B.C Kuo Solutions 1

Ninth Edition
l
ua
an
M
ns
tio
lu
So
Farid Golnaraghi • Benjamin C. Kuo

Automatic Control Systems, 9th Edition
A Chapter 2 Solution
ns Golnarraghi, Kuo

C
Chapter 2
2
2‐1 (a) −10;
Poless: s = 0, 0, −1, − (b) Poles: s = −2,, −2;
Zeross: s = −2, ∞, ∞, ∞. Zeros: s = 0.
The pole and zero at s = −1 ccancel each otther.

( Poles: s = 0, −1 + j, −1 − j;
(c) (d) Poles: s = 0, −1, −2, ∞.
Zeross: s = −2.

2
2-2) a)
b)
c)
2
2-3)
M
MATLAB code
e:
2‐1

Automatic Control Systems, 9th Edition Chapter 2 Solutions Golnaraghi, Kuo

clear all;
s = tf('s')
'Generated transfer function:'
Ga=10*(s+2)/(s^2*(s+1)*(s+10))
'Poles:'
pole(Ga)
'Zeros:'
zero(Ga)
Gb=10*s*(s+1)/((s+2)*(s^2+3*s+2))
'Poles:';
pole(Gb)
'Zeros:'
zero(Gb)
Gc=10*(s+2)/(s*(s^2+2*s+2))
'Poles:';
pole(Gc)
'Zeros:'
zero(Gc)
Gd=pade(exp(-2*s),1)/(10*s*(s+1)*(s+2))
'Poles:';
pole(Gd)
'Zeros:'
zero(Gd)
2‐2


Poles and zeros of the above functions:
(a)
Poles: 0 0 ‐10 ‐1
Zeros: ‐2
(b)
Poles: ‐2.0000 ‐2.0000 ‐1.0000
Zeros: 0 ‐1
(c)
Poles:
0
‐1.0000 + 1.0000i
‐1.0000 ‐ 1.0000i
Zeros: ‐2
Generated transfer function:
(d) using first order Pade approximation for exponential term
Poles:
0
‐2.0000
‐1.0000 + 0.0000i
‐1.0000 ‐ 0.0000i
Zeros:
1
2‐3


2-4) Mathematical representation:
In all cases substitute and simplify. The use MATLAB to verify.
10 22 + ω 2 1 + ω 2 102 + ω 2
R= ;
−ω 2 (ω 2 + 1)(ω 2 + 100)
ω
10( jω + 2)
φ1 = tan −1 2 + ω
2 2
−ω ( jω + 1)( jω + 10)
2
2
10( jω + 2) ( − jω + 1)(− jω + 10) 22 + ω 2
= ×
−ω ( jω + 1)( jω + 10) (− jω + 1)(− jω + 10)
2
−ω
10( jω + 2)( − jω + 1)(− jω + 10)
φ2 = tan −1 1 + ω
2
a) =
−ω (ω + 1)(ω + 100)
2 2 2 1
jω + 2 − jω + 1 − jω + 10 1+ ω2
=R
22 + ω 2 1 + ω 2 102 + ω 2 −ω
φ3 = tan −1 10 + ω
2 2
= R(e jφ1 e jφ2 e jφ3 )
10
102 + ω 2
φ = φ1 + φ2 + φ3
10 1 + ω 2 9 + ω 2
R= ;
(ω 2 + 1) 2 (ω 2 + 9)
−ω
10
−1 1 + ω
2
( jω + 1) ( jω + 3)
2 φ1 = tan
1
10 (− jω + 1)(− jω + 1)(− jω + 3) 1+ ω2
= ×
( jω + 1)( jω + 1)( jω + 3) ( − jω + 1)(− jω + 1)(− jω + 3) −ω
10(− jω + 1)(− jω + 1)(− jω + 3) −1 1 + ω
2
b) = φ = tan
1

(ω 2 + 1) 2 (ω 2 + 9)
2
− jω + 1 − jω + 1 − jω + 3 1+ ω2
= R
1+ ω2 1+ ω2 9 + ω2 −ω
φ3 = tan −1 9 + ω
2
= R (e jφ1 e jφ2 e jφ3 )
3
9 + ω2
φ = φ1 + φ2 + φ3
2‐4


10
jω ( j 2ω + 2 − ω 2 )
−10 j (2 − ω 2 − j 2ω )
= ×
ω ( j 2ω + 2 − ω 2 ) (2 − ω 2 − j 2ω )
10(−2ω − (2 − ω 2 ) j )
c) =
ω (4ω 2 + (2 − ω 2 ) 2 )
−2ω − (2 − ω 2 ) j
=R
4ω 2 + (2 − ω 2 ) 2
jφ
= R (e )
10 4ω 2 + (2 − ω 2 ) 2 10
R= = ;
ω (4ω + (2 − ω ) ) ω 4ω + (2 − ω 2 ) 2
2 2 2 2
−2 − ω 2
4ω 2 + (2 − ω 2 ) 2
φ = tan −1 −2ω
4ω + (2 − ω 2 ) 2
2
1
R= ;
10ω 2 + ω 2 1 + ω 2
2
e −2 jω
−ω
10 jω ( jω + 1)( jω + 2)
φ = tan −1 22 + ω 2
− j (− jω + 1)(− jω + 2) −2 jω 1 2
= e
d) 10ω (ω 2 + 1)(ω 2 + 2) 2 + ω2
2
− jω + 2 − jω + 1 −2 jω − jπ / 2 −ω
=R e
2 + ω 1+ ω
2 2 2
φ2 = tan −1 1 + ω
2
1
= R (e jφ1 e jφ2 e jφ3 )
1+ ω2
φ = φ1 + φ2 + φ3
MATLAB code:
clear all;
s = tf('s')
Ga=10*(s+2)/(s^2*(s+1)*(s+10))
figure(1)
2‐5


Nyquist(Ga)
Gb=10*s*(s+1)/((s+2)*(s^2+3*s+2))
figure(2)
Nyquist(Gb)
Gc=10*(s+2)/(s*(s^2+2*s+2))
figure(3)
Nyquist(Gc)
Gd=pade(exp(-2*s),1)/(10*s*(s+1)*(s+2))
figure(4)
Nyquist(Gd)
Nyquist plots (polar plots):
Part(a)
2‐6


Nyquist Diagram
15
10
5
Imaginary Axis
-5
-10
-15
-300 -250 -200 -150 -100 -50 0
Real Axis

Part(b)
Nyquist Diagram
1.5
0.5
Imaginary Axis
-0.5
-1
-1.5
-1 -0.5 0 0.5 1 1.5 2 2.5
Real Axis

Part(c)
2‐7


Nyquist Diagram
80
60
40
20
Imaginary Axis
-20
-40
-60
-80
-7 -6 -5 -4 -3 -2 -1 0
Real Axis

Part(d)
Nyquist Diagram
2.5
1.5
0.5
Imaginary Axis
-0.5
-1
-1.5
-2
-2.5
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4
Real Axis

2‐8


2-5) In all cases find the real and imaginary axis intersections.
10 10(− jω + 2) 10 2 − jω
G ( jω ) = = = ;
( jω − 2) (ω + 4)
2
(ω 2 + 4) (ω 2 + 4)
2
Re {G ( jω )} = cos φ = ,
(ω 2 + 4)
−ω
a) Im {G ( jω )} = sin φ = ,
(ω 2 + 4)
2
(ω 2 + 4)
φ = tan −1 −ω
(ω 2 + 4)
10
R=
(ω 2 + 4)
lim ω →0 G ( jω ) = 5; φ = tan −1 1 = −90o

−0
lim ω →∞ G ( jω ) = 0; φ = tan −1 0 = −180o
−1
Real axis intersection @ jω = 0
Imaginary axis int er sec tion does not exist.
b&c) =1 0o
∞ =0 -180o
Therefore:
Re{ G(jω) } =
Im {G(jω)} =
2‐9


If Re{G(jω )} = 0 Ö
0
If Im{ G(jω )} = 0 Ö 0
∞
If ω = ωn Ö
90
If ω = ωn and ξ = 1 Ö
If ω = ωn and ξ 0 Ö
If ω = ωn and ξ ∞ Ö 0
d) G(jω) =
limω G jω = - 90o
limω ∞G jω = -180o
e) | |
√
G(jω) = + = tan-1 (ω T) – ω L
2‐10


2‐6
MATLAB code:
clear all;
s = tf('s')
%Part(a)
Ga=10/(s-2)
figure(1)
nyquist(Ga)
%Part(b)
zeta=0.5; %asuuming a value for zeta <1
wn=2*pi*10 %asuuming a value for wn
Gb=1/(1+2*zeta*s/wn+s^2/wn^2)
figure(2)
nyquist(Gb)
%Part(c)
zeta=1.5; %asuuming a value for zeta >1
wn=2*pi*10
Gc=1/(1+2*zeta*s/wn+s^2/wn^2)
figure(3)
nyquist(Gc)
%Part(d)
T=3.5 %assuming value for parameter T
Gd=1/(s*(s*T+1))
figure(4)
nyquist(Gd)
2‐11


%Part(e)
T=3.5
L=0.5
Ge=pade(exp(-1*s*L),2)/(s*T+1)
figure(5)
hold on;
nyquist(Ge)
notes: In order to use Matlab Nyquist command, parameters needs to be assigned with values, and Pade
approximation needs to be used for exponential term in part (e).
Nyquist diagrams are as follows:
2‐12


Part(a)
Nyquist Diagram
2.5
1.5
0.5
Imaginary Axis
-0.5
-1
-1.5
-2
-2.5
-5 -4 -3 -2 -1 0 1
Real Axis

Part(b)
Nyquist Diagram
1.5
0.5
Imaginary Axis
-0.5
-1
-1.5
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
Real Axis

2‐13


Part(c)
Nyquist Diagram
0.8
0.6
0.4
0.2
Imaginary Axis
-0.2
-0.4
-0.6
-0.8
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
Real Axis

Part(d)
2‐14


Nyquist Diagram
60
40
20
Imaginary Axis
-20
-40
-60
-3.5 -3 -2.5 -2 -1.5 -1 -0.5 0
Real Axis

Part(e)
Nyquist Diagram
0.8
0.6
0.4
0.2
Imaginary Axis
-0.2
-0.4
-0.6
-0.8
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
Real Axis

2‐15


2-7) a) G(jω) =
. .
Steps for plotting |G|:
(1) For ω < 0.1, asymptote is
Break point: ω = 0.5

Slope = -1 or -20 dB/decade
(2) For 0.5 < ω < 10
Break point: ω = 10
Slope = -1+1 = 0 dB/decade
(3) For 10 < ω < 50:
Break point: ω = 50
(4) For ω > 50
Steps for plotting G
(1) = -90o
0: 90
(2) =
∞ 0
0: 0
.
(3) =
.
∞: 90
.
0: 90
.
(4) =
.
∞ 0
.
b) Let’s convert the transfer function to the following form:
G(jω) = . Ö G(s) =
.
Steps for plotting |G|:
(1) Asymptote: ω < 1 |G(jω)| 2.5 / ω

Slope: -1 or -20 dB/decade
| | 2.5
2‐16


(2) ωn =2 and ξ = 0.1 for second-order pole
break point: ω = 2
slope: -3 or -60 dB/decade
| | 5
Steps for plotting G(jω):
(1) for term 1/s the phase starts at -90o and at ω = 2 the phase will be -180o
(2) for higher frequencies the phase approaches -270o
c) Convert the transfer function to the following form:
0.01 1
0.01 1
9
for term , slope is -2 (-40 dB/decade) and passes through | | 1
(1) the breakpoint: ω = 1 and slope is zero

(2) the breakpoint: ω = 2 and slope is -2 or -40 dB/decade
|G(jω)|ω = 1 = 2ξ = 0.01 below the asymptote
|G(jω)|ω = 1 = = = 50 above the asymptote
ξ .
Steps for plotting G:
(1) phase starts from -180o due to
(2) G(jω)|ω =1 = 0
(3) G(jω)|ω = 2 = -180o
d) G(jω) = ω ω
ξ ω ω
Steps for plotting the |G|:
(1) Asymptote for <1 is zero
(2) Breakpoint: = 1, slope = -1 or -10 dB/decade
(3) As ξ is a damping ratio, then the magnitude must be obtained for various ξ when
0≤ξ≤1
2‐17


The high frequency slope is twice that of the asymptote for the single-pole case
Steps for plotting G:
(1) The phase starts at 0o and falls -1 or -20 dB/decade at = 0.2 and approaches -180o
at = 5. For > 5, the phase remains at -180o.
(2) As ξ is a damping ratio, the phase angles must be obtained for various ξ when
0≤ξ≤1
2‐8) Use this part to confirm the results from the previous part.
MATLAB code:
s = tf('s')
Ga=2000*(s+0.5)/(s*(s+10)*(s+50))
figure(1)
bode(Ga)
grid on;
Gb=25/(s*(s+2.5*s^2+10))
figure(2)
bode(Gb)
grid on;
Gc=(s+100*s^2+100)/(s^2*(s+25*s^2+100))
figure(3)
bode(Gc)
grid on;
2‐18


zeta = 0.2
wn=8
Gd=1/(1+2*zeta*s/wn+(s/wn)^2)
figure(4)
bode(Gd)
grid on;
t=0.3
'from pade approzimation:'
exp_term=pade(exp(-s*t),1)
Ge=0.03*(exp_term+1)^2/((exp_term-1)*(3*exp_term+1)*(exp_term+0.5))
figure(5)
bode(Ge)
grid on;
Part(a)
2‐19


Bode Diagram
60
40
Magnitude (dB)
20
-20
-40
-60
0
-45
Phase (deg)
-90
-135
-180
-2 -1 0 1 2 3
10 10 10 10 10 10
Frequency (rad/sec)

Part(b)
Bode Diagram
50
Magnitude (dB)
-50
-100
-90
-135
Phase (deg)
-180
-225
-270
-1 0 1 2
10 10 10 10
Frequency (rad/sec)

2‐20


Part(c)
Bode Diagram
60
40
Magnitude (dB)
20
-20
-40
0
-45
Phase (deg)
-90
-135
-180
-1 0 1
10 10 10
Frequency (rad/sec)

Part(d)
2‐21


Bode Diagram
10
Magnitude (dB)
-10
-20
-30
-40
-50
0
-45
Phase (deg)
-90
-135
-180
-1 0 1 2
10 10 10 10
Frequency (rad/sec)

Part(e)
Bode Diagram
0
-20
Magnitude (dB)
-40
-60
-80
-100
-120
0
-90
Phase (deg)
-180
-270
-1 0 1 2 3
10 10 10 10 10
Frequency (rad/sec)
2‐22


2-9)
a)
⎡ −1 2 0 ⎤ ⎡0 0 ⎤
⎡ u1 (t ) ⎤
A = 0 −2 3 ⎥
⎢ B = 1 0⎥
⎢ u (t ) = ⎢ ⎥
⎢ ⎥ ⎢ ⎥ ⎣ u 2 (t ) ⎦
⎢⎣ −1 −3 −1⎥⎦ ⎢⎣ 0 1 ⎥⎦
b)
1 2 0 2 0
2 0 1 0 1
3 4 1 0 0
2-10) We know that:

∞
∞
1
2
2
∞
Partial integration of equation (1) gives:

∞ ∞
1 ′
Ö sG s g 0 g′ t
Ö sG s – g 0
Differentiation of both sides of equation (1) with respect to s gives:

∞ ∞
∞ ∞
Comparing with equation (1), we conclude that:
2‐23


2-11) Let g(t) = ∞

then
Using Laplace transform and differentiation property, we have X(s) = sG(s)
Therefore G(s) = , which means:
∞
1
2-12) By Laplace transform definition:

∞
Now, consider τ = t - T, then:

∞ ∞
Which means: ↔
2-13) Consider:
f(t) = g1(t) g2(t) =
By Laplace transform definition:
By using time shifting theorem, we have:
2‐24


Let’s consider g(t) = g1(t) g2(t)
By inverse Laplace Transform definition, we have
1
2
Then
Where
therefore:
1
G s G s
2
2-14) a) We know that
= = sG(s) + g(0)
When s Æ ∞ , it can be written as:
2‐25


As
Therefore: 0
b) By Laplace transform differentiation property:
As
∞ 0
Therefore
0 ∞ 0
which means:
2‐15)
MATLAB code:
clear all;
syms t
s=tf('s')
f1 = (sin(2*t))^2
L1=laplace(f1)
2‐26


% f2 = (cos(2*t))^2 = 1-(sin(2*t))^2 ===> L(f2)=1/s-L(f1) ===>
L2= 1/s - 8/s/(s^2+16)
f3 = (cos(2*t))^2
L3=laplace(f3)
'verified as L2 equals L3'
2
MATLAB solution for L sin 2t is: { }
8/s/(s^2+16)
2
{ 2
}
Calculating L cos 2t based on L sin 2t { }
{ }
L cos 2 2t = (s^^3 + 8 s)/( s^4 + 16 s^2)
2
{
verifying L cos 2t : }
(8+s^2)/s/(s^2+16)
2‐16) (a) (b) (c)
5 4s 1 4
G ( s ) = G ( s ) = + G ( s ) =
( s + 5) 2
(s 2
+4 ) s+2 2
s + 4s + 8
(d) (e)
∞
1 1
G ( s ) = 2
s +4
G ( s) = ∑e kT ( s + 5 )
=
1− e
− T ( s +5 )

k =0
2‐27


2‐17) Note: %section (e) requires assignment of T and a numerical loop calculation
MATLAB code:
clear all;
syms t u
f1 = 5*t*exp(-5*t)
L1=laplace(f1)
f2 = t*sin(2*t)+exp(-2*t)
L2=laplace(f2)
f3 = 2*exp(-2*t)*sin(2*t)
L3=laplace(f3)
f4 = sin(2*t)*cos(2*t)
L4=laplace(f4)
%section (e) requires assignment of T and a numerical loop calculation
(a) g (t ) = 5te −5t u s (t )

Answer: 5/(s+5)^2
(b) g ( t ) = ( t sin 2t + e −2t )us ( t )
Answer: 4*s/(s^2+4)^2+1/(s+2)
(c) g ( t ) = 2e −2t sin 2t u s (t )
Answer: 4/(s^2+4*s+8)
2‐28


(d) g ( t ) = sin 2t cos 2t u s (t )
Answer: 2/(s^2+16)
∞
(e) g ( t ) = ∑e −5 kT
δ ( t − kT ) where δ(t) = unit‐impulse function
k =0
%section (e) requires assignment of T and a numerical loop calculation
2‐18 (a)
g (t ) = u s (t ) − 2u s (t − 1) + 2u s ( t − 2) − 2u s (t − 3) + L
−s
1 1− e
G(s) = (1 − 2e − s + 2e −2 s − 2e −3s + L) =

s (
s 1+ e
−s
)
gT (t ) = u s (t ) − 2u s (t − 1) + u s (t − 2) 0≤t≤2
1 1
(1 − 2e − s + e −2 s ) = (1 − e − s )
2
GT ( s ) =
s s
∞ ∞ −s
1 1− e
g ( t ) = ∑g T
( t − 2 k )u s ( t − 2 k ) G( s) = ∑ s (1 − e − s 2 −2 ks
) e =
s(1 + e
−s
)

k =0 k =0
(b)
g (t ) = 2tu s (t ) − 4(t − 0.5)u s (t − 0.5) + 4(t − 1)u s (t − 1) − 4(t − 1.5)u s (t − 1.5) + L
2 ( −0.5s )
2 1− e
G ( s ) = ( 1 − 2e
−0.5 s
+ 2e
−s
− 2e
−1.5 s
) s 2 1 + e−0.5s
+L =
s
2
( )
gT ( t ) = 2tu s ( t ) − 4( t − 0.5)u s ( t − 0.5) + 2( t − 1)u s ( t − 1) 0 ≤ t ≤ 1
2 2
(1 − 2e−0.5 s + e− s ) = s 2 (1 − e−0.5 s )
2
GT ( s ) = 2

s
∞ ∞
2 ( −0.5 s
2 1− e )
g (t ) = ∑ gT (t − k )us (t − k ) G ( s) = ∑ (1 − e )
−0.5 s 2
e
− ks
= 2
s (1 + e
k =0 k =0 s
2 −0.5 s
)

2‐19)
g ( t ) = ( t + 1)u s ( t ) − ( t − 1)u s ( t − 1) − 2u s ( t − 1) − ( t − 2 )u s ( t − 2) + ( t − 3)u s ( t − 3) + u s ( t − 3)
2‐29


1 1
G ( s ) =
s
2 (1 − e − s − e −2 s + e −3s ) + s (1 − 2e − s + e −3s )
2-20)
1 1
1
2-21)
1
1
2-22) 0 0 0
0 0
Ö 2 2 2 2
2‐30


Ö 2 2 2 2
2‐23
MATLAB code:
clear all;
syms t u s x1 x2 Fs
f1 = exp(-2*t)
L1=laplace(f1)/(s^2+5*s+4);
Eq2=solve('s*x1=1+x2','s*x2=-2*x1-3*x2+1','x1','x2')
f2_x1=Eq2.x1
f2_x2=Eq2.x2
f3=solve('(s^3-s+2*s^2+s+2)*Fs=-1+2-(1/(1+s))','Fs')
Here is the solution provided by MATLAB:
Part (a): F(s)=1/(s+2)/(s^2+5*s+4)

Part (b): X1(s)= (4+s)/(2+3*s+s^2)
X2(s)= (s‐2)/(2+3*s+s^2)

Part (c): F(s) = s/(1+s)/(s^3+2*s^2+2)
2‐31


2‐24)
MATLAB code:

clear all;
syms s Fs
f3=solve('s^2*Fs-Fs=1/(s-1)','Fs')
Answer from MATLAB: Y(s)= 1/(s-1)/(s^2-1)
2‐25)
MATLAB code:
clear all;
syms s CA1 CA2 CA3
v1=1000;
v2=1500;
v3=100;
k1=0.1
k2=0.2
k3=0.4
f1='s*CA1=1/v1*(1000+100*CA2-1100*CA1-k1*v1*CA1)'
f2='s*CA2=1/v2*(1100*CA1-1100*CA2-k2*v2*CA2)'
f3='s*CA3=1/v3*(1000*CA2-1000*CA3-k3*v3*CA3)'
Sol=solve(f1,f2,f3,'CA1','CA2','CA3')
CA1=Sol.CA1
CA3=Sol.CA2
CA4=Sol.CA3
2‐32


Solution from MATLAB:
CA1(s) =
1000*(s*v2+1100+k2*v2)/(1100000+s^2*v1*v2+1100*s*v1+s*v1*k2*v2+1100*s*v2+1100*k2*v2+k1*v1*s*v2+
1100*k1*v1+k1*v1*k2*v2)
CA3(s) =
1100000/(1100000+s^2*v1*v2+1100*s*v1+s*v1*k2*v2+1100*s*v2+1100*k2*v2+k1*v1*s*v2+1100*k1*v1+k1*
v1*k2*v2)
CA4 (s)=
1100000000/(1100000000+1100000*s*v3+1000*s*v1*k2*v2+1100000*s*v1+1000*k1*v1*s*v2+1000*k1*v1*k
2*v2+1100*s*v1*k3*v3+1100*s*v2*k3*v3+1100*k2*v2*s*v3+1100*k2*v2*k3*v3+1100*k1*v1*s*v3+1100*k1
*v1*k3*v3+1100000*k1*v1+1000*s^2*v1*v2+1100000*s*v2+1100000*k2*v2+1100000*k3*v3+s^3*v1*v2*v3+
1100*s^2*v1*v3+1100*s^2*v2*v3+s^2*v1*v2*k3*v3+s^2*v1*k2*v2*v3+s*v1*k2*v2*k3*v3+k1*v1*s^2*v2*v3
+k1*v1*s*v2*k3*v3+k1*v1*k2*v2*s*v3+k1*v1*k2*v2*k3*v3)
2-26) (a)
1 1 1 1 1 −2t 1 −3t
G( s) = − + g (t ) = − e + e t ≥ 0
3s 2( s + 2) 3( s + 3) 3 2 3
(b)
−2.5 5 2.5 −t −t −3t

G( s) = + 2
+ g ( t ) = −2.5e + 5te + 2.5e t ≥ 0
s + 1 ( s + 1) s+3
(c)
G(s) = ( 50
s
−
s +1
20
−
30 s + 20
s +4
2 ) e
−s
[
g ( t ) = 50 − 20 e
− ( t −1)
− 30 cos 2(t − 1) − 5 sin 2(t − 1) ] us (t − 1)
(d)
1 s −1 1 1 s
G( s) = − 2
= + 2
− 2
Taking the inverse Laplace transform,
s s + s+2 s s + s+2 s + s+2
g (t ) = 1 + 1.069e
−0.5t
[sin 1.323t + sin (1.323t − 69.3o )] = 1 + e−0.5t (1.447 sin 1.323t − cos1.323t ) t ≥ 0
2 −t
(e) g ( t ) = 0.5t e t ≥ 0
(f)Try using MATLAB
>> b=num*2
2‐33


b=
2 2 2
>>num =
1 1 1
>> denom1=[1 1]
denom1 =
1 1
>> denom2=[1 5 5]
denom2 =
1 5 5
>> num*2
ans =
2 2 2
>> denom=conv([1 0],conv(denom1,denom2))
denom =
1 6 10 5 0
>> b=num*2
b=
2 2 2
>> a=denom
a=
1 6 10 5 0
>> [r, p, k] = residue(b,a)
r=
-0.9889
2‐34


2.5889
-2.0000
0.4000
p=
-3.6180
-1.3820
-1.0000
k=[]
If there are no multiple roots, then
The number of poles n is
b r r r
= 1 + 2 + ... + n + k
a s + p1 s + p2 s + pn
In this case, p1 and k are zero. Hence,
0.4 0.9889 2.5889 2

G ( s) = − + −
s s + 3.6180 s + 1.3820 s + 1
g (t ) = 0.4 − 0.9889e −3.618t + 1.3820e −2.5889t − 2e − t
(g)
Ö G s 2e 2e 2e 1
(h)
Ö 3
2‐35


(i)
2‐27
MATLAB code:

clear all;
syms s
f1=1/(s*(s+2)*(s+3))
F1=ilaplace(f1)
f2=10/((s+1)^2*(s+3))
F2=ilaplace(f2)
f3=10*(s+2)/(s*(s^2+4)*(s+1))*exp(-s)
F3=ilaplace(f3)
f4=2*(s+1)/(s*(s^2+s+2))
F4=ilaplace(f4)
f5=1/(s+1)^3
F5=ilaplace(f5)
f6=2*(s^2+s+1)/(s*(s+1.5)*(s^2+5*s+5))
F6=ilaplace(f6)
s=tf('s')
f7=(2+2*s*pade(exp(-1*s),1)+4*pade(exp(-2*s),1))/(s^2+3*s+2) %using Pade command

for exponential term
2‐36


[num,den]=tfdata(f7,'v') %extracting the polynomial values
syms s
f7n=(-2*s^3+6*s+12)/(s^4+6*s^3+13*s^2+12*s+4) %generating sumbolic function for

ilaplace
F7=ilaplace(f7n)
f8=(2*s+1)/(s^3+6*s^2+11*s+6)
F8=ilaplace(f8)
f9=(3*s^3+10^s^2+8*s+5)/(s^4+5*s^3+7*s^2+5*s+6)
F9=ilaplace(f9)
Solution from MATLAB for the Inverse Laplace transforms:
1
Part (a): G ( s) =
s( s + 2)( s + 3)
G(t)=‐1/2*exp(‐2*t)+1/3*exp(‐3*t)+1/6
To simplify:
syms t
digits(3)
vpa(‐1/2*exp(‐2*t)+1/3*exp(‐3*t)+1/6)
ans =‐.500*exp(‐2.*t)+.333*exp(‐3.*t)+.167
10
Part (b): G ( s) =
( s + 1) 2 ( s + 3)
G(t)= 5/2*exp(‐3*t)+5/2*exp(‐t)*(‐1+2*t)
2‐37


100( s + 2)
Part (c): G( s) = 2
e− s
s( s + 4 )( s + 1)
G(t)=Step(t‐1)*(‐4*cos(t‐1)^2+2*sin(t‐1)*cos(t‐1)+4*exp(‐1/2*t+1/2)*cosh(1/2*t‐1/2)‐4*exp(‐t+1)‐cos(2*t‐2)‐
2*sin(2*t‐2)+5)
2( s + 1)
Part (d): G ( s) =
s( s 2 + s + 2 )
G(t)= 1+1/7*exp(‐1/2*t)*(‐7*cos(1/2*7^(1/2)*t)+3*7^(1/2)*sin(1/2*7^(1/2)*t))
To simplify:
syms t
digits(3)
vpa(1+1/7*exp(‐1/2*t)*(‐7*cos(1/2*7^(1/2)*t)+3*7^(1/2)*sin(1/2*7^(1/2)*t)))
ans = 1.+.143*exp(‐.500*t)*(‐7.*cos(1.32*t)+7.95*sin(1.32*t))
1
Part (e): G ( s) =
( s + 1) 3
G(t)= 1/2*t^2*exp(‐t)
2( s 2 + s + 1)
Part (f): G( s) =
. )( s 2 + 5s + 5)
s( s + 15
G(t)= 4/15+28/3*exp(‐3/2*t)‐16/5*exp(‐5/2*t)*(3*cosh(1/2*t*5^(1/2))+5^(1/2)*sinh(1/2*t*5^(1/2)))
2 + 2 se − s + 4e −2 s
Part (g): G ( s ) =
s 2 + 3s + 2
2‐38


G(t)= 2*exp(‐2*t)*(7+8*t)+8*exp(‐t)*(‐2+t)
2s + 1
Part (h): G ( s) =
s + 6 s 2 + 11s + 6
3
G(t)= ‐1/2*exp(‐t)+3*exp(‐2*t)‐5/2*exp(‐3*t)
3s 3 + 10 s 2 + 8s + 5
Part (i): G ( s) = 4
s + 5 s 3 + 7 s 2 + 5s + 6
G(t)= ‐7*exp(‐2*t)+10*exp(‐3*t)‐
1/10*ilaplace(10^(2*s)/(s^2+1)*s,s,t)+1/10*ilaplace(10^(2*s)/(s^2+1),s,t)+1/10*sin(t)*(10+dirac(t)*(‐exp(‐
3*t)+2*exp(‐2*t)))
2‐39


2-28) Ax t Bu t
a)
b)
3 2
2 2
2-29) (a) (b)

Y (s) 3s + 1 Y (s) 5
= 3 2
= 4 2

R( s) s + 2 s + 5s + 6 R( s) s + 10 s + s + 5
(c) (d)
−s
Y (s) s ( s + 2) Y (s) 1 + 2e
= 4 3 2
= 2

R( s) s + 10 s + 2 s + s + 2 R( s) 2s + s + 5
e) 1
Ö 5 2 2
By using Laplace transform, we have:
4 5 2
As X s e Y s , then
2‐40


2 1
4
Then:
f) By using Laplace transform we have:
2
2 2 2
As a result:
2‐30)
After taking the Laplace transform, the equation was solved in terms of Y(s), and consecutively was divided by
input R(s) to obtain Y(s)/R(s):
MATLAB code:
clear all;
syms Ys Rs s
sol1=solve('s^3*Ys+2*s^2*Ys+5*s*Ys+6*Ys=3*s*Rs+Rs','Ys')
Ys_Rs1=sol1/Rs
sol2=solve('s^4*Ys+10*s^2*Ys+s*Ys+5*Ys=5*Rs','Ys')
Ys_Rs2=sol2/Rs
sol3=solve('s^3*Ys+10*s^2*Ys+2*s*Ys+2*Ys/s=s*Rs+2*Rs','Ys')
Ys_Rs3=sol3/Rs
sol4=solve('2*s^2*Ys+s*Ys+5*Ys=2*Rs*exp(-1*s)','Ys')
Ys_Rs4=sol4/Rs
2‐41


%Note: Parts E&F are too complicated with MATLAB, Laplace of integral is not
executable in MATLAB.....skipped
MATLAB Answers:
Part (a): Y(s)/R(s)= (3*s+1)/(5*s+6+s^3+2*s^2);
Part (b): Y(s)/R(s)= 5/(10*s^2+s+5+s^4)
Part (c): Y(s)/R(s)= (s+2)*s/(2*s^2+2+s^4+10*s^3)
Part (d): Y(s)/R(s)= 2*exp(‐s)/(2*s^2+s+5)
%Note: Parts E&F are too complicated with MATLAB, Laplace of integral is not
executable in MATLAB.....skipped
2‐31
MATLAB code:
clear all;
s=tf('s')
%Part a
Eq=10*(s+1)/(s^2*(s+4)*(s+6));
[num,den]=tfdata(Eq,'v');
[r,p] = residue(num,den)
%Part b
Eq=(s+1)/(s*(s+2)*(s^2+2*s+2));
%Part c
Eq=5*(s+2)/(s^2*(s+1)*(s+5));
2‐42


