
-20
-50
| 1/ |dB
-2 -1 0 1 2
10 10 10 10 10
Frequency (rad/sec)
220
Phase-Lag Compensator
Implementation
Passive circuit: 1 R2C1 s
Eo (s ) Ei (s )
1 ( R1 R2 )C1 s
R1
ei R2 then
eo
C1
1 s
1 ' s
Eo ( s ) Ei ( s )
1
R2 where
s C1
Eo ( s ) Ei ( s ) R2
R1 R2
1 RC 2 1 , '
R1
s C1 R2
221
Bode Plot Design of Phase-Lag
1 s
Compensators G (s) K
1 ' s
c c
222
Bode Plot Design of Phase-Lag
Compensators: Design Steps
1. Loop gain requirement:
– Determine the loop gain required to meet the steady-
state error specifications.
2. Bode plot:
– Plot the loop transfer function including the required
gain increase from step 1.
3. Required zero-crossing frequency: (calculate )
– Find the frequency, wc where the specified phase
margin + 5 would be met, if it were the zero-
crossing frequency. 223
Bode Plot Design of Phase-Lag
Compensators: Design Steps
– Determine the gain reduction required to make wc
the zero-crossing frequency.
– Calculate from the required gain reduction =
| |dB .
4. Required :
– Select such that the zero, 1/ is one decade
below the new zero-crossing frequency.
1 w 'c
10
224
Bode Plot Design of Phase-Lag
Compensators: Design Steps
5. Required Kc :
– Select Kc to give the required low-frequency
gain increase.
6.Implement:
1 s
Gc (s) Kc
1 ' s
225
Phase-Lag Design Example 1
• The plant • Steady-state error:
1 1 1
G p ( s) E ( s) R( s )
s(s 2) 1 KGp (s) s 2
sK
s2
• Specifications:
ess lim s E( s)
1
< 0.05
s0
– steady-state error <5% K /2
for a ramp input. K > 40
– Phase margin >45°
• Draw the Bode plot of
• The first part of the 40
G p ( s)
s(s 2)
design is the same as
the phase-lead design. 226
Phase-Lag Design Example 1
• Initial fm = 18° at wc =
Phase-Lag Design Example
60
6.5 rad/s.
Magnitude (dB)
• Frequency at which fm
40
20
20dB = 45° + 5° = 50° (where
0
1.7 6.5
the phase plot = –130°)
-100
wc = 1.7 rad/s
Phase (deg)
5.88
228
Phase-Lag Design Example 1
Final Bode Plot
Phase-Lag Design Example
60
uncompensated
Magnitude (dB)
40
20
0
compensated 20dB
-20
-100
Phase (deg)
-120
-135
-140
-160
10--2 10--1 10 0 10 1
Frequency (rad/sec)
229
Phase-Lag Design Example 1
Closed-Loop Step Response
Lead and Lag Design Step Response
1.6 uncompensated, K=40
1.4 lead compensated
1.2 lag compensated
Amplitude
0.8
0.6
uncompensated, K=1
0.4
0.2
0
0 1 2 3 4 5
Time (sec.)
230
Characteristics of Lead and
Lag Compensators
Lead Compensator Lag Compensator
– uses phase advance at – uses gain reduction to
the zero-crossing shift the zero-crossing
to the left
– increases high
frequency gain and – introduces phase lag at
shifts wc to the right frequencies below wc
– increases bandwidth – reduces bandwidth
– limited amount of – very large amount of
compensation possible compensation possible
231
Characteristics of Lead-Lag
Compensators G (s) K 1 s 1 s '
c c
1 s 1
' '
s
Lead-Lag Compensator Characteristics
0 lead lag
Magnitude (dB)
-5 1
'
1 1 = 0.4472 22.36
-10
dB
' dB 2 dB
=5 = 8
-15
1 1 1 1
= 2.236 178.9
40
' '
'
m m = 40°
Phase (deg)
wm = 1
20
0
1 =1
-20
-40 For a constrained
10 -3 10 -2 10 -1 10 0 10 1 10 2 implementation, =
Frequency (rad/sec)
232
Lead-Lag Compensator
Implementation ( = ´)
C1 (constrained)
1
R1 R2
s C2
ei
R2
eo Eo ( s ) Ei ( s )
R1
C2 R2 1
1 s R1C1 s C2
' R C
1 1
R2C2
• An unconstrained lead-lag
R1C1 R2C2 R1C2 network, ´ can be
implemented as separate
• This results in a lead and lag networks
constrained lead-lag buffered by an op-amp.
network, = ´ .
