Dynamic Response

Prof. Enver Tatlicioglu

Department of Electrical & Electronics Engineering

Ege University

Chapter 3

Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 1 / 71

Dynamic response

We can now model dynamic systems with differential equations.

Q. What do these equations mean?

Q. How does the system respond to certain inputs?

Q. If we add dynamics (i.e., a controller), how will the system respond?

Q. How should the system respond (i.e., design specifications)?

A. Essential tool is Laplace transform.

Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 2 / 71

Some important input signals

Several signals recur throughout this course.

The unit step function:

1, t ≥ 0
1 (t) =
0, otherwise

1 (t)
1

Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 3 / 71

Some important input signals

The unit ramp function:


t, t ≥ 0
r (t) =
0, otherwise

r (t)

Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 4 / 71

Some important input signals

The unit parabola function:

t2

2, t≥0
p (t) =
0, otherwise

p (t)

Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 5 / 71

Some important input signals
The cosine/sine functions:

cos (t)

sin (t)

Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 6 / 71

Some important input signals
The impulse function:

The impulse function is a very strange generalized function, only

defined under an integral
Z∞
δ (t) = 0 for t 6= 0 and δ (t) dt = 1.
| {z }
zero duration −∞
| {z }
unit area

R∞
It satisfies the useful sifting property u (τ ) δ (t − τ ) dτ = u (t).
−∞
Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 7 / 71
Response of linear time–invariant systems
Let y (t) be the output of an LTI system with input u (t)

y (t) = T [u (t)]
 ∞ 
Z
= T u (τ ) δ (t − τ ) dτ 
−∞
Z∞
= u (τ ) T [δ (t − τ )] dτ.
−∞

Let
Z∞
h (t, τ ) , T [δ (t − τ )] ⇒ y (t) = u (τ ) h (t, τ ) dτ
−∞

where h (t) is the impulse response that is the time function

corresponding to the transfer function.
The impulse response is also called the natural response of the system.
Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 8 / 71
Response of linear time–invariant systems

Note that, for LTI systems

h (t, τ ) = h (t − τ )

then
Z∞
y (t) = u (τ ) h (t − τ ) dτ
−∞
= u (t) ∗ h (t) .

The output of an LTI system is equal to the convolution of its

impulse response with the input.
This makes life easy.

Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 9 / 71

BIBO stability

There are many classifications of stability in system analysis.

Figure: Stability, marginal stability and instability

For LTI systems, they all are basically same which is bounded input
bounded output (BIBO) stability.
If input is bounded (i.e., |u (t)| < K1 ), then output is also bounded
(i.e., |y (t)| < K2 ).

Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 10 / 71

BIBO stability
In the time domain, the output has the form
+∞
Z
y (t) = h (τ ) u (t − τ ) dτ.
−∞

Bounding the output results in

+∞
Z
|y (t)| = h (τ ) u (t − τ ) dτ
−∞
+∞
Z
≤ |h (τ ) u (t − τ )| dτ
−∞
+∞
Z
= |h (τ )| |u (t − τ )| dτ.
−∞

Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 11 / 71

BIBO stability

After noting that |u (t)| < K1 , we obtain

+∞
Z
|y (t)| ≤ K1 |h (τ )| dτ.
−∞

So if an LTI system is BIBO stable, then |y (t)| < K2 is ensured

provided that
+∞
Z
K2
|h (τ )| dτ < < ∞.
K1
−∞

Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 12 / 71

Finding an impulse response
Example
Consider a first order system

ẏ (t) + ky (t) = u (t)

with y (0− ) = 0 and u (t) = δ (t) where k is a positive constant.

So for positive time ẏ (t) + ky (t) = 0.
Recall from previous courses that y (t) = A exp (st) with A and s
being constants.
We need to solve for A and s.
Time derivative of y (t) = A exp (st) is found as

ẏ (t) = As exp (st) .

Substituting the above expression results in

As exp (st) + kA exp (st) = 0 ⇒ s + k = 0 ⇒ s = −k.

Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 13 / 71

Finding an impulse response
Example

We have solved for s, now we will solve for A

Z0+ Z0+ Z0+

ẏ (t) dt + k y (t) dt = δ (t) dt
0− − 0−
| {z } |0 {z } | {z }
+ 0 1
y (t)|00−

⇒ A exp −k0+ − 0 + 0 = 1

⇒ A = 1.

So the response of this system to impulse function is

h (t) = exp (−kt) , t ≥ 0

= exp (−kt) 1 (t) .

Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 14 / 71

Finding an impulse response
Example

Response of this system to general input:

Z∞
y (t) = u (τ ) h (t − τ ) dτ
−∞
Z∞
= u (t − τ ) h (τ ) dτ
−∞
Z∞
= u (t − τ ) exp (−kτ ) 1 (τ ) dτ
−∞
Z∞
= u (t − τ ) exp (−kτ ) dτ.
0

Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 15 / 71

Transfer function

Response to impulse is called the impulse response

h (t) .

Response to general input u (t) is found by the below convolution

u (t) ∗ h (t) .

Response to sinusoid or cosinusoid

A
A cos (ωt) = (exp (jωt) + exp (−jωt))
2
reduces to response to exponential.

Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 16 / 71

Transfer function
Let u (t) = exp (st) with s being complex. The response of the
system is found as
Z∞
y (t) = u (t − τ ) h (τ ) dτ
−∞
Z∞
= exp (s (t − τ )) h (τ ) dτ
−∞
Z∞
= exp (st) exp (−sτ ) h (τ ) dτ
−∞
Z∞
= exp (st) exp (−sτ ) h (τ ) dτ
−∞
= exp (st) H (s) .

Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 17 / 71

Transfer function

An input of the form exp (st) decouples the convolution into two
independent parts:
? a part depending on exp (st),
? a part depending on h (t).
Recall that, a signal for which the system output is a (possibly
complex) constant times the input is referred to as an eigenfunction
of the system, and the amplitude factor is referred to as the system’s
eigenvalue.
Complex exponentials are eigenfunctions of all LTI systems.
Notice that H (s) (which is a constant for a specific value of s) is the
eigenvalue associated with the eigenfunction exp (st).

Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 18 / 71

Transfer function
Example
Consider the following system model

ẏ (t) + ky (t) = u (t)

with u (t) = exp (st).

We know that

y (t) = H (s) exp (st) ⇒ ẏ (t) = sH (s) exp (st) .

Substituting the above expression into the system model, results in

sH (s) exp (st) + kH (s) exp (st) = exp (st)

sH (s) + kH (s) = 1
1
⇒ H (s) =
s +k
exp (st)
⇒ y (t) = .
s +k
Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 19 / 71
Response to cosinusoid

For s = jω,

u (t) = exp (jωt) ⇒ y (t) = H (jω) exp (jωt) .

For s = −jω,

u (t) = exp (−jωt) ⇒ y (t) = H (−jω) exp (−jωt) .

A
Thus, for u (t) = A cos (ωt) = 2 (exp (jωt) + exp (−jωt)),

A
y (t) = [H (jω) exp (jωt) + H (−jω) exp (−jωt)] .
2

Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 20 / 71

Response to cosinusoid
Consider a real h (t) with

H (jω) = L exp (jφ)

so we know that
H (−jω) = L exp (−jφ) .
Substitute these into y (t) to obtain

A
y (t) = [L exp (jφ) exp (jωt) + L exp (−jφ) exp (−jωt)]
2
AL
= [exp (j (ωt + φ)) + exp (−j (ωt + φ))]
2
= AL cos (ωt + φ) .

The response of an LTI system to a sinusoid is a sinusoid (of the

same frequency).

Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 21 / 71

Frequency response of a first order system
For
1 1
H (s) = ⇒ H (s) |s=jω = H (jω) = .
s +k jω + k
Magnitude of H (jω)
1
L = |H (jω)| = √ .
ω2 + k 2
Phase angle of H (jω)
ω 
φ = ∠H (jω) = − tan−1 .
k
The output can be written as
A   ω 
y (t) = √ cos ωt − tan−1 .
ω2 + k 2 k
Can we use these results to simplify convolution and get an easier way
to understand dynamic response?
Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 22 / 71
Laplace L− transform

We have seen that if a system has an impulse response h (t), we can

compute its transfer function H (s)
Z∞
H (s) = h (t) exp (−st) dt.
−∞

Since we deal with causal systems (possibly with an impulse at

t = 0), we can integrate from 0− instead of negative infinity
Z∞
H (s) = h (t) exp (−st) dt.
0−

This is called the one–sided (uni–lateral) Laplace transform of h (t).

Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 23 / 71

Laplace L− transform

Name Time function f (t) Laplace transform F (s)

Unit impulse δ (t) 1
Unit step 1 (t) 1/s
Unit ramp t1 (t) 1/s 2
nth order ramp t n 1 (t) n!/s n+1

Sine sin (bt) 1 (t) b/ s 2 + b 2

Cosine cos (bt) 1 (t) 2 2
s/ s + b 
Damped sine exp (−at) sin (bt) 1 (t) b/ (s + a)2 + b 2
 
Damped cosine exp (−at) cos (bt) 1 (t) (s + a) / (s + a)2 + b 2

2 
Diverging sine t sin (bt) 1 (t) 2bs/ s 2 + b 2


2 
Diverging cosine t cos (bt) 1 (t) s 2 − b2 / s 2 + b2

Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 24 / 71

Properties of Laplace transform

Superposition

L {af1 (t) + bf2 (t)} = aF1 (s) + bF2 (s) .

Time delay
L {f (t − τ )} = exp (−sτ ) F (s) .
Time scaling
1 s 
L {f (at)} = F .
|a| a
? Time scaling property is useful if original equations are expressed
poorly in time scale.
? For example, measuring disk–drive seek speed in hours.

Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 25 / 71

Properties of Laplace transform

Differentiation
n o
L f˙ (t) = sF (s) − f 0−


n o
L f¨ (t) = s 2 F (s) − sf 0− − f˙ 0−

..
n o .
L f (m) (t) = s m F (s) − s m−1 f 0− − · · · ... − f (m−1) 0− .

Integration  t 
Z  1
L f (τ ) dτ = F (s) .
  s
0−

Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 26 / 71

Properties of Laplace transform

Convolution

Y (s) = L {y (t)}
= L {h (t) ∗ u (t)}
 t 
 Z 
= L h (τ ) u (t − τ ) dτ
 
τ =0−
Z∞ Zt
= h (τ ) u (t − τ ) dτ exp (−st) dt
t=0− τ =0−
Z∞ Z∞
= h (τ ) u (t − τ ) exp (−st) dtdτ.
τ =0− t=τ −

Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 27 / 71

Properties of Laplace transform
Multiply the right–hand side of the above expression with
exp (−sτ ) exp (sτ ) to obtain
Z∞ Z∞
Y (s) = h (τ ) exp (−sτ ) u (t − τ ) exp (−s (t − τ )) dtdτ.
τ =0− t=τ −

0
Let t = t − τ
Z∞ Z∞  0   0
0
Y (s) = h (τ ) exp (−sτ ) dτ u t exp −st dt
τ =0− t 0 =0−
= H (s) U (s) .

Laplace transform unwraps convolution for general input signals.

This makes systems easy to analyze.

Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 28 / 71

Inverse Laplace transform
Inverse Laplace transform converts F (s) to f (t).
Once we get an intuitive feel for F (s), we won’t need to do this often.
The main tool for inverse Laplace transform is partial fraction
expansion.
For most physical systems1
b0 s m + b1 s m−1 + ... + bm
F (s) =
s n + a1 s n−1 + ... + an
with n ≥ m.
If we write F (s) in the poles/zeros form
m
Q
(s − zi )
F (s) = b0 i=1
n .
Q
(s − pi )
i=1
1
A polynomial in which the coefficient of the highest order term is 1 is called a monic
polynomial.
Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 29 / 71
Inverse Laplace transform
If pi are distinct
c1 c2 cn
F (s) = + + ... +
s − p1 s − p2 s − pn
c2 (s − p1 ) cn (s − p1 )
⇒ (s − p1 ) F (s) = c1 + + ... + .
s − p2 s − pn

Let s = p1 then c1 = [(s − p1 ) F (s)] |s=p1 .

Similarly s = pi then ci = [(s − pi ) F (s)] |s=pi .
Since, we know that
1
L {exp (kt) 1 (t)} =
s −k
then
n
X
f (t) = ci exp (pi t) 1 (t) .
i=1

Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 30 / 71

Inverse Laplace transform
Example

Find the inverse Laplace transform of

5 5
F (s) = = .
s2 + 3s + 2 (s + 1) (s + 2)

Using partial fraction expansion

5
c1 = [(s + 1) F (s)] |s=−1 = |s=−1 = 5
s +2
5
c2 = [(s + 2) F (s)] |s=−2 = |s=−2 = −5.
s +1
Then, f (t) is found as

f (t) = [5 exp (−t) − 5 exp (−2t)] 1 (t) .

Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 31 / 71

Inverse Laplace transform
Repeated Roots
If F (s) has repeated roots, we must modify the procedure.
For example, consider below F (s) which includes a third order pole
k
F (s) = 3
(s − p1 ) (s − p2 ) · · ·
c1,1 c1,2 c1,3 c2
= + 2
+ 3
+ + ··· .
s − p1 (s − p1 ) (s − p1 ) s − p2
The modified formula to compute the constants is
h i
c1,3 = (s − p1 )3 F (s) |s=p1
  
d
c1,2 = (s − p1 )3 F (s) |s=p1
ds
 2  
d 3
c1,1 = (s − p1 ) F (s) |s=p1
ds 2
c2 = [(s − p2 ) F (s)] |s=p2 .
Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 32 / 71
Inverse Laplace transform
Example
Find the inverse Laplace transform of
s +3
F (s) = .
(s + 1) (s + 2)2
Let
c1,1 c1,2 c2
F (s) = + 2
+ .
s + 2 (s + 2) s +1
Using partial fraction expansion
 
h
2
i s +3
c1,2 = (s + 2) F (s) |s=−2 = |s=−2 = −1
s +1
     
d d s +3
c1,1 = (s + 2)2 F (s) |s=−2 = |s=−2
ds ds s + 1
 
1 s +3
= − |s=−2 = −1 − 1 = −2.
s + 1 (s + 1)2

Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 33 / 71

Inverse Laplace transform
Example

Using partial fraction expansion

 
s +3
c2 = [(s + 1) F (s)] |s=−1 = |s=−1 = 2.
(s + 2)2

Then, f (t) is found as

f (t) = [2 exp (−t) − 2 exp (−2t) − t exp (−2t)] 1 (t) .

Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 34 / 71

Using Laplace transform to solve problems
Example
We can use Laplace transform to solve both homogeneous and forced
differential equations.
Solve for y (t)
ÿ (t) + y (t) = 0
with y (0− ) = α, ẏ (0− ) = β.
Using Laplace transform yields

s 2 Y (s) − αs − β + Y (s) = 0.

Thus, Y (s) is found as

αs + β s 1
Y (s) = 2
=α 2 +β 2 .
s +1 s +1 s +1
From tables,
y (t) = [α cos (t) + β sin (t)] 1 (t) .
Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 35 / 71
Using Laplace transform to solve problems
Example
Solve for y (t)
ÿ (t) + 5ẏ (t) + 4y (t) = u (t)
with u (t) = 2 exp (−2t) 1 (t) and y (0− ) = 0, ẏ (0− ) = 0.
Using Laplace transform results in
2
s 2 Y (s) + 5sY (s) + 4Y (s) = .
s +2
Thus, Y (s) is found as
2 2
Y (s) = =
(s +2) (s 2
+ 5s + 4) (s + 2) (s + 1) (s + 4)
−1 2/3 1/3
= + + .
s +2 s +1 s +4
From tables,
 
2 1
y (t) = − exp (−2t) + exp (−t) + exp (−4t) 1 (t) .
3 3
Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 36 / 71
Time response vs. pole locations

Once the transfer function is determined, we can start to analyze the

response of the system it represents.
When the system equations are ordinary differential equations, the
transfer function will be a ratio of polynomials

b (s)
H (s) =
a (s)

where we assume that a (s) and b (s) have no common factors.

The values of s such that a (s) = 0 represents points where H (s) is
infinity and these s values are called poles of H (s).
The values of s such that b (s) = 0 are points where H (s) = 0 and
these s values are called zeros of H (s).

Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 37 / 71

Time response vs. pole locations

Poles and zeros are used to compute corresponding time response of

the system.
? Poles qualitatively determine the behavior of the system.
? Poles identify the classes of signals contained in the impulse response.
? Zeros quantify this relationship.
? A zero near a pole tends to cancel the effect of that pole.

Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 38 / 71

First order system

Consider a first order system

1
H (s) = .
s +σ
From the tables, h (t) can be found as

h (t) = exp (−σt) 1 (t) .

If σ > 0, pole is at s < 0, stable (i.e., impulse response decays, and

any bounded input produces bounded output).
If σ < 0, pole is at s > 0, unstable.

Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 39 / 71

First order system

h (t)
1.0

0.8

0.6 exp (−σt)

0.4 ← 1
e

0.2

Time (sec ×τ )
1 2 3 4 5

Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 40 / 71

First order system

Figure: Unit step response of a stable first order system

Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 41 / 71

Second order systems without any zeros
Standard form of a second order system without any zeros
b0 K ωn2
H (s) = =
s 2 + a1 s + a2 s 2 + 2ζωn s + ωn2
where ζ is the damping ratio, ωn is the natural/undamped frequency,
K is the gain with b0 , a1 , a2 being real constants.
For unity gain (i.e., K = 1), from the tables, h (t) can be found as
ωn
h (t) = p exp (−σt) sin (ωd t) 1 (t)
1 − ζ2
p
where σ = ζωn and ωd = ωn 1 − ζ 2 is the damped frequency.

Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 42 / 71

Second order systems without any zeros

Figure: Damping ratios vs. pole locations

Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 43 / 71

Second order systems without any zeros

Figure: Envelope of sinusoid decays as exp (−σt)

Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 44 / 71

Second order systems without any zeros

Figure: Impulse responses of second order systems

Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 45 / 71

Second order systems without any zeros

Figure: Step responses of second order systems

Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 46 / 71

Second order systems without any zeros

ζ = 0 undamped.
0 < ζ < 1 underdamped.
? Low damping, ζ ≈ 0, oscillatory.
? High damping, ζ ≈ 1, no oscillations.
ζ = 1 critically damped.
ζ > 1 overdamped.

Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 47 / 71

Second order systems without any zeros

Figure: Impulse responses vs. pole locations

Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 48 / 71

Second order systems without any zeros

Figure: Step responses vs. pole locations

Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 49 / 71

Time domain specifications

We have seen impulse and step responses for first and second order
systems.
Our control problem may be to specify exactly what the response
should be.
Usually expressed in terms of the step response.

Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 50 / 71

Time domain specifications
tr is the rise time and is equal to the time to reach vicinity of set
point.
ts is the settling time and is the time for transients to decay (to 5%,
1%).
Mp is the percent overshoot.
tp is the time to peak.

Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 51 / 71

Rise time

All step responses rise in roughly the same amount of time.

Take ζ = 0.5 to be average.
Time from 0.1 to 0.9 is approximately equal to
1.8 1.8
tr ≈ ⇒ ωn ≈ .
ωn tr
We could make this more accurate, but note that this is only valid for
second order systems with no zeros.
Use this as approximate design “rule of thumb” and iterate design
until specifications are met.

Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 52 / 71

Peak time
Step response can be found from ILT of H (s) /s as
 
σ
y (t) = 1 − exp (−σt) cos (ωd t) + sin (ωd t)
ωd
p
where σ = ζωn and ωd = ωn 1 − ζ 2 .
Peak occurs when ẏ (t) = 0
 
σ
ẏ (t) = σ exp (−σt) cos (ωd t) + sin (ωd t)
ωd
− exp (−σt) (−ωd sin (ωd t) + σ cos (ωd t))
 2
σ + ωd2

= exp (−σt) sin (ωd t) = 0.
ωd

So,
π π
ωd tp = π ⇒ tp = = p .
ωd ωn 1 − ζ 2
Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 53 / 71
Percent overshoot vs. damping ratio
Percent overshoot is
 p 
Mp = 100 exp −ζπ/ 1 − ζ 2
where common values are Mp = 16% for ζ = 0.5 or Mp = 5% for
ζ = 0.7.

Figure: Percent overshoot vs. damping ratio

Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 54 / 71
Settling time
Determined mostly by decaying exponential

exp (−ωn ζts ) = 

with  = 0.01, 0.02, or 0.05.

For  = 0.01, we obtain
4.6 4.6
exp (−ωn ζts ) = 0.01 ⇒ ωn ζts = 4.6 ⇒ ts = = .
ωn ζ σ
For  = 0.02, we obtain
3.9 3.9
exp (−ωn ζts ) = 0.02 ⇒ ωn ζts = 3.9 ⇒ ts = = .
ωn ζ σ
For  = 0.05, we obtain
3.6 3.6
exp (−ωn ζts ) = 0.05 ⇒ ωn ζts = 3.6 ⇒ ts = = .
ωn ζ σ

Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 55 / 71

Design synthesis

Specifications on tr , ts , Mp determine pole locations.

? ωn ≥ 1.8/tr .
? ζ ≥ function (Mp ) (read from Mp versus ζ graph).
? σ ≥ 4.6/ts (for  = 0.01).

Figure:

Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 56 / 71

Converting specifications to s–plane
Example
Specifications are: tr ≤ 0.6, Mp ≤ 10%, ts ≤ 3 sec at 1%.
? ωn ≥ 1.8/tr = 1.8/0.6 = 3.0 rad/sec.
? From Mp versus ζ graph, ζ ≥ 0.6.
? σ ≥ 4.6/ts = 4.6/3 ≥ 1.5 sec for  = 0.01.