%Part d
Eq=5*(pade(exp(-2*s),1))/(s^2+s+1); %Pade approximation oreder 1 used
%Part e
Eq=100*(s^2+s+3)/(s*(s^2+5*s+3));
%Part f
Eq=1/(s*(s^2+1)*(s+0.5)^2);
%Part g
Eq=(2*s^3+s^2+8*s+6)/((s^2+4)*(s^2+2*s+2));
%Part h
Eq=(2*s^4+9*s^3+15*s^2+s+2)/(s^2*(s+2)*(s+1)^2);
The solutions are presented in the form of two vectors, r and p, where for each case, the partial fraction
expansion is equal to:
2‐43


b( s ) r r r
= 1 + 2 + ... + n
a( s ) s − p1 s − p 2 s − pn
Following are r and p vectors for each part:
Part(a):
r =0.6944
‐0.9375
0.2431
0.4167
p =‐6.0000
‐4.0000
0
0
Part(b):
r =0.2500
‐0.2500 ‐ 0.0000i
‐0.2500 + 0.0000i
0.2500
p =‐2.0000
‐1.0000 + 1.0000i
2‐44


‐1.0000 ‐ 1.0000i
0
Part(c):
r =0.1500
1.2500
‐1.4000
2.0000
p = ‐5
‐1
0
0
Part(d):
r =10.0000
‐5.0000 ‐ 0.0000i
‐5.0000 + 0.0000i
p =‐1.0000
‐0.5000 + 0.8660i
‐0.5000 ‐ 0.8660i
2‐45


Part(e):
r =110.9400
‐110.9400
100.0000
p =‐4.3028
‐0.6972
0
Part(f):
r =0.2400 + 0.3200i
0.2400 ‐ 0.3200i
‐4.4800
‐1.6000
4.0000
p =‐0.0000 + 1.0000i
‐0.0000 ‐ 1.0000i
‐0.5000
‐0.5000
0
2‐46


Part(g):
r =‐0.1000 + 0.0500i
‐0.1000 ‐ 0.0500i
1.1000 + 0.3000i
1.1000 ‐ 0.3000i
p =0.0000 + 2.0000i
0.0000 ‐ 2.0000i
‐1.0000 + 1.0000i
‐1.0000 ‐ 1.0000i
Part(h):
r =5.0000
‐1.0000
9.0000
‐2.0000
1.0000
p =‐2.0000
‐1.0000
‐1.0000
0
0
2‐47


2‐32)
MATLAB code:
clear all;
syms s
%Part a
Eq=10*(s+1)/(s^2*(s+4)*(s+6));
ilaplace(Eq)
%Part b
Eq=(s+1)/(s*(s+2)*(s^2+2*s+2));
ilaplace(Eq)
%Part c
Eq=5*(s+2)/(s^2*(s+1)*(s+5));
ilaplace(Eq)
%Part d
exp_term=(-s+1)/(s+1) %pade approcimation
Eq=5*exp_term/((s+1)*(s^2+s+1));
ilaplace(Eq)
%Part e
Eq=100*(s^2+s+3)/(s*(s^2+5*s+3));
ilaplace(Eq)
%Part f
Eq=1/(s*(s^2+1)*(s+0.5)^2);
ilaplace(Eq)
2‐48


%Part g
Eq=(2*s^3+s^2+8*s+6)/((s^2+4)*(s^2+2*s+2));
ilaplace(Eq)
%Part h
Eq=(2*s^4+9*s^3+15*s^2+s+2)/(s^2*(s+2)*(s+1)^2);
ilaplace(Eq)
MATLAB Answers:
Part(a):
G(t)= ‐15/16*exp(‐4*t)+25/36*exp(‐6*t)+35/144+5/12*t
To simplify:
syms t
digits(3)
vpa(‐15/16*exp(‐4*t)+25/36*exp(‐6*t)+35/144+5/12*t)
ans =‐.938*exp(‐4.*t)+.694*exp(‐6.*t)+.243+.417*tPart(b):
G(t)= 1/4*exp(‐2*t)+1/4‐1/2*exp(‐t)*cos(t)
Part(c):
G(t)= 5/4*exp(‐t)‐7/5+3/20*exp(‐5*t)+2*t
2‐49


Part(d):
G(t)= ‐5*exp(‐1/2*t)*(cos(1/2*3^(1/2)*t)+3^(1/2)*sin(1/2*3^(1/2)*t))+5*(1+2*t)*exp(‐t)
Part(e):
G(t)= 100‐800/13*exp(‐5/2*t)*13^(1/2)*sinh(1/2*t*13^(1/2))
Part(f):
G(t)= 4+12/25*cos(t)‐16/25*sin(t)‐8/25*exp(‐1/2*t)*(5*t+14)
Part(g):
G(t)= ‐1/5*cos(2*t)‐1/10*sin(2*t)+1/5*(11*cos(t)‐3*sin(t))*exp(‐t)
Part(h):
G(t)= ‐2+t+5*exp(‐2*t)+(‐1+9*t)*exp(‐t)
2‐50


2-33) (a) Poles are at s = 0, − 15
. + j16583
. , − 15
. − j16583
. One poles at s = 0. Marginally stable.
(b) Poles are at s = −5, − j 2 , j 2 Two poles on jω axis. Marginally stable.
(c) Poles are at s = −0.8688, 0.4344 + j 2.3593, 0.4344 − j 2.3593 Two poles in RHP. Unstable.
(d) Poles are at s = −5, − 1 + j , − 1 − j All poles in the LHP. Stable.
(e) Poles are at s = −13387
. , 16634
. + j 2.164, 16634
. − j 2.164 Two poles in RHP. Unstable.
(f) Poles are at s = −22.8487 ± j 22.6376, 213487

. ± j 22.6023 Two poles in RHP. Unstable.
2-34) Find the Characteristic equations and then use the roots command.
(a)
p= [ 1 3 5 0]
sr = roots(p)
p=
1 3 5 0
sr =
-1.5000 + 1.6583i
-1.5000 - 1.6583i
(b) p=conv([1 5],[1 0 2])
sr = roots(p)
p=
1 5 2 10
2‐51


sr =
-5.0000
0.0000 + 1.4142i
0.0000 - 1.4142i
(c)
>> roots([1 5 5])
ans =
-3.6180
-1.3820
(d) roots(conv([1 5],[1 2 2]))
ans =
-5.0000
-1.0000 + 1.0000i
-1.0000 - 1.0000i
(e) roots([1 -2 3 10])
ans =
1.6694 + 2.1640i
1.6694 - 2.1640i
-1.3387
2‐52


(f) roots([1 3 50 1 10^6])
-22.8487 +22.6376i
-22.8487 -22.6376i
21.3487 +22.6023i
21.3487 -22.6023i
Alternatively
Problem 2‐34
MATLAB code:
% Question 2-34,
clear all;
s=tf('s')
%Part a
Eq=10*(s+2)/(s^3+3*s^2+5*s);
roots(den)
%Part b
Eq=(s-1)/((s+5)*(s^2+2));
roots(den)
%Part c
Eq=1/(s^3+5*s+5);
roots(den)
2‐53


%Part d
Eq=100*(s-1)/((s+5)*(s^2+2*s+2));
roots(den)
%Part e
Eq=100/(s^3-2*s^2+3*s+10);
roots(den)
%Part f
Eq=10*(s+12.5)/(s^4+3*s^3+50*s^2+s+10^6);
roots(den)
MATLAB answer:
Part(a)

0
‐1.5000 + 1.6583i
‐1.5000 ‐ 1.6583i
Part(b)

‐5.0000
2‐54


‐0.0000 + 1.4142i
‐0.0000 ‐ 1.4142i
Part(c)

0.4344 + 2.3593i
0.4344 ‐ 2.3593i
‐0.8688
Part(d)

‐5.0000
‐1.0000 + 1.0000i
‐1.0000 ‐ 1.0000i
Part(e)

1.6694 + 2.1640i
1.6694 ‐ 2.1640i
‐1.3387
Part(f)

2‐55


‐22.8487 +22.6376i
‐22.8487 ‐22.6376i
21.3487 +22.6023i
21.3487 ‐22.6023i
2-35)
(a) s 3 + 25 s 2 + 10 s + 450 = 0 Roots: −25.31, 0.1537 + j 4.214, 0.1537 − 4.214
Routh Tabulation:
3
s 1 10
2
s 25 450
250 − 450 Two sign changes in the first column. Two roots in RHP.
1
s = −8 0
25
0
s 450
3 2
(b) s + 25 s + 10 s + 50 = 0 Roots: −24.6769, − 0.1616 + j1.4142, − 0.1616 − j1.4142
Routh Tabulation:
3
s 1 10
2
s 25 50
250 − 50 No sign changes in the first column. No roots in RHP.
1
s =8 0
25
0
s 50
(c) s 3 + 25 s 2 + 250 s + 10 = 0 Roots: −0.0402, − 12.48 + j 9.6566, − j 9.6566

Routh Tabulation:
2‐56


3
s 1 250
2
s 25 10
6250 − 10 No sign changes in the first column. No roots in RHP.
1
s = 249.6 0
25
0
s 10
4 3 2
(d) 2 s + 10 s + 5.5 s + 5.5 s + 10 = 0 Roots: −4.466, − 1116
. , 0.2888 + j 0.9611, 0.2888 − j 0.9611

Routh Tabulation:
4
s 2 5.5 10
3
s 10 5.5
2 55 − 11
s = 4.4 10
10
1 24.2 − 100
s = −75.8
4.4
0
s 10
Two sign changes in the first column. Two roots in RHP.
(e) s 6 + 2 s 5 + 8 s 4 + 15 s 3 + 20 s 2 + 16 s + 16 = 0 Roots: −1.222 ± j 0.8169, 0.0447 ± j1153

. , 0.1776 ± j 2.352

Routh Tabulation:
6
s 1 8 20 16
5
s 2 15 16
16 − 15 40 − 16
4
s = 0.5 = 12
2 2
3
s − 33 − 48
2‐57


2 −396 + 24
s = 11.27 16
−33
1 −5411
. + 528
s = −116
. 0
11.27
0
s 0
Four sign changes in the first column. Four roots in RHP.
4 3 2
(f) s + 2 s + 10 s + 20 s + 5 = 0 Roots: −0.29, − 1788
. , 0.039 + j 3105
. , 0.039 − j 3105
.

Routh Tabulation:
4
s 1 10 5
3
s 2 20
20 − 20
2
s =0 5
2
2
s ε 5 Replace 0 in last row by ε
1 20ε − 10 10
s ≅−
ε ε Two sign changes in first column. Two roots in RHP.
0
s 5
(g)
s8 1 8 20 16 0
s7 2 12 16 0 0
s6 2 12 16 0 0
s5 0 0 0 0 0
2 12 16
12 60 64
2‐58


s5 12 60 64 0
s4 2 16 0 0
3
s3 28 64 0 0
s2 0.759 0 0 0
s1 28 0
s0 0
2-36) Use MATLAB roots command
a) roots([1 25 10 450])
ans =
-25.3075
0.1537 + 4.2140i
0.1537 - 4.2140i
b) roots([1 25 10 50])
ans =
-24.6769
-0.1616 + 1.4142i
-0.1616 - 1.4142i
c) roots([1 25 250 10])
ans =
-12.4799 + 9.6566i
-12.4799 - 9.6566i
2‐59


-0.0402
d) roots([2 10 5.5 5.5 10])
ans =
-4.4660
-1.1116
0.2888 + 0.9611i
0.2888 - 0.9611i
e) roots([1 2 8 15 20 16 16])
ans =
0.1776 + 2.3520i
0.1776 - 2.3520i
-1.2224 + 0.8169i
-1.2224 - 0.8169i
0.0447 + 1.1526i
0.0447 - 1.1526i
f) roots([1 2 10 20 5])
ans =
0.0390 + 3.1052i
0.0390 - 3.1052i
-1.7881
-0.2900
g) roots([1 2 8 12 20 16 16])
2‐60


ans =
0.0000 + 2.0000i
0.0000 - 2.0000i
-1.0000 + 1.0000i
-1.0000 - 1.0000i
0.0000 + 1.4142i
0.0000 - 1.4142i
Alternatively use the approach in this Chapter’s Section 2‐14:
1. Activate MATLAB
2. Go to the directory containing the ACSYS software.
3. Type in
Acsys

4. Then press the “transfer function Symbolic button
2‐61


5. Enter the characteristic equation in the denominator and press the “Routh‐Hurwitz” push‐
button.
RH =
[ 1, 10]
[ 25, 450]
[ -8, 0]
[ 450, 0]
Two sign changes in the first column. Two roots in RHP=> UNSTABLE
2-37) Use the MATLAB “roots” command same as in the previous problem.
2‐62


2-38) To solve using MATLAB, set the value of K in an iterative process and find the roots such that at
least one root changes sign from negative to positive. Then increase resolution if desired.
Example: in this case 0<K<12 ( increase resolution by changing the loop to: for K=11:.1:12)
for K=0:12
K
roots([1 25 15 20 K])
end
K=
ans =
-24.4193
-0.2904 + 0.8572i
-0.2904 - 0.8572i
K=
ans =
2‐63


-24.4192
-0.2645 + 0.8485i
-0.2645 - 0.8485i
-0.0518
K=
ans =
-24.4191
-0.2369 + 0.8419i
-0.2369 - 0.8419i
-0.1071
K=
2‐64


ans =
-24.4191
-0.2081 + 0.8379i
-0.2081 - 0.8379i
-0.1648
K=
ans =
-24.4190
-0.1787 + 0.8369i
-0.1787 - 0.8369i
-0.2237
K=
2‐65


ans =
-24.4189
-0.1496 + 0.8390i
-0.1496 - 0.8390i
-0.2819
K=
ans =
-24.4188
-0.1215 + 0.8438i
-0.1215 - 0.8438i
-0.3381
K=
2‐66


ans =
-24.4188
-0.0951 + 0.8508i
-0.0951 - 0.8508i
-0.3911
K=
ans =
-24.4187
-0.0704 + 0.8595i
-0.0704 - 0.8595i
-0.4406
K=
2‐67


9
ans =
-24.4186
-0.0475 + 0.8692i
-0.0475 - 0.8692i
-0.4864
K=
10
ans =
-24.4186
-0.0263 + 0.8796i
-0.0263 - 0.8796i
-0.5288
K=
2‐68


11
ans =
-24.4185
-0.0067 + 0.8905i
-0.0067 - 0.8905i
-0.5681
K=
12
ans =
-24.4184
0.0115 + 0.9015i
0.0115 - 0.9015i
-0.6046
Alternatively use the approach in this Chapter’s Section 2‐14:
2‐69


1. Activate MATLAB
3. Type in
Acsys

4. Then press the “transfer function Symbolic button
2‐70


button.
RH =

[ 1, 15, k]
[ 25, 20, 0]
[ 71/5, k, 0]
[ ‐125/71*k+20, 0, 0]
[ k, 0, 0]
6. Find the values of K to make the system unstable following the next steps.
2‐71


Alternative Problem 2‐36
Using ACSYS toolbar under “Transfer Function Symbolic”, the Routh‐Hurwitz option can be used to generate RH
matrix based on denominator polynhomial. The system is stable if and only if the first column of this matrix
contains NO negative values.
MATLAB code: to calculate the number of right hand side poles
%Part a
den_a=[1 25 10 450]
roots(den_a)
%Part b
den_b=[1 25 10 50]
roots(den_b)
%Part c
den_c=[1 25 250 10]
roots(den_c)
%Part d
den_d=[2 10 5.5 5.5 10]
roots(den_d)
%Part e
den_e=[1 2 8 15 20 16 16]
roots(den_e)
%Part f
den_f=[1 2 10 20 5]
roots(den_f)
2‐72


%Part g
den_g=[1 2 8 12 20 16 16 0 0]
roots(den_g)

using ACSYS, the denominator polynomial can be inserted, and by clicking on the “Routh‐Hurwitz” button, the R‐
H chart can be observed in the main MATLAB command window:
Part(a): for the transfer function in part (a), this chart is:
RH chart =
[ 1, 10]
[ 25, 450]
[ ‐8, 0]
[ 450, 0]
Unstable system due to ‐8 on the 3rd row.
2 complex conjugate poles on right hand side. All the poles are:
‐25.3075
0.1537 + 4.2140i and 0.1537 ‐ 4.2140i
2‐73


Part (b):
RH chart:
[ 1, 10]
[ 25, 50]
[ 8, 0]
[ 50, 0]
Stable system >> No right hand side pole
Part (c):
2‐74


RH chart:
[ 1, 250]
[ 25, 10]
[ 1248/5, 0]
[ 10, 0]
Part (d):
RH chart:
[ 2, 11/2, 10]
[ 10, 11/2, 0]
[ 22/5, 10, 0]
[ ‐379/22, 0, 0]
[ 10, 0, 0]
Unstable system due to ‐379/22 on the 4th row.
‐4.4660
‐1.1116
0.2888 + 0.9611i
0.2888 ‐ 0.9611i
2‐75


Part (e):
RH chart:
[ 1, 8, 20, 16]
[ 2, 15, 16, 0]
[ 1/2, 12, 16, 0]
[ ‐33, ‐48, 0, 0]
[ 124/11, 16, 0, 0]
[ ‐36/31, 0, 0, 0]
[ 16, 0, 0, 0]
Unstable system due to ‐33 and ‐36/31 on the 4th and 6th row.
0.1776 + 2.3520i
0.1776 ‐ 2.3520i
‐1.2224 + 0.8169i
‐1.2224 ‐ 0.8169i
0.0447 + 1.1526i
0.0447 ‐ 1.1526i
Part (f):
RH chart:
[ 1, 10, 5]
[ 2, 20, 0]
2‐76


[ eps, 5, 0]
[ (‐10+20*eps)/eps, 0, 0]
[ 5, 0, 0]
Unstable system due to ((‐10+20*eps)/eps) on the 4th.
2 complex conjugate poles slightly on right hand side. All the poles are:
0.0390 + 3.1052i
0.0390 ‐ 3.1052i
‐1.7881
‐0.2900
Part (g):
RH chart:
[ 1, 8, 20, 16, 0]
[ 2, 12, 16, 0, 0]
[ 2, 12, 16, 0, 0]
[ 12, 48, 32, 0, 0]
[ 4, 32/3, 0, 0, 0]
[ 16, 32, 0, 0, 0]
[ 8/3, 0, 0, 0, 0]
[ 32, 0, 0, 0, 0]
[ 0, 0, 0, 0, 0]
2‐77


6 poles wt zero real part:
0
0
0.0000 + 2.0000i
0.0000 ‐ 2.0000i
‐1.0000 + 1.0000i
‐1.0000 ‐ 1.0000i
0.0000 + 1.4142i
0.0000 ‐ 1.4142i
2‐78


(a) s 4 + 25 s 3 + 15 s 2 + 20 s + K = 0
Routh Tabulation:
4
s 1 15 K
3
s 25 20
2 375 − 20
s = 14.2 K
25
1 284 − 25 K
s = 20 − 176
. K 20 − 176
. K > 0 or K < 1136
.
14.2
0
s K K>0
Thus, the system is stable for 0 < K < 11.36. When K = 11.36, the system is marginally stable. The
2 2
auxiliary equation is A( s ) = 14.2 s + 11.36 = 0. The solution of A(s) = 0 is s = −0.8. The
frequency of oscillation is 0.894 rad/sec.
4 3 2
(b) s + Ks + 2 s + ( K + 1) s + 10 = 0
Routh Tabulation:
4
s 1 2 10
3
s K K +1 K>0
2 2K − K − 1 K −1
s = 10 K >1
K K
2
1 −9 K − 1 2
s − 9K − 1 > 0
K −1
0
s 10

2 2
The conditions for stability are: K > 0, K > 1, and −9 K − 1 > 0 . Since K is always positive, the
last condition cannot be met by any real value of K. Thus, the system is unstable for all values of K.
2‐79


3 2
(c) s + ( K + 2 ) s + 2 Ks + 10 = 0
Routh Tabulation:
3
s 1 2K
2
s K+2 10 K > −2
2
2 K + 4 K − 10
1 2
s K + 2K − 5 > 0
K+2
0
s 10

2
The conditions for stability are: K > −2 and K + 2 K − 5 > 0 or (K +3.4495)(K − 1.4495) > 0,
or K > 1.4495. Thus, the condition for stability is K > 1.4495. When K = 1.4495 the system is
2 2
marginally stable. The auxiliary equation is A( s ) = 3.4495 s + 10 = 0. The solution is s = −2.899 .
The frequency of oscillation is 1.7026 rad/sec.
3 2
(d) s + 20 s + 5 s + 10 K = 0
Routh Tabulation:
3
s 1 5
2
s 20 10 K
100 − 10 K
1
s = 5 − 0.5K 5 − 0.5K > 0 or K < 10
20
0
s 10 K K>0
The conditions for stability are: K > 0 and K < 10. Thus, 0 < K < 10. When K = 10, the system is
2
marginally stable. The auxiliary equation is A( s ) = 20 s + 100 = 0. The solution of the auxiliary
2
equation is s = −5. The frequency of oscillation is 2.236 rad/sec.
2‐80


4 3 2
(e) s + Ks + 5 s + 10 s + 10 K = 0
Routh Tabulation:
4
s 1 5 10 K
3
s K 10 K>0
2 5 K − 10
s 10 K 5K − 10 > 0 or K > 2
K
50 K − 100 2
− 10 K 3
1 K 50 K − 100 − 10 K 3
s = 5 K − 10 − K > 0
5K − 10 5 K − 10
K
0
s 10 K K>0
3
The conditions for stability are: K > 0, K > 2, and 5 K − 10 − K > 0.
Use Matlab to solve for k from last condition
>> syms k
>> kval=solve(5*k‐10+k^3,k);
>> eval(kval)
kval =
1.4233
‐0.7117 + 2.5533i
‐0.7117 ‐ 2.5533i
So K>1.4233.
Thus, the conditions for stability is: K > 2
2‐81


4 3 2
(f) s + 12.5 s + s + 5 s + K = 0
Routh Tabulation:
4
s 1 1 K
3
s 12.5 5
2 12.5 − 5
s = 0.6 K
12.5
1 3 − 12.5 K
s = 5 − 20.83 K 5 − 20.83 K > 0 or K < 0.24
0.6
0
s K K>0
The condition for stability is 0 < K < 0.24. When K = 0.24 the system is marginally stable. The auxiliary
2 2
equation is A( s ) = 0.6 s + 0.24 = 0. The solution of the auxiliary equation is s = −0.4. The frequency of
oscillation is 0.632 rad/sec.
2-39)
3 2
The characteristic equation is Ts + ( 2T + 1) s + ( 2 + K ) s + 5 K = 0
Routh Tabulation:
3
s T K+2 T >0

2
s 2T + 1 5K T > −1 / 2
1 ( 2T + 1)( K + 2 ) − 5 KT
s K (1 − 3T ) + 4T + 2 > 0
2T + 1
0
s 5K K>0

4T + 2
The conditions for stability are: T > 0, K > 0, and K < . The regions of stability in the
3T − 1
2‐82


T‐versus‐K parameter plane is shown below.

2‐40 Use the approach in this Chapter’s Section 2‐14:
1. Activate MATLAB
3. Type in
Acsys
2‐83


4. Then press the “transfer function Symbolic button.”
2‐84


button.
RH =

[1, 50000, 24*k]
[600, k, 80*k]
[‐1/600*k+50000, 358/15*k, 0]
[ (35680*k‐1/600*k^2)/(‐1/600*k+50000), 80*k, 0]
[ 24*k*(k^2‐21622400*k+5000000000000)/(k‐30000000)/(35680*k‐1/600*k^2)*(‐1/600*k+50000),
0, 0]
[80*k, 0, 0]
6. Find the values of K to make the system unstable following the next steps.
(a) Characteristic equation: s 5 + 600 s 4 + 50000 s 3 + Ks 2 + 24 Ks + 80 K = 0
Routh Tabulation:
5
s 1 50000 24 K
4
s 600 K 80 K
7
3 3 × 10 − K 14320 K 7
s K < 3 × 10
600 600
2
2 21408000 K − K
s 7
80 K K < 21408000
3 × 10 − K
16 11 2
1 −7.2 × 10 + 3113256
. × 10 K − 14400 K 2 7 12
s K − 2.162 × 10 K + 5 × 10 < 0
600(21408000 − K )
0
s 80 K K>0
2‐85


Conditions for stability:
3 7
From the s row: K < 3 × 10
2 7
From the s row: K < 2.1408 × 10
1 2 7 12 5 7
From the s row: K − 2.162 × 10 K + 5 × 10 < 0 or ( K − 2.34 × 10 )( K − 2.1386 × 10 ) < 0
5 7
Thus, 2.34 × 10 < K < 2.1386 × 10
0
From the s row: K > 0
5 7
Thus, the final condition for stability is: 2.34 × 10 < K < 2.1386 × 10
5
When K = 2.34 × 10 ω = 10.6 rad/sec.
7
When K = 2.1386 × 10 ω = 188.59 rad/sec.
(b) Characteristic equation: s 3 + ( K + 2 ) s 2 + 30 Ks + 200 K = 0

Routh tabulation:
3
s 1 30 K
2
s K+2 200 K K > −2
2
1 30 K − 140 K
s K > 4.6667
K+2
0
s 200 K K>0
Stability Condition: K > 4.6667
2 2
When K = 4.6667, the auxiliary equation is A( s ) = 6.6667 s + 933.333 = 0 . The solution is s = −140.
3 2
(c) Characteristic equation: s + 30 s + 200 s + K = 0

2‐86


Routh tabulation:
3
s 1 200
2
s 30 K
6000 − K
1
s K < 6000
30
0
s K K>0
Stabililty Condition: 0 < K < 6000
2 2
When K = 6000, the auxiliary equation is A( s ) = 30 s + 6000 = 0. The solution is s = −200.
3 2
(d) Characteristic equation: s + 2 s + ( K + 3) s + K + 1 = 0
Routh tabulation:
3
s 1 K+3
2
s 2 K +1
K +5
1
s K > −5
30
0
s K +1 K > −1
Stability condition: K > −1. When K = −1 the zero element occurs in the first element of the
0
s row. Thus, there is no auxiliary equation. When K = −1, the system is marginally stable, and one
of the three characteristic equation roots is at s = 0. There is no oscillation. The system response
would increase monotonically.
2‐87


2‐42 State equation: Open‐loop system: x& ( t ) = Ax ( t ) + Bu ( t )
⎡1 −2 ⎤ ⎡0 ⎤
A=⎢ ⎥ B= ⎢ ⎥
⎣10 0 ⎦ ⎣1 ⎦
Closed‐loop system: x& ( t ) = ( A − BK )x ( t )
⎡ 1 −2 ⎤
A − BK = ⎢
⎣10 − k1 − k 2 ⎥⎦
Characteristic equation of the closed‐loop system:
s −1 2
sI − A + BK = = s + ( k 2 − 1) s + 20 − 2k1 − k 2 = 0
2

−10 + k1 s + k2
Stability requirements:
k 2 − 1 > 0 or k 2 > 1
20 − 2k1 − k 2 > 0 or k 2 < 20 − 2k1
Parameter plane:

2‐43) Characteristic equation of closed‐loop system:
s −1 0
sI − A + BK = 0 −1 = s + ( k 3 + 3 ) s + ( k 2 + 4 ) s + k1 = 0
3 2
s
k1 k2 + 4 s + k3 + 3

2‐88


Routh Tabulation:
k2 + 4
3
s 1
k3 + 3 k 3 +3>0 or k 3 > −3
2
s k1
(k + 3 )( k 2 + 4 ) − k1
s
1 3
(k + 3) ( k + 4 ) − k > 0
k3 + 3
3 2 1
k >0
0
s k 1 1
Stability Requirements:
k3 > −3, k1 > 0, (k 3

+ 3 )( k 2 + 4 ) − k1 > 0
2‐44 (a) Since A is a diagonal matrix with distinct eigenvalues, the states are decoupled from each other. The
second row of B is zero; thus, the second state variable, x 2 is uncontrollable. Since the uncontrollable
state has the eigenvalue at −3 which is stable, and the unstable state x3 with the eigenvalue at −2 is
controllable, the system is stabilizable.
(b) Since the uncontrollable state x1 has an unstable eigenvalue at 1, the system is no stabilizable.
2-45) a)
If , then or
If , then 1 . As a result:
1
1 1
2‐89


1
b Ö
As a result:
1 1
Y ( s) G ( s) H (s) ( K p + K d s)
= =
X ( s ) (1 + G ( s ) H ( s )) ((τ s + 1)( s 2 − g / l ) + K p + K d s )
( K p + K d s)
=
(τ s + (τ (− g / l ) + 1) s 2 + K d s − g / l + K p )
3
c) lets choose 10 0.1.
Use the approach in this Chapter’s Section 2‐14:
1. Activate MATLAB
3. Type in
Acsys
2‐90


2‐91


button.
RH =
[ 1/10, kd]
[ eps, kp-10]
[ (-1/10*kp+1+kd*eps)/eps, 0]
[ kp-10, 0]
For the choice of g/l or τ the system will be unstable. The quantity τ g/l must be >1.
Increase τ g/l to 1.1 and repeat the process.
2‐92


d) Use the ACSYS toolbox as in section 2-14 to find the inverse Laplace transform. Then plot
the time response by selecting the parameter values. Or use toolbox 2-6-1.
Use the approach in this Chapter’s Section 2‐14:
1. Activate MATLAB
3. Type in
Acsys

2‐93


5. Enter the characteristic equation in the denominator and press the “Inverse Laplace Transform”
push‐button.
----------------------------------------------------------------
Inverse Laplace Transform
----------------------------------------------------------------
2‐94


G(s) =
[ kd kp ]
[------------------------ ------------------------------]
[ 3 3 ]
[1/10 s + s kd + kp - 10 1/10 s + s kd + kp - 10]
G(s) factored:
[ kd kp ]
[10 -------------------------- 10 --------------------------]
[ 3 3 ]
[ s + 10 s kd + 10 kp - 100 s + 10 s kd + 10 kp - 100]
Inverse Laplace Transform:
g(t) =
matrix([[10*kd*sum(1/(3*_alpha^2+10*kd)*exp(_alpha*t),_alpha=RootOf(_Z^3+10*_Z*kd
+10*kp-
100)),10*kp*sum(1/(3*_alpha^2+10*kd)*exp(_alpha*t),_alpha=RootOf(_Z^3+10*_Z*kd+1
0*kp-100))]])
While MATLAB is having a hard time with this problem, it is easy to see the solution will be unstable
for all values of Kp and Kd. Stability of a linear system is independent of its initial conditions. For
different values of g/l and τ, you may solve the problem similarly – assign all values (including Kp and
Kd) and then find the inverse Laplace transform of the system. Find the time response and apply the
initial conditions.
Lets chose g/l=1 and keep τ=0.1, take Kd=1 and Kp=10.
2‐95


Y (s) G (s) H (s) ( K p + K d s)
= =
X ( s ) (1 + G ( s ) H ( s )) ((τ s + 1)( s 2 − g / l ) + K p + K d s )
(10 + s ) (10 + s )
= =
(0.1s + (0.1(−1) + 1) s + s − 1 + 10) (0.1s + 0.9 s 2 + s + 9)
3 2 3
Using ACSYS:
RH =
[ 1/10, 1]
[ 9/10, 9]
[ 9/5, 0]
[ 9, 0]
Hence the system is stable
----------------------------------------------------------------
Inverse Laplace Transform
----------------------------------------------------------------
G(s) =
s + 10
-------------------------
3 2
1/10 s + 9/10 s + s + 9
2‐96