234
Bode Plot Design of Lead-Lag
Compensators G (s) K 1 s 1 s '
c c
1 s 1
' '
s
6. Required Kc : ' 10
– Select Kc to give the required low-frequency
gain increase.
s 1 ' s
7.Implement: Gc ( s) K c
1
1 s 1
' '
s
238
Bode Plot Design of Lead-Lag
Compensators: ( = ´ )
• For the constrained design ( = ´ ) steps 3
and 4 may require some iteration similar to
the lead design.
• In this case the gain reduction is
| 1/ |dB + ½| |dB = – ½| |dB
239
Lead-Lag Design Example 1
• The plant • Steady-state error:
1 1 1
G p ( s) E ( s) R( s )
s(s 2) 1 KGp (s) s 2
sK
s2
• Specifications:
ess lim s E( s)
1
< 0.05
s0
– steady-state error <5% K /2
for a ramp input. K > 40
– Phase margin >45°
• Draw the Bode plot of
• The first part of the 40
G p ( s)
s(s 2)
design is the same as
the phase-lead design. 240
Lead-Lag Design Example 1
Lead-Lag Design (unconstrained)
• Choose m = 20° , then
= 2.0, ½| |dB = 3dB.
60
Magnitude (dB)
40
20
• fm(uncomp)
0
9 dB 45° – 20° + 5° = 30°
wc = 3.5 rad/s
-100
• Gain reduction = -9dB
Phase (deg)
-120
= | 1/ |dB+½| |dB
-140
-150
-160
\ | 1/ |dB = -12dB
10-2 10-1 100 3.5 101 and 4
Frequency (rad/sec)
241
Lead-Lag Design Example 1
• Lead time constant • Controller gain
1
wc = 3.5 =
2
Kc = 40
= 0.20
• Final Controller
• Lag time constant
1 wc' 1 0.4 s 1 2.86 s
= 0.35 Gc ( s ) 40
' 10 1 0.2 s 1 11.44 s
2.86
242
Lead-Lag Design Example 1
m = 45 wc´ = 3.5 rad/s
60
uncompensated
Magnitude (dB)
40
20
0 compensated
-20
-40
-100
Phase (deg)
-120
-135
-140
-160
-180
10 -2 10 -1 10 0 3.5 10 1 10 2
Frequency (rad/sec)
243
Lead-Lag Design Example 1
Constrained ( = ´ )
Lead-Lag Design (unconstrained)
• For m = 20° , = 2.0
60
4
Magnitude (dB)
20
9 dB
= 2.7 2.7
0
• The gain reduction is
-100 – ½| |dB = – 4.3 dB
Phase (deg)
-100
Phase (deg)
-120
compensated -133
-140
-160 uncompensated
-180
10-2 10-1 100 4.7 101 102
Frequency (rad/sec)
246
Closed-Loop Step Responses
Step Response
1.2
1
• Specifications: ess lim s E( s ) 0.02
s 0 1 10K
– steady-state error <2%
for a step input. K 4.9
– Phase margin >60°
• Draw the Bode plot of
245
G p (s)
( s 0.2)( s 2 5s 25)
248
Compensator Design Example 2
Bode Diagram for Example 2 • fm(uncomp) = – 27° at
wc = 6.7 rad/s
20
15
Magnitude (dB)
• A lead compensator
10
5 13dB
0
-5
would need to
-10 provide > 87° phase
-100
advance.
-115
Phase (deg)
-120
-140 • A lag compenstor
-160
-180
fm = – 27°
needs to introduce a
-200
-220 high frequency gain
100 2.3 6.7 101 reduction of 13dB .