Dr. Enver Tatlicioglu (EEE@EgeU) Figure:

Control Systems 1 Chapter 3 57 / 71
Converting specifications to s–plane
Designing a motor compensator

Suppose a servo motor system for a pen plotter has transfer function

0.5Ka ωn2
=
s 2 + 2s + 0.5Ka s 2 + 2ζωn s + ωn2

where there is only one adjustable parameter which is Ka .

So, we can choose only one specification (i.e., either of tr , ts or Mp ).
Let the design requirement be to allow no overshoot.
Thus, Mp = 0 results in ζ = 1.
From the denominator of the transfer function, 2 = 2ζωn , thus
ωn = 1.
Thus 0.5Ka = 1 resulting in Ka = 2.
Note that, for  = 0.01, we obtain ts = 4.6 sec.
We will need a better controller than this for a pen plotter!

Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 58 / 71

Second order systems without any zeros

We have looked at first order and second order systems without zeros,
and with unity gain.

Non–unity gain:
? If we multiply by K , the dc gain becomes K .
? Note that, tr , ts , Mp , tp are not affected.

Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 59 / 71

Second order systems with a zero

Adding a zero to a second order system:

? Consider the following transfer functions
2 2 2
H1 (s) = = −
(s + 1) (s + 2) s +1 s +2
2 (s + 1.1) 0.18 1.64
H2 (s) = = + .
1.1 (s + 1) (s + 2) s +1 s +2

? Same dc gain (at s = 0).

? Coefficient of (s + 1) pole is greatly reduced.
General conclusion:
? A zero near a pole tends to cancel the effect of that pole.

Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 60 / 71

Second order systems with a zero

How about transient response?

s
αζωn +1
H (s) =  2
s 2ζs
ωn + ωn +1

? Zero at −ασ.
? Poles at R (s) = −σ.
Large α, zero far from poles → no effect.
α ≈ 1, large effect.
Overshoot goes up as α → 0.

Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 61 / 71

Second order systems with a zero

Figure: Step responses for a second order system with a zero

Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 62 / 71

Second order systems with a zero

Figure: Overshoot versus normalized zero location for a second order system with
a zero

Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 63 / 71

Second order systems with a zero
Set ωn = 1
s
αζ +1
H (s) =
s 2 + 2ζs + 1
1 1 s
= 2 + 2
s + 2ζs + 1 αζ s + 2ζs + 1
= Ho (s) + Hd (s) .

where Ho (s) is the original response (i.e., without the zero) and
Hd (s) is the added term due to the zero.
Notice that
1
Hd (s) = sHo (s) .
αζ
The time response of Hd (s) is a scaled version of the derivative of
the time response of Ho (s).
If any of the zeros are in RHP, the system is called non–minimum
phase.
Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 64 / 71
Second order systems with a zero

Figure: Step response for a second order minimum phase system

Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 65 / 71

Second order systems with a zero

Figure: Step response for a second order non–minimum phase system

Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 66 / 71

Adding a pole to a second order system

Consider the transfer function

1
H (s) =    2 .
s s 2ζ
αζωn +1 ωn + ωn s +1

? Original poles are at R (s) = −σ = −ζωn .

? New pole is at s = −αζωn = −ασ.
Major effect is an increase in rise time.

Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 67 / 71

Adding a pole to a second order system

Figure: Step response for a pole added second order system (i.e., a third order
system)

Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 68 / 71

Adding a pole to a second order system

Figure: Normalized rise time vs. normalized pole location

Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 69 / 71

Summary of higher order approximations

Extra zero in LHP will increase overshoot if the zero is within a factor
of ≈ 4 from the real part of complex poles.
Extra zero in RHP depresses overshoot, and may cause step response
to start in wrong direction.
Extra pole in LHP increases rise time if extra pole is within a factor of
≈ 4 from the real part of complex poles.

Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 70 / 71

Summary of higher order approximations

Matlab step and impulse commands can plot higher order system
responses.
Since a model is an approximation of a true system, it may be all
right to reduce the order of the system to a first or second order
system (if higher order poles and zeros are a factor of 5 or 10 times
farther from the imaginary axis).
Analysis and design are much easier.
Numerical accuracy of simulations are better for lower order models.
First and second order models provide us with great intuition into
how the system works.
May be just as accurate as higher order model, since higher order
model itself may be inaccurate.

Dr. Enver Tatlicioglu (EEE@EgeU) Control Systems 1 Chapter 3 71 / 71