G factored:
Zero/pole/gain:
10 (s+10)
-----------------
(s+9) (s^2 + 10)
Inverse Laplace Transform:
g(t) = -10989/100000*exp(-
2251801791980457/40564819207303340847894502572032*t)*cos(79057/25000*t)+868757373/25000
0000*exp(-
2251801791980457/40564819207303340847894502572032*t)*sin(79057/25000*t)+10989/100000*ex
p(-9*t)
Use this MATLAB code to plot the time response:
for i=1:1000
t=0.1*i;
tf(i)=‐10989/100000*exp(‐
2251801791980457/40564819207303340847894502572032*t)*cos(79057/25000*t)+868757373/250
000000*exp(‐
2251801791980457/40564819207303340847894502572032*t)*sin(79057/25000*t)+10989/100000*e
xp(‐9*t);
end
figure(3)
plot(1:1000,tf)
2‐97


2‐52) USE MATLAB
syms t
f=5+2*exp(‐2*t)*sin(2*t+pi/4)‐4*exp(‐2*t)*cos(2*t‐pi/2)+3*exp(‐4*t)
F=laplace(f)
cltF=F/(1+F)
f =
5+2*exp(‐2*t)*sin(2*t+1/4*pi)‐4*exp(‐2*t)*sin(2*t)+3*exp(‐4*t)

F =
(8*s^3+44*s^2+112*s+160+8*2^(1/2)*s^2+16*2^(1/2)*s+2^(1/2)*s^3)/s/(s^2+4*s+8)/(s+4)
cltF =
2‐98


(8*s^3+44*s^2+112*s+160+8*2^(1/2)*s^2+16*2^(1/2)*s+2^(1/2)*s^3)/s/(s^2+4*s+8)/(s+4)/(1+(8*s^3
+44*s^2+112*s+160+8*2^(1/2)*s^2+16*2^(1/2)*s+2^(1/2)*s^3)/s/(s^2+4*s+8)/(s+4))
syms s
cltFsimp=simplify(cltF)
Next type the denominator into ACSYS Routh‐Hurwitz program.
char=collect(s^4+16*s^3+68*s^2+144*s+160+8*2^(1/2)*s^2+16*2^(1/2)*s+2^(1/2)*s^3)
char =
160+s^4+(16+2^(1/2))*s^3+(8*2^(1/2)+68)*s^2+(16*2^(1/2)+144)*s
>> eval(char)
ans =
160+s^4+4901665356903357/281474976710656*s^3+2790603031573437/35184372088832*s^2+293
1340519928765/17592186044416*s
>> sym2poly(ans)
ans =
1.0000 17.4142 79.3137 166.6274 160.0000
Hence the Characteristic equation is:
Δ = s 4 + 17.4142 s 3 + 79.3137 s 2 + 166.6274 s + 160
USE ACSYS Routh‐Hurwitz tool as described in previous problems and this Chapter’s section 2‐14.
RH =

[ 1, 5581205465083989*2^(‐46), 160]
[87071/5000, 5862680441794645*2^(‐45), 0]
[427334336632381556219/6127076924293382144, 160, 0]
2‐99


[ 238083438912827127943602680401244833403/1879436288300987963959490983755776000,
0, 0]
[ 160, 0, 0]
The first column is all positive, and the system is STABLE.
For the other section
syms s
G=(s+1)/(s*(s+2)*(s^2+2*s+2))
g=ilaplace(G)
G =
(s+1)/s/(s+2)/(s^2+2*s+2)
g =
1/4‐1/2*exp(‐t)*cos(t)+1/4*exp(‐2*t)
cltG=G/(1+G)
cltG =
(s+1)/s/(s+2)/(s^2+2*s+2)/(1+(s+1)/s/(s+2)/(s^2+2*s+2))

cltGsimp=simplify(cltG)
cltGsimp =
(s+1)/(s^4+4*s^3+6*s^2+5*s+1)
Next type the denominator into ACSYS Routh‐Hurwitz program.
2‐100


RH =

[ 1, 6, 1]
[ 4, 5, 0]
[ 19/4, 1, 0]
[ 79/19, 0, 0]
[ 1, 0, 0]
STABLE
2‐101


Chapter 3__________________________________________________________________________
3-1) a)
b)
c)
G H
d) Feedback ratio =
e)
3-2)
Characteristic equation: 1 2 1 0
Ö 3 2 1 0
3‐1


3-3)
G 1
1 − G1 H 1
H2
G2
G1
1 − G1 H 1
G1G2 H2
G3 +
1 − G1 H 1 G2
1
1 1
1
3‐2


3-4)
G2
1 + G 2 G3 H 3
+ G2
X G1 G3 Y
1 + G2G3 H 3 + G2 H 2
-
H1
3‐3

ns Golnarraghi, Kuo

+ G1G2 G3
X Y
1 + G 2 G3 H 3 + G 2 H 2
-
H1
G3
3
3-5)
3‐4


3-6) MATLAB
syms s
G=[2/(s*(s+2)),10;5/s,1/(s+1)]
H=[1,0;0,1]
A=eye(2)+G*H
B=inv(A)
Clp=simplify(B*G)
G=
[ 2/s/(s+2), 10]
[ 5/s, 1/(s+1)]
H=
1 0
0 1
A=
[ 1+2/s/(s+2), 10]
[ 5/s, 1+1/(s+1)]
B=
[ s*(s+2)/(s^2-48*s-48), -10/(s^2-48*s-48)*(s+1)*s]
[ -5/(s^2-48*s-48)*(s+1), (s^2+2*s+2)*(s+1)/(s+2)/(s^2-48*s-48)]
Clp =
[ -2*(24+25*s)/(s^2-48*s-48), 10/(s^2-48*s-48)*(s+1)*s]
[ 5/(s^2-48*s-48)*(s+1), -(49*s^2+148*s+98)/(s+2)/(s^2-48*s-48)]
3‐5

ns Golnarraghi, Kuo

3-7)
33-8)
3‐6

ns Golnarraghi, Kuo

3-9)
3
3-10)
3
3-11)
3‐7

ns Golnarraghi, Kuo

3-12)
3-13)
syms t
f=100*(1‐0.33*exp(‐6*t)‐0
0.7*exp(‐10*t))
F=laplace(f)
syms s
F=eval(F)
Gc=F*s
M=30000
syms K
Olp=simplify(KK*Gc/M/s)
Kt=0.15
Clp= simplify(Olp/(1+Olp*KKt))
s=0
Ess=eval(Clp)

f =
100‐30*exp(‐6 6*t)‐70*exp(‐‐10*t)

F =
80*(11*s+75)/s/(s+6)/(s+110)

ans =
3‐8

ns Golnarraghi, Kuo

(880*s+6000)/s/(s+6)/(s+110)

Gc =
(880*s+6000)/(s+6)/(s+100)

M =
30000

Olp =
1/375*K*(11**s+75)/s/(s+66)/(s+10)

Kt =
0.1500

Clp =
2
20/3*K*(11*s s+75)/(2500*s^3+40000*ss^2+150000*ss+11*K*s+75*K)

s
s =
0

E
Ess =
2
20/3
3
3-14)
3‐9


3-15)
‐1 ‐1
Note: If G(s) = g(t), then {e‐asG(s)} = u(t ‐ a) • g(t ‐ a)
syms t s
f=100*(1‐0.3*exp(‐6*(t‐0.5)))
F=laplace(f)*exp(‐0.5*s)
F=eval(F)
Gc=F*s
M=30000
syms K
Olp=simplify(K*Gc/M/s)
Kt=0.15
Clp= simplify(Olp/(1+Olp*Kt))
s=0
Ess=eval(Clp)
digits (2)
Fsimp=simplify(expand(vpa(F)))
Gcsimp=simplify(expand(vpa(Gc)))
Olpsimp=simplify(expand(vpa(Olp)))
Clpsimp=simplify(expand(vpa(Clp)))

f =
100‐30*exp(‐6*t+3)

F =
(100/s‐30*exp(3)/(s+6))*exp(‐1/2*s)

F =
(100/s‐2650113767660283/4398046511104/(s+6))*exp(‐1/2*s)

Gc =
(100/s‐2650113767660283/4398046511104/(s+6))*exp(‐1/2*s)*s

M =
30000

Olp =
‐1/131941395333120000*K*(2210309116549883*s‐2638827906662400)/s/(s+6)*exp(‐1/2*s)

Kt =
0.1500

Clp =
3‐10

ns Golnarraghi, Kuo

2
20/3*K*(2210 03091165498 883*s‐263882 279066624000)*exp(‐1/2*s)/(‐87960930 022208000000*s^2‐
5
52776558133 324800000*s+ +2210309116 6549883*K*eexp(‐1/2*s)*s‐‐2638827906 6662400*K*exxp(‐1/2*s))

s =
0

Ess =
2
20/3

Fsimp =
‐.10e3*exp(‐.550*s)*(5.*s‐6
6.)/s/(s+6.)

Gcsimp =
‐.10e3*exp(‐.550*s)*(5.*s‐6
6.)/(s+6.)

Olpsimp =
‐.10e‐2*K*exp p(‐.50*s)*(17
7.*s‐20.)/s/(s++6.)

Clpsimp =
5
5.*K*exp(‐.50
0*s)*(15.*s‐117.)/(‐.44e4*ss^2‐.26e5*s+1
11.*K*exp(‐.550*s)*s‐13.*K
K*exp(‐.50*s)))
3-16)
3‐11


3-17)
3-18)
1
5 6 3 0.5
1
0.5
1
0.5 1.5 0.5 0.5 0.5
0.5 0.5
0.5 0.5
0.5
u1 0.5 z1
1
3
0.5
1/s -1 1/s 1/s x1 0.5
-6
0.5 0.5 x3 x2 0.5
1 -5
1.5 1
u2 0.5 0.5
-0.5 z2
3‐12


3-19)
Ö
Ö
B1
B0 1/s 1 1/s
u y
x
-A1
-A0
3-20)
3‐13

ns Golnarraghi, Kuo

3
3-21)
3
3-22)
3‐14

ns Golnarraghi, Kuo

33-23)
3‐15

ns Golnarraghi, Kuo

3-24)
3
3-25)
3
3-26)
3
3-27)
3‐16

ns Golnarraghi, Kuo

33-28)
3‐17

ns Golnarraghi, Kuo

33-29)
3‐18

ns Golnarraghi, Kuo

3-30) Use Mason’ss formula:
3-31) MATLAB
syms s K
G=100//(s+1)/(s+5)
g=ilaplace(G/s)
H=K/s
YN=simmplify(G/(1+G*H))
Yn=ilapplace(YN/s)

G =
100/(ss+1)/(s+5)

g =
‐25*exxp(‐t)+5*expp(‐5*t)+20

H =
K/s
3‐19


YN =

100*s/(s^3+6*s^2+5*s+100*K)

Apply Routh‐Hurwitz within Symbolic tool of ACSYS (see chapter 3)

RH =

[ 1, 5]
[ 6, 100*k]
[ ‐50/3*k+5, 0]
[ 100*k, 0]
Stability requires: 0<k<3/10.

3‐20

ns Golnarraghi, Kuo

3‐32)

3‐21

ns Golnarraghi, Kuo

3‐33) M
MATLAB solu
utions are in
n 3‐34.

3‐34) M
MATLAB
clear all
p = [1
1 5 6 10] % Define polynomial
p s^3+5*s^2
2+6*s+10=0
roots(
(p)
G=tf(1
1,p)
step(G
G)

3‐22


p =
1 5 6 10

ans =
‐4.1337
‐0.4331 + 1.4938i
‐0.4331 ‐ 1.4938i
Transfer function:
1
‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐
s^3 + 5 s^2 + 6 s + 10

3‐23


Alternatively:
clear all
syms s
G=1/( s^3 + 5*s^2 + 6*s + 10)
y=ilaplace(G/s)
s=0
yfv=eval(G)
G =
1/(s^3+5*s^2+6*s+10)
y =
1/10+1/5660*sum((39*_alpha^2-91+160*_alpha)*exp(_alpha*t),_alpha =
RootOf(_Z^3+5*_Z^2+6*_Z+10))
s =
0
yfv =
0.1000

Problem finding the inverse Laplace.
Use Toolbox 2‐5‐1 to find the partial fractions to better find inverse Laplace
clear all
B=[1]
A = [1 5 6 10 0] % Define polynomial s*(s^3+5*s^2+6*s+10)=0
[r,p,k]=residue(B,A)
B =
1
A =
1 5 6 10 0
r =
-0.0152
-0.0424 + 0.0333i
-0.0424 - 0.0333i
0.1000
p =
-4.1337
-0.4331 + 1.4938i
-0.4331 - 1.4938i
0
k =
[]
1 −0.0152 −0.0424 + 0.0333i −0.0424 - 0.0333i
So partial fraction of Y is: + + +
s s − 4.1337 s − 0.4331 + 1.4938i s − 0.4331 - 1.4938i
3‐24

ns Golnarraghi, Kuo

3‐35) M
MATLAB solu
utions are in
n 3‐36.

3‐25


3‐36)
clear all
p = [1 4 3 5 1] % Define polynomial s^4+4*s^3+3*s^2+5*s+1=0
roots(p)
G=tf(1,p)
step(G)
p =
1 4 3 5 1
ans =
-3.5286
-0.1251 + 1.1250i
-0.1251 - 1.1250i
-0.2212
Transfer function:
1
-----------------------------
s^4 + 4 s^3 + 3 s^2 + 5 s + 1
3‐26


Alternatively:
clear all
syms s t
G=1/(s^4+4*s^3+3*s^2+5*s+1)
y=ilaplace(G/s)
s=0
yfv=eval(G)
G =
1/(s^4+4*s^3+3*s^2+5*s+1)
y =
1-
1/14863*sum((3955*_alpha^3+16873+14656*_alpha^2+7281*_alpha)*exp(_alpha*t),_alp
ha = RootOf(_Z^4+4*_Z^3+3*_Z^2+5*_Z+1))
s =
0
yfv =
1
Problem finding the inverse Laplace.
Use Toolbox 2‐5‐1 to find the partial fractions to better find inverse Laplace
clear all
B=[1]
A = [1 4 3 5 1] % Define polynomial s^4+4*s^3+3*s^2+5*s+1=0
B =
1
A =
1 4 3 5 1
r =
-0.0235
-0.1068 + 0.0255i
-0.1068 - 0.0255i
0.2372
p =
-3.5286
-0.1251 + 1.1250i
-0.1251 - 1.1250i
-0.2212
k =
[]
3‐27

ns Golnarraghi, Kuo

3-37)
3-38) MATLAB
M
Use TFFcal in ACSY
YS (go to ACSYS foldeer and type in
n TFcal in th
he MATLAB
B Command
d
Windoow).
TFcal
3‐28


Alternatively use toolboxes 3-3-1 and 3-3-2

clear all
syms s
G1=100
G2=(s+1)/(s+2)
G3=10/s/(s+20)
G4=(101*s^3+2122*s^2+2040*s)/10/(s+1)
H1=1
H2=1
simplify(G1*G2*G3/(1+G1*G2*H1+G1*G2*H2+G1*G2*G3))
G1 =
100
G2 =
(s+1)/(s+2)
G3 =
10/s/(s+20)
G4 =
(101/10*s^3+1061/5*s^2+204*s)/(s+1)
H1 =
1
H2 =
1
ans =
1000*(s+1)/(201*s^3+4222*s^2+5040*s+1000)
clear all
TF=tf([1000 1000],[201 4222 5040 1000])
step(TF)
3‐29


3-39)
clear all
syms s
P1 = 2*s^6+9*s^5+15*s^4+25*s^3+25*s^2+14*s+6 % Define polynomial
P2 = s^6+8*s^5+23*s^4+36*s^3+38*s^2+28*s+16 % Define polynomial
solve(P1, s)
solve(P2, s)
collect(P2-P1)
collect(P2+P1)
collect((P1-P2)*P1)
P1 =
2*s^6+9*s^5+15*s^4+25*s^3+25*s^2+14*s+6
P2 =
s^6+8*s^5+23*s^4+36*s^3+38*s^2+28*s+16
ans =
3‐30


-1
-3
i*2^(1/2)
-i*2^(1/2)
-1/4+1/4*i*7^(1/2)
-1/4-1/4*i*7^(1/2)
ans =
-2
-4
i
-i
-1+i
-1-i
ans =
-s^6-s^5+8*s^4+11*s^3+13*s^2+14*s+10
ans =
3*s^6+17*s^5+38*s^4+61*s^3+63*s^2+42*s+22
ans =
-60+2*s^12+11*s^11+8*s^10-54*s^9-195*s^8-471*s^7-796*s^6-1006*s^5-1027*s^4-
848*s^3-524*s^2-224*s
Alternative:
clear all
P1 = [2 9 15 25 25 14 6] % Define polynomial
roots(P1)
P2 = [1 8 23 36 38 28 16] % Define polynomial
roots(P2)
P1 =
2 9 15 25 25 14 6
ans =
-3.0000
-0.0000 + 1.4142i
-0.0000 - 1.4142i
-1.0000
-0.2500 + 0.6614i
-0.2500 - 0.6614i
3‐31


P2 =
1 8 23 36 38 28 16
ans =
-4.0000
-2.0000
-1.0000 + 1.0000i
-1.0000 - 1.0000i
0.0000 + 1.0000i
0.0000 - 1.0000i
3-40)
clear all
syms s
P6 = (s+1)*(s^2+2)*(s+3)*(2*s^2+s+1) % Define polynomial
P7 = (s^2+1)*(s+2)*(s+4)*(s^2+s+1) % Define polynomial
digits(2)
vpa(solve(P6, s))
vpa(solve(P7, s))
collect(P6)
collect(P7)
P6 =
(s+3)*(s+1)*(2*s^2+s+1)*(s^2+2)
P7 =
(s^2+1)*(s+2)*(s+4)*(s^2+s+1)
ans = -1.
-3.
1.4*i
-1.4*i
-.25+.65*i
-.25-.65*i
ans = -2.
-4.
i
-1.*i
-.50+.85*i
-.50-.85*i
ans =
2*s^6+9*s^5+15*s^4+25*s^3+25*s^2+14*s+6
ans =
8+s^6+7*s^5+16*s^4+21*s^3+23*s^2+14*s
3‐32


3‐41)
Use Toolbox 2‐5‐1 to find the partial fractions
clear all
B= conv(conv(conv([1 1],[1 0 2]),[1 4]),[1 10])
A= conv(conv(conv([1 0],[1 2]),[1 2 5]),[2 1 4])
B =
1 15 56 70 108 80
A =
2 9 26 45 46 40 0
r =
-1.0600 - 1.7467i
-1.0600 + 1.7467i
0.9600
-0.1700 + 0.7262i
-0.1700 - 0.7262i
2.0000
p =
-1.0000 + 2.0000i
-1.0000 - 2.0000i
-2.0000
-0.2500 + 1.3919i
-0.2500 - 1.3919i
0
k =
[]
3‐33


3‐42) Use toolbox 3‐3‐2

clear all
B= conv(conv(conv([1 1],[1 0 2]),[1 4]),[1 10])
A= conv(conv(conv([1 0],[1 2]),[1 2 5]),[2 1 4])
G1=tf(B,A)
YR1=G1/(1+G1)
pole(YR1)
B =
1 15 56 70 108 80
A =
2 9 26 45 46 40 0
Transfer function:
s^5 + 15 s^4 + 56 s^3 + 70 s^2 + 108 s + 80
‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐
2 s^6 + 9 s^5 + 26 s^4 + 45 s^3 + 46 s^2 + 40 s
Transfer function:
2 s^11 + 39 s^10 + 273 s^9 + 1079 s^8 + 3023 s^7 + 6202 s^6 + 9854 s^5 + 12400 s^4
+ 11368 s^3 + 8000 s^2 + 3200 s
‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐
4 s^12 + 38 s^11 + 224 s^10 + 921 s^9 + 2749 s^8 + 6351 s^7 + 11339 s^6 + 16074 s^5
+ 18116 s^4 + 15048 s^3 + 9600 s^2 + 3200 s
ans =
0
‐0.7852 + 3.2346i
‐0.7852 ‐ 3.2346i
‐2.5822
‐1.0000 + 2.0000i
‐1.0000 ‐ 2.0000i
‐2.0000
‐0.0340 + 1.3390i
‐0.0340 ‐ 1.3390i
‐0.2500 + 1.3919i
‐0.2500 ‐ 1.3919i
‐0.7794
C= [1 12 47 60]
D= [4 28 83 135 126 62 12]
G2=tf(D,C)
YR2=G2/(1+G2)
pole(YR2)
C =
1 12 47 60
3‐34


D =
4 28 83 135 126 62 12
Transfer function:
4 s^6 + 28 s^5 + 83 s^4 + 135 s^3 + 126 s^2 + 62 s + 12
-------------------------------------------------------
s^3 + 12 s^2 + 47 s + 60
Transfer function:
4 s^9+76 s^8+607 s^7+2687 s^6+7327 s^5+12899 s^4+14778 s^3+10618 s^2

+ 4284 s + 720
-------------------------------------------------------------------------------
4 s^9+76 s^8+607 s^7+2688 s^6+7351 s^5+13137 s^4+16026 s^3+14267 s^2+9924
s+4320
ans =
-5.0000
-4.0000
0.0716 + 0.9974i
0.0716 - 0.9974i
-1.4265 + 1.3355i
-1.4265 - 1.3355i
-3.0000
-2.1451 + 0.3366i
-2.1451 - 0.3366i
3-43) Use Toolbox 3-3-1

G3=G1+G2
G4=G1-G2
G5=G4/G3
G6=G4/(G1*G2)
G3=G1+G2
G4=G1-G2
G5=G4/G3
G6=G4/(G1*G2)
Transfer function:
8 s^12 + 92 s^11 + 522 s^10 + 1925 s^9 + 5070 s^8 + 9978 s^7 + 15154 s^6 +
18427 s^5
+ 18778 s^4 + 16458 s^3 + 13268 s^2 + 10720 s +

4800
3‐35


-------------------------------------------------------------------------------
2 s^9 + 33 s^8 + 228 s^7 + 900 s^6 + 2348 s^5 + 4267 s^4 + 5342 s^3 + 4640 s^2
+
2400 s
Transfer function:
-8 s^12 - 92 s^11 - 522 s^10 - 1925 s^9 - 5068 s^8 - 9924 s^7 - 14588 s^6
- 15413 s^5 - 9818 s^4 - 406 s^3 + 7204 s^2 + 9760 s +

4800
-------------------------------------------------------------------------------
2 s^9 + 33 s^8 + 228 s^7 + 900 s^6 + 2348 s^5 + 4267 s^4 + 5342 s^3 + 4640 s^2
+
2400 s
Transfer function:
-16 s^21 - 448 s^20 - 5904 s^19 - 49252 s^18 - 294261 s^17 - 1.346e006 s^16
- 4.906e006 s^15 - 1.461e007 s^14 - 3.613e007 s^13 - 7.482e007 s^12
- 1.3e008 s^11 - 1.883e008 s^10 - 2.234e008 s^9 - 2.078e008 s^8 -

1.339e008 s^7
- 2.674e007 s^6 + 6.595e007 s^5 + 1.051e008 s^4 + 8.822e007 s^3 +

4.57e007 s^2
+
1.152e007 s
-------------------------------------------------------------------------------
16 s^21 + 448 s^20 + 5904 s^19 + 49252 s^18 + 294265 s^17 + 1.346e006 s^16
+ 4.909e006 s^15 + 1.465e007 s^14 + 3.643e007 s^13 + 7.648e007 s^12
+ 1.369e008 s^11 + 2.105e008 s^10 + 2.803e008 s^9 + 3.26e008 s^8 +

3.343e008 s^7
+ 3.054e008 s^6 + 2.493e008 s^5 + 1.788e008 s^4 + 1.072e008 s^3 +

4.8e007 s^2
3‐36


+
1.152e007 s
Transfer function:
-16 s^21 - 448 s^20 - 5904 s^19 - 49252 s^18 - 294261 s^17 - 1.346e006 s^16
- 4.906e006 s^15 - 1.461e007 s^14 - 3.613e007 s^13 - 7.482e007 s^12
- 1.3e008 s^11 - 1.883e008 s^10 - 2.234e008 s^9 - 2.078e008 s^8 -

1.339e008 s^7
- 2.674e007 s^6 + 6.595e007 s^5 + 1.051e008 s^4 + 8.822e007 s^3 +

4.57e007 s^2
+
1.152e007 s
-------------------------------------------------------------------------------
8 s^20 + 308 s^19 + 5270 s^18 + 54111 s^17 + 379254 s^16 + 1.955e006 s^15
+ 7.778e006 s^14 + 2.471e007 s^13 + 6.416e007 s^12 + 1.383e008 s^11
+ 2.504e008 s^10 + 3.822e008 s^9 + 4.919e008 s^8 + 5.305e008 s^7 +

4.73e008 s^6
+ 3.404e008 s^5 + 1.899e008 s^4 + 7.643e007 s^3 + 1.947e007 s^2 +

2.304e006 s
3‐37

Chapter 4__________________________________________________________________________
4-1) When the mass is added to spring, then the spring will stretch from position O to position L.
The total potential energy is:
1
2
where y is a displacement from equilibrium position L.
The gravitational energy is:
The kinetic energy of the mass-spring system is calculated by:
1
2
As we know that , then
By differentiating from above equation, we have:
0
since KL = mg, therefore:
0
As y cannot be zero because of vibration, then 0
4‐1

4-2)
KU
KL
4-3) a) Rotational kinetic energy:
Translational kinetic energy:
Relation between translational displacement and rotational displacement:
1
2
Potential energy:
As we know , then:
4‐2

1 1 1
2 2 2
By differentiating, we have:
0
Since cannot be zero, then 0
b)
c)
1 1 1
2 2 2
where at the maximum energy.
1 1
2 2
Then:
1 1
2 2
Or:
4‐3

J
d) G (s) =
(ms + K )
2
% select values of m, J and K

%Step input
K=10;
J=10;
M=1;
G=tf([10],[1 0 10])
step(G,10)
xlabel( 'Time(sec)');
ylabel('Amplitude');
Transfer function:
10
‐‐‐‐‐‐‐‐
s^2 + 10

4‐4

Automatic Conttrol Systems, 9th Edition Chapteer 4 Solutionss Golnarraghi, Kuo
4-44)
(a) Force equaations:
⎛ dy1 − dy2 ⎞ + K y − y
2
d y1 dy1
f (t ) = M 1 + B1 + B3 ⎜ ⎟ ( 1 2)
dt
2
dt ⎝ dt dt ⎠

⎛ d 2⎞
2
dy1 dy d y2 dy2
B3 ⎜ − ⎟ + K ( y1 − y2 ) + M 2 2 + B2
⎝ dt ddt ⎠ dt dt
Rearraange the equattions as follows:
2
d y1 (B + B3 ) dy1 B3 dyy 2 K f
2
=− 1
+ − (y 1
− y2 ) +
dt M1 dt M 1 dt
d M1 M1

2
d y2 B3 dy1 (B + B3 ) dy2 K
2
= − 2
+ (y1
− y2 )
dt M 2 dt M2 dtt M2
(i) Staate diagram: SSince y1 − y 2 appears as one
a unit, the minimum number of integrators is three.
4‐5

dy 2 dy1
State equations: Define the state variables as x1 = y1 − y 2 , x 2 = , x3 = .
dt dt
dx1 dx2 K (B + B3 ) B3 dx3 K B3 (B + B3 ) 1

= − x2 + x3 = x1 − 2
x2 + x3 =− x1 + x2 − 1
x3 + f
dt dt M2 M2 M2 dt M1 M1 M1 M
dy 2 dy1
(ii) State variables: x1 = y 2 , x 2 = , x3 = y1 , x 4 = .
dt dt
State equations:
dx1 dx 2 K B2 + B3 K B3
= x2 =− x1 − x2 + x3 + x4
dt dt M2 M2 M2 M2
dx3 dx 4 K B3 K B1 + B3 1
= x4 = x1 + x2 − x3 − x4 + f
dt dt M1 M1 M1 M1 M1
State diagram:
Transfer functions:
M 2 s + ( B2 + B3 ) s + K
2
Y1 ( s )
=
F (s) {
s M 1 M 2 s + [( B1 + B3 ) M 2 + ( B2 + B3 ) M 1 ] s + [ K ( M 1 + M 2 ) + B1 B2 + B2 B3 + B1 B3 ] s + ( B1 + B2 ) K
3 2
}

Y2 ( s ) B3 s + K
=
F (s) {
s M 1 M 2 s + [ ( B1 + B3 ) M 2 + ( B2 + B3 ) M 1 ] s + [ K ( M 1 + M 2 ) + B1 B2 + B2 B3 + B1 B3 ] s + ( B1 + B2 ) K
3 2
}
(b) Force equations:
4‐6

2
d y1 (B + B2 ) dy1 B2 dy 2 1 dy2 dy1 K
2
=− 1
+ + f = − y2
dt M dt M dt M dt dt B2
(i) State diagram:

Define the outputs
of the integrators
as state variables,
dy
x1 = y 2 , x 2 = 1
dt
.