Frequency (rad/sec)
249
Lag Compensator Design
Example 2
• The gain reduction at • The time constant is
wc´ = 2.3 rad/s to make 1/´ = wc´/10 = 0.23
this the zero crossing is
´ = 4.35
–13dB. The phase at
this frequency is –115°. • The gain is
(–180° + 60° + 5°) Kc = 4.9
• |1/´|dB = –13dB • The final controller is
´ = 4.6 1 4.35 s
Gc ( s ) 4.9
1 20 . 0 s
250
Lag Compensator Design
Example 2
Lag Design for Example 2
Magnitude (dB) 20
10
13dB
0
-10
-20
2.3
-100
-120
Phase (deg)
-140
fm = – 60°
-160
-180
-200
-220
-240
10 0 101
Frequency (rad/sec)
251
Lead-Lag Compensator Design
Example 2
Bode Diagram for Example 2 • Choose m = 65° , then
= 20.3, and
20
15
Magnitude (dB)
10
5
½| |dB = 13.1dB.
5.5dB
0
-5
• fm(uncomp)
-10 60° – 65° + 5° = 0°
-100
-115
wc = 5.1 rad/s
Phase (deg)
-120
-140
-160
• Gain reduction = – 5.5dB
-180
-200
fm = – 27° = | 1/ |dB + ½| |dB
-220
\ | 1/ |dB = – 18.6dB
100 2.3 5.1 6.7 101
Frequency (rad/sec) and 8.5 252
Lead-Lag Compensator Design
Example 2
• Lead time constant • Controller gain
1
wc = 5.1 =
20.3
Kc = 4.9
= 0.0435
• Final Controller
• Lag time constant
1 wc' 1 0.883 s 1 1.96 s
= 0.51 Gc ( s ) 4.9
' 10 1 0.0435 s 1 16.66 s
1.96
253
Lead-Lag Compensator Design
Example 2
Lead-Lag Design for Example 2
20
15 uncompensated
Magnitude (dB)
10
5 compensated
0
-5
-10
5.1
-80
-100
Phase (deg)
-120
-140
fm = 60°
-160
-180
-200
-220
-240 0 1
10 10
Frequency (rad/sec)
254
Design Comparison
Example 2
Bode Diagrams for Example 2
20
Magnitude (dB)
0
-20 uncompensated
-40
lead-lag design
-60
-80
lag design
2.3 5.1 6.5
0
-50
Phase (deg)
-100
-120
-150
-200
-250
10 -2 10 -1 10 0 10 1 10 2
Frequency (rad/sec)
255
Design Comparison
Example 2
Step Responses for Example 2
0.8
___ uncompensated
Amplitude
0.6
___ lead-lag design
___ lag design
0.4
0.2
0
0 2 4 6 8 10
Time (sec.)
256
Lead-Lead Compensation Design
• One of the limitations of lead compensators
is that the phase advance is limited to about
70 - 80°.
• To overcome this limit, consider cascading
two lead compensators.
• The design follows as with a single lead
except the phase advance is doubled, to 2m
and the gain increase at the new zero
crossing is also doubled, to | |dB .
257
Lead-Lead Design for Example 2
• Initial fm = – 27° , so
Lead-Lead Design for Example 2
20
-60
fm(uncomp) = – 69°
-200
-250 -249
1 1 0.283 s
Gc ( s) 70.8
fm= – 75°+140° = 65 ° 32 1 0.00884 s
-40
-60
-50
-100
-115
fm = 65°
Phase (deg)
-150
-200
-250
10 0 101 20 10 2
Frequency (rad/sec)
260
Closed-loop System
Step Responses for Example 2
Step Response Comparison
0.8
Amplitude
0
0 1 2 3 4 5
Time (sec.)
261
PID Control Design In The
Frequency Domain
Kp (1 s ) where K
d Gc ( j w) Kp (1 j w)
Kp
• This is a lead Gc ( j w) Kp 1 2w 2
Gc ( j ) tan ( w)
compensator with the
w 1
pole at w .
263
PD Frequency Domain Design
PD Controller Reference Plot : Kd = 1 , Kp = 1
40 __ ideal
• PD controller with a
high frequency cut-off.
Magnitude (dB)
1 d s
K K sK s
10
|Kp|dB 0
p p d d
1 s
1/ 1/d 1/d
80 d
1 s
Phase (deg)
60
K
1 s
40 p
d
20
where (
K d
10
-1
10 0 101 102 d )
Frequency (rad/sec) Kp
264
PD Frequency Domain Design
• The design process is • Check the new zero
similar to the lead crossing frequency, wc
compensator design. and the corresponding
phase margin (iterate).
• Choose the phase
advance and determine • Determine from
w from tan1 ( w) tan ( wc )
1
• Determine the gain • Determine Kp from the
increase from low frequency gain
Gc ( j w) Kp 1 2w 2
requirements.
• 1/d = 100 1/ 265
PI Frequency Domain Design
• PI Controller Kp
(1 s) • Note that the PI
Ki Ki compensator introduces
Gc ( s ) K p Ki
s s
(1 i s )
a pole at the origin
Ki which increases the
s
system type by 1.
• This is a lag
compensator with the • The high frequency
pole at w = 0 and the gain Kp .
zero at Kp
i
Ki
266
PI Frequency Domain Design
40
PI Controller : Ki = 1 , Kp = 1
• The design is similar to
the lag compensator.
Magnitude (dB)
30
s (1 d s )
K (1 ( K K ) s 1 K ( K K ) s ) 2
i p i d i p d d
s (1 s ) d
K (1 s )(1 s )
where, ( K K ) K
s (1 s )
i 1 2 i
1 2 p d d
d K K 1 2 p i d
268
PID Frequency Domain Design
PID Controller: 1 =10 , 2 = 1 , Ki = 1 • The PID controller is
60
similar to a lead-lag
Magnitude (dB)
50
40 compensator, with
30 the lag pole at zero
1 / 2 w2
|Kp|dB and the lead pole at
1 / 1
20
50
1/d .
• w2 represents a
Phase (deg)
-50
frequency relative to
1/2 for a phase
10 -3 10 -2 10 -1 10 0 10 1
Frequency (rad/sec) advance of 70° .
269
PID Frequency Domain Design
• For the PID design, first • Select 1/d two decades
choose a phase advance above 1/2 .
and calculate w2from • Plot the compensated
tan (2 w)
1
system with Ki = 1 and
• Then determine the determine the gain
frequency, wc´ where reduction required to
fm(uncomp) fm(spec) – + 5 .
° make wc
´ the zero
crossing frequency.
• Calculate 2 from wc and .
´
10
w 2 = tan(70 °) = 2.75 .
5
0
-5
-10 • Find the frequency
-100
where, fm(uncomp)
60° – 70° + 5° = – 5°
Phase (deg)
-120
-140
-160
-180 -185
wc = 5.2 rad/s
• Find 2 from wc´ 2 =
-200
-220
5.2
2 = 0.529
100 101
Frequency (rad/sec) 2.75 271
PID Frequency Domain Design:
Example 2
• Choose 1/1 one • Plot the systems with
decade below 1/2 . the compensator
1/1 = 1/102 Gc ( s) Ki
(1 5.29s )(1 0.529s)
1 = 5.29 . s (1 0.00529s )
20 uncompensated
28.5dB
margin of 60°. wc = 5.2
10
0
rad/s
-10
compensated, Ki = 0.0376
• Find the required gain
reduction to make this
the zero crossing
Phase (deg)
-100
-120
-150 frequency. 28.5dB
-200 • Set Ki = the gain
10 0 5.2 101
reduction. Ki = 0.0376
Frequency (rad/sec)
273
PID Frequency Domain Design:
Step Responses for Example 2
Step Responses for PID and Lead-Lag
1
0.9
0.8
0.7
__ PID
Amplitude
0.6
0.5
__ Lead-Lag
0.4
0.3
0.2
0.1
0
0 1 2 3 4 5
Time (sec.)
274
Eceg 402
Section 8.
Root Locus Analysis jw
p2
1 KGH ( s ) 0 X
z1 s
1 O Xp
GH ( s ) –4 1
K
p3X
Step Responses
K jw
1.6 K = 50.0
1.4 K = 15.0
1.2
K = 2.0
Amplitude
1 K=1
0.8
K = 1.0 s
0.6
X X
K = 0.5 –2
0.4
K=0
0.2
0
0 2 4 6 8 10
Time (sec.)
279
Root Locus Example:
Some Observations
• This is a second-order system and there two loci.
• The root loci start at the open-loop poles.
• The root loci tend towards the open-loop zeros at
infinity as K . (Note: the number of zeros is
equal to the number of poles, when the zeros at
infinity are included.)