State equations:
dx1 K dx 2 K B1 1
=− x1 + x 2 =− x1 − x2 + f
dt B2 dt M M M
dy1
(ii) State equations: State variables: x1 = y 2 , x 2 = y1 , x3 = .
dt
dx1 K dx 2 dx3 K B1 1
=− x1 + x3 = x3 =− x1 − x3 + f
dt B2 dt dt M M M

Transfer functions:
Y1 ( s) B2 s + K Y2 ( s ) B2
= =
F ( s ) s ⎡⎣ MB2 s + ( B1 B2 + KM ) s + ( B1 + B2 ) K ⎤⎦
2
F (s) MB2 s + ( B1 B2 + KM ) s + ( B1 + B2 ) K
2
(c) Force equations:
dy1 dy2 1
2
d y2 (B + B2 ) dy2 B1 dy2 B1 dy1 K
= + f 2
=− 1
+ + − y2
dt dt B1 dt M dt M dt M dt M
(i) State diagram:
4‐7


State equatio
ons: Define thee outputs of integrators as sttate variables.
dx1 dxx 2 K B2 1
= x2 =− x1 − x2 + f
dt d
dt M M M
dy 2
(ii) State equations: state variaables: x1 = y1 , x 2 = y 2 , x3 = .
dt
dx1 1 dx 2 dxx3 K B2 1
= x3 + f = x3 =− x2 − x3 + f
dt B1 dt d
dt M M M
State diagram
m:
Transfer funcctions:
Ms + ( B1 + B2 ) s + K
2
Y1 ( s ) Y2 ( s ) 1
= =
F (s) (
B1 s Ms + B2 s + K
2
) F (s) Ms + B2 s + K
2
4-55)
4‐8

(a) Force equaations:
2
1 d y2 B dy 2 K1 + K 2 K2
y1 = ( f + Mg ) + y 2 2
=− − y2 + y1
K2 dt M dt M M
Staate diagram:

Staate equations:
Deffine the state
varriables as:
dy
x1 = y 2 , x 2 = 2
d
dt
.

dx1 d 2
dx K1 B 1
= x2 =− x1 − x2 + ( f + Mg )
dtt dt M M M
Transfer function
ns:
2
Y1 ( s ) s + Bs + K1 + K 2 Y2 ( s ) 1
= 2
= 2

F ( s) K 2 ( Ms + Bs + K1 ) F ( s) Mss + Bs + K1
(b) Forcce equations:
⎛ dy1 − dy2 ⎞ + K1 y − y − B2 y − y − B2 dy2

2
dy1 1 dy 2 K1 d y2 B1
= [ f (t ) + Mg ] + − (y − y2 ) = ⎜ ⎟ ( 1 2) ( 1 2)
M ⎝ dt dt ⎠ M
1 2
dt B1 dt B1 dt M M dt
Statte diagram: (W
With minimum number of inttegrators)
4‐9

To obtain the transsfer functions Y1 ( s ) / F ( s ) and Y2 ( s ) / F ( s ), we need to redefine the sttate variables aas:
x1 = y 2 , x 2 = dy 2 / dtt , and x3 = y1 .
Statte diagram:
Ms + ( B1 + B2 ) s + K1 Bs + K1
2
Y1 ( s ) Y2 ( s )
= =
F (s) s
2
[ MB
M s + (B B
1 1 2
+ MK1 )] F (s) s
2
[ MB
M s + (B B
1 1 2
+ MK1 )]
4-66) a)
y1
y2
K ( y1 − y 2 )
m K ( y1 − y 2 ) M F
μmgy& 2 μMgy&1
b) From Newtton’s Law:
If y1 and y2 are considerred as a posiition and v1 and

a v2 as vellocity variabbles
Then:
The output equation cann be the veloocity of the engine,

e whicch means
4‐10

c)
Obtaining requires solving above equation with respect to Y2(s)
From the first equation:
Substituting into the second equation:
By solving above equation:
1
2
d)
0 0 1 0
0 0 0 1 0
0
0 1
0 0
0001 0
4-7) (a) Force equations:
⎛ dy1 − dy2 ⎞ ⎛ dy1 − dy2 ⎞ = M d y2 + B dy2

2
f (t ) = K h ( y1 − y2 ) + Bh ⎜ ⎟ K h ( y1 − y 2 ) + Bh ⎜ ⎟
⎝ dt dt ⎠ ⎝ dt dt ⎠ 2 t
dt dt
dy 2
(b) State variables: x1 = y1 − y 2 , x 2 =
dt
4‐11

State equations:
dx1 Kh 1 dx 2 Bt 1
=− x1 + f (t ) =− x2 + f (t )
dt Bh Bh dt M M
4-8)
α2
α1
For the left pendulum:
sin
sin sin cos sin sin cos
Ö 0
Ö sin sin sin cos
For the right pendulum, we can write the same equation:
sin sin sin cos
since the angles are small:
sin
cos 1 0
Ö Ö
sin 0
cos 1
4-9) a)
4‐12

Fy
θ m
x(t) Fx
mg
M
f
θ
b) If we consider the coordinate of centre of gravity of mass m as (xg, yg),
Then sin and cos
From force balance, we have:
Ö cos sin
From the torque balance, we have:
cos sin sin
Where:
cos
cos sin
Substituting these equation:
cos sin
4-10) a)
4‐13

Fy1 Fy 2
Fx1 Fx 2
m1 g
m2 g
θ1 θ2
b) Kinetic energy
(i) For lower pendulum:
1 1
sin cos
2 2
For upper pendulum:
1 1
sin cos
2 2
For the cart:
(ii) Potential energy:
For lower pendulum: cos
For upper pendulum: cos
For the cart: 0
(iii) Total kinetic energy:
Total potential energy:
4‐14

The Lagrangian equation of motion is:
Substituting T and U into the Lagrangian equation of motion gives:
cos cos sin sin

cos sin
cos sin
4-11) a) From the Lagrangian equation of motion:
sin 0
b) As:
Then
sin 0
If we linearize the equation about beam angle α = 0, then sinα ≈ α and sinθ ≈ θ
Then:
4‐15

c) Considering
Then the state-space equation is described as:
0
0 1
0 0
mgd
d) G ( s) =
( s L( J / r 2 + m))
2
clear all
% select values of m, d, r, and J
%Step input
g=10;
J=10;
M=1;
D=0.5;
R=1;
L=5;
G=tf([M*g*D],[L*(J/R^2+M) 0 0])
step(G,10)
Transfer function:
5
------
55 s^2
4‐16

4-12) If the aircraft is at a constant altitude and velocity, and also the change in pitch angle does not
change the speed, then from longitudinal equation, the motion in vertical plane can be written as:
sin
cos
Where u is axial velocity, ω is vertical velocity, q is pitch rate, and θ is pitch angle.
Converting the Cartesian components with polar inertial components and replace x, y, z by T, D,
and L. Then we have:
1
cos sin
1
sin cos
Where α = θ – γ is an attack angle, V is velocity, and γ is flight path angle.
4‐17

It should be mentioned that T, D, L and M are function of variables α and V.
Refer to the aircraft dynamics textbooks, the state equations can be written as:
b) The Laplace transform of the system is:
By using Laplace transform, we have:
1
2
3
From equation (1):
Substituting in equation (2) and solving for q(s):
Substituting above expression in equation (3) gives:
If we consider sin , then
By using Laplace transform:
Which gives:
4‐18

For plotting
g state flow diagram, equation
u (4) muust be rewritten as:
or
S the state flow diagram

So, m will plotteed as:
ml 1/Ms 1 1/s
u y
x
-B/M
-K
Also look at seection 4-11
4-113) (a) Torque equation: State d
diagram:
2
d θ B dθ 1
2
=− + T (t )
dt J dt J
State equations:
dx1 dx 2 B 1
= x2 =− x2 + T
dt dt J J
Transsfer function:
Θ( s ) 1
=
T ( s) s( Js + B )
(b) Torque equations:
d θ1 dθ 2
2
K 1
2
=− (θ 1
− θ2 ) + T K (θ 1 − θ 2 ) = B
dt J J dt
State diagram: (minimum numbeer of integratorrs)
4‐19

State equations:
dx1 K dx 2 K 1
=− x1 + x 2 =− x1 + T
dt B dt J J
dθ 1
State equations: Let x1 = θ 2 , x 2 = θ 1 , and x3 = .
dt
dx1 K K dx 2 dx3 K K 1
=− x1 + x2 = x3 = x1 − x2 + T
dt B B dt dt J J J
State diagram:
Tran
nsfer functionss:
Θ1 ( s ) B +K
Bs Θ2 ( s) K
= =
T (s) (
s BJs + JKs + BK
2
) T (s) (
s BJs + JK
2
Ks + BK )
(c) Torqque equationss:
d θ1 d θ2
2 2
T (t ) = J 1 2
+ K (θ 1 − θ 2 ) K (θ 1 − θ 2 ) = J 2 2
dt dt
Statte diagram:
4‐20

dθ 2 dθ 1
e equations: sttate variables:: x1 = θ 2 , x 2 =
State , x3 = θ 1 , x 4 = .
dt d
dt
dx1 dx 2 K K d 3
dx d 4
dx K K 1
= x2 =− x1 + x3 = x4 = x1 − x3 + T
dt dt J2 J2 dt dt J1 J1 J1
nsfer functionss:
Tran
Θ1 ( s ) J2s + K Θ2 ( s)
2
K
= =
s ⎡⎣ J 1 J 2 s + K ( J 1 + J 2 ) ⎤⎦ s ⎡⎣ J 1 J 2 s + K ( J 1 + J 2 ) ⎤⎦
2 2 2 2
T (s) T (s)
(d) Torque equationss:
d θm d θ1 d θ2
2 2 2
T (t ) = J m 2
+ K1 (θ m − θ 1 ) + K 2 (θ m − θ 2 ) K1 (θ m − θ 1 ) = J 1 2
K 2 (θ m − θ 2 ) = J 2 2

dt dt dt

4‐21

State diiagram:

dθ 1 dθ m dθ 2
ons: x1 = θ m − θ 1 , x 2 =
State equatio , x3 = , x 4 = θ m − θ 2 , x5 = .
dtt dt dt
dx1 dx 2 K1 dx3 K1 K2 1 dx 4 dx5 K2

= − x 2 + x3 = x1 =− x1 − x4 + T = x3 − x5 = x4
dt dt J1 dt Jm Jm Jm dt dtt J2
Θ1 ( s ) K1 ( J 2 s + K 2 )
2
=
s ⎣⎡ s + ( K1 J 2 J m + K 2 J 1 J m + K1 J 1 J 2 + K 2 J 1 J 2 ) s + K1 K 2 ( J m + J 1 + J 2 ) ⎤⎦
2 4 2
T (s)

Θ2 (s) K 2 ( J 1 s + K1 )
2
=
s ⎡⎣ s + ( K1 J 2 J m + K 2 J 1 J m + K1 J 1 J 2 + K 2 J 1 J 2 ) s + K1 K 2 ( J m + J 1 + J 2 ) ⎤⎦
2 4 2
T (s)
(e) Torqque equationss:
d 2θ m K1 K2 1 d θ1
2
K1 B1 dθ 1 d θ2
2
K2 B2 dθ 2

2
=− (θ m − θ1 ) − (θ m − θ 2 ) + T = (θ m
− θ1 ) − 2
= (θ m
− θ1 ) −
dt Jm Jm Jm d
dt
2
J1 J 1 dt dt J2 J 2 dt
State diagram
m:
4‐22

dθ 1 dθ m dθ 2
SState variabless: x1 = θ m − θ 1 , x 2 = , x3 = , x 4 = θ m − θ 2 , x5 = .
dt dt dt
State equatio
ons:
dx1 dx 2 K1 B dx3 K1 K2 1 dx4 dx5 K2 B2
= − x 2 + x3 = x1 − 1 x 2 =− x1 − x4 + T = x3 − x 5 = x4 − x5
dt dt J1 J1 dt Jm Jm Jm dt dt J2 J2

Θ1 ( s )
=
(
K1 J 2 s + B2 s + K 2
2
) Θ2 (s)
=
(
K 2 J 1 s + B1 s + K1
2
)
T (s) Δ(s) T (s) Δ(s)
Δ ( s ) = s { J 1 J 2 J m s + J m ( B1 + B2 ) s + [( K1 J 2 + K 2 J 1 ) J m + ( K1 + K 2 ) J 1 J 2 + B1 B2 J m ] s
2 4 3 2

+ [( B1 K 2 + B2 K1 ) J m + B1 K 2 J 2 + B2 K1 J 1 ] s + K1 K 2 ( J m + J 1 + J 2 )}
4-114)
d θ1 d θ3
2 2
N1 N3 N1
Tm (t ) = J m 2
+ T1 T1 = T2 T3 = T4 T4 = J L 2
T2 = T3 θ2 = θ1
dt N2 N4 dt N2
(a))
N1 N 3 N3 N3 d θ3
2
d θ1
2
⎡
N1 N 3 ⎡ N1 N 3 ⎤ ⎤ d 2θ 1
2
θ3 = θ1 T2 = T4 = JL Tm = J m + T4
= ⎢ m ⎢
J + ⎥ JL ⎥ 2
⎢⎣ ⎣ N 2 N 4 ⎦ ⎥⎦ dt
2 2
N2 N4 N4 N4 dt dt N2 N4

4‐23

(b)
d θ1 d θ2 d θ3
2 2 2
N1 N3
Tm = J m 2
+ T1 T2 = J 2 2
+ T3 T4 = ( J 3 + J L ) 2
T1 = T2 T3 = T4
dt dt dt N2 N4

d θ2 d θ2 d θ3
2 2 2
N1 N1 N 3 N3 N3
θ2 = θ1 θ3 = θ1 T2 = J 2 2
+ T4 = J 2 2
+ (J 3
+ JL ) 2
N2 N2 N4 dt N4 dt N4 dt
d θ1
2
N1 ⎛ ⎡ ⎛ N1 ⎞
d θ2
2
2
⎛ N1 N 3 ⎞
N3
2
⎤ d 2θ1 d θ3 ⎞
2
Tm (t ) = J m 2 + ⎜ J 2 2 + ( J 3 + J 4 ) 2 ⎟ = ⎢ J m + ⎜ ⎟ J 2 + ⎜ ⎟ ( J 3 + J L ) ⎥ dt 2
dt N2 ⎝ dt N4 dt ⎠ ⎣⎢ ⎝ N2 ⎠ ⎝ N2 N4 ⎠ ⎦⎥
4-15) (a)
2 2
d θm d θL N1
Tm = J m 2
+ T1 T2 = J L 2
+ TL T1 = T2 = nT2 θ m N1 = θ L N 2
dt dt N2
d θm d θL ⎛ J m + nJ ⎞ α + nT nTm − n TL
2 2 2
Tm = J m + nJ L + nTL = ⎜ L ⎟
Thus, α L =
⎝ n ⎠
L L
Jm + n JL
2 2 2
dt dt
∂α L J mTL Jm
Set
∂n
= 0. (Tm ( 2
)
− 2 nTL ) J m + n J L − 2 nJ L nTm − n J L = 0 Or, n + ( 2
) 2
n− = 0
J LTm JL

2 2 2
∗ J mTL J mTL + 4 J m J LTm
Optimal gear ratio: n =− + where the + sign has been chosen.
2 J LTm 2 J LTm

(b) When TL = 0 , the optimal gear ratio is
∗
n = Jm / J L
4-16) (a) Torque equation about the motor shaft: Relation between linear and rotational displacements:
2 2
d θm 2 d θm dθ m
Tm = J m 2
+ Mr 2
+ Bm y = rθ m
dt dt dt
(b) Taking the Laplace transform of the equations in part (a), with zero initial conditions, we have
Tm ( s ) = J m + Mr ( 2
)s Θ 2
m
( s ) + Bm sΘ m ( s ) Y (s) = rΘm (s)
4‐24

Transffer function:
Y ( s) r
=
Tm ( s ) (
s ⎡⎣ J m + Mr
r
)s + B m
⎤⎦
4-117) (a)
d θm
2
Tm = J m 2
+ r ( T1 − T2 ) T1 = K 2 ( rθ m − rθ p ) = K 2 ( rθ m − y ) T2 = K1 ( y − rθ m )
dt
d θm
2 2 2
d y d y
T1 − T2 = M 2
us, Tm = J m
Thu 2
+ r ( K1 + K 2 )( rθ m − y ) M 2
= ( K1 + K 2 )( rθ m − y )
dt dtt dt
(c) State equations:
dx1 dx2 K1 + K 2 dx3 − r ( K1 + K 2 ) 1

= rx3 − x2 = x1 = x1 + Tm
dt dt M dt Jm Jm
(d) Transfer function:
Y ( s) r ( K1 + K 2 )
=
s ⎡⎣ J m Ms + ( K1 + K 2 )(( J m + rM ) ⎤⎦
2 2
Tm ( s )
(e) Charaacteristic equaation:
s
2
⎡⎣ J m Ms 2 + ( K1 + K 2 )( J m + rM ) ⎤⎦ = 0
4-118) (a) Systeem equations:
4‐25

dω m dia
Tm = K i ia = ( J m + J L ) + Bmω m ea = Ra ia + La + K bω m y = nθ m y = y ( t − TD )
dt dt
d
TD = (sec) e = r −b b = Ks y E a ( s ) = KGc ( s ) E ( s )
V
Blockk diagram:
(b) Forw
ward‐path transfer function
n:
− TD s
Y (s) KK i nGc ( s )e
=
E (s) s {( Ra + La s ) [( J m + J L ) s + Bm ] + K b K i }
Clossed‐loop transsfer function:
− TD s
Y ( s) KK
K i nGc ( s )e
=
s ( Ra + La s ) [ ( J m + J L ) s + Bm ] + K b K i s + KGc ( s ) K i ne
− TD s
R(s)
4-119) (a) Torquue equations:
d θm dθ m d θL dθ L
2 2
Tm (t ) = J m 2
+ Bm + K (θ m − θ L ) K (θ m − θ L ) = J L 2
+ BL
dt dt dt dt

State diagram:
4‐26

(b) Transsfer functions:
Θ L ( s) Θm (s) J L s + BL s + K
2
K
= = Δ ( s ) = s ⎡⎣ J m J L s + ( Bm J L + BL J m ) s + ( KJ m + KJ L + Bm BL ) s + Bm K ⎤⎦
3 2
Tm ( s ) Δ(s) Tm ( s ) Δ( s)
(c) Charaacteristic equaation: Δ( s ) = 0
Tm
dy‐state perforrmance: Tm ( t ) = Tm = constant.
(d) Stead t Tm ( s ) = .
s
J L s + BL s + K
2
1
lim ω m (t ) = lim sΩ m ( s ) = lim =
J m J L s + ( Bm J L + BL J m ) s + ( KJ
K m + KJ L + Bm BL ) s + Bm K
3 2
t →∞ s →0 s →0 Bm
Thuss, in the steadyy state, ω m = ω L .
(e) Thee steady‐state vvalues of ω m and
a ω L do not depend on J m and J L .
4-220) (a) Torqque equation: (About the ceenter of gravityy C)
2
d θ
J 2
= Ts d 2 sin δ + Fd d1 Fa d1 = J α α 1 = K F d1θ sin δ ≅ δ
dt
2 2
d θ d θ
Thuss, J 2
= Ts d 2δ + K F d1θ J 2
− K F d1θ = Ts d 2δ
dt dt
2
(b) Js Θ ( s ) − K F d1Θ ( s ) = Ts d 2 Δ ( s )
(c) With
h C and P intercchanged, the torque equation about C is:
dθ dθ
2 2
Ts ( d1 + d 2 ) δ + Fα d 2 = J 2
Ts ( d1 + d 2 ) δ + K F d 2θ = J 2

dt dt
4‐27

Θ( s ) Ts ( d1 + d 2 )
Js Θ ( s ) − K F d 2 Θ ( s ) = Ts ( d1 + d 2 ) Δ ( s ) =
2

Δ(s) Js − K F d 2
2
4-21) (a) Nonlinear differential equations:
dx ( t ) dv ( t )
= v(t ) = − k ( v ) − g ( x ) + f ( t ) = − Bv ( t ) + f ( t )
dt dt
e( t ) e( t 0
With Ra = 0 , φ ( t ) = = K f i f ( t ) = K f i f ( t ) = K f ia ( t ) Then, ia ( t ) =
Kb v ( t ) Kb K f v ( t )
2
Ki e ( t ) dv ( t ) Ki 2
f ( t ) = Kiφ ( t )ia ( t ) = 2 2
. Thus, = − Bv ( t ) + 2 2
e (t )
Kb K f v (t ) dt Kb K f v (t )
(b) State equations: ia ( t ) as input.
dx ( t ) dv ( t ) 2
= v (t ) = − Bv ( t ) + Ki K f ia ( t )
dt dt
(c) State equations: φ ( t ) as input.
2 φ (t )
f ( t ) = K i K f ia ( t ) ia ( t ) = i f ( t ) =
Kf

dx ( t ) dv ( t ) Ki 2
= v (t ) = − Bv ( t ) + φ (t )
dt dt Kf
4‐28

4-22) Define as the angle between mass m and the horizontal axis (positive in c..c.w. direction):
θ=ωt ym
Use Newton’s second law:
y + &&
m( && ym ) = − Fm
( M − m) &&
y = Fm − By& − Ky
ym = −eω 2 sin ωt
&&
⇒
My&& + By& + Ky = meω 2 sin ωt
Where M is the Mass of the overall block system. M-m is the mass of the block alone.
Y meω 2
G ( s) = =
R Ms 2 + Bs + K
Zero i.c. and input r (t ) = sin ωt
Note θ = ωt . So in case of a step response as asked in the question, ω is a step input and angle θ increases
with time – i.e. it is a ramp function. Hence, ym is a sinusoidal function, where the Laplace transform of a
ω
sine function is sin(ωt ) = 2
s + ω2
Pick values of the parameters and run MATLAB. See toolbox 5-8-2
clear all
m=20.5 %kg
M=60 %kg
K=100000 %N/m
Om=157 %rad/s
B=60 %N-m/s
e=0.15 %m
G=tf([m*e*Om^2],[M B K])
t=0:0.01:1;
u=1*sin(Om*t);
lsim(G,u,t)
4‐29

m =
20.5000
M =
60
K =
100000
Om =
157
B =
60
e =
0.1500
Transfer function:
7.58e004
----------------------
60 s^2 + 60 s + 100000
4‐30

4-23) a) summation of vertical forces gives:
2 1
0 2
If we consider and , then:
2
0
The state-space model is:
0 0 1 0
0 0 0 1 0
2 0
0 0 1
0 0 0
b)
By applying Laplace transform for equations (1) and (2), we obtain:
Which gives:
and
Therefore:
2 2
4-24) a) Summation of vertical forces gives:
Consider and , then
4‐31

So, the state-space model of the system is:

0 0 1 0
0 0 0 1 0
0
1
b) The Laplace transform of the system is defined by:
where
as a result:
Substituting into above equation:
4-25) a) According to the circuit:
0
2 2
0
2 2
0
2 2
By using Laplace transform we have:
4‐32

V s V s V s V s
CsV s 0
2R 2R
Cs V s Cs
V s V s V s V s 0
2 R 2
Cs V s V s
V s V s 0
2 2R
From above equations:
1
V s V s V s
2 RCS 1
RCS
V s V s V s
2 RCS 1
Substituting V1(s) and V2(s) into preceding equations, we obtain:
1
4 1
b) Measuring Vout requires a load resistor, which means:
i1 i2 RL
Vin VC1 VC 2
Then we have:
When
If RL >>R, then
4‐33

By using Laplace transform we have:
Therefore:
can be obtained by substituting above expressions into the first equation of the state variables
of the system.
4-26) a) The charge q is related to the voltage across the plate:
The force fv produced by electric field is:
2
Since the electric force is opposes the motion of the plates, then the equation of the motion is written
as:
The equations for the electric circuit are:
As we know, and , then:
4‐34

Since , then :
b) As then
If 1
Then the transfer function is:
4‐35

If 1
Then the transfer function is:
4-27) a) The free body diagram is:
where F is required force for holding the core in the equilibrium point against magnetic
field
b) The current of inductor, i, and the force, F, are function of flux, Φ, and displacement, x.
Also, we know that
4‐36

The total magnetic field is:
,
2
where W is a function of electrical and mechanical power exerted to the inductor, so:
∂W Φ
i
∂Φ L x
∂W Φ
F
∂x 2L x
As v = , then:
1
2
c) Changing the flux requires a sinusoidal movement, and then we can conclude that:
if the inductance is changing relatively, then L(x) = Lx, where L is constant.
Also, the current is changing with the rate of changes in displacement. It means:
i Bx
So:
Substituting these equations into the state-space equations gives:
1
2
Therefore:
V s 2
tan ω 1
F s B
4‐37

4-28) a) The free body diagram is:
where F is the external force required for holding the plate in the equilibrium point against the
electrical field.
b) The voltage of capacitors, , and the force, , are function of charge, , and displacement, .
Also, we know
The total electrical force between plates is:
,
2
Where W is a function of electrical and mechanical power exerted to the capacitor, so:
As , then:
1
2
c) The same as Problem 4.28,
Consider:
Then solve the equations.
4‐38

4-29) According to the circuit:

_ _
or
As an op-amp is modeled with the following equation:

_
1
Then:
1
1 1
1
4-30) a) Positive feedback ratio:
b) Negative feedback ratio:
c) According to the circuit:
Therefore:
4‐39

As
10
1
then:
10
1
1
which gives:
1
10 1 10
It is stable when 1 10 0 which means: 10
4-31) a) If the drop voltage of Rin is called v1
Then:
Also:
0
Then:
Substituting this expression into the above equation gives:
1
1 0
As a result:
4‐40

1 1
1
Or
1
b) If the dropped voltage across resistor Rf is called vf, then

0
0
As a result:
Substituting into the second equation gives:
or
As a result:
1
1
4-32) The heat flow-in changes with respect to the electric power as:
where R is the resistor of the heater.
The heat flow-out can be defined as:
where Kf is the heat flow coefficient between actuator and air, T1 and T2 are temperature of
actuator and ambient.
4‐41

Since the temperature changes with the differences in heat flows:
1 1 1
where C is the thermal capacitor.
The displacement of actuator is changing proportionally with the temperature differences:
If we consider the T2 is a constant for using inside a room, then
Therefore:
By linearizing the right hand side of the equation around point

1
2 1
Or
2 1
If we consider the right hand side of the above equation as two inputs to the system as: 2
and 1 or , then:
2
1
4-33) Due to insulation, there is no heat flow through the walls. The heat flow through the sides is:
4‐42

2
, 1
ln
2
, 2
ln
Where T1 and T2 are the temperature at the surface of each cylinder.
As , , , then from equation (1) and (2), we obtain:
ln
, 3
2
The conduction or convection at:
: 4
: 5
: 6
The thermal capacitance dynamics gives:
, 7
Where
According to the equation (7) and (8), T1 and Tf are state variables.
Substituting equation (3), (4), (5) and (6) into equation (7) and (8) gives the model of the system.
4-34) As heat transfer from power supply to enclosure by radiation and conduction, then:
2
1 1 1 1
4‐43

Also the enclosure loses heat to the air through its top. So:
Where
And Ct is the convective heat transfer coefficient and At is the surface area of the enclosure.
The changes if the temperature of heat sink is supposed to be zero, then:
Therefore where , as a result:
According to the equations (1) and (4), Tp and Te are state variables. The state model of the system
is given by substituting equations (2), (3), and (6) into these equations give.
4-35) If the temperature of fluid B and A at the entrance and exit are supposed to be and , and
TAN and TAX, respectively. Then:
1
2
The thermal fluid capacitance gives:
From thermal conductivity:
5
ln
1 1
2
4‐44

Where Ci and Co arre convectivve heat transffer coefficiennt of the inner and outerr tube; Ai andd Ao are
the surfface of innerr and outer tuube; Ri and Ro are the raadius of inneer and outer tube.
t
Substittuting equations (1), (2), and (5) intoo equations (3) and (4) giives the statee model of thhe
system
m.
4-336) (a) Blockk diagram:
(b) Transfer function:
−τ D s
Ω( s ) K1 K 4 e
=
Js + ( JK L + B ) s + K 2 B + K 3 K 4 e
−τ D s
α ( s)
2
(c) Characteristic equattion:
Js + ( JK L + B ) s + K 2 B + K 3 K 4 e
−τ D s
= 0
2

(d) Transfer function:
Ω( s ) K1 K 4 ( 2 − τ D s )
≅
α (s) Δ( s)
Charaacteristic equaation:
Δ ( s ) ≅ J τ D s + ( 2 J + JK 2τ D + Bτ D ) s + ( 2 JK 2 + 2 B − τ D K 2 B − τ D K 3 K 4 ) s + 2 ( K 2 B + K 3 K 4 ) = 0
3 2

4-337) The tottal potential energy is:
1 1
2 2
The tottal kinetic en
nergy is:
4‐45

Therefore:
2
As a result:
2
So, the natural frequency of the system is calculated by:
Also, by assuming sin and substituting into yields the same result
when calculated for maximum displacement.
4-38) If the height of the reservoir, the surge tank and the storage tank are assumed to be H, h1 and h2,
then potential energy of reservoir and storage tank are:
For the pipeline we have:
The surge tank dynamics can be written as:
At the turbine generator, we have:
where I is a known input and Q2-v is the fluid flow transfer between point 2 and valve. The
behaviour of the valve in this system can be written as:
4‐46

| |
Regarding Newton’s Law:
According to above equations, it is concluded that Q and h1 are state variables of the system.
The state equations can be rewritten by substituting P2, Pv, Ps and Q2-v from other equations.
4-39)
If the beam rotate around small angle of α cos 1 , then
where A and E are cross sectional area and elasticity of the cable; H is the distance between point
O and the bottom of well, and y is the displacement.
On the other hand, Newton’s Law gives:
4‐47

Pa Aω
Bv 2
Pb Aω
where B is the viscous friction coefficient, Aw is the cross sectional area of the well; P1 and P2 are
pressures above and below the mass m.
The dynamic for the well can be written as two pipes separating by mass m:
Pb Aω
Ps Aω
Q
Q y
Ff
Pa Aω P1 Aω
a b)
0
8
0
Where D is the distance between point O and ground, Ps is the pressure at the surface and known.
If the diameter of the well is assumed to be r, the Ff for the laminar flow is
32
Therefore:
4‐48

32
32
The state variables of the system are ω, v, y, Q, Q1.
4-40) For the hydraulic amplifier, we have:
As a result
where N is a constant and A is the cross sectional area.
For the walking beam:
For the spring:
The angular velocity of the lever is assumed as:
0 0
The moments of inertia of the lever are calculated as:
sin
cos sin
cos
0
cos sin
where L is the length of lever and r is the offset from the center of rotation.
According to the equation of angular motion:
4‐49

Also:
sin cos
2
Due to force balance, we can write:
sin sin
2
Therefore can be calculated form above equations.
On the other hand, sin , and and , the dynamic of the system is:
where B is the viscous friction coefficient, and n1 and n2 are constant.
The state variables of the systems are α ,yp, ω and ω2.
4‐50

4-41) If the capacitances of the tanks are assumed to be C1 and C2 respectively, then
Therefore:
1
Asa result:
1 1 1
0
1 1
0
4-42) The equation of motion is:
Considering gives:
or
4-43) (a) Block diagram:

4‐51

(b) Transffer function:
TAO ( s ) KM KR 3.51
= =
Tr ( s ) (1 + τ s )((1 + τ s ) + K
c s m
KR 20 s + 122 s + 4.51
2
4-444)
Sysstem equations:
d θm dθ m d θL dθ L
2 2
Tm (t ) = J m 2
+ Bm + K (θ m − θ L ) K (θ m − θ L ) = J L 2
+ Bp
dt dt dt d
dt
Eθ L
uation: eo =
Output equ
20π
State diiagram:
Transferr
fun
nction:

ΘL (s) K
=
s ⎡⎣ J m J L s + ( Bm J L + B p J m ) s + ( J m K + J L K + Bm B p ) s + Bm K ⎤⎦
3 2
Tm ( s )
Eo ( s ) KE / 20π
=
s ⎡⎣ J m J L s + ( Bm J L + B p J m ) s + ( J m K + J L K + Bm B p ) s + Bm K ⎤⎦
3 2
Tm ( s )
4-445) (a) Statte equations:
dθ L dω L K2 K2 dθ t dω t K1 K1
=ωL = θm − θL = ωt = θm − θt
dt dtt JL JL dt dtt Jt Jt
4‐52

dθ m dω m Bm (K + K2 ) K1 K2 1
= ωm =− ωm − 1
θm + θt + θL + Tm
dt dtt Jm Jm Jm Jm Jm
(b) State ddiagram:

(c) Transffer functions:

Θ L (s)
=
(
K 2 J t s + K1
2
) Θt ( s)
=
(
K1 J L s + K 2
2
) Θm (s)
=
J t J L s + ( K1 J L + K 2 J t ) s + K1 K 2
4 2

Tm ( s ) Δ(s) Tm ( s ) Δ(s) Tm ( s ) Δ(s)
Δ ( s ) = s[ J m J L s + Bm J L J t s + ( K1 J L J t + K 2 J L J t + K1 J m J L + K 2 J m J t ) s
5 4 3

+ Bm J L ( K1 + K 2 ) s + K1 K 2 ( J L + J t + J m ) s + Bm K1 K 2 ] = 0
2
(d) Charaacteristic equattion: Δ s ) = 0 .
Δ(
4-446)
(a) Transffer function:
Ec ( s ) 1 + R2 Cs
G(s) = =
E (s) 1 + ( R1 + R2 ) Cs
(b) Block diagram:
4‐53

(c) Forward‐path transfer function:
Ωm (s) K (1 + R2 Cs )
=
E (s) [1 + ( R 1
+ R2 ) Cs ] ( K b K i + Ra J L s )
(d) Closed‐loop transfer function:
Ωm (s) K φ K (1 + R2 Cs )
=
Fr ( s ) [1 + ( R 1
+ R2 ) Cs ] ( K b K i + Ra J L s ) + K φ KK e N (1 + R2 Cs )
Ec ( s ) (1 + R Cs )
(e) Gc ( s ) = = 2

E (s) R1Cs

Forward‐path transfer function:
Ωm (s) K (1 + R2 Cs )
=
E (s) R1Cs ( K b K i + Ra J L s )
Closed‐loop transfer function:
Ωm (s) K φ K (1 + R2 Cs )
=
Fr ( s ) R1Cs ( K b K i + Ra J L s ) + K φ KK e N (1 + R2 Cs )
Ke = 36 pulses / rev = 36 / 2π pulses / rad = 5.73 pulses / rad.