• The relationship between the characteristic
responses and the increasing gain is seen through
the root loci.
280
The General Root Locus Method
• Consider the general • The characteristic
system equation is
R(s) C(s)
G(s) 1 GH ( s) 0
+ or
–
GH ( s) 1
H(s)
where or
GH ( s) 1
C (s) G ( s)
GH ( s) (2k 1)
R ( s ) 1 GH ( s )
k 0, 1, 2
281
The General Root Locus Method
• All values of s which satisfy
GH ( s) 1 ; GH ( s) (2k 1) ; k 0, 1, 2
are roots of the closed-loop characteristic
equation.
• Consider the following general form
K ( s z1 )( s z2 ) ( s zm )
GH ( s )
( s p1 )( s p2 ) ( s pn )
Note : pi may be zero.
282
The General Root Locus Method
• Then
m
K sz i
GH ( s ) n
i 1 1
s p i
i 1
m n
GH ( s ) ( s zi ) ( s pi ) (2k 1)
i 1 i 1
k 0, 1, 2
283
Root Locus Method:
Geometric Interpretation
p3 jw
• Consider the example
-p3 X B s1
K ( s z1 )
GH ( s )
s ( s p2 )( s p3 ) A C
z1 p1 s
• Then the values of s = s1
O X
which satisfy -z1 D -p1
K s z1
1 p2
s s p2 s p3 -p2 X
( s z1 ) (s ( s p2 ) ( s p3 )) (2k 1)
are on the loci and are roots of the characteristic equation.
284
Root Locus Method:
Geometric Interpretation
p3 jw
• In terms of the vectors,
-p3 X B s1
the condition for s = s1
to be on the root loci are C
A
K A z1 p1 s
1 or A 1 O X
BCD BCD K -z1 D -p1
and
p2
z1 ( p1 p 2 p 3 ) (2k 1) -p2 X
k 0, 1, 2,
285
Root Locus Method
• When plotting the loci • The magnitude condition
of the roots as may then be used to
K = 0 , the determine the gain K
magnitude condition is corresponding to that
always satisfied. value s1 .
• Therefore, a value of • How can we easily
s = s1 that satisfies the determine if the angle
angle condition, is a condition is satisfied?
point of the root loci.
286
Root Locus Construction Rules
1. The loci start (K = 0) • For our example system
at the poles of the s z1 1
GH ( s)
open-loop system. s s p2 s p3 K
There are n loci .
• Therefore, as K 0 ,
2. The loci terminate GH(s) , the poles of
(K ) at the zeroes the loop transfer function.
of the open-loop
system (include zeroes • As K , GH(s) 0 ,
at infinity). the zeroes of the loop
transfer function.
287
Root Locus Construction Rules
3. The root loci are • The roots with imaginary
symmetrical about the parts always occur in
real axis. conjugate complex pairs.
4. As K the loci • When the loci approach
approach asymptotes. infinity, the angles from
There are q = n – m all the poles and zeroes
asymptotes and they are equal. The angle
intersect the real axis at condition then is
angles defined by m – n = (2k + 1)
(2k 1)
, k 0, 1, 2,
q 288
Root Locus Construction Rules
5. The asymptotes
intersection point on the
real is defined by
• The angles from poles
sa poles of GH (s) zeroes of GH (s) and zeroes to the left
q of s1 are zero. The
6. Real axis sections of the angles from poles and
root loci exist only where zeroes to the right are
there is an odd number of – .An odd number
poles and zeroes to the are required to satisfy
right. the angle condition.
289
Root Locus Construction Rules
Example
• Consider our example jw asymptote
with z1 = 4 , p12 = 1 2j
K ( s 4) real axis X 2j
GH ( s ) locus
s ( s 1 2 j )( s 1 2 j ) s
• Asymptotes: O X
–4 –2 –1 +1
(2k 1)
angles =
3 1 2 X –2j
sa
0 (1 2 j ) (1 2 j ) (4)
1
3 1
290
Root Locus Construction Rules
7. The angles of departure, • The angle of arrival at
d from poles and the zero, -z1 is obtained
arrival, a to zeroes may from
m
be found by applying az1 ( -z1 zi )
the angle condition to a i 2
n
point very near the pole
or zero. (-z p ) (2k 1)
1 i
i 1
291
Root Locus Construction Rules
Example
jw
p2
• Departure angle from p2 .