(f) f r = 120 pulses / sec ω m = 200 RPM = 200(2π / 60 ) rad / sec
f ω = NK eω m = 120 pulses / sec = N ( 36 / 2π ) 200(2π / 60 ) = 120 N pulses / sec
Thus, N = 1. For ω m = 1800 RPM, 120 = N ( 36 / 2π )1800(2π / 60) = 1080 N . Thus, N = 9.
4-47) If the incremental encoder provides a pulse at every edge transition in the two signals of channels
A and B, then the output frequency is increased to four times of input frequency.
4-48) (a)
−1 ⎛ K H (s) ⎞ − K1 ⎛ H i (s) ⎞
⎜ 1 + K1 H e ( s ) + 1 i ⎟ ⎜ H e (s) + ⎟
Ωm (s) B + Js ⎝ Ra + La s ⎠ B + Js ⎝ Ra + La s ⎠
= ≅ =0
TL ( s ) ωr =0
Δ( s) Δ(s)
Thus,
4‐54

H i (s) H i (s)
H e (s) = − = − ( Ra + La s )
Ra + La s H e (s)
K1 K i
Ωm ( s) (R + La s )( B + Js )
(b) = a

Ωr (s) TL = 0
Δ( s)
K1 K b K1 H i ( s ) K1 K i K b H e ( s )
Δ ( s ) = 1 + K1 H e ( s ) + + +
(R a
+ La s )( B + Js ) Ra + La s (R a
+ La s )( ( B + Js )

K1 K b K1 K i
= 1+ +
(R a
+ La s )( B + Js ) (R a
+ La s )( B + Js )
Ωm (s) K1 K i 1
= ≅
Ωr ( s) TL = 0
(R a
+ La s )( B + Js ) + K i K b + K1 K i K b H e ( s ) Kb H e (s)
4-49) (a) Cause‐and‐effect equations: θe = θ r −θ o e = K sθ e ea = Ke
dia Ra 1
=− ia + (e a
− eb ) Tm = K i ia
dt La La
d θm Bm dθ m
2
1 nK L Tm
2
=− + Tm − ( nθ m
− θo ) T2 = θ 2 = nθ m
dt J m dt J Jm n
d θo
2
KL
2
= (θ 2
− θo )
dt JL
State variables: x1 = θ o , x 2 = ω o , x3 = θ m , x 4 = ω m , x5 = ia
State equations:
dx1 dx 2 KL nK L dx3
= x2 =− x1 + x3 = x4
dt dt JL JL dt

2
dx 4 nK L n KL Bm Ki dx5 KK s Kb Ra KK s
=− x1 − x3 − x4 + x5 =− x1 − x4 − x5 + θr
dt Jm Jm Jm Jm dt La La La La
(b) State diagram:
4‐55

(c) Forw
ward‐path tran
nsfer function:
Θo ( s) KK s K i nK L
=
Θ e ( s )
4 3
(
s ⎡⎣ J m J L La s + J L ( Ra J m + Bm J m + Bm La ) s + n K L La J L + K L J m La + Bm Ra J L s +
2
) 2
(n R K J
2
a L L )
+ Ra K L J m + Bm K L La s + K i K b K L + Ra Bm K L ⎤⎦
Clossed‐loop transsfer function:
Θo ( s) KK s K i nK L
=
Θ r ( s )
5 4
(
J m J L La s + J L ( Ra J m + Bm J m + Bm La ) s + n K L La J L + K L J m La + Bm Ra J L s +
2
) 3
(n R K J
2
a L L )
+ Ra K L J m + Bm K L La s + ( K i K b K L + Ra Bm K L ) s + nKK s K i K L
2
(d) K L = ∞, θ o = θ 2 = nθ m . J L is reflected to motor side so J T = J m + n 2 J L .

Staate equations::
dω m Bm Ki dθ m dia Ra KK s KK s Kb
=− ωm + ia = ωm =− ia + θr − nθ m − ωm
dtt JT JT dt dt La La La La
State diagram:
4‐56

Forrward‐path traansfer function
n:
Θo ( s ) KK s K i n
=
Θe ( s) s ⎡⎣ J T La s + ( Ra J T + Bm La ) s + Ra Bm + K i K b ⎤⎦
2
Closed‐loop transsfer function:
Θo ( s) KK s K i n
=
Θr (s) J T La s + ( Ra J T + Bm La ) s + ( Ra Bm + K i K b ) s + KK s K i n
3 2
From paart (c), when K L = ∞, all the terms withoutt K L in Θ o ( s ) / Θ e ( s ) and Θ o ( s ) / Θ r ( s ) cann be neglectedd.
The sam
me results as ab
bove are obtained.
4-550) (a) System equations:
dv dia dis dis dia

f = K i ia = M T + BT v ea = Ra ia + ( La + Las ) − Las + eb 0 = Rs is + ( Ls + Las ) − Las
dt dt dt dt dt
(b) Take the Laplace traansform on bo
oth sides of thee last three equ
uations, with zero initial cond
ditions, we havve
K i I a ( s ) = ( M T s + BT ) V ( s ) Ea ( s ) = [ Ra + ( La + Las ) s ] I a ( s ) − Las sI s ( s ) + K bV ( s )

0 = − Las sI a ( s ) + [ Rs + s ( Ls + Las )] I s ( s )
Rearrranging these equations, wee get
Ki V (s) Ki
V (s) = I a (s) Y (s) = = I a (s)
M T s + BT s s ( M T s + BT )

1 Las s
I a (s) = [E ( s ) + Las sI s ( s ) − K bV ( s ) ] I s (s) = I a (s)
Ra + ( La + Las ) s Ra + ( La + Las ) s
a

diagram:
Block d
4‐57

(c) Trannsfer function::
Y (s) K i [ Rs + ( Ls + Las ) s ]
=
s [ Ra + ( La + Las ) s ] [ Rs + ( Ls + Las ) s ] ( M T s + BT ) + K i K b [ Rs + ( La + Las ) s ] − Las s (M s + BT )
2 2
Ea ( s ) T
4-551) (a) Causee‐and‐effect eqquations:
ea − eb
θe = θr −θL e = K sθ e K s = 1 V/rrad ea = Ke ia =
Ra

dω m 1 Bm KL dω L KL
Tm = K i ia = Tm − ω− (θ m
−θL ) = (θ m
−θL ) eb = K bω m
dt
d Jm Jm Jm dt JL
15.5
Kb = 15.5 V / KRPM = = 0.148 V / rad / sec
1000 × 2π / 60
Statte equations:
dθ L dω L KL KL dθ m dω m Bm KL 1 Ki
= ωL = θm − θL = ωm =− ωm − θL + ( KK θs e
− K bω m )
dt dt JL JL dt dt Jm Jm J m Ra
(b) Statte diagram:
(c) Forw
ward‐path tran
nsfer function:
K i KK
Ks KL
G(ss ) =
s ⎡⎣ J m J L Ra s + ( Bm Ra + K i K b ) J L s + Ra K L ( J L + J m ) s + K L ( Bm Ra + K i K b ) ⎤⎦
3 2
4‐58

J m Ra J L = 0.03 × 115
. × 0.05 = 0.001725 Bm Ra J L = 10 × 115
. × 0.05 = 0.575 Ki Kb J L = 21 × 0.148 × 0.05 = 0.1554
Ra K L J L = 115
. × 50000 × 0.05 = 2875 Ra K L J m = 115
. × 50000 × 0.03 = 1725 Ki KK s K L = 21 × 1 × 50000 K = 1050000 K
K L ( Bm Ra + K i K b ) = 50000(10 × 1.15 + 21 × 0.148) = 730400
608.7 × 10 K
6
G(s) =
(
s s + 423.42 s + 2.6667 × 10 s + 4.2342 × 10
3 2 6 8
)
(d) Closed‐loop transfer function:
ΘL (s) G(s) K i KK s K L
M ( s ) = = =
Θr ( s) 1 + G(s) J m J L Ra s + ( Bm Ra + K i K b ) J L s + Ra K L ( J L + J m ) s + K L ( Bm Ra + K i K b ) s + K i KK s K L
4 3 2
8
6.087 × 10 K
M ( s) = 4 3 6 2 8 8

s + 423.42 s + 2.6667 × 10 s + 4.2342 × 10 s + 6.087 × 10 K
Characteristic equation roots:
K =1 K = 2738 K = 5476
s = −1.45 s = ± j1000 s = 405 ± j1223.4

s = −159.88 s = −2117
. ± j1273.5 s = −617.22 ± j1275
s = −13105
. ± j1614.6
4-52) (a) Nonlinear differential equations:
dx ( t ) dv ( t )
= v(t ) = − k ( v ) − g ( x ) + f ( t ) = − Bv ( t ) + f ( t )
dt dt
e( t ) e( t 0
With Ra = 0 , φ ( t ) = = K f i f ( t ) = K f i f ( t ) = K f ia ( t ) Then, ia ( t ) =
Kb v ( t ) Kb K f v ( t )
2
Ki e ( t ) dv ( t ) Ki 2
f ( t ) = Kiφ ( t )ia ( t ) = 2 2
. Thus, = − Bv ( t ) + 2 2
e (t )
Kb K f v (t ) dt Kb K f v (t )
(b) State equations: ia ( t ) as input.
dx ( t ) dv ( t ) 2
= v (t ) = − Bv ( t ) + Ki K f ia ( t )
dt dt
(c) State equations: φ ( t ) as input.
4‐59

2 φ (t )
f ( t ) = K i K f ia ( t ) ia ( t ) = i f ( t ) =
Kf

dx ( t ) dv ( t ) Ki 2
= v (t ) = − Bv ( t ) + φ (t )
dt dt Kf
4-553) (a) Diffeerential equations:
d θm dθ m ⎛ dθ m − dθ L ⎞
2
K i ia = J m + Bm + K (θ m − θ L ) + B ⎜ ⎟
dt
2
dt ⎝ dt
d dt ⎠
⎛ dθ m − dθ L ⎞ = ⎛ J d θ L + B dθ L ⎞
2
K (θ m − θ L ) + B ⎜ ⎟ ⎜ L 2 ⎟ + TL
⎝ dt dt ⎠ ⎝
L
dt dt ⎠
(b) Take the Laplace traansform of thee differential eq
quations with zero initial con
nditions, we geet
( )
K i I a ( s ) = J m s + Bm s + Bs + K Θ m ( s ) + ( Bs + K ) Θ L ( s )
2

( Bs + K ) Θ m ( )
( s ) − ( Bs + K ) Θ L ( s ) = J L s + BL sΘ L ( s ) + TL ( s)
2
s
a Θ L ( s ) from the last two
Solving for Θ m ( s ) and o equations, wee have
Ki Bs + K
Θm (s) = I a (s) + Θ L (s)
J m s + ( Bm + B ) s + K J m s + ( Bm + B ) s + K
2 2

Bss + K TL ( s )
Θ L (s) = Θm (s) −
J L s + ( BL + B ) s + K J L s + ( BL + B ) s + K
2 2
Signaal flow graph:
4‐60

(c) Transfer matrix:
⎡Θm (s) ⎤ 1 ⎡ K i ⎡⎣ J L s + ( BL + B ) s + K ⎤⎦ Bs + K ⎤ ⎡ I a (s) ⎤

2
⎢ Θ (s) ⎥ = Δ (s) ⎢ K i ( Bs + K )
⎥⎢ ⎥
J m s + ( Bm + B ) s + K ⎦ ⎣ −TL ( s ) ⎦
⎣ L ⎦ ⎣
2
o
Δ o ( s ) = J L J m s + [ J L ( Bm + B ) + J m ( BL + B )] s + [ BL Bm + ( BL + BM ) B + ( J m + J L ) K ] s + K ( BL + B ) s
3 3 2

4-54) As can be estimated by:
1 2
2
1 2
2
Therefore:
2
2 1 1
As a result:
Poles: , ,
zeros:
4-55) By approximating :
1
2
1
2
a)
1
2
1 1
2
Therefore:
1
2
1 1
2
4‐61

b)
1 1
2 2 2 4
1 1
2
3 2
2 2 1 2 1 4 1 2
2 1
4 4
2 1
4-56) MATLAB
clear all
L=1
T=0.1
G1=tf([‐1/2 1],conv([0.1 1],[1/2 1]))
figure(1)
step(G1)
G2=tf([‐1 ‐1 4 4], conv (conv ([1 2],[1 2]),conv([1 1],[1 1])))
figure(2)
step(G2)

L =
1

T =
0.1000

Transfer function:
‐0.5 s + 1
‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐
0.05 s^2 + 0.6 s + 1

Transfer function:
‐s^3 ‐ s^2 + 4 s + 4
‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐
s^4 + 6 s^3 + 13 s^2 + 12 s + 4
4‐62

⎡ L⎤
4-57) 4‐23 (a) Differential equations: ⎢ L ( y ) =
⎣ y ⎥⎦
d L ( y )i ( t ) dL( y ) dy ( t ) L di ( t ) L dy ( t ) L di ( t )
e ( t ) = Ri ( t ) + = Ri ( t ) + i ( t ) + = Ri ( t ) − 2
i(t ) +
dt dy dt y dt y dt y dt
2 2
Ki ( t ) di ( t ) dy ( t ) d y(t )
My ( t ) = Mg − 2
At equilibrium, = 0, = 0, 2
= 0
y (t ) dt dt dt
4‐63

Eeq dy eq Eeq K
Thus, ieq = =0 yeq =
R dt R Mg
dy
(b) Define the state variables as x1 = i, x 2 = y, and x3 = .
dt
Eeq Eeq K
Then, x1eq = x 2eq = x3eq = 0
R R Mg
The differential equations are written in state equation form:
2
dx1 R x1 x3 x2 dx 2 dx3 K x1
=− x1 x 2 + + e = f1 = x3 = f 2 = g− 2
= f3
dt L x2 L dt dt M x2
(c) Linearization:
∂f1 R x3eq Eeq K ∂f1 R x1 x3 Eeq ∂f1 x1eq Mg

=− x 2eq + =− =− x1eq − 2
+ =0 = =
∂x1 L x 2eq L Mg ∂x 2 L x2 L ∂x 3 x 2eq K
∂f1 x 2 eq 1 K Eeq ∂f 2 ∂f 2 ∂f 2 ∂f 2
= = =0 =0 =1 = 0
∂e L L Mg R ∂x1 ∂x 2 ∂x 3 ∂e
2
∂f 3 2 K x1eq 2 Rg ∂f 3 2 K x1eq 2 Rg Mg ∂f 3
=− 2
=− = 3
= = 0
∂x1 M x 2 eq Eeq ∂x 2 M x 2eq Eeq K ∂e
∗ ∗
The linearized state equations about the equilibrium point are written as: Δx& = A Δx + B Δe
⎡ Eeq K Mg ⎤ ⎡ Eeq K ⎤
⎢− 0 ⎥ ⎢ ⎥
⎢ L Mg K ⎥
⎢ RL Mg ⎥
A =⎢ 0 ⎥ B =⎢ ⎥
∗ ∗
0 0 0
⎢ ⎥ ⎢ ⎥
⎢ 2 Rg 2 Rg Mg ⎥ ⎢ 0 ⎥
⎢ − E 0 ⎥ ⎢ ⎥
⎣ eq
Eeq K ⎦ ⎣ ⎦
4-58)
(a) Differential equations:
2 2
d y1 ( t ) dy1 ( t ) Ki ( t ) 2 1
M1 2
= M1 g − B − 2
+ Ki ( t ) 2

dt dt y1 ( t ) y 2 ( t ) − y1 ( t )
4‐64

2 2
d y2 (t ) 2 dy 2 ( t ) Ki ( t )
M2 2
= M2 g − B − 2

dt dt y 2 ( t ) − y1 ( t )
dy1 dy 2
Define the state variables as x1 = y1 , x 2 = , x3 = y 2 , x 4 = .
dt dt
The state equations are:
2 2 2
dx1 dx2 Ki Ki dx3 dx4 Ki
= x2 M1 = M 1 g − Bx2 − + = x4 M2 = M 2 g − Bx4 −
dt dt x1
2
(x 3
− x1 )
2
dt dt (x 3
− x1 )
2
dx1 dx 2 dx3 dx 4
At equilibrium, = 0, = 0, = 0, = 0. Thus, x 2eq = 0 and x 4 eq = 0.
dt dt dt dt
2 2 2
KI KI KI
M1g − + =0 M2g − = 0
X1
2
(X 3
− X1 )
2
(X 3
− X1 )
2
Solving for I, with X 1 = 1, we have
⎛ ( M1 + M 2 ) g ⎞
1/ 2
⎛ M + M2 ⎞
1/ 2
Y2 = X 3 = 1 + ⎜ 1 ⎟ I =⎜ ⎟
⎝ M2 ⎠ ⎝ K ⎠
(b) Nonlinear state equations:
2 2
dx1 dx2 B K Ki dx3 dx4 B Ki
= x2 =g− x2 − i + = x4 =g− x4 −
2

M 1 ( x3 − x1 ) M 2 ( x3 − x1 )
2 2 2
dt dt M1 M 1 x1 dt dt M2

(c) Linearization:
∂ f1 ∂ f1 ∂ f1 ∂ f1 ∂ f1
=0 =0 =0 =0 =0
∂ x1 ∂ x2 ∂ x3 ∂ x4 ∂i

∂ f2 ∂ f2 ∂ f2 −2 KI ∂ f2
2 2 2
2 KI 2 KI B
= + =− = =0
∂ x1 M1 ( X 3 − X1 ) ∂ x2 ∂ x3 M1 ( X 3 − X1 ) ∂ x4
3 3 3
M 1 x1 M1
∂ f2 2 KI⎛ −1 1 ⎞ ∂ f3 ∂ f3 ∂ f3 ∂ f3 ∂ f3
= ⎜ 2+ 2 ⎟
=0 =0 =0 =1 =0
∂i M1 ⎝ X1 ( X 3 − X1 ) ⎠ ∂ x1 ∂ x2 ∂ x3 ∂ x4 ∂i
∂ f4 −2 KI ∂ f4 ∂ f4 ∂ f4 ∂ f4 −2 KI
2 2
2 KI B
= =0 = =− =
∂ x1 M 2 ( X 3 − X1 ) ∂ x2 ∂ x3 M 2 ( X 3 − X1 ) ∂ x4 ∂i M 2 ( X 3 − X1 )
3 3 2
M2
Linearized state equations: M1 = 2, M 2 = 1, g = 32.2, B = 0.1, K = 1.
4‐65

1/ 2
⎛ 32.2(1 + 2) ⎞ X = 96.6 X = 9.8285 X
I = ⎜ X1 =
1
= 1
⎟
⎝ ⎠
1 1 1
1 9.8285
( )
X 3 = 1 + 1 + 2 X 1 = 2.732 X 1 = Y2 = 2.732 X 3 − X 1 = 1.732
⎡ 0 1 0 0 ⎤
⎢ ⎥
⎢ 2 KI ⎛⎜ 1 + ⎞ −B −2 KI ⎡ 0 0 ⎤
2 2
1 1 0
3 ⎟
0 ⎥
⎢ M1 ⎝ X1 ( X 3 − X1 ) ⎠
3
M1 M1 ( X 3 − X1 )
3
⎥ ⎢ 115.2 −0.05 −18.59 0 ⎥
A = ⎢
∗
⎥ = ⎢ ⎥
⎢ 0 0 0 1 ⎥ ⎢ 0 0 0 1 ⎥
⎢ ⎢ ⎥
−2 KI − B ⎥ ⎣ −37.18 0 37.18 −0.1⎦
2 2
2 KI
⎢ 0 ⎥
⎢⎣ M 2 ( X 3 − X1 ) M 2 ( X 3 − X1 ) M 2 ⎥⎦
3 3
⎡ 0 ⎤
⎢ ⎥
⎢ 2 KI ⎛⎜ −1 + 1 ⎞⎥ ⎡ 0 ⎤
⎢ M 1 ⎝ X 12 ( X 3 − X 1 )2 ⎟⎠ ⎥ ⎢ −6.552 ⎥
⎥=⎢ ⎥
∗
B = ⎢
⎢ 0 ⎥ ⎢ 0 ⎥
⎢ −2 KI ⎥ ⎣ −6.552 ⎥⎦
⎢
⎢ ⎥
⎣⎢ M 2 ( X 3 − X1 ) ⎦⎥
2
4-59) a)
F3
θ
F1 F2
b) The equation of the translational motion is:
4‐66

The equation of rotational motion is:
where
Also, the relation between rotational and translational motion defines:
Therefore, substituting above expression into the first equation gives:
2
The resulted state space equations are:
2
2
c) According to generalized elements:

1) Viscous friction can be replaced by a resistor where R = B
2) Spring can be replaced by a capacitor where
3) Mass M and m can be replaced by two inductors where and . Then the
angular velocity is measured as a voltage of the inductor L2
4) The gear will be replaced by a transformer with the ratio of
5) The term Mg is also replaced by an input voltage of
4‐67

4-60) As the base is not moving then the model can be reduced to:
Therefore:
1) As , they can be replaced by a inductor with L = m

2) Friction B can be replaced by a resistor where R = B
3) Spring can be replaced by a capacitor where
4) The force F is replaced by a current source where Is = F
4-61)
V2
| |
A
C
ρg
4‐68

4-62) Recall Eq. (4-324)
Z (s) −1
= 2
&&
Y ( s ) s + 2ζω n s + ω n 2
ω = 1,
Set Y&&( s ) = impulse , pick n for simplicity.
ς =1
clear all
G=tf([-1],[1 2 1])
figure(1)
impulse(G)
Transfer function:
-1
-------------
s^2 + 2 s + 1
4‐69

4-63)
Use Eq. (4-329).
Kmr
Ra
Z (s) = Va ( s )
⎛ La ⎞ 2 K m Kb
⎜ s + 1⎟ ( Js + Bs + K ) + s
⎝ Ra ⎠ Ra
⎛ La ⎞
⎜ s + 1⎟ r
− ⎝ Ra ⎠ mrY&&( s )
⎛ La ⎞ 2 K m Kb
⎜ s + 1⎟ ( Js + Bs + K ) + s
R
⎝ a ⎠ R a
La
For =0 (very small) the format of the equation is similar to Eq. (4-324), and we expect the same
Ra
Kmr
Ra
response for the disturbance input. Except, Z ( s ) = Va ( s ) can be used to reduce the
K K
( Js + Bs + K ) + m b s
2
Ra
effects of disturbance. See Chapter 6.
4‐70

Chapter 5
5‐1 (a) ζ ≥ 0.707 ω n ≥ 2 rad / sec (b) 0 ≤ ζ ≤ 0.707 ω n ≤ 2 rad / sec

(c) ζ ≤ 0.5 1 ≤ ω n ≤ 5 rad / sec (d) 0.5 ≤ ζ ≤ 0.707 ω n ≤ 0.5 rad / sec

5‐2 (a) Type 0 (b) Type 0 (c) Type 1 (d) Type 2 (e) Type 3 (f) Type 3
(g) type 2 (h) type 1

2
5‐3 (a) K p = lim G ( s ) = 1000 K v = lim sG ( s ) = 0 K a = lim s G ( s ) = 0
s→ 0 s→ 0 s→ 0

2
(b) K p = lim G ( s ) = ∞ K v = lim sG ( s ) = 1 K a = lim s G ( s ) = 0
s→ 0 s→ 0 s→ 0
5‐1


2
(c) K p = lim G ( s ) = ∞ K v = lim sG ( s ) = K K a = lim s G ( s ) = 0
s→ 0 s→ 0 s→ 0

2
(d) K p = lim G ( s ) = ∞ K v = lim sG ( s ) = ∞ K a = lim s G ( s ) = 1
s→ 0 s→ 0 s→ 0

2
(e) K p = lim G ( s ) = ∞ K v = lim sG ( s ) = 1 K a = lim s G ( s ) = 0
s→ 0 s→ 0 s→ 0
2
(f) K p = lim G ( s ) = ∞ K v = lim sG ( s ) = ∞ K a = lim s G ( s ) = K
s→ 0 s→ 0 s→ 0
5‐4 (a) Input Error Constants Steady‐state Error
________________________________________________________________________________
u s ( t ) K p = 1000 1 1001
tu s ( t ) K v = 0 ∞
∞
2
t us (t ) / 2 K a = 0
(b) Input Error Constants Steady‐state Error
________________________________________________________________________________
u s ( t ) K p = ∞ 0
tu s ( t ) K v = 1 1
∞
2
t us (t ) / 2 K a = 0
(c) Input Error Constants Steady‐state Error
________________________________________________________________________________
u s ( t ) K p = ∞ 0
5‐2

tu s ( t ) K v = K 1/ K
∞
2
t us (t ) / 2 K a = 0
The above results are valid if the value of K corresponds to a stable closed‐loop system.
(d) The closed‐loop system is unstable. It is meaningless to conduct a steady‐state error analysis.
(e) Input Error Constants Steady‐state Error
________________________________________________________________________________
u s ( t ) K p = ∞ 0
tu s ( t ) K v = 1 1
∞
2
t us (t ) / 2 Ka = 0
(f) Input Error Constants Steady‐state Error
________________________________________________________________________________
u s ( t ) Kp = ∞ 0
tu s ( t ) K v = ∞ 0
2
t us (t ) / 2 K a = K 1/ K
The closed‐loop system is stable for all positive values of K. Thus the above results are valid.
G( s) s +1
5‐5 (a) K H = H ( 0) = 1 M ( s) = = 3 2

1+ G( s) H ( s) s + 2 s + 3s + 3
a0 = 3, a1 = 3, a2 = 2, b0 = 1, b1 = 1.
Unit‐step Input:
1⎛ b0 K H ⎞ 2
ess = ⎜1 − a ⎟= 3
KH ⎝ 0 ⎠
5‐3

Unit‐ramp input:
a0 − b0 K H = 3 − 1 = 2 ≠ 0. Thus e ss = ∞.

Unit‐parabolic Input:
a0 − b0 K H = 2 ≠ 0 and a1 − b1 K H = 1 ≠ 0. Thus e ss = ∞.

(b) K H = H ( 0) = 5
G( s) 1
M ( s) = = 2
a0 = 5, a1 = 5, b0 = 1, b1 = 0.
1+ G( s) H ( s) s + 5s + 5
Unit‐step Input:
1 ⎛ b0 K H ⎞ 1⎛ 5⎞
ess = ⎜1 − a ⎟ = 5 ⎜1 − 5 ⎟ = 0
KH ⎝ 0 ⎠ ⎝ ⎠
Unit‐ramp Input:
i = 0: a0 − b0 K H = 0 i = 1: a1 − b1 K H = 5 ≠ 0
a1 − b1 K H 15
e ss = = =
a0 K H 25 5
e ss = ∞
(c) K H = H ( 0) = 1 / 5
G( s) s+5
M ( s ) = = 4 3 2
The system is stable.
1+ G( s) H ( s) s + 15s + 50 s + s + 1
a0 = 1, a1 = 1, a2 = 50, a3 = 15, b0 = 5, b1 = 1
Unit‐step Input:
1 ⎛ b0 K H ⎞ ⎛ 5/5⎞
ess = ⎜1 − a ⎟ = 5 ⎜1 − 1 ⎟ = 0
KH ⎝ 0 ⎠ ⎝ ⎠
Unit‐ramp Input:
5‐4

i = 0: a0 − b0 K H = 0 i = 1: a1 − b1 K H = 4 / 5 ≠ 0
a1 − b1 K H 1− 1/ 5
e ss = = = 4
a0 K H 1/ 5
e ss = ∞
(d) K H = H ( 0 ) = 10
G( s) 1
M ( s ) = = 3 2
The system is stable.
1+ G( s) H ( s) s + 12 s + 5s + 10
a0 = 10, a1 = 5, a2 = 12, b0 = 1, b1 = 0, b2 = 0
Unit‐step Input:
1 ⎛ b0 K H ⎞ 1 ⎛ 10 ⎞
ess = ⎜1 − a ⎟ = 10 ⎜ 1 − 10 ⎟ = 0
KH ⎝ 0 ⎠ ⎝ ⎠
Unit‐ramp Input:
i = 0: a0 − b0 K H = 0 i = 1: a1 − b1 K H = 5 ≠ 0
a1 − b1 K H 5
e ss = = = 0.05
a0 K H 100
e ss = ∞
s+4
5‐6 (a) M ( s ) = 4 3 2
K H = 1 The system is stable.
s + 16 s + 48 s + 4 s + 4
a0 = 4, a1 = 4, a2 = 48, a3 = 16, b0 = 4, b1 = 1, b2 = 0, b3 = 0
Unit‐step Input:
1 ⎛ b0 K H ⎞ ⎛ 4⎞
ess = ⎜1 − a ⎟ = ⎜1 − 4 ⎟ = 0
KH ⎝ 0 ⎠ ⎝ ⎠
Unit‐ramp input:
i = 0: a0 − b0 K H = 0 i = 1: a1 − b1 K H = 4 − 1 = 3 ≠ 0
5‐5

a1 − b1 K H 4 −1 3
e ss = = =
a0 K H 4 4
e ss = ∞
K ( s + 3)
(b) M ( s ) = 3 2
K H = 1 The system is stable for K > 0.
s + 3s + ( K + 2) s + 3K
a0 = 3 K , a1 = K + 2, a2 = 3, b0 = 3 K , b1 = K
Unit‐step Input:
1 ⎛ b0 K H ⎞ ⎛ 3K ⎞
ess = ⎜1 − a ⎟ = ⎜ 1 − 3K ⎟ = 0
KH ⎝ 0 ⎠ ⎝ ⎠
Unit‐ramp Input:
i = 0: a0 − b0 K H = 0 i = 1: a1 − b1 K H = K + 2 − K = 2 ≠ 0
a1 − b1 K H K + 2− K 2
e ss = = =
a0 K H 3K 3K
e ss = ∞
The above results are valid for K > 0.
s+5 10 s H ( s)
(c) M ( s ) = 4 3 2
H ( s) = K H = lim = 2
s + 15 s + 50 s + 10 s s+5 s→ 0 s
a0 = 0, a1 = 10, a2 = 50, a3 = 15, b0 = 5, b1 = 1
Unit‐step Input:
1 ⎛ a2 − b1 K H ⎞ 1 ⎛ 50 − 1 × 2 ⎞
ess = ⎜ a ⎟= ⎜ ⎟ = 2.4
KH ⎝ 1 ⎠ 2⎝ 10 ⎠
Unit‐ramp Input:
5‐6

e ss = ∞
e ss = ∞
K ( s + 5)
(d) M ( s ) = 4 3 2
K H = 1 The system is stable for 0 < K < 204.
s + 17 s + 60 s + 5 Ks + 5 K
a0 = 5 K , a1 = 5K , a2 = 60, a3 = 17, b0 = 5K , b1 = K

Unit‐step Input:
⎛ b0 K H
1 ⎞ ⎛ 5K ⎞
ess = ⎜1 − a ⎟ = ⎜ 1 − 5K ⎟ = 0
KH ⎝ 0 ⎠ ⎝ ⎠
Unit‐ramp Input:
i = 0: a0 − b0 K H = 0 i = 1: a1 − b1 K H = 5 K − K = 4 K ≠ 0
a1 − b1 K H 5K − K 4
e ss = = =
a0 K H 5K 5
e ss = ∞

The results are valid for 0 < K < 204.
5‐7

5-7)
1
3 2 5 1
5 1 5 11 5
1
2 3
Ö Type of the system is zero
Pole: s = -2.2013 + 1.8773i, s = -2.2013 - 1.8773i, and s = -0.5974
Zero: s = -1
5-8)
5( s + 1)
s ( s + 2)( s + 3)
5 1
2 3
a) Position error: lim G s lim ∞
b) Velocity error: lim lim
c) Acceleration error: lim lim 0
5-9) a) Steady state error for unit step input:
Referring to the result of problem 5-8, ∞Î 0
b) Steady state error for ramp input:
5‐8

Regarding the result of problem 5-8, Î ∞
c) Steady state error for parabolic input:
Regarding the result of problem 5-8, 0Î ∞ ∞
5-10)
4( s + 1)
s 2 ( s + 2)
a) Step error constant: lim ∞
b) Ramp error constant: lim ∞
c) Parabolic error constant: lim 2
5-11) 3 4
where x1 is a unit step input, x2 is a ramp input, and x3 is a unit parabola input. Since the system
is linear, then the effect of X(s) is the summation of effect of each individual input.
That is: ∞ ∞ 3 ∞ 4 ∞
So:
Ö 4 2
5‐9

5-12) The step input response of the system is:
1 1 1 1
1
Therefore:
1
1
1
The rise time is the time that unit step response value reaches from 0.1 to 0.9. Then:
1 . .
1
It is obvious that tr > 0, then:
1 . .
0
1
As |k| < 1, then 0
. . . .
Therefore 0 or
which yields: 0
5‐13)
Y ( s) KG p ( s ) 20 s 100 K
G( s) = = = Type‐1 system.
E ( s) 1 + Kt G p ( s ) 20 s(1 + 0.2 s + 100 Kt )
5K
Error constants: K p = ∞, K v = , Ka = 0
1 + 100 Kt

1
(a) r ( t ) = u s ( t ): e ss = = 0
1+ K p
1 1 + 100 Kt
(b) r ( t ) = tu s ( t ): e ss = =
Kv 5K
1
(c) r ( t ) = t 2u s ( t ) / 2: e ss = = ∞
Ka
5‐10

5‐14
100 Y ( s) KG p ( s )
G p ( s) = G( s) = =
(1 + 0.1s )(1 + 0.5s ) E ( s) 20 s 1 + Kt G p ( s )
100 K
G( s) =
20 s (1 + 0.1s )(1 + 0.5 s ) + 100 Kt
5K
Error constants: K p = ∞, K v = , Ka = 0
1 + 100 Kt
1
(a) r ( t ) = u s ( t ): e ss = = 0
1+ K p
1 1 + 100 Kt
(b) r ( t ) = tu s ( t ): e ss = =
Kv 5K
1
(c) r ( t ) = t 2u s ( t ) / 2: e ss = = ∞
Ka
Since the system is of the third order, the values of K and Kt must be constrained so that the system is
stable. The characteristic equation is
s + 12 s + ( 20 + 2000 K t ) s + 100 K = 0
3 2

Routh Tabulation:
3
s 1 20 + 2000 Kt
2
s 12 100 K
240 + 24000 Kt − 100 K
1
s
12
0
s 100 K
1 + 100 K t 1
Stability Conditions: K > 0 12 (1 + 100 K t ) − 5 K > 0 or >
5K 12
Thus, the minimum steady‐state error that can be obtained with a unit‐ramp input is 1/12.