z1 = tan-1(2/3) = 33.7° -p2
X 2j
p1 = tan-1(–2/1) = 116.6°
-z1 33.7°
116.6° s
p3 = 90° O X -p
–4 –2 1 +1
• Then 90°
296
Root Locus Construction Rules
9. The angle of breakaway • Note: K GH ( s )1
2 d GH ( s )
is 180°/k where k is the dK
number of converging GH ( s) 0
ds ds
loci. • Also,
The location of the
d GH ( s) d N ( s) D( s)
breakaway point is ds ds
found from N ( s) N ( s) D( s)
0
d GH ( s)
2
D( s ) D( s )
dK
0 or 0
ds ds D( s) N ( s) N ( s) D( s) 0
297
Root Locus Plot:
Breakaway Point Example
• Consider the following loop jw
transfer function. asymptotes
K
GH ( s) 2j
s( s 3) 2
• Real axis loci exist for the
60° s
XX X
full negative axis. –4 –2 +1
• Asymptotes:
–2j
angles = (2k+1)= /3 , , 5/3
3
( 3 3 0 ) ( 0 )
sa 2
3 298
Root Locus Plot:
Breakaway Point Example
• Determine the breakaway jw
points from
2j
d K d K
2
3
ds s ( s 3) ds s 6 s 2 9 s s
K (3s 2 12 s 9) XX X
0 –4 –2 +1
( s 6s 9s)
3 2 2
then –2j
s 2 4 s 3 ( s 1)( s 3) 0
s 1 , 3
299
Root Locus Construction Rules
10. For a point on the • Alternately, K may be
root locus, s =s1 determined graphically
calculate the gain, K from the root locus plot
from
B s jw
s1 p1 s1 p2
X 1
K
s1 z 1 s1 z 2 A C
O X
s
D
BCD
K
X A
300
Summary of Root Locus Plot
Construction
• Plot the poles and zeros of the open-loop system.
• Find the section of the loci on the real axis (odd
number of poles an zeroes to the right).
• Determine the asymptote angles and intercepts.
(2k 1)
angles , q n m, k 0, 1, 2,
q
sa poles zeroes
q
301
Summary of Root Locus Plot
Construction
• Determine departure angles. For a pole -p1
( -p1 z1) (-p1 z2 ) p1 (-p1 p2 ) (2k 1)
• Check for imaginary axis crossings using the
Routh-Hurwitz criterion.
• Determine breakaway points.
angle / k , k # of converging loci
d GH ( s )
location from 0
ds
• Complete the plot.
302
Root Locus Plot Example 3
jw
• Loop Transfer function: X 4j
K
GH ( s ) asymptotes
s ( s 4)( s 2 4 s 20)
2j
• Roots:
s = 0, s = –4, s = –2 4j 45°
s
X X
• Real axis segments: –4 –2 +1
between 0 and –4 .
• Asymptotes: –2j
(2k 1) 3 5 7
angles = , , ,
40 4 4 4 4 X –4j
s a ( 4 2 2 0 ) 2
4 303
Root Locus Plot Example 3
jw
X 4j
• Breakaway points:
d K 2j
ds s 4 8s3 36s 2 80s 45°
s
K ( 4 s 3
24s 2 72s 80) X X
4 0 –4 –2 +1
(s 8s 36s 80s)
3 2 2
or s3 6s 2 18s 20 0 –2j
solving, sb 2, 2 2.45 j
• Three points that X –4j
breakaway at 90° . 304
Root Locus Plot Example 3
• The imaginary axis
• Condition for critical
crossings:
stability
Characteristic equation 80-8K/26 > 0 or K<260
s 8s 36s 80s K 0
4
3
2
The auxiliary equation
Routh table 26 s 2 + 260 = 0
s4 1 36 K solving
s3 8 80 0 s 10 j 3.16 j
s2 26 K 0
s 80-8K/26 0 0
s0 K 0 0 305
Root Locus Plot Example 3
jw
• The final plot is shown X 4j
on the right. 3.16j
2j
= 45°
–2j
X –4j
308
Root Locus Plot Example 3
Time Responses
jw • Examine the responses for
X K=260 4j
the various gains and relate
3.16j them to the location of the
K=64 K=100 2j closed-loop roots.
s • K = 64, roots are –2, –2, –
X X 2±3.46j
–4 –2
• K = 100, roots are –2±2.45j,
–2j –2±2.45j
• K = 260, roots are ±3.16j, –
X –4j 4±3.16j
309
Root Locus Plot Example 3
Time Responses
Step Response, K = 64
s = –2±3.46j
1
s = –2, –2
0.8
Amplitude
0.6
whole response
0.4
0.2
0
0 1 2 3 4 5
Time (sec.)
310
Root Locus Plot Example 3
Time Responses
Step Response, K = 100
0.8
s = –2±2.45j (repeated)
Amplitude
0.6
whole response
0.4
0.2
0
0 1 2 3 4 5
Time (sec.)
311
Root Locus Plot Example 3
Time Responses
Step Response K = 260
2
s = ±3.16j
1.8
1.6
1.4
whole response
Amplitude
1.2
0.8
s = –4±3.16j
0.6
0.4
0.2
0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
Time (sec.)
312
Root Locus with Other Parameters
• Can we plot the root • Consider the loop transfer
locus in terms of the function (1 s )
GH ( s ) 10
variation of a s ( s 2)
parameter other than • characteristic equation
gain, for example a
s 2 2 s 10 s 10 0
time constant? • regrouping the terms
• To achieve this we
( s 2 2 s 10) 10 s 0
apply a mathematical • equivalent loop transfer
10 s
trick. function
G H (s)
s 2 2 s 10
313
Root Locus with Other Parameters
• The open-loop poles are = 0 jw
s = –1 ± 3j and the zero is = 0.294 X 3j
x
s=0.
• Asymptotes: 180° = 0.4325 = 0.9
s
• Angle of departure: x x O
–4 –2 +1
108 – (90 + d) = 180
° ° °
d = – 198°
x
• Break-in point sb = – 3.16 X –3j
d 10 s
2 0 (s 2 10) 0
ds s 2s 10 314
Root Locus with Other Parameters
Step Responses
= 0 , s = – 1 ± 3j
= 0.293 ,
1.2
1
s = – 2.5 ± 2j
Amplitude
0.8
0.6 = 0.4325 ,
0.4 s = – 3.16, s = – 3.16
0.2
0
= 0.90 ,
0 1 2 3 4 5 6 s = – 1, s = – 10
Time (sec.)
315
Root Locus with Two Parameters
• The root locus method • To construct the root
focuses on the roots of locus for a characteristic
the characteristic equation which has two
equation. parameters, we construct
• There can be several fictitious loop transfer
different loop transfer functions and apply the
functions that have the normal methods.
same closed-loop
characteristic equation.
316
Root Locus with Two Parameters
• Consider the following • This will allow the
characteristic equation. plotting of the root locus
s3 + s2 + b s + = 0 . with respect to the gain b
• Write this in the general for any given value of .
form of 1 + GH(s) = 0 • The roots of the
with b as a multiplying characteristic equation of
gain. GH´(s) define the starting
bs points for the root loci.
1 3 2 0
s s Consider the loci of these
GH´(s) roots.
317
Root Locus with Two Parameters
• The characteristic jw
equation of GH´(s) is
s3 + s2 + = 0 , which 2
may be written as 2 1 1
s
1 0 X XX
s 2 ( s 1) –2 –1
GH´´(s) 1 = 0.3
s = –1.2, 0.1±0.5j
• Now construct the root 2 = 1.8
locus of GH´´(s) in terms s = –1.66, 0.33±1.0j
of the gain .
318
Root Locus with Two Parameters
• Now construct the root • Asymptotes: ± 90°
locus for 0.3
bs sa (1.20.10.1)/2
GH´(s) 3
s s2 0.5
where the open-loop 1.2
poles correspond to sa (1.660.330.33)/2
the previous root locus 0.5
for varying .