5‐11

5‐15 (a) From Figure 3P‐29,
K1 K 2 K i K b + KK1 K i K t
1+ +
Θo ( s) Ra + La s (R + La s )( Bt + J t s )
= a

Θr (s) K1 K 2 K i K b + KK1 K i K t KK s K1 K i N
1+ + +
Ra + La s (Ra
+ La s )( Bt + J t s ) s ( Ra + La s )( Bt + J t s )
Θo (s) s [( Ra + La s )( Bt + J t s ) + K1 K 2 ( Bt + J t s ) + K i K b + KK1 K i K t ]
=
Θr (s) La J t s + ( La Bt + Ra J t + K1 K 2 J t ) s + ( Ra Bt + K i K b + KK i K1 K t + K1 K 2 Bt ) s + KK s K1 K i N
3 2
1
θ r ( t ) = u s ( t ), Θ r ( s ) = lim sΘ e ( s ) = 0
s s→ 0
Provided that all the poles of sΘ e ( s ) are all in the left‐half s‐plane.
2
(b) For a unit‐ramp input, Θ r ( s ) = 1 / s .
Ra Bt + K1 K 2 Bt + Ki Kb + KK1 Ki Kt
e ss = lim θ e ( t ) = lim sΘ e ( s ) =
t →∞ s→ 0 KK s K1 Ki N
if the limit is valid.

5‐12

5‐16 (a) Forward‐path transfer function: [n(t) = 0]:
K (1 + 0.02 s )
Y ( s) s ( s + 25)
2
K (1 + 0.02 s )
G (s) = = = Type‐1 system.
E (s)
1+ 2
KK t s (
s s + 25s + KK t
2
)
s ( s + 25)
1
Error Constants: K p = ∞, Kv = , Ka = 0
Kt
1 1
For a unit‐ramp input, r ( t ) = tu s ( t ), R( s ) = 2
, e ss = lim e( t ) = lim sE ( s ) = = Kt
s t →∞ s→ 0 Kv
Routh Tabulation:
KK t + 0.02 K
3
s 1
2
s 25 K

1 25 K ( K t + 0.02 ) − K
s
25
0
s K
Stability Conditions: K >0 25 ( K t + 0.02 ) − K > 0 or K t > 0.02
(b) With r(t) = 0, n( t ) = u s ( t ), N ( s ) = 1 / s.
System Transfer Function with N(s) as Input:
K
Y (s) s ( s + 25) K
2
= = 3
K (1 + 0.02 s ) KK t s s + 25 s + K ( K t + 0.02 ) s + K
2
N (s)
1+ 2 + 2
s ( s + 25) s ( s + 25)
Steady‐State Output due to n(t):
y ss = lim y ( t ) = lim sY ( s ) = 1 if the limit is valid.

t →∞ s→ 0
5‐13

5‐17 You may use MATLAB in all Routh Hurwitz calculations.
1. Activate MATLAB
3. Type in
Acsys
4. Then press the “transfer function Symbolic” and enter the Characteristic equation
5. Then press the “Routh Hurwitz” button
6. For example look at below Figures

5‐14

(a) n( t ) = 0, r ( t ) = tu s ( t ).
Forward‐path Transfer function:
Y (s) K ( s + α )( s + 3)
G(s) = = Type‐1 system.
E ( s) n=0 (
s s −1
2
)
Ramp‐error constant: K v = lim sG ( s ) = −3 Kα
s→ 0
1 1
Steady‐state error: e ss = =−
Kv 3Kv
3 2
Characteristic equation: s + Ks + [ K ( 3 + α ) − 1] s + 3αK = 0
Routh Tabulation:
3
s 1 3 K + αK − 1
2
s K 3αK
K ( 3 K + αK − 1) − 3αK
1
s
K
0
s 3αK
1 + 3K
3 K + αK − 1 − 3α > 0 or K>
Stability Conditions: 3+α
αK > 0
(b) When r(t) = 0, n( t ) = u s ( t ), N ( s ) = 1 / s.
K ( s + 3)
Y (s) s −1
2
Ks ( s + 3)
Transfer Function between n(t) and y(t): = =
N (s) K ( s + α )( s + 3) s 3 + Ks 2 + [ K ( s + α ) − 1]s + 3α K
r =0
1+
s s −1
2
( )
Steady‐State Output due to n(t):
y ss = lim y ( t ) = lim sY ( s ) = 0 if the limit is valid.

t →∞ s→ 0

5‐15

5‐18
− πζ
2
1−ζ
Percent maximum overshoot = 0.25 = e
Thus
πζ 1 − ζ = − ln 0.25 = 1.386 π ζ = 1.922 (1 − ζ )

2 2 2 2

Solving for ζ from the last equation, we have ζ = 0.404.
π π
Peak Time t max = = 0.01 sec. Thus, ω n = = 343.4 rad / sec
2 2
ω n 1− ζ 0.01 1 − ( 0.404)
Transfer Function of the Second‐order Prototype System:
2
Y ( s) ωn 117916
= 2 2
= 2

R( s ) s + 2ζω n s + ω n s + 277.3s + 117916

5‐16

Extended MATLAB solutions of problems similar to 5‐19‐5‐27 appear later
on – e.g. 5‐58
5‐19 Closed‐Loop Transfer Function: Characteristic equation:
Y (s) 25 K
= s + ( 5 + 500 K t ) s + 25 K = 0
2

s + ( 5 + 500 K t ) s + 25 K
2
R(s)
For a second‐order prototype system, when the maximum overshoot is 4.3%, ζ = 0.707.
ω n = 25 K , 2ζω n = 5 + 500 Kt = 1.414 25 K
Rise Time:
2
1 − 0.4167ζ + 2.917ζ 2.164
tr = = = 0.2 sec Thus ω n = 10.82 rad / sec
ωn ωn
2
ωn (10.82)
2
10.3
Thus, K= = = 4.68 5 + 500 Kt = 1.414ω n = 15.3 Thus Kt = = 0.0206
25 25 500
With K = 4.68 and Kt = 0.0206, the system transfer function is
Y ( s) 117
= 2

R( s ) s + 15.3s + 117
Unit‐step Response:

y = 0.1 at t = 0.047 sec.
y = 0.9 at t = 0.244 sec.
t r = 0.244 − 0.047 = 0.197 sec.

y max = 0.0432 ( 4.32% max. overshoot)
5‐17

5‐20 Closed‐loop Transfer Function: Characteristic Equation:
Y (s) 25 K
= s + ( 5 + 500 K t ) s + 25 K = 0
2

s + ( 5 + 500 K t ) s + 25 K
2
R(s)
πζ
= − ln 0.1 = 2.3 π ζ = 5.3 (1 − ζ )
2 2 2
When Maximum overshoot = 10%,
1−ζ
2
Solving for ζ , we get ζ = 0.59.
The Natural undamped frequency is ω n = 25K Thus, 5+ 500 Kt = 2ζω n = 118

. ω n
Rise Time:
2
1 − 0.4167ζ + 2.917ζ 17696
.
tr = = 0.1 = sec. Thus ω n = 17.7 rad / sec
ωn ωn
2
ωn 15.88
K= = 12.58 Thus Kt = = 0.0318
25 500
Y ( s) 313
= 2

R( s ) s + 20.88 s + 314.5
y = 0.1 when t = 0.028 sec.
y = 0.9 when t = 0.131 sec.
t r = 0.131 − 0.028 = 0.103 sec.

y max = 11
. (10% max. overshoot )
5‐18

5‐21 Closed‐Loop Transfer Function: Characteristic Equation:
Y ( s) 25 K
= s + ( 5 + 500 K t ) s + 25 K = 0
2
2

R( s ) s + ( 5 + 500 Kt ) s + 25 K
πζ
= − ln 0.2 = 1.61 π ζ = 2.59 (1 − ζ )
2 2 2
When Maximum overshoot = 20%,
1−ζ
2
Solving for ζ , we get ζ = 0.456.
The Natural undamped frequency ω n = 25 K 5 + 500 Kt = 2ζω n = 0.912ω n
Rise Time:
2
1 − 0.4167ζ + 2.917ζ 1.4165 1.4165
tr = = 0.05 = sec. Thus, ω n = = 28.33
ωn ωn 0.05
2
ωn
K= = 32.1 5 + 500 Kt = 0.912ω n = 25.84 Thus, Kt = 0.0417
25
Y ( s) 802.59
= 2

R( s ) s + 25.84 s + 802.59
y = 0.1 when t = 0.0178 sec.
y = 0.9 when t = 0.072 sec.
t r = 0.072 − 0.0178 = 0.0542 sec.
y max = 1.2 ( 20% max. overshoot )
5‐19

Y (s) 25 K
= s + ( 5 + 500 K t ) s + 25 K = 0
2

s + ( 5 + 500 K t ) s + 25 K
2
R(s)
2
. + 0.125ζ + 0.469ζ
11
Delay time t d ≅ = 0.1 sec.
ωn
1.423
When Maximum overshoot = 4.3%, ζ = 0.707. td = = 0.1 sec. Thus ω n = 14.23 rad/sec.
ωn
2 2
K =⎜
⎛ ω n ⎞ = ⎛ 14.23 ⎞ = 8.1 5 + 500 K = 2ζω = 1.414ω = 20.12 Thus K = 1512
.
= 0.0302
⎟ ⎜ ⎟
⎝ 5 ⎠ ⎝ 5 ⎠ t
t n n
500
With K = 20.12 and Kt = 0.0302 , the system transfer function is
Y ( s) 202.5
= 2

R( s ) s + 20.1s + 202.5
Unit‐Step Response:
When y = 0.5, t = 0.1005 sec.
Thus, t d = 0.1005 sec.

y max = 1043
. ( 4.3% max. overshoot )

5‐20

Y (s) 25 K
= s + ( 5 + 500 K t ) s + 25 K = 0
2

s + ( 5 + 500 K t ) s + 25 K
2
R(s)
2
. + 0.125ζ + 0.469ζ
11 1337
. 1337
.
Delay time t d ≅ = 0.05 = Thus, ω n = = 26.74
ωn ωn 0.05
2 2
⎛ ω ⎞ ⎛ 26.74 ⎞ = 28.6 5 + 500 K = 2ζω = 2 × 0.59 × 26.74 = 3155
K =⎜ n ⎟ =⎜ Thus Kt = 0.0531
⎟ .
⎝ 5 ⎠ ⎝ 5 ⎠ t n
Y ( s) 715
= 2

R( s ) s + 3155
. s + 715
y = 0.5 when t = 0.0505 sec.
Thus, t d = 0.0505 sec.
y max = 11007
. (10.07% max. overshoot )

5‐21

Y (s) 25 K
= s + ( 5 + 500 K t ) s + 25 K = 0
2

s + ( 5 + 500 K t ) s + 25 K
2
R(s)
For Maximum overshoot = 0.2, ζ = 0.456 .
2
. + 0.125ζ + 0.469ζ
11 1.2545
Delay time t d = = = 0.01 sec.
ωn ωn
Natural Undamped Frequency ω n =
1.2545 ⎛ ω ⎞ 15737.7 = 629.5
= 125.45 rad/sec. Thus, K = ⎜ n ⎟ =
0.01 ⎝ 5 ⎠ 25
5 + 500 Kt = 2ζω n = 2 × 0.456 × 125.45 = 114.41 Thus, Kt = 0.2188
With K = 629.5 and Kt = 0.2188 , the system transfer function is
Y ( s) 15737.7
= 2

R( s ) s + 114.41s + 15737.7
y = 0.5 when t = 0.0101 sec.
Thus, t d = 0.0101 sec.
y max = 1.2 ( 20% max. overshoot )
5‐22

Y (s) 25 K
= s + ( 5 + 5000 K t ) s + 25 K = 0
2

s + ( 5 + 500 K t ) s + 25 K
2
R(s)
ζ = 0.6 2ζω n = 5 + 500 Kt = 1.2ω n
3.2 3.2 3.2

Settling time t s ≅ = = 0.1 sec. Thus, ω n = = 53.33 rad / sec
ζω n 0.6ω n 0.06
2
1.2ω n − 5 ωn
Kt = = 0.118 K= = 113.76
500 25
System Transfer Function:
Y ( s) 2844
= 2

R( s ) s + 64 s + 2844
y(t) reaches 1.00 and never exceeds this
value at t = 0.098 sec.
Thus, t s = 0.098 sec.

5‐23

5‐26 (a) Closed‐Loop Transfer Function: Characteristic Equation:
Y (s) 25 K
= s + ( 5 + 500 K t ) s + 25 K = 0
2

s + ( 5 + 500 K t ) s + 25 K
2
R(s)
For maximum overshoot = 0.1, ζ = 0.59. 5 + 500 Kt = 2ζω n = 2 × 0.59ω n = 118

. ω n
3.2 3.2 3.2

Settling time: t s = = = 0.05 sec. ωn = = 108.47
ζω n 0.59ω n 0.05 × 0.59
2
. ωn −5
118 ωn
Kt = = 0.246 K= = 470.63
500 25
Y ( s) 11765.74
= 2

R( s ) s + 128 s + 11765.74
y(t) reaches 1.05 and never exceeds
this value at t = 0.048 sec.
Thus, t s = 0.048 sec.
(b) For maximum overshoot = 0.2, ζ = 0.456. 5 + 500 Kt = 2ζω n = 0.912ω n
3.2 3.2 3.2

Settling time t s = = = 0.01 sec. ω n = = 701.75 rad / sec
ζω n 0.456ω n 0.456 × 0.01
0.912ω n − 5
Kt = = 1.27
500
Y ( s) 492453
= 2

R( s ) s + 640 s + 492453
5‐24


y(t) reaches 1.05 and never
exceeds this value at t = 0.0074 sec.
Thus, t s = 0.0074 sec. This is less
than the calculated value of 0.01 sec.

5‐25

Y (s) 25 K
= s + ( 5 + 500 K t ) s + 25 K = 0
2

s + ( 5 + 500 K t ) s + 25 K
2
R(s)
4.5ζ 3.1815
Damping ratio ζ = 0.707. Settling time t s = = = 0.1 sec. Thus, ω n = 31815
. rad/sec.
ωn ωn
2
ωn
5 + 500 Kt = 2ζω n = 44.986 Thus, Kt = 0.08 K= = 40.488
2ζ
Y ( s) 1012.2
= 2

R( s ) s + 44.986 s + 1012.2
Unit‐Step Response: The unit‐step response reaches 0.95 at t = 0.092 sec. which is the measured t s .

5‐26

5‐28 (a) When ζ = 0.5 , the rise time is

2
1 − 0.4167ζ + 2.917ζ 1521
.
tr ≅ = = 1 sec. Thus ω n = 1.521 rad/sec.
ωn ωn
The second‐order term of the characteristic equation is written
2 2 2
s + 2ζω n s + ω n = s + 1521
. s + 2.313 = 0
3 2
The characteristic equation of the system is s + ( a + 30 ) s + 30 as + K = 0
2
Dividing the characteristic equation by s + 1521
. s + 2.313, we have

For zero remainders, 28.48 a = 45.63 Thus, a = 16

. K = 65.874 + 2.313a = 69.58
Forward‐Path Transfer Function:
69.58
G( s) =
s( s + 16
. )( s + 30)
y = 0.1 when t = 0.355 sec.
y = 0.9 when t = 1.43 sec.

Rise Time:
t r = 1.43 − 0.355 = 1075

. sec.
5‐27


(b) The system is type 1.
(i) For a unit‐step input, e ss = 0.
K 60.58 1
(ii) For a unit‐ramp input, K v = lim sG ( s ) = = = 1.45 e ss = = 0.69
s→ 0 30 a 30 × 16
. Kv
5‐29 (a) Characteristic Equation:
3 2
s + 3 s + ( 2 + K ) s − K = 0
Apply the Routh‐Hurwitz criterion to find the range of K for stability.
Routh Tabulation:
3
s 1 2+ K
2
s 3 −K
6 + 4K
1
s
3
0
s −K
Stability Condition: ‐1.5 < K < 0 This simplifies the search for K for two equal roots.
When K = −0.27806, the characteristic equation roots are: −0.347,
−0.347, and −2.3054.
(b) Unit‐Step Response: (K = −0.27806)

5‐28

(c) Unit‐Step Response (K = −1)

The step responses in (a) and (b) all have a negative undershoot for small values of t. This is due to the
zero of G(s) that lies in the right‐half s‐plane.

5‐29

5‐30 (a) The state equations of the closed‐loop system are:
dx1 dx 2
= − x1 + 5x 2 = −6 x1 − k1 x1 − k 2 x 2 + r
dt dt
The characteristic equation of the closed‐loop system is
s +1 −5
Δ= = s + (1 + k 2 ) s + ( 30 + 5k1 + k 2 ) = 0
2

6 + k1 s + k2
2
For ω n = 10 rad / sec, 30 + 5k1 + k 2 = ω n = 100. Thus 5k1 + k 2 = 70
k2
(b) For ζ = 0.707, 2ζω n = 1 + k 2 . Thus ω n = 1 + .
1.414
(1 + k ) 2
ωn = = 30 + 5k1 + k 2 k 2 = 59 + 10 k1
2 2 2
Thus
2
(c) For ω n = 10 rad / sec and ζ = 0.707,
5k1 + k 2 = 100 and 1 + k 2 = 2ζω n = 14.14 Thus k 2 = 1314

.
Solving for k1 , we have k1 = 1137

. .
(d) The closed‐loop transfer function is
Y (s) 5 5
= =
s + ( k 2 + 1) s + ( 30 + 5k1 + k 2 ) s + 14.14 s + 100
2 2
R(s)
5
For a unit‐step input, lim y ( t ) = lim sY ( s ) = = 0.05
t →∞ s→ 0 100
(e) For zero steady‐state error due to a unit‐step input,
30 + 5k1 + k 2 = 5 Thus 5k1 + k 2 = −25
5‐30

Parameter Plane k1 versus k 2 :
5‐31 (a) Closed‐Loop Transfer Function (b) Characteristic Equation:
Y (s) 100 ( K P + K D s ) 2
= s + 100 K D s + 100 K P = 0
s + 100 K D s + 100 K P
2
R(s)
The system is stable for K P > 0 and K D > 0.
(b) For ζ = 1, 2ζω n = 100 K D .
ω n = 10 K P Thus 2ω n = 100 K D = 20 K P K D = 0.2 K P
(c) See parameter plane in part (g).
5‐31

(d) See parameter plane in part (g).
−2
(e) Parabolic error constant K a = 1000 sec
K a = lim s G ( s ) = lim100 ( K P + K D s ) = 100 K P = 1000 Thus K P = 10

2

s→0 s →0
(f) Natural undamped frequency ω n = 50 rad/sec.
ω n = 10 K P = 50 Thus K P = 25

(g) When K P = 0,
100 K D s 100 K D
G( s) = 2
= (pole‐zero cancellation)
s s

5‐32

5‐32 (a) Forward‐path Transfer Function:
Y (s) KK i 10 K
G ( s ) = = =
E (s) s [ Js (1 + Ts ) + K i K t ] (
s 0.001s + 0.01s + 10 K t
2
)
K 1 Kt
When r ( t ) = tu s ( t ), K v = lim sG ( s ) = e ss = =
s→ 0 Kt Kv K

(b) When r(t) = 0
Y (s) 1 + Ts 1 + 0.1s
= =
Td ( s ) s [ Js (1 + Ts ) + K i K t ] + KK i (
s 0.001s + 0.01s + 10 K t + 10 K
2
)
1 1
For Td ( s ) = lim y ( t ) = lim sY ( s ) = if the system is stable.
s t →∞ s→ 0 10 K
(c) The characteristic equation of the closed‐loop system is
3 2
0.001s + 0.01s + 0.1s + 10 K = 0
The system is unstable for K > 0.1. So we can set K to just less than 0.1. Then, the minimum value of
the steady‐state value of y(t) is
1 +
=1
10 K K = 0.1
−
However, with this value of K, the system response will be very oscillatory. The maximum overshoot
will be nearly 100%.
(d) For K = 0.1, the characteristic equation is
3 2 3 2 4
0.001s + 0.01s + 10 Kt s + 1 = 0 or s + 10 s + 10 Kt s + 1000 = 0
5‐33

For the two complex roots to have real parts of −2/5. we let the characteristic equation be written as
(
( s + a ) s + 5s + b = 0
2
) or s + ( s + 5) s + (5a + b) s + ab = 0
3 2
4
Then, a + 5 = 10 a=5 ab = 1000 b = 200 5a + b = 10 Kt Kt = 0.0225
The three roots are: s = − a = −5 s = − a = −5 s = −2.5 ± j13.92
. . . . .
5-33) Rise time: 0.56
.
Peak time: 0.785
√ .
.
Maximum overshoot: . 0.095
.
Settling time: 0 0.69
.
Ö 1.067
.
5‐34

5-34)
K
s ( Js + a + KK f )
Ö exp 0.2 Î ξ = 0.456

Ö 0.1 Î ω n = 0.353
0.125
Ö
2 5.42
. .
Ö 5.49
.
Ö 19.88
5‐35

5-35) a)
6.5
1
6.5 2
.
Ö 1
Ö 6.5 1
Ö
. .
Substituting into equation (2) gives:

7.5 6.5
6.5 6.5
Since the system is multi input and multi output, there are 4 transfer functions as:
, , ,
To find the unit step response of the system, let’s consider

6.5
6.5 2
6.5 √6.5
1 1
2 1
2 2√6.5
By looking at the Laplace transform function table:
1
1 sin 1
1
where cos
b)
5‐36

Therefore:
As a result:
which means:
1
1
The unit step response is:
1
1
Therefore as a result:
1
1 sin 1
1
where ω n = 1 and ξ ω n = 1 / 2
c)
Therefore:
1 1
As a result, the step response of the system is:
1
1
By looking up at the Laplace transfer function table:
5‐37

where ω n = 1, and 2ξ ω n = 1 Î ξ = 1 / 2
sin 1
where θ = cos-1 (2ξ 2 -1) = cos-1(0.5), therefore,
2 √3
1 sin
√3 2
5-36) MATLAB CODE

(a)
clear all
Amat=[-1 -1;6.5 0]
Bmat=[1 1;1 0]
Cmat=[1 0;0 1]
Dmat=[0 0;0 0]
disp(' State-Space Model is:')
Statemodel=ss(Amat,Bmat,Cmat,Dmat)
[mA,nA]=size(Amat);
rankA=rank(Amat);
disp(' Characteristic Polynomial:')
chareq=poly(Amat);
% p = poly(A) where A is an n-by-n matrix returns an n+1 element

%row vector whose elements are the coefficients of the characteristic
%polynomialdet(sI-A). The coefficients are ordered in descending powers.
[mchareq,nchareq]=size(chareq);
syms 's';
poly2sym(chareq,s)
disp(' Equivalent Transfer Function Model is:')
Hmat=Cmat*inv(s*eye(2)-Amat)*Bmat+Dmat
Since the system is multi input and multi output, there are 4 transfer functions as:
, , ,
To find the unit step response of the system, let’s consider
6.5
6.5
5‐38

Let’s obtain this term and find Y2(s) time response for a step input.
H22=Hmat(2,2)
ilaplace(H22/s)
Pretty(H22)
H22poly=tf([13/2],chareq)
step(H22poly)
H22 =
13/(2*s^2+2*s+13)
ans =
1-1/5*exp(-1/2*t)*(5*cos(5/2*t)+sin(5/2*t))
13
---------------
2
2 s + 2 s + 13
Transfer function:
6.5
-------------
s^2 + s + 6.5
To find the step response H11, H12, and H21 follow the same procedure.
Other parts are the same.
5‐39

5-37) Impulse response:

.
a) and 1 , therefore,
.
sin 1
1
b)
sin 1
1
c)
1
1 sin 1
1
where α = cos-1 ξ
5-38) Use the approach in 5-36 except:

H22=Hmat(2,2)
ilaplace(H22)
Pretty(H22)
H22poly=tf([13/2],chareq)
impulse(H22poly)
H22 =
13/(2*s^2+2*s+13)
ans =
13/5*exp(-1/2*t)*sin(5/2*t)
13
---------------
2
2 s + 2 s + 13
Transfer function:
6.5
-------------
s^2 + s + 6.5
5‐40

Other parts are the same.
5-39) a) The displacement of the bar is:
sin
Then the equation of motion is:
As x is a function of θ and changing with time, then
cos
If θ is small enough, then sin θ ≈ θ and cos θ ≈ 1. Therefore, the equation of motion is
rewritten as:
5‐41

(b) To find the unit step response, you can use the symbolic approach shown in Toolbox 2-1-1:
clear all
%s=tf(‘s’);
syms s B L K
Theta=B*s/s/L/(B*s+K)
ilaplace(Theta)
Theta =
B/L/(B*s+K)
ans =
1/L*exp(-K*t/B)
Alternatively, assign values to B L K and find the step response – see solution to problem 5-36.
5‐40 (a) Kt = 10000 oz - in / rad
The Forward‐Path Transfer Function:
9 × 10 K
12
G(s) =

(
s s + 5000 s + 1.067 × 10 s + 50.5 × 10 s + 5.724 × 10
4 3 7 2 9 12
)
9 × 10 K
12
=
s ( s + 116)( s + 4883)( s + 41.68 + j 3178.3)( s + 41.68 − j 3178.3)
Routh Tabulation:
5 7 12
s 1 1.067 × 10 5.724 × 10
4 9 12
s 5000 50.5 × 10 9 × 10 K

3 5 12 9
s 5.7 × 10 5.72 × 10 − 18
. × 10 K 0
2 8 7 12
s 2.895 × 10 + 1579
. × 10 K 9 × 10 K
13 12 9 2
1 16.6 × 10 + 8.473 × 10 K − 2.8422 × 10 K
s
29 + 1579
. K
0 12
s 9 × 10 K
5‐42


1 2
From the s row, the condition of stability is 165710 + 8473 K − 2.8422 K > 0
2
or K − 298114
. K − 58303.427 < 0 or ( K + 19.43)( K − 3000.57 ) < 0
Stability Condition: 0 < K < 3000.56
The critical value of K for stability is 3000.56. With this value of K, the roots of the characteristic
equation are: −4916.9, −41.57 + j3113.3, −41.57 + j3113.3, −j752.68, and j752.68
(b) K L = 1000 oz‐in/rad. The forward‐path transfer function is
9 × 10 K
11
G(s) =

(
s s + 5000 s + 1.582 × 10 s + 5.05 × 10 s + 5.724 × 10
4 3 6 2 9 11
)
9 × 10 K
11
=
s (1 + 116.06)( s + 4882.8)( s + 56.248 + j1005)( s + 56.248 − j1005)
(c) Characteristic Equation of the Closed‐Loop System:
5 4 6 3 9 2 11 11
s + 5000 s + 1.582 × 10 s + 5.05 × 10 s + 5.724 × 10 s + 9 × 10 K = 0

Routh Tabulation:
5 6 11
s 1 1.582 × 10 5.724 × 10
4 9 11
s 5000 5.05 × 10 9 × 10 K

3 5 11 8
s 5.72 × 10 5.724 × 10 − 18
. × 10 K 0
2 7 6 11
s 4.6503 × 10 + 15734
. × 10 K 9 × 10 K
18 15 14 2
1 26.618 × 10 + 377.43 × 10 K − 2.832 × 10 K
s 7 6
4.6503 × 10 + 15734
. × 10 K
0 11
s 9 × 10 K
5‐43

1 4 2
From the s row, the condition of stability is 26.618 × 10 + 3774.3 K − 2.832 K > 0
2
Or, K − 1332.73 K − 93990 < 0 or ( K − 1400 )( K + 67.14 ) < 0
Stability Condition: 0 < K < 1400
The critical value of K for stability is 1400. With this value of K, the characteristic equation root are:
−4885.1, −57.465 + j676, −57.465 − j676, j748.44, and −j748.44
(c) K L = ∞.
Forward‐Path Transfer Function:
nK s K i K
G(s) = JT = J m + n J L
2

s ⎡⎣ La J T s + ( Ra J T + Rm La ) s + Ra Bm + K i K b ⎤⎦
2
891100 K 891100 K
= =
(
s s + 5000 s + 566700
2
) s ( s + 116)( s + 4884)
Characteristic Equation of the Closed‐Loop System:

3 2
s + 5000 s + 566700 s + 891100 K = 0

Routh Tabulation:
3
s 1 566700
2
s 5000 891100 K

1
s 566700 − 178.22 K
0
s 8991100 K
1
From the s row, the condition of K for stability is 566700 − 178.22K > 0.
5‐44

Stability Condition: 0 < K < 3179.78
The critical value of K for stability is 3179.78. With K = 3179.78, the characteristic equation roots are
−5000, j752.79, and −j752.79.
When the motor shaft is flexible, K L is finite, two of the open‐loop poles are complex. As the shaft
becomes stiffer, K L increases, and the imaginary parts of the open‐loop poles also increase. When
K L = ∞ , the shaft is rigid, the poles of the forward‐path transfer function are all real. Similar effects
are observed for the roots of the characteristic equation with respect to the value of K L .
5‐41 (a)
100( s + 2 )
Gc ( s ) = 1 G( s) = 2
K p = lim G ( s ) = −200
s −1 s→ 0
When d(t) = 0, the steady‐state error due to a unit‐step input is
1 1 1
e ss = = =− = −0.005025
1+ K p 1 − 200 199
(b)
s +α 100( s + 2)( s + α )
Gc ( s ) = G(s) = Kp = ∞ ess = 0
s (
s s −1
2
)

(c)
α =5 maximum overshoot = 5.6%
α = 50 maximum overshoot = 22%
α = 500 maximum overshoot = 54.6%
5‐45

As the value of α increases, the maximum overshoot increases because the damping effect of the
zero at s = −α becomes less effective.
Unit‐Step Responses:
1
(d) r ( t ) = 0 and Gc ( s ) = 1. d ( t ) = u s ( t ) D( s ) =
s
System Transfer Function: (r = 0)

Y ( s) 100( s + 2)
= 3 2

D( s ) r =0
s + 100 s + (199 + 100α ) s + 200α
Output Due to Unit‐Step Input:
100( s + 2 )
Y ( s) =
3 2
s s + 100 s + (199 + 100α ) s + 200α
200 1
y ss = lim y ( t ) = lim sY ( s ) = =
t →∞ s→ 0 200α α
(e) r ( t ) = 0, d ( t ) = u s ( t )
s +α
Gc ( s ) =
s
5‐46

System Transfer Function [r(t) = 0]
Y ( s) 100 s( s + 20 1
= 3 2
D( s ) =
D( s ) r =0
s + 100 s + (199 + 100α ) s + 200α s

y ss = lim y ( t ) = lim sY ( s ) = 0
t →∞ s→ 0
(f)
Y ( s) 100 s( s + 2 )
α =5 = 3 2
D( s ) r =0
s + 100 s + 699 s + 1000
Y ( s) 100 s( s + 2)
α = 50 = 3 2

D( s ) r =0
s + 100 s + 5199 s + 10000
Y ( s) 100 s( s + 2)
α = 5000 = 3 2
D( s ) r =0
s + 100 s + 50199 s + 100000

Unit‐Step Responses:
(g) As the value of α increases, the output response y(t) due to r(t) becomes more oscillatory, and the
overshoot is larger. As the value of α increases, the amplitude of the output response y(t) due to d(t)
becomes smaller and more oscillatory.
5‐47

5‐42 (a) Forward‐Path Transfer function: Characteristic Equation:
H ( s) 10 N N 2
G ( s ) = = ≅ s + s + N = 0
E ( s) s( s + 1)( s + 10) s( s + 1)

2
N=1: Characteristic Equation: s + s + 1 = 0 ζ = 0.5 ω n = 1 rad/sec.
− πζ
1−ζ
2
π
Maximum overshoot = e = 0.163 (16.3%) Peak time t max = = 3.628 sec.
2
ω n 1− ζ
2
N=10: Characteristic Equation: s + s = 10 = 0 ζ = 0.158 ω n = 10 rad/sec.
− πζ
1−ζ
2
π
Maximum overshoot = e = 0.605 (60.5%) Peak time t max = = 1006
. sec.
2
ω n 1− ζ
(b) Unit‐Step Response: N = 1

5‐48


Second‐order System Third‐order System
Maximum overshoot 0.163 0.206
Peak time 3.628 sec. 3.628 sec.
Unit‐Step Response: N = 10
Second‐order System Third‐order System
Maximum overshoot 0.605 0.926
Peak time 1.006 sec. 1.13 sec.

5‐49

5‐43 Unit‐Step Responses:
When Tz is small, the effect is lower overshoot due to improved damping. When Tz is very large, the
overshoot becomes very large due to the derivative effect. Tz improves the rise time, since 1 + Tz s is a
derivative control or a high‐pass filter.
5‐50

5‐44 Unit‐Step Responses
1
The effect of adding the pole at s = − to G(s) is to increase the rise time and the overshoot. The system is
Tp
less stable. When T p > 0.707 , the closed‐loop system is stable.

5‐51

5-45) You may use the ACSYS software developed for this book. For description refer to Chapter 9. We
use a MATLAB code similar to toolbox 2-2-1 and those in Chapter 5 to solve this problem.
(a) Using Toolbox 5-9-3

clear all
num = [];
den = [0 -0.55 -1.5];
G=zpk(num,den,1)
t=0:0.001:15;
step(G,t);
hold on;
for Tz=[1 5 20];
t=0:0.001:15;
num = [-1/Tz];
den = [0 -0.55 -1.5];
G=zpk(num,den,1)
step(G,t);
hold on;
end
xlabel('Time(secs)')
ylabel('y(t)')
title('Unit-step responses of the system')
Zero/pole/gain:
1
------------------
s (s+0.55) (s+1.5)
Zero/pole/gain:
(s+1)
------------------
s (s+0.55) (s+1.5)
Zero/pole/gain:
(s+0.2)
------------------
s (s+0.55) (s+1.5)
Zero/pole/gain:
(s+0.05)
------------------
s (s+0.55) (s+1.5)
5‐52

(b)
clear all
for Tz=[0 1 5 20];
t=0:0.001:15;
num = [Tz 1];
den = [1 2 2];
G=tf(num,den)
step(G,t);
hold on;
end
ylabel('y(t)')
Transfer function:
1
-------------
s^2 + 2 s + 2
Transfer function:
s + 1
-------------
s^2 + 2 s + 2
Transfer function:
5 s + 1
-------------
s^2 + 2 s + 2
5‐53

Transfer function:
20 s + 1
-------------
s^2 + 2 s + 2
Follow the same procedure for other parts.
5-46) Since the system is linear we use superposition to find Y, for inputs X and D
First, consider D = 0
k K
Js + a
Kf
Then
;
1
5‐54

Kk
s ( Js + a + KK f k )

According to above block diagram:
1
1 1
Now consider X = 0, then:
k
K Kf
+1
Js + a s
Accordingly,
and
In this case, E(s) = D-KY(s)

1
1 1
5‐55

Now the steady state error can be easily calculated by:

1
1/ : lim
1
1
lim lim 0
1
1 1
: lim –
1
1
lim
1
lim –
(c) The overall response is obtained through superposition
Y ( s) = Y ( s) D =0 + Y ( s ) X =0
y (t ) = y (t ) d (t ) =0 + y (t ) x (t ) =0
MATLAB
clear all
syms s K k J a Kf
X=1/s;
D=1/s^2
Y=K*k*X/(s*(J*s+a+K*k*Kf)+K*k)+k*(Kf+s)*D/(s*(J*s+a)+k*(Kf+s)*K)
ilaplace(Y)
D =
1/s^2
Y =
K*k/s/(s*(J*s+a+K*k*Kf)+K*k)+k*(Kf+s)/s^2/(s*(J*s+a)+k*(Kf+s)*K)
ans =
1+t/K+1/k/K^2/Kf/(a^2+2*a*K*k+K^2*k^2-
4*J*K*k*Kf)^(1/2)*sinh(1/2*t/J*(a^2+2*a*K*k+K^2*k^2-4*J*K*k*Kf)^(1/2))*exp(-
1/2*(a+K*k)/J*t)*(a^2+a*K*k-2*J*K*k*Kf)-cosh(1/2*t/J*(a^2+2*a*K*k*Kf+K^2*k^2*Kf^2-
4*J*K*k)^(1/2))*exp(-1/2*(a+K*k*Kf)/J*t)-(a+K*k*Kf)/(a^2+2*a*K*k*Kf+K^2*k^2*Kf^2-
4*J*K*k)^(1/2)*sinh(1/2*t/J*(a^2+2*a*K*k*Kf+K^2*k^2*Kf^2-4*J*K*k)^(1/2))*exp(-
1/2*(a+K*k*Kf)/J*t)+1/k/K^2/Kf*a*(-1+exp(-
1/2*(a+K*k)/J*t)*cosh(1/2*t/J*(a^2+2*a*K*k+K^2*k^2-4*J*K*k*Kf)^(1/2)))
5‐56

∞
5-47) (a) Find the ∫ e(t )dt when e(t) is the error in the unit step response.
0
∞
As the system is stable then ∫0
e(t )dt will converge to a constant value:
∞ E ( s)
∫
0
e(t )dt = lim s
s →0 s
= lim E ( s )
s →0
Y ( s) G ( s) ( A s + 1)( A2 s + 1)...( An s + 1)
= = 1 n≤m
X ( s) 1 + G ( s) ( B1 s + 1)( B2 s + 1)...( Bm s + 1)
Try to relate this to Equation (5-40).
G ( s) 1
E ( s ) = X ( s) − Y ( s) = X (s) − X (s) = X (s)
1 + G ( s) 1 + G (s)
⎛ ( A s + 1)( A2 s + 1)...( An s + 1) ⎞ ⎛ ( B1s + 1)( B2 s + 1)...( Bm s + 1) − ( A1s + 1)( A2 s + 1)...( An s + 1) ⎞
= ⎜1 − 1 ⎟ X (s) = ⎜ ⎟ X (s)
⎝ ( B1s + 1)( B2 s + 1)...( Bm s + 1) ⎠ ⎝ ( B1s + 1)( B2 s + 1)...( Bm s + 1) ⎠
1 ⎛ ( B s + 1)( B2 s + 1)...( Bm s + 1) − ( A1s + 1)( A2 s + 1)...( An s + 1) ⎞
lim E ( s ) = lim ⎜ 1 ⎟
s →0 s →0 s
⎝ ( B1s + 1)( B2 s + 1)...( Bm s + 1) ⎠
= ( B1 + B2 + L + Bm ) − ( A1 + A2 + L + An )
⎛ ( B1s + 1)( B2 s + 1)...( Bm s + 1) ⎞

G ( s) = ⎜ ⎟ −1
⎝ ( B1s + 1)( B2 s + 1)...( Bm s + 1) − ( A1s + 1)( A2 s + 1)...( An s + 1) ⎠
⎛ ( A1s + 1)( A2 s + 1)...( An s + 1) ⎞
G ( s) = ⎜ ⎟
⎝ ( B1s + 1)( B2 s + 1)...( Bm s + 1) − ( A1s + 1)( A2 s + 1)...( An s + 1) ⎠
1 ⎛ ( A s + 1)( A2 s + 1)...( An s + 1) ⎞
ess = lim s ⎜ 1 − 1 ⎟=0
s →0 s ⎝ ( B1s + 1)( B2 s + 1)...( Bm s + 1) ⎠
5‐57

1 1
(b) Calculate =
K lim sG ( s )
s→ 0
Recall
G ( s) 1
E ( s ) = X ( s) − Y ( s) = X ( s) − X (s) = X (s)
1 + G(s) 1 + G(s)
Hence

1 1 1 1
lim E ( s ) = lim X ( s ) = lim = lim =
s →0 s →0 1 + G ( s ) s → 0 s + sG ( s ) s →0 sG ( s ) Kv
Ramp Error Constant
5‐58

5-48)
10 10
25 10 35 25 10
Comparing with the second order prototype system and matching denominators:
25 10
35 2
exp 0.25 1.386 0.210

1 1
3.2
0.1 80, 0 0.69
Let ξ = 0.4 and ω n = 80
Then
2 35 29
56.25
clear all
p=29;
K=56.25;
num = [10 10*K];
den = [1 35+p 25*p+10*K];
G=tf(num,den)
step(G);
ylabel('y(t)')
Transfer function:
10 s + 562.5
-----------------
s^2 + 64 s + 1288
5‐59

5-49) According to the maximum overshoot:
which should be less than t, then
or 1
jω
π /t
ωn 1 − ξ 2
−π /t ωn 1 − ξ 2
5‐60

5-50) Using a 2nd order prototype system format, from Figure 5-15, ωn is the radial distance from the
complex conjugate roots to the origin of the s-plane, then ωn with respect to the origin of the
shown region is ωn≈ 3.6.
Therefore the natural frequency range in the region shown is around 2.6 ≤ ωn ≤ 4.6
On the other hand, the damping ratio ζ at the two dashed radial lines is obtained from:
ζ cos /2 sin
ζ cos /2 sin
The approximation from the figure gives:
ζ 0.56
ζ 0.91
Therefore 0.56 ≤ ζ ≤ 0.91
b)
As Kp=2, then:
2
2 1 2
If the roots of the characteristic equations are assumed to be lied in the centre of the shown
region:
3 2
Ö 6 13 0
3 2
Comparing with the characteristic equation:
2 1 6 2
2 13 3.25
c) The characteristic equation
2 0
is a second order polynomial with two roots. These two roots can be determined by two terms
2(p+KKp) and KKpKI which includes four parameters. Regardless of the p and Kp values, we can
always choose K and so that to place the roots in a desired location.
5‐61

5-51) a)
By substituting the values:
0.2
0.109
b) Speed of the motor is
0.2
0.109
.
lim 10 19.23
.
(c)
0.2
0.109
d)
0.2
0.109
0.2
0.109
As a result:
0.2
0.109 0.2
e) As exp 0.2 , then, ξ = 0.404

.
According to the transfer function, 2 0.109, then 0.14 /
.
where 0.2 Ö K<0.0845
. . .
f) As , then 0.6,
as 0.2 , therefore; 1
5‐62

g) MATLAB
clear all
for K=[0.5 1 2];
t=0:0.001:15;
num = [0.2*K];
den = [1 0.109 0.2*K];
G=tf(num,den)
step(G,t);
hold on;
end
ylabel('y(t)')
Transfer function:
0.1
-------------------
s^2 + 0.109 s + 0.1
Transfer function:
0.2
-------------------
s^2 + 0.109 s + 0.2
Transfer function:
0.4
-------------------
s^2 + 0.109 s + 0.4
5‐63

Rise time decreases with K increasing.
Overshoot increases with K.
5‐64

5-52) exp 0.1, therefore, ξ = 0.59
. .
As , then, 3.62
. .
1 3
3 3
Therefore:
3 3 2
If p is a non-dominant pole; therefore comparing both sides of above equation and:
2 3
2 3
If we consider p = 10a (non-dominant pole), ξ = 0.6 and ωn = 4, then:
4.8 10 3
48 16 3
160 160
8.3
89.8
83
clear all
K=160;
a=8.3;
b=89.8;
p=83;
num = [K K*a];
den = [1 3+b 3*b+K K*a];
G=tf(num,den)
step(G);
ylabel('y(t)')
Transfer function:
160 s + 1328
-------------------------------
s^3 + 92.8 s^2 + 429.4 s + 1328
5‐65

Both Overshoot and settling time values are met. No need to adjust parameters.
5-53) For the controller and the plant to be in series and using a unity feedback loop we have:
MATLAB
USE toolbox 5-8-3
clear all
num=[‐1 ‐2 ‐3];
denom=[‐3+sqrt(9‐40) ‐3‐sqrt(9‐40) ‐0.02+sqrt(.004‐.07) ‐0.02‐sqrt(.004‐.07) ‐10];
G=zpk(num,denom,60)
rlocus(G)

Zero/pole/gain:
60 (s+1) (s+2) (s+3)
‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐
(s+10) (s^2 + 0.04s + 0.0664) (s^2 + 6s + 40)

5‐66


Note the system has two dominant complex poles close to the imaginary axis. Lets zoom in the root locus diagram and
use the cursor to find the parameter values.
5‐67


As shown for K=0.933 the dominant closed loop poles are at ‐0.46±j 0.626 AND OVERSHOOT IS
ALMOST 10%.

Increasing K will push the poles closer towards less dominant zeros and poles. As a process the design process becomes less
trivial and more difficult.

To confirm use
M=feedback(G*.933,1) %See toolbox 5‐4‐2
step(M)
Zero/pole/gain:
55.98 (s+3) (s+2) (s+1)
‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐
(s+7.048) (s^2 + 0.9195s + 0.603) (s^2 + 8.072s + 85.29)
5‐68


To reduce rise time, the poles have to move to left to make the secondary poles more dominant. As
a result the little bump in the left hand side of the above graph should rise. Try K=3:

>> M=feedback(G*3,1)

Zero/pole/gain:
180 (s+3) (s+2) (s+1)
‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐
(s+5.01) (s^2 + 1.655s + 1.058) (s^2 + 9.375s + 208.9)

>> step(M)
**Try a higher K value, but looking at the root locus and the time plots, it appears that the
overshoot and rise time criteria will never be met simultaneously.
5‐69


K=5
M=feedback(G*5,1) %See toolbox 5-4-2

step(M)
Zero/pole/gain:
300 (s+3) (s+2) (s+1)
---------------------------------------------------------
(s+4.434) (s^2 + 1.958s + 1.252) (s^2 + 9.648s + 329.1)
5‐70

5‐71

5-54) Forward‐path Transfer Function:
M ( s) K
G( s) = = 3 2

1− M ( s) s + ( 20 + a ) s + ( 200 + 20a ) s + 200a − K
For type 1 system, 200a − K = 0 Thus K = 200a
Ramp‐error constant:
K 200 a
K v = lim sG ( s ) = = = 5 Thus a = 10 K = 2000
s→ 0 200 + 20a 200 + 20 a
MATLAB Symbolic tool can be used to solve above. We use it to find the roots for the next part:
>> syms s a K
>>solve(5*200+5*20*a‐200a)
ans =
10
>> D=(s^2+20*s+200)*s+a))
D =
(s^2+20*s+200)*(s+a)
>> expand(D)
ans =
s^3+s^2*a+20*s^2+20*s*a+200*s+200*a
>> solve(ans,s)
ans =
‐a
‐10+10*i
‐10‐10*i
The forward‐path transfer function is The controller transfer function is
5‐72

G ( s ) =
2000
Gc ( s ) =
G(s)
=
(
20 s + 10 s + 100
2
)
(
s s + 30 s + 400
2
) Gp (s) (s 2
+ 30 s + 400 )

The maximum overshoot of the unit‐step response is 0 percent.
MATLAB
clear all
K=2000;
a=10;
num = [];
den = [-10+10i -10-10i -a];
G=zpk(num,den,K)
step(G);
ylabel('y(t)')
Zero/pole/gain:
2000
-------------------------
(s+10) (s^2 + 20s + 200)
Clearly PO=0.
5‐73

5-55)
Forward‐path Transfer Function:
M ( s) K
G( s) = = 3 2

1− M ( s) s + ( 20 + a ) s + ( 200 + 20a ) s + 200a − K
K 200 a
K v = lim sG ( s ) = = = 9 Thus a = 90 K = 18000
s→ 0 200 + 20a 200 + 20 a
MATLAB Symbolic tool can be used to solve above. We use it to find the roots for the next part:
>> syms s a K
solve(9*200+9*20*a‐200*a)
ans =
90
>>D=(s^2+20*s+200)*s+a))
D =
(s^2+20*s+200)*(s+a)
>> expand(D)
ans =
s^3+s^2*a+20*s^2+20*s*a+200*s+200*a
>> solve(ans,s)
ans =
‐a
‐10+10*i
‐10‐10*i
5‐74

G ( s ) =
18000
Gc ( s ) =
G(s)
=
(
180 s + 10 s + 100
2
)
(
s s + 110 s + 2000
2
) Gp ( s) ( s + 110s + 2000 )
2
The maximum overshoot of the unit‐step response is 4.3 percent.
From the expression for the ramp‐error constant, we see that as a or K goes to infinity, Kv approaches 10.
Thus the maximum value of Kv that can be realized is 10. The difficulties with very large values of K and
a are that a high‐gain amplifier is needed and unrealistic circuit parameters are needed for the controller.
clear all
K=18000;
a=90;
num = [];
den = [-10+10i -10-10i -a];
G=zpk(num,den,K)
step(G);
ylabel('y(t)')
Zero/pole/gain:
18000
-------------------------
(s+90) (s^2 + 20s + 200)
PO is less than 4.
5‐75

5-56) (a) Ramp‐error Constant:
MATLAB
clear all
syms s Kp Kd kv
Gnum=(Kp+Kd*s)*1000
Gden= (s*(s+10))
G=Gnum/Gden
Kv=s*G
s=0
eval(Kv)

Gnum =
1000*Kp+1000*Kd*s

Gden =
s*(s+10)

G =
(1000*Kp+1000*Kd*s)/s/(s+10)

Kv =
(1000*Kp+1000*Kd*s)/(s+10)

s =
0

ans =
100*Kp

1000 ( K P + K D s ) 1000 K P
K v = lim s = = 100 K P = 1000 Thus K P = 10
s →0 s ( s + 10) 10
Kp=10
clear s
syms s
Mnum=(Kp+Kd*s)*1000/s/(s+10)
Mden=1+(Kp+Kd*s)*1000/s/(s+10)
5‐76

Kp =
10
Mnum =
(10000+1000*Kd*s)/s/(s+10)
Mden =
1+(10000+1000*Kd*s)/s/(s+10)
ans =
(s^2+10*s+10000+1000*Kd*s)/s/(s+10)
s + (10 + 1000 K D ) s + 1000 K P = 0

2
Characteristic Equation:
Match with a 2nd order prototype system
ω n = 1000 K P = 10000 = 100 rad/sec 2ζω n = 10 + 1000 K D = 2 × 0.5 × 100 = 100
solve(10+1000*Kd‐100)
ans =
9/100
90
Thus KD = = 0.09
1000
Use the same procedure for other parts.
(b) For K v = 1000 and ζ = 0.707 , and from part (a), ω n = 100 rad/sec,
131.4
2ζω n = 10 + 1000 K D = 2 × 0.707 × 100 = 141.4 Thus KD = = 0.1314
1000
(c) For K v = 1000 and ζ = 1.0 , and from part (a), ω n = 100 rad/sec,
190
2ζω n = 10 + 1000 K D = 2 × 1 × 100 = 200 Thus KD = = 0.19
1000
5‐77

5-57) The ramp‐error constant:
1000 ( K P + K D s )
K v = lim s = 100 K P = 10, 000 Thus K P = 100
s →0 s ( s + 10)
1000 (100 + K D s )
The forward‐path transfer function is: G(s) =
s ( s + 10)
clear all
for KD=0.2:0.2:1.0;
num = [-100/KD];
den = [0 -10];
G=zpk(num,den,1000);
M=feedback(G,1)
step(M);
hold on;
end
ylabel('y(t)')
Zero/pole/gain:
1000 (s+500)
‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐
(s^2 + 1010s + 5e005)
Zero/pole/gain:
1000 (s+250)
‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐
(s+434.1) (s+575.9)
Zero/pole/gain:
1000 (s+166.7)
‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐
(s+207.7) (s+802.3)
5‐78

Zero/pole/gain:
1000 (s+125)
‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐
(s+144.4) (s+865.6)
Zero/pole/gain:
1000 (s+100)
‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐
(s+111.3) (s+898.7)

Use the cursor to obtain the PO and tr values.
For part b the maximum value of KD results in the minimum overshoot.
5‐79

5-58) (a) Forward‐path Transfer Function:
4500 K ( K P + K D s )
G ( s ) = Gc ( s )G p ( s ) =
s ( s + 361.2)
4500 KK P
Ramp Error Constant: K v = lim sG ( s ) = = 12.458 KK P
s→ 0 361.2
1 0.0802
e ss = = ≤ 0.001 Thus KK P ≥ 80.2 Let K P = 1 and K = 80.2
Kv KK P
clear all
KP=1;
K=80.2;
figure(1)
num = [-KP];
den = [0 -361.2];
G=zpk(num,den,4500*K)
M=feedback(G,1)
step(M)
hold on;
for KD=00.0005:0.0005:0.002;
num = [-KP/KD];
den = [0 -361.2];
G=zpk(num,den,4500*K*KD)
M=feedback(G,1)
step(M)
end
ylabel('y(t)')
Zero/pole/gain:
360900 (s+1)
‐‐‐‐‐‐‐‐‐‐‐‐
s (s+361.2)
Zero/pole/gain:
360900 (s+1)
‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐
(s+0.999) (s+3.613e005)
Zero/pole/gain:
180.45 (s+2000)
‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐
s (s+361.2)
Zero/pole/gain:
180.45 (s+2000)
5‐80

‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐
(s^2 + 541.6s + 3.609e005)
Zero/pole/gain:
360.9 (s+1000)
‐‐‐‐‐‐‐‐‐‐‐‐‐‐
s (s+361.2)
Zero/pole/gain:
360.9 (s+1000)
‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐
(s^2 + 722.1s + 3.609e005)
Zero/pole/gain:
541.35 (s+666.7)
‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐
s (s+361.2)
Zero/pole/gain:
541.35 (s+666.7)
‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐
(s^2 + 902.5s + 3.609e005)
Zero/pole/gain:
721.8 (s+500)
‐‐‐‐‐‐‐‐‐‐‐‐‐
s (s+361.2)
Zero/pole/gain:
721.8 (s+500)
‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐
(s^2 + 1083s + 3.609e005)
5‐81

KD tr (sec) ts (sec) Max Overshoot (%)
0 0.00221 0.0166 37.1
0.0005 0.00242 0.00812 21.5
0.0010 0.00245 0.00775 12.2
0.0015 0.0024 0.0065 6.4
0.0016 0.00239 0.00597 5.6
0.0017 0.00238 0.00287 4.8
0.0018 0.00236 0.0029 4.0
0.0020 0.00233 0.00283 2.8
5‐82

5-59) The forward‐path Transfer Function: N = 20
200 ( K P + K D s )
G(s) =
s ( s + 1)( s + 10)
To stabilize the system, we can reduce the forward‐path gain. Since the system is type 1, reducing the
gain does not affect the steady‐state liquid level to a step input. Let K P = 0.05
200 ( 0.05 + K D s )
G(s) =
s ( s + 1)( s + 10)
ALSO try other Kp values and compare your results.
clear all
figure(1)
KD=0
num = [];
den = [0 -1 -10];
G=zpk(num,den,200*0.05)
M=feedback(G,1)
step(M)
hold on;
for KD=0.01:0.01:0.1;
KD
num = [-0.05/KD];
G=zpk(num,den,200*KD)
M=feedback(G,1)
step(M)
end
ylabel('y(t)')

KD =
0

Zero/pole/gain:
10
‐‐‐‐‐‐‐‐‐‐‐‐‐‐
s (s+1) (s+10)

Zero/pole/gain:
10
‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐
(s+10.11) (s^2 + 0.8914s + 0.9893)

KD =
0.0100

5‐83

Zero/pole/gain:
2 (s+5)
‐‐‐‐‐‐‐‐‐‐‐‐‐‐
s (s+1) (s+10)

Zero/pole/gain:
2 (s+5)
‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐
(s+9.889) (s^2 + 1.111s + 1.011)

KD =
0.0200

Zero/pole/gain:
4 (s+2.5)
‐‐‐‐‐‐‐‐‐‐‐‐‐‐
s (s+1) (s+10)

Zero/pole/gain:
4 (s+2.5)
‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐
(s+9.658) (s^2 + 1.342s + 1.035)

KD =
0.0300

Zero/pole/gain:
6 (s+1.667)
‐‐‐‐‐‐‐‐‐‐‐‐‐‐
s (s+1) (s+10)

Zero/pole/gain:
6 (s+1.667)
‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐
(s+9.413) (s^2 + 1.587s + 1.062)

KD =
0.0400

Zero/pole/gain:
8 (s+1.25)
‐‐‐‐‐‐‐‐‐‐‐‐‐‐
s (s+1) (s+10)

Zero/pole/gain:
8 (s+1.25)
‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐
(s+9.153) (s^2 + 1.847s + 1.093)

5‐84

KD =
0.0500

Zero/pole/gain:
10 (s+1)
‐‐‐‐‐‐‐‐‐‐‐‐‐‐
s (s+1) (s+10)

Zero/pole/gain:
10 (s+1)
‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐
(s+8.873) (s+1.127) (s+1)

KD =
0.0600

Zero/pole/gain:
12 (s+0.8333)
‐‐‐‐‐‐‐‐‐‐‐‐‐‐
s (s+1) (s+10)

Zero/pole/gain:
12 (s+0.8333)
‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐
(s+8.569) (s+1.773) (s+0.6582)

KD =
0.0700

Zero/pole/gain:
14 (s+0.7143)
‐‐‐‐‐‐‐‐‐‐‐‐‐‐
s (s+1) (s+10)

Zero/pole/gain:
14 (s+0.7143)
‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐
(s+8.232) (s+2.221) (s+0.547)

KD =
0.0800

Zero/pole/gain:
16 (s+0.625)
‐‐‐‐‐‐‐‐‐‐‐‐‐‐
s (s+1) (s+10)

Zero/pole/gain:
16 (s+0.625)
‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐
(s+7.85) (s+2.673) (s+0.4765)

5‐85

KD =
0.0900

Zero/pole/gain:
18 (s+0.5556)
‐‐‐‐‐‐‐‐‐‐‐‐‐‐
s (s+1) (s+10)

Zero/pole/gain:
18 (s+0.5556)
‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐
(s+7.398) (s+3.177) (s+0.4255)

KD =
0.1000

Zero/pole/gain:
20 (s+0.5)
‐‐‐‐‐‐‐‐‐‐‐‐‐‐
s (s+1) (s+10)

Zero/pole/gain:
20 (s+0.5)
‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐
(s+0.3861) (s+3.803) (s+6.811)

5‐86

Unit‐step Response Attributes:
KD ts (sec) Max Overshoot (%)
0.01 5.159 12.7
0.02 4.57 7.1
0.03 2.35 3.2
0.04 2.526 0.8
0.05 2.721 0
0.06 3.039 0
0.10 4.317 0
When K D = 0.05 the rise time is 2.721 sec, and the step response has no overshoot.
5‐87

5-60) (a) For e ss = 1,
200 ( K P + K D s )
K v = lim sG ( s ) = lim s = 20 K P = 1 Thus K P = 0.05
s →0 s→0 s ( s + 1)( s + 10)

200 ( 0.05 + K D s )
G ( s ) =
s ( s + 1)( s + 10)
Because of the choice of Kp this is the same as previous part.
5-61)
(a) Forward‐path Transfer Function:
⎛
100 ⎜ K P +
KI ⎞
⎟ 100 ( K P s + K I )
G ( s ) =
⎝ s ⎠ For K v = 10, K v = lim sG ( s ) = lim s = K I = 10
s + 10 s + 100
2
s →0 s →0
(
s s + 10 s + 100
2
)

Thus K I = 10.
100 ( K P s + K I ) 100 ( K P s + 10 )
Gcl ( s ) = =
s + 10 s + 100 s + 100 ( K P s + K I )
2 2
s + 10 s + 100(1 + K P ) s + 1000
3 3
(b) Let the complex roots of the characteristic equation be written as s = −σ + j15 and s = −σ − j15.
The quadratic portion of the characteristic equation is
2
(
s + 2σ s + σ + 225 = 0
2
)
s + 10 s + (100 + 100 K P ) s + 1000 = 0
3 2
The characteristic equation of the system is
The quadratic equation must satisfy the characteristic equation. Using long division and solve for zero
remainder condition.
5‐88

s + (10 − 2σ )
s + 2σ s + σ + 225 s + 10 s + (100 + 100 K P ) s + 1000

2 2 3 2

s + 2σ s + σ + 225 s
3 2
( 2
)
(
(10 − 2σ ) s + 100 K P − σ − 125 s + 1000
2
) 2

(10 − 2σ ) s + ( 20σ − 4σ ) s + (10 − 2σ ) ( s + 225 )
2 2 2
(100 K P )
+ 3σ − 20σ − 125 s + 2σ − 10σ + 450σ − 1250
2 3 2
3 2
For zero remainder, 2σ − 10σ + 450σ − 1250 = 0 (1)
2
and 100 K P + 3σ − 20σ − 125 = 0 (2)
The real solution of Eq. (1) is σ = 2.8555 . From Eq. (2),
2
125 + 20σ − 3σ
KP = = 15765
.
100
The characteristic equation roots are: s = −2.8555 + j15, − 2.8555 − j15, and s = −10 + 2σ = −4.289
(c) Root Contours:
Dividing both sides of s + 10 s + (100 + 100 K P ) s + 1000 = 0 by the terms that do not contain Kp we have:

3 2

100 K P s
1+ = 1 + Geq
s + 10 s + 100 s + 1000
3 2
100 K P s 100 K P s
Geq ( s ) = =
s + 10 s + 100 s + 1000
3 2
( s + 10 ) ( s 2
+ 100 )
Root Contours: See Chapter 9 toolbox 9‐5‐2 for more information
clear all
Kp =.001;
num = [100*Kp 0];
den = [1 10 100 1000];
rlocus(num,den)
5‐89

5‐90

⎛
100 ⎜ K P +
KI ⎞
⎟ 100 ( K P s + K I )
G ( s ) =
⎝ s ⎠ For K = 10, K v = lim sG ( s ) = lim s = K I = 10
s + 10 s + 100
2 v
s →0 s →0
(
s s + 10 s + 100
2
)
Thus the forward‐path transfer function becomes
100 (10 + K P s )
G(s) =
(
s s + 10 s + 100
2
)
100 ( K P s + K I ) 100 ( K P s + 10 )
Gcl ( s ) = =
s + 10 s + 100 s + 100 ( K P s + K I )
2 2
s + 10 s + 100(1 + K P ) s + 1000
3 3

clear all
for Kp=.4:0.4:2;
num = [100*Kp 1000];
den =[1 10 100 0];
[numCL,denCL]=cloop(num,den);
GCL=tf(numCL,denCL);
step(GCL)
hold on;
end

5‐91

Use the curser to find the maximum overshoot and rise time. For example when Kp = 2, PO=43 and tr100%=0.152
sec.
Transfer function:
200 s + 1000
‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐
s^3 + 10 s^2 + 300 s + 1000
5‐92

5-63)
100 ( K P s + K I )
G(s) =
(
s s + 10 s + 100
2
)
For K v = 100,
100 ( K P s + K I )
K v = lim sG ( s ) = lim s = K I = 100 Thus K I = 100.
s →0 s →0
(
s s + 10 s + 100
2
)
s + 10 s + (100 + 100 K P ) s + 100 K I = 0
3 2
(b) The characteristic equation is
Routh Tabulation:
3
s 1 100 + 100 K P
2
s 10 10,000
For stability, 100 K P − 900 > 0 Thus KP > 9
1
s 100 K P − 900 0
0
s 10,000
7. Activate MATLAB
9. Type in
Acsys
10. Then press the “transfer function Symbolic” and enter the Characteristic equation
11. Then press the “Routh Hurwitz” button
5‐93


RH =

[ 1, 100+100*kp]
[ 10, 10000]
[ ‐900+100*kp, 0]
[ (‐9000000+1000000*kp)/(‐900+100*kp), 0]

5‐94

Root Contours:
100 K P s 100 K P s
Geq ( s ) = 3 2
=
s + 10 s + 100 s + 10,000 ( s + 23.65)( s − 6.825 + j19.4)( s − 6.825 − j19.4)
Root Contours: See Chapter 9 toolbox 9‐5‐2 for more information
clear all
Kp =.001;
num = [100*Kp 0];
den = [1 10 100 10000];
rlocus(num,den)
5‐95

(c) K I = 100
100 ( K P s + 100 )
G(s) =
(
s s + 10 s + 100
2
)
The following maximum overshoots of the system are computed for various values of K P .
clear all
Kp=[15 20 22 24 25 26 30 40 100 1000];
[N,M]=size(Kp);
for i=1:M
num = [100*Kp(i) 10000];
den = [1 10 100 0];
GCL=tf(numCL,denCL);
figure(i)
step(GCL)
end