319
Root Locus with Two Parameters
• Imaginary axis
jw
crossings:
s3 + s2 + bs + = 0
2
s3 1 b 0
2 1 1
s
s2 1 0 O
–2 –1
s b 0 0
s0 0 0
324
Root Locus Lead Design
• Three effects of the pole
zero combination: jw
Design Point
– modifies the real axis s1 4j
segments
– shifts the asymptote real
p z s
axis intersection to the X
X X
left –8 –4 –2
– modifies the angle
criterion by – 4j
b z p
325
Root Locus Lead Design
• Design steps:
– place the zero below the jw
design point, s1 (approx.) Design Point
s1 4j
– determine the angle
difference, b 90° p b
p s
that will satisfy the angle
X X X
criterion for the design p –4 –2
point, s1 .
– locate the pole location – 4j
from the angle, p and
the location of the zero
326
Root Locus Lead Design
• calculate p
b 135° 116.6° 180° jw
Design Point
b 431.6° or 71.6° s1 4j
\ p 4/0.333 4 16
327
Root Locus Lead Design
• The asymptote intersection:
sa (0 2 16) (4) jw
31
\ sa 7 Design Point s1 4j
s
X X X
– 16 –8 –4 –2
• Determine the gain at the
design point: – 4j
1.5
20
GH ( s)
78.6( s 4) 1 s( s 2)
GH ( s)
( s 16) s( s 2)
1
Amplitude
1
GH ( s)
s( s 2)
0.5
0
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
Time (sec.)
330
Root Locus Lag Design
• Pole/zero placement: jw
– The lag compensator is
represented by a pole near
the imaginary axis and a
zero further to the left. s
• Low frequency gain: X
–1 –z –p
– The compensator low
frequency gain is Lag compensator
|z| sz
s p
| p|
331
Root Locus Lag Design
s1
| z|| p| L < 2
0.035L w1
L
then 0.035L2
| p | ( 1) f
w1 s
X
0.035(s 1 w1 ) – s1 –z –p
2 2
| p | f
w1 ( 1)
333
Root Locus Lag Design
• Design steps:
– Determine a point, s1 on the – Calculate the compensator
uncompensated loci that pole location using s1 and
satisfies the dynamic in the relationship
requirements.
0.035(s 1 w1 )
2 2
– Find the gain at s1 and then | p |
the low frequency gain of the w1 ( 1)
system. – Finally, calculate the
– Determine the low frequency compensator zero from
gain, of the compensator |z|
required to meet the system
| p|
steady-state error requirement.
334
Root Locus Lag Design Example
jw
• Consider again the K = 2 • 2 /1 = 4
system defined by
K=4 s1
K 1.73j
GH ( s)
s( s 2) K=1
60° s
• Control design X X
–2
specifications:
0.50
ess 5% for a ramp
input.
335
Root Locus Lag Design Example
• The velocity steady- • The compensator pole
state error constant is magnitude is
4
2 ( 1 1.73 2
)
K v lim s
s 0 s ( s 2 )
| p | 0.035
1.73 (10 1)
For ess 0.05, Kv 20 = 0.0128
• The compensator zero
• Therefore the compensator magnitude is
low-frequency gain must | z | = | p | = 0.128
be 10 .
336
Root Locus Lag Design Example
• The final compensator is • The final closed-loop
s 0.128 systems is
Gc 4 .0
s 0.0128
s 0.128
G4 3
• The compensated open- s 2.013s 4.026s 0.512
2
loops system is
s 0.128 1 • The roots of the final
GcG p 4.0 closed-loop systems are
s 0.0128 s( s 2)
s = -0.9386 ± 1.7000i
s = -0.1358
337
Root Locus of Final System
2
1.5
0.5
Imag Axis
0 x
0.25
-0.5 0.2
0.15
0.1
-1
0.05
Imag Axis
0
xx
-1.5 -0.05
-0.1
-0.15
-2 -0.2
-3 -2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2
-0.25
Real Axis
-0.25 -0.2 -0.15 -0.1 -0.05 0 0.05
Real Axis
338
Step Responses
Step Response
1.2
compensated
1
uncompensated
0.8
Amplitude
0.6
0.4
0.2
0
0 2 4 6 8 10 12 14 16 18 20
Time (sec.) 339
Ramp Response
Ramp Response
10
9
8
5
uncompensated
4
2
1
0
0 1 2 3 4 5 6 7 8 9 10
Time (sec.) 340