KP 15 20 22 24 25 26 30 40 100 1000
5‐96

ymax 1.794 1.779 1.7788 1.7785 1.7756 1.779 1.782 1.795 1.844 1.859
When KP = 25, minimum ymax = 1.7756
Use: close all to close all the figure windows.
5‐97

5-64) MATLAB solution is the same as 5-63.
100 ( K P s + K I ) 100 K I
G(s) = For K v = = 10, K I = 10
(
s s + 10 s + 100
2
) 100
(b) Characteristic Equation: ( )
s3 + 10 s 2 + 100 K + 1 s + 1000 = 0
P
Routh Tabulation:

3
s 1 100 + 100 K P
2
s 10 1000
For stability, KP > 0
1
s 100 K P 0
0
s 1000
Root Contours:
100 K P s
Geq ( s ) = 3 2

s + 10 s + 100 s + 1000

5‐98

(c) The maximum overshoots of the system for different values of K P ranging from 0.5 to 20 are
computed and tabulated below.
KP 0.5 1.0 1.6 1.7 1.8 1.9 2.0 3.0 5.0 10 20
ymax 1.393 1.275 1.2317 1.2416 1.2424 1.2441 1.246 1.28 1.372 1.514 1.642

When KP = 1.7, maximum ymax = 1.2416
5-65)
KDs + KPs + KI ⎛ KI 2 ⎞
2
KI
Gc ( s ) = K P + K D s + = = (1 + K D 1 s ) ⎜ K P 2 + ⎟
s s ⎝ s ⎠
where
K P = K P 2 + K D1 K I 2 K D = K D1 K P 2 KI = KI 2
G ( s ) = Gc ( s )G p ( s ) =
100 (K D s2 + KP s + KI )
(
s s + 10 s + 100
2
)
And rename the ratios: K D / K P = A, KI / KP = B
Thus
KI
K v = lim sG ( s ) = 100 = 100
s →0 100
K I = 100
5‐99

ForKD being sufficiently small:
Forward‐path Transfer Function: Characteristic Equation:
100 ( K P s + 100 )
G ( s ) = s + 10 s + (100 + 100 K P ) s + 10, 000 = 0
3 2

(
s s + 10 s + 100
2
)
For stability, Kp>9. Select Kp =10 and observe the response.
clear all
Kp=10;
num = [100*Kp 10000];
den =[1 10 100 0];
GCL=tf(numCL,denCL)
step(GCL)
Transfer function:
1000 s + 10000
-----------------------------
s^3 + 10 s^2 + 1100 s + 10000
Obviously by increasing Kp more oscillations will occur. Add KD to reduce oscillations.
clear all
Kp=10;
Kd=2;
num = [100*Kd 100*Kp 10000];
den =[1 10 100 0];
GCL=tf(numCL,denCL)
step(GCL)

5‐100

Transfer function:
200 s^2 + 1000 s + 10000
------------------------------
s^3 + 210 s^2 + 1100 s + 10000

Unit‐step Response
The rise time seems reasonable. But we need to increase Kp to improve approach to steady state.
Increase Kp to Kp=30.
clear all
Kp=30;
Kd=1;
num = [100*Kd 100*Kp 10000];
den =[1 10 100 0];
GCL=tf(numCL,denCL)
step(GCL)
Transfer function:
100 s^2 + 3000 s + 10000
------------------------------
s^3 + 110 s^2 + 3100 s + 10000

5‐101

To obtain a better response continue adjusting KD and KP.
5‐102

5-66) This problem has received extended treatment in Chapter 6, Control Lab – see Section 6-6.
For the sake simplicity, this problem we assume the control force f(t) is applied in parallel to the
spring K and damper B. We will not concern the details of what actuator or sensors are used.
Lets look at Figure 4-84 and equations 4-322 and 4-323.
xc
mc
mc
ks cs
ks cs x
xw m
mw
mw k c
kw kw cw
y y
cw
(a) (b) (c)
Figure 4-84 Quarter car model realization: (a) quarter car, (b) 2 degree of
freedom, and (c) 1 degree of freedom model.
The equation of motion of the system is defined as follows:
mx&&(t ) + cx& (t ) + kx (t ) = cy& (t ) + ky (t ) (4-322)
which can be simplified by substituting the relation z(t) = x(t)‐y(t) and non‐dimensionalizing the coefficients to
the form
z (t ) + 2ζω n z& (t ) + ω n 2 z (t ) = − &&

&& y (t ) (4-323)
The Laplace transform of Eq. (4‐323) yields the input output relationship
5‐103

Z (s) −1
= 2 (4-324)
Y&&( s ) s + 2ζω n s + ω n 2
Now let’s apply control – see section 6-6 for more detail.
x
m
k c
k f(t)
y
For simplicity and better presentation, we have scaled the control force as kf(t) we rewrite (4-324) as:
mx&&(t ) + cx& (t ) + kx(t ) = cy& (t ) + ky (t ) + kf (t )

z (t ) + 2ζωn z& (t ) + ωn 2 z (t ) = − &&
&& y (t ) + ωn 2 f (t )
(4-324)
s 2 + 2ζωn s + ωn 2 = − A( s ) + ωn 2 F ( s )
A( s ) = Y&&( s )
Setting the controller structure such that the vehicle bounce Z ( s ) = X ( s ) − Y ( s ) is minimized:
⎛ K ⎞
F (s) = 0 − ⎜ K P + K D s + I ⎟ Z (s)
⎝ s ⎠
Z (s) −1
=
A( s ) ⎛ K ⎞
s 2 + 2ζωn s + ωn 2 ⎜ 1 + K P + K D s + I ⎟
⎝ s ⎠
Z (s) −s
= 3
A( s ) s + 2ζωn s + ωn ( (1 + K P ) s + K D s 2 + K I )
2 2
See Equation (6-4).
For proportional control KD=KI=0.
5‐104

Pick ς = 0.707 and ωn = 1 for simplicity. This is now an underdamped system.
Use MATLAB to obtain response now.

clear all
Kp=1;
Kd=0;
Ki=0;
num = [-1 0];
den =[1 2*0.707+Kd 1+Kp Ki];
G=tf(num,den)
step(G)
Transfer function:
-s
---------------------
s^3 + 1.414 s^2 + 2 s
Adjust parameters to get the desired response if necessary.
The process is the same for parts b, c and d.
5‐105

5-67) Replace F(s) with
⎛ K ⎞
F ( s ) = X ref − ⎜ K P + K D s + I ⎟ X ( s )
⎝ s ⎠
B
2ζωn =
M
K
ωn 2 =
M
X ( s) 1
=
X ref ( s ) ⎛ K ⎞
s 2 + 2ζωn s + ⎜ ωn 2 + K P + K D s + I ⎟
⎝ s ⎠
Use MATLAB to obtain response now.

clear all
Kp=1;
Kd=0;
Ki=0;
B=10;
K=20;
M=1;
omega=sqrt(K/M);
zeta=(B/M)/2/omega;
num = [1 0];
den =[1 2*zeta*omega+Kd omega^2+Kp Ki];
G=tf(num,den)
step(G)
ransfer function:
s
-------------------
s^3 + 10 s^2 + 21 s
T0 achieve the proper response, adjust controller gains accordingly.
5‐106

5-68) From problem 4-3
a) Rotational kinetic energy:
Translational kinetic energy:
Relation between translational displacement and rotational displacement:
1
2
Potential energy:
As we know , then:
1 1 1
2 2 2
By differentiating, we have:
0
Since cannot be zero, then 0
b)
c)
5‐107

1 1 1
2 2 2
where at the maximum energy.
1 1
2 2
Then:
1 1
2 2
Or:
J
d) G (s) =
(ms + K )
2

K=100;
J=5;
m=25;
G=tf([J],[m 0 K])
Pole(G)
impulse(G,10)
Transfer function:
5
------------
25 s^2 + 100
ans =
0 + 2.0000i
0 - 2.0000i
5‐108

Uncontrolled
With a proportional controller one can adjust the oscillation amplitude the transfer function is
rewritten as:
JK p
Gcl ( s ) =
(ms + K + JK p )
2

Kp=0.1
K=100;
J=5;
m=25;
G=tf([J*Kp],[m 0 (K+J*Kp)])
Pole(G)
impulse(G,10)
5‐109

Kp =
0.1000
Transfer function:
0.5
--------------
25 s^2 + 100.5
ans =
0 + 2.0050i
0 - 2.0050i
A PD controller must be used to damp the

oscillation and reduce overshoot. Use Example
5-11-1 as a guide.
5‐110

5-69) From Problem 4-6 we have:
a)
y2 y1
K ( y1 − y 2 )
K ( y1 − y 2 )
μmgy& 2 μMgy&1
b) From Newton’s Law:
If y1 and y2 are considered as a position and v1 and v2 as velocity variables
Then:
The output equation can be the velocity of the engine, which means
c)
Obtaining requires solving above equation with respect to Y2(s)
From the first equation:
Substituting into the second equation:
5‐111

By solving above equation:
1
2
Replace Force F with a proportional controller so that F=K(Z-Zref):
KP
Zref _ F Z
1
2
1
2 1
5-70) Also see derivations in 4-9.
L
x(t)
F
L
M
θ
Here is an alternative representation including friction (damping) μ. In this case the angle θ is
measured differently.
Let’s find the dynamic model of the system:
5‐112

1)
2)
Let . If Φ is small enough then 1 and , therefore
which gives:
Φ s
F s
Ignoring friction 0.
Φ s
F s
where
Ignoring actuator dynamics (DC motor equations), we can incorporate feedback control using a
series PD compensator and unity feedback. Hence,
KP+KDs
R _ F Φ
F ( s) = K p ( R ( s ) − Φ ) − K D s ( R ( s ) − Φ )
The system transfer function is:
Φ A( K p + KDs)
= 2
R (s + K D s + A ( K p − B )
Control is achieved by ensuring stability (Kp>B)
5‐113

Use Routh Hurwitz to establish stability first. Use Acsys to do that as

demonstrated in this chapter problems. Also Chapter 2 has many examples.
Use MATLAB to simulate response:

clear all
Kp=10;
Kd=5;
A=10;
B=8;
num = [A*Kd A*Kp];
den =[1 Kd A*(Kp-B)];
G=tf(num,den)
step(G)
Transfer function:
50 s + 100
--------------
s^2 + 5 s + 20
Adjust parameters to achieve desired response. Use THE PROCEDURE in Example 5-11-1.
You may look at the root locus of the forward path transfer function to get a better perspective.
5‐114

Φ A ( K p + K D s ) AK D ( z + s )
= =
E s 2 − AB s 2 − AB
fix z and vary K D .
clear all
z=100;
Kd=0.01;
A=10;
B=8;
num = [A*Kd A*Kd*z];
den =[1 0 -(A*B)];
G=tf(num,den)
rlocus(G)
Transfer function:
0.1 s + 10
----------
s^2 – 80
For z=10, a large KD=0.805 results in:
5‐115

clear all
Kd=0.805;
Kp=10*Kd;
A=10;
B=8;
num = [A*Kd A*Kp];
G=tf(num,den)
pole(G)
zero(G)
step(G)
Transfer function:
8.05 s + 80.5
-------------------
s^2 + 0.805 s + 0.5
ans =
-0.4025 + 0.5814i
-0.4025 - 0.5814i
ans =
-10
Looking at dominant poles we expect to see an oscillatory response with overshoot

close to desired values.
For a better design, and to meet rise time criterion, use Example 5-
11-1.
5‐116

Chapter 6 THE CONTROL LAB
Part 1) Solution to Lab questions within Chapter 6
6-4-1 Open Loop Speed
1. Open loop speed response using SIMLab:

a. +5 V input:
The form of response is like the one that we expected; a second order system response with overshoot and
oscillation. Considering an amplifier gain of 2 and K b = 0.1 , the desired set point should be set to 2.5 and as
seen in the figure, the final value is approximately 50 rad/sec which is armature voltage divided by K b . To find
the above response the systems parameters are extracted from:
Ra J m Ra J m − k b k mτ m
τm = , B= = 0.000792kg ⋅ m 2 / sec
Ra B + k b k m Raτ m
6-1
b. +15 V input:
c. –10 V input:
6-1
2. Study of the effect of viscous friction:
The above figure is plotted for three different friction coefficients (0, 0.001, 0.005) for 5 V armature input. As
seen in figure, two important effects are observed as the viscous coefficient is increased. First, the final steady
state velocity is decreased and second the response has less oscillation. Both of these effects could be predicted
from Eq. (5-114) by increasing damping ratio ζ.
6-1
3. Additional load inertia effect:
As the overall inertia of the system is increased by 0.005 / 5.2 2 and becomes 1.8493 × 10 −3 kg.m2, the
mechanical time constant is substantially increased and we can assume the first order model for the motor
(ignoring the electrical sub-system) and as a result of this the response is more like an exponential form. The
above results are plotted for 5 V armature input.
6-1
4. Reduce the speed by 50% by increasing viscous friction:
As seen in above figure, if we set B=0.0075 N.s/m the output speed drop by half comparing with the case that
B=0 N.s/m. The above results are plotted for 5 V armature input.
6-1
5. Study of the effect of disturbance:
Repeating experiment 3 for B=0.001 N.s/m and TL =0.05 N.m result in above figure. As seen, the effect of
disturbance on the speed of open loop system is like the effect of higher viscous friction and caused to decrease
the steady state value of speed.
6-1
6. Using speed response to estimate motor and load inertia:
Using first order model we are able to identify system parameters based on unit step response of the system. In
above plot we repeated the experiments 3 with B=0.001 and set point voltage equal to 1 V. The final value of the
speed can be read from the curve and it is 8.8, using the definition of system time constant and the cursor we read
63.2% of speed final value 5.57 occurs at 0.22 sec, which is the system time constant. Considering Eq. (5-116),
and using the given value for the rest of parameters, the inertia of the motor and load can be calculated as:
τ m ( Ra B + K m K b ) 0.22(1.35 × 0.001 + 0.01)

J= = = 1.8496 × 10 −3 kg.m2
Ra 1.35
We also can use the open loop speed response to estimate B by letting the speed to coast down when it gets to the
steady state situation and then measuring the required time to get to zero speed. Based on this time and energy
conservation principle and knowing the rest of parameters we are able to calculate B. However, this method of
identification gives us limited information about the system parameters and we need to measure some parameters
directly from motor such as Ra , K m , K b and so on.
So far, no current or voltage saturation limit is considered for all simulations using SIMLab software.
6-1
7. Open loop speed response using Virtual Lab:

a. +5 V:
6-1
b. +15 V:
c. –10 V:
6-1
Comparing these results with the part 1, the final values are approximately the same but the shape of responses is
closed to the first order system behavior. Then the system time constant is obviously different and it can be
identified from open loop response. The effect of nonlinearities such as saturation can be seen in +15 V input with
appearing a straight line at the beginning of the response and also the effects of noise and friction on the response
can be observed in above curves by reducing input voltage for example, the following response is plotted for a 0.1
V step input:
6-1
6-1
8. Identifying the system based on open loop response:
Open loop response of the motor to a unit step input voltage is plotted in above figure. Using the definition of
time constant and final value of the system, a first order model can be found as:
9
G(s) = ,
0.23s + 1
where the time constant (0.23) is found at 5.68 rad/sec (63.2% of the final value).
6-1
6-4-2 Open Loop Sine Input
9. Sine input to SIMLab and Virtual Lab (1 V. amplitude, and 0.5, 5, and 50 rad/sec frequencies)
a. 0.5 rad/sec (SIMLab):
6-1
b. 5 rad/sec (SIMLab):
c. 50 rad/sec (SIMLab):
6-1
d. 0.5 rad/sec (Virtual Lab):
e. 5 rad/sec (Virtual Lab):
6-1
f. 50 rad/sec (Virtual Lab):
6-1
10. Sine input to SIMLab and Virtual Lab (20 V. amplitude)

a. 0.5 rad/sec (SIMLab):
b. 5 rad/sec (SIMLab):
6-1
c. 50 rad/sec (SIMLab):
d. 0.5 rad/sec (Virtual Lab):
6-1
e. 5 rad/sec (Virtual Lab):
6-1
f. 5 rad/sec (Virtual Lab):
In both experiments 9 and 10, no saturation considered for voltage and current in SIMLab software. If
we use the calculation of phase and magnitude in both SIMLab and Virtual Lab we will find that as
input frequency increases the magnitude of the output decreases and phase lag increases. Because of
existing saturations this phenomenon is more sever in the Virtual Lab experiment (10.f). In this
experiments we observe that M = 0.288 and ϕ = −93.82 o for ω = 50.
6-1
6-4-3 Speed Control

11. Apply step inputs (SIMLab)
In this section no saturation is considered either for current or for voltage.
a. +5 V:
6-1
b. +15 V:
c. -10 V:
6-1
6-1
12. Additional load inertia effect:

a. +5 V:
b. +15 V:
6-1
c. -10 V:
13. Study of the effect of viscous friction:
6-1
As seen in above figure, two different values for B are selected, zero and 0.0075. We could change the
final speed by 50% in open loop system. The same values selected for closed loop speed control but as
seen in the figure the final value of speeds stayed the same for both cases. It means that closed loop
system is robust against changing in system’s parameters. For this case, the gain of proportional
controller and speed set point are 10 and 5 rad/sec, respectively.
14. Study of the effect of disturbance:
6-1
Repeating part 5 in section 6-4-1 for B=0.001 and TL =0.05 N.m result in above figure. As seen, the effect of
disturbance on the speed of closed loop system is not substantial like the one on the open loop system in part 5,
and again it is shown the robustness of closed loop system against disturbance. Also, to study the effects of
conversion factor see below figure, which is plotted for two different C.F. and the set point is 5 V.
6-1
By decreasing the C.F. from 1 to 0.2, the final value of the speed increases by a factor of 5.
15. Apply step inputs (Virtual Lab)

a. +5 V:
6-1
b. +15 V:
c. –10 V:
6-1
As seen the responses of Virtual Lab software, they are clearly different from the same results of
SIMLab software. The nonlinearities such as friction and saturation cause these differences. For
example, the chattering phenomenon and flatness of the response at the beginning can be considered as
some results of nonlinear elements in Virtual Lab software.
6-4-4 Position Control
16. 160 o step input (SIMLab)
6-1
6-1
17. –0.1 N.m step disturbance
18. Examine the effect of integral control
6-1
In above figure, an integral gain of 1 is considered for all curves. Comparing this plot with the previous
one without integral gain, results in less steady state error for the case of controller with integral part.
19. Additional load inertia effect (J=0.0019, B=0.004):
6-1
6-1
20. Set B=0:
21. Study the effect of saturation
6-1
The above figure is obtained in the same conditions of part 20 but in this case we considered ± 10 V. and
± 4 A. as the saturation values for voltage and current, respectively. As seen in the figure, for higher
proportional gains the effect of saturations appears by reducing the frequency and damping property of
the system.
22. Comments on Eq. 5-126
After neglecting of electrical time constant, the second order closed loop transfer function of position
control obtained in Eq. 5-126. In experiments 19 through 21 we observe an under damp response of a
second order system. According to the equation, as the proportional gain increases, the damped
frequency must be increased and this fact is verified in experiments 19 through 21. Experiments16
through 18 exhibits an over damped second order system responses.
23. In following, we repeat parts 16 and 18 using Virtual Lab:
Study the effect of integral gain of 5:
6-1
6-1
Ch. 6 Problem Solutions
Part 2) Solution to Problems in Chapter 6
6-1. In order to find the current of the motor, the motor constant has to be separated from the electrical
component of the motor.
The response of the motor when 5V of step input is applied is:
a) The steady state speed: 41.67rad/sec

b) It takes 0.0678 second to reach 63% of the steady state speed (26.25rad/sec). This is the time
constant of the motor.
c) The maximum current: 2.228A
6-1
11.2
The steady state speed at 5V step input is 50rad/sec.
a) It takes 0.0797 seconds to reach 63% of the steady state speed (31.5rad/sec).
b) The maximum current: 2.226A
c) 100rad/sec
6-1
6-3
a) 50rad/sec
b) 0.0795 seconds
c) 2.5A. The current
d) When Jm is increased by a factor of 2, it takes 0.159 seconds to reach 63% of its steady state
speed, which is exactly twice the original time period . This means that the time constant has
been doubled.
6-1
6-4
Part 1: Repeat problem 6-1 with TL = -0.1Nm
a) It changes from 41.67 rad/sec to 25 rad/sec.

b) First, the speed of 63% of the steady state has to be calculated.
41.67 - (41.67 - 25) × 0.63 = 31.17 rad/sec.
The motor achieves this speed 0.0629 seconds after the load torque is applied
c) 2.228A. It does not change
a) It changes from 50 rad/sec to 30 rad/sec.

b) The speed of 63% of the steady state becomes
50 - (50 - 30) × 0.63 = 37.4 rad/sec.
The motor achieves this speed 0.0756 seconds after the load torque is applied
c) 2.226A. It does not change.
6-1
a) It changes from 50 rad/sec to 30 rad/sec.

b) 50 - (50 - 30) × 0.63 = 37.4 rad/s
The motor achieves this speed 0.0795 seconds after the load torque is applied. This is the same
as problem 6-3.
c) 2.5A. It does not change
d) As TL increases in magnitude, the steady state velocity decreases and steady state current
increases; however, the time constant does not change in all three cases.
6-1
6-4 The steady state speed is 4.716 rad/sec when the amplifier input voltage is 5V:
6-6
a) 6.25 rad/sec.
b) 63% of the steady state speed: 6.25 × 0.63 = 3.938 rad/sec
It takes 0.0249 seconds to reach 63% of its steady state speed.
c) The maximum current drawn by the motor is 1 Ampere.
6-1
6-7
a) 9.434 rad/sec.
b) 63% of the steady state speed: 9.434 × 0.63 = 5.943 rad/sec
It takes 0.00375 seconds to reach 63% of its steady state speed.
c) The maximum current drawn by the motor is 10 Amperes.
d) When there is no saturation, higher Kp value reduces the steady state error and decreases the rise
time. If there is saturation, the rise time does not decrease as much as it without saturation.
Also, if there is saturation and Kp value is too high, chattering phenomenon may appear.
6-8
a) The steady state becomes zero. The torque generated by the motor is 0.1 Nm.
b) 6.25 - (6.25 - 0) × 0.63 = 2.31 rad/sec. It takes 0.0249 seconds to reach 63% of its new steady
state speed. It is the same time period to reach 63% of its steady state speed without the load
torque (compare with the answer for the Problem 6-6 b).
11-9 The SIMLab model becomes
The sensor gain and the speed input are reduced by a factor of 5. In order to get the same result as
Problem 6-6, the Kp value has to increase by a factor of 5. Therefore, Kp = 0.5.
The following graphs illustrate the speed and current when the input is 2 rad/sec and Kp = 0.5.
6-1
6-10
6-1
a) 1 radian.
b) 1.203 radians.
c) 0.2215 seconds.
6-11
6-1
a) The steady state position is very close to 1 radian.

b) 1.377 radians.
c) 0.148 seconds.
It has less steady state error and a faster rise time than Problem 6-10, but has larger overshoot.
6-1
6-12
Different proportional gains and their corresponding responses are shown on the following graph.
As the proportional gain gets higher, the motor has a faster response time and lower steady state error,
but if it the gain is too high, the motor overshoot increases. If the system requires that there be no
overshoot, Kp = 0.2 is the best value. If the system allows for overshoot, the best proportional gain is
dependant on how much overshoot the system can have. For instance, if the system allows for a 30%
overshoot, Kp = 1 is the best value.
6-1
6-13 Let Kp = 1 is the best value.
As the derivative gain increases, overshoot decreases, but rise time increases.
6-1
6-14
6-1
6-15
There could be many possible answers for this problem. One possible answer would be
Kp= 100
Ki= 10
Kd= 1.4
The Percent Overshoot in this case is 3.8%.
6-1
6-16
0.1 Hz
0.2 Hz
6-1
0.5 Hz
1 Hz
6-1
2Hz
5Hz
6-1
10Hz
50Hz
6-1
As frequency increases, the phase shift of the input and output also increase. Also, the amplitude of the
output starts to decrease when the frequency increases above 0.5Hz.
6-17
6-1
As proportional gain increases, the steady state error decreases.
6-1
6-18
Considering fast response time and low overshoot, Kp=1 is considered to be the best value.
6-1
6-19 It was found that the best Kp = 1
As Kd value increases, the overshoot decreases and the rise time increases.
6-1
6-20 From the “Experiment” Menu select the Open Loop Sine Input” option.
6-1
Double click on the “Sine wave” block and choose the input values.
6-1
Run simulation in SIMULINK.

Change run time to 20 sec (default is
5 sec)
6-1
Calculate the Gain and Phase values by entering the input frequency. Repeat the process for other
frequency values and use the calculated gain and phase values to plot the frequency response of the
system.
6-1
6-21)
Select the Speed Control option in Virtual Lab Experiment Window.

Enter the Step Input and Controller Gain values by double clicking on their respective blocks.
6-1
Run simulation, and repeat the process for different gain values. Observe the steady state value change
with K.
6-1
6-1
6-22)
Select the Position Control option in Virtual Lab Experiment Window.

Enter the Step Input and Controller Gain values by double clicking on their respective blocks.
6-1
Run simulation, and repeat the process for different gain values.
6-1
6-1
6-23) If we choose Kp= 1, Ki= 0, and Kd= 0, we get
Increase Kd=0.1 we get:
6-1
6-1
6-24) Here Kp=275 and Kd=12.
Next Kp=275 and Kd=20.
6-1
You may try other parts and make observations.
6-1
6-25) a)
Change variables displayed using “Setup Axes” if desired.

b) Next use the Model Parameters button and change k to 15 as shown. Simulate the response and
show the desire variables.
6-1
6-1
6-1
c) A sample response
d)
6-1
Other parts are trivial and follow section 6-6.
6-1
6-1

Chapter 7
7-1 (a) P ( s ) = s 4 + 4 s 3 + 4 s 2 + 8 s Q( s ) = s + 1
Finite zeros of P(s): 0, −3.5098, −0.24512 ± j1.4897
Finite zeros of Q(s): −1
Asymptotes: K > 0: 60 o , 180 o , 300 o K < 0: 0 o , 120 o , 240 o
Intersect of Asymptotes:
−3.5 − 0.24512 − 0.24512 − ( −1)
σ1 = = −1
4 −1
(b) P ( s ) = s 3 + 5 s 2 + s Q( s ) = s + 1
Finite zeros of P(s) : 0, −4.7912, −0.20871
Asymptotes: K > 0: 90 o , 270 o K < 0: 0 o , 180 o
−4.7913 − 0.2087 − ( −1)
σ1 = = −2
3−1
(c) P ( s ) = s 2 3 2
Q( s ) = s + 3s + 2 s + 8
Finite zeros of P(s): 0, 0
Finite zeros of Q(s): −3.156, 0.083156 ± j15874

.
Asymptotes: K > 0: 180 o K < 0: 0 o
7‐1


(d) P ( s ) = s 3 + 2 s 2 + 3 s ( )
Q( s) = s − 1 ( s + 3)
2
Finite zeros of P(s): 0, − 1 ± j1.414
Finite zeros of Q(s): 1, −1, −3
Asymptotes: There are no asymptotes, since the number of zeros of P(s) and
Q(s) are
equal.
(e) P ( s ) = s 5 + 2 s 4 + 3 s 3 2
Q( s ) = s + 3s + 5
Finite zeros of P(s): 0, 0, 0, − 1 ± j1.414

. ± j16583
.
−1 − 1 − ( −15)
. − ( −15)
. 1
σ1 = =
5− 2 3
(f) P ( s ) = s 4 + 2 s 2 + 10 Q( s ) = s + 5
Finite zeros of P(s): −1.0398 ± j1.4426, 10398

. ± j1.4426
−10398
. − 10398
. + 10398
. + 10398
. − ( −5) −5
σ1 = =
4 −1 3

7‐2


7-2(a) MATLAB code:

s = tf('s')
num_GH=(s+1);
den_GH=(s^4+4*s^3+4*s^2+8*s);
GH_a=num_GH/den_GH;
figure(1);
rlocus(GH_a)
GH_p=pole(GH_a)
GH_z=zero(GH_a)
n=length(GH_p) %number of poles of G(s)H(s)
m=length(GH_z) %number of zeros of G(s)H(s)
%Asymptotes angles:
k=0;
Assymp1_angle=+180*(2*k+1)/(n-m)
Assymp2_angle=-180*(2*k+1)/(n-m)
k=1;
%Asymptotes intersection point on real axis:
sigma=(sum(GH_p)-sum(GH_z))/(n-m)
Root Locus
6
2
Imaginary Axis
-2
-4
-6
-8 -6 -4 -2 0 2 4
Real Axis
Assymp1_angle = 60
Assymp2_angle = -60
Assymp3_angle = 180
7‐3


sigma = -1.0000 (intersect of asymptotes)

7-2(b) MATLAB code:
s = tf('s')
'Generating the transfer function:'
num_GH=(s+1);
den_GH=(s^3+5*s^2+s);
GH_a=num_GH/den_GH;
figure(1);
rlocus(GH_a)
GH_p=pole(GH_a)
GH_z=zero(GH_a)
%Asymptotes angles:
k=0;
Root Locus
10
2
Imaginary Axis
-2
-4
-6
-8
-10
-5 -4 -3 -2 -1 0 1
Real Axis
Assymp1_angle = 90
Assymp2_angle = -90
7‐4


sigma = -2 (intersect of asymptotes)
7-2(c) MATLAB code:

s = tf('s')
num_GH=(s^3+3*s^2+2*s+8);
den_GH=(s^2);
GH_a=num_GH/den_GH;
figure(1);
rlocus(GH_a)
GH_p=pole(GH_a)
GH_z=zero(GH_a)
%Asymptotes angles:
k=0;
Root Locus
2
1.5
0.5
Imaginary Axis
-0.5
-1
-1.5
-2
-6 -5 -4 -3 -2 -1 0 1
Real Axis
7‐5


Assymp1_angle = 180
sigma = -3.0000 (intersect of asymptotes)
7-2(d) MATLAB code:

s = tf('s')
num_GH=((s^2-1)*(s+3));
den_GH=(s^3+2*s^2+3*s);
GH_a=num_GH/den_GH;
figure(1);
rlocus(GH_a)
GH_p=pole(GH_a)
GH_z=zero(GH_a)
Root Locus
1.5
0.5
Imaginary Axis
-0.5
-1
-1.5
-3.5 -3 -2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5
Real Axis
No asymptotes
7‐6


7-2(e) MATLAB code:

s = tf('s')
num_GH=(s^2+3*s+5);
den_GH=(s^5+2*s^4+3*s^3);
GH_a=num_GH/den_GH;
figure(1);
rlocus(GH_a)
GH_p=pole(GH_a)
GH_z=zero(GH_a)
%Asymptotes angles:
k=0;
k=1;
Root Locus
6
2
Imaginary Axis
-2
-4
-6
-8 -6 -4 -2 0 2 4
Real Axis
Assymp1_angle = 60
7‐7


Assymp2_angle = -60
Assymp3_angle = 180
sigma = 0.3333 (intersect of asymptotes)
7-2(f) MATLAB code:

s = tf('s')
num_GH=(s+5);
den_GH=(s^4+2*s^2+10);
GH_a=num_GH/den_GH;
figure(1);
rlocus(GH_a)
xlim([-20 20])
ylim([-20 20])
GH_p=pole(GH_a)
GH_z=zero(GH_a)
%Asymptotes angles:
k=0;
k=1;
7‐8


Root Locus
20
15
10
5
Imaginary Axis
-5
-10
-15
-20
-20 -15 -10 -5 0 5 10 15 20
Real Axis
Assymp1_angle = 60
Assymp2_angle = -60
Assymp3_angle = 180
sigma = 1.6667 (intersect of asymptotes)
7-3) Consider
As the asymptotes are the behavior of G(s)H(s) when |s|Æ∞ , then
|s| > |zi| for i = 1,2,…,m and |s| > |pi| for i = 1,2,…,n
therefore ‫ ס‬G(s)H(s)
According to the condition on angles:
‫ ס‬G(s)H(s) =
7‐9


If we consider , then:
‫ ס‬G(s)H(s) =
or
7-4) If , then each point on root locus must satisfy the characteristic equation of
If and , then
or
If the roots of above expression is considered as si for i = 1,2,…,(n-m), then
since the intersect of (n-m) asymptotes lies on the real axis of the s-plane and is
real, therefore
7-5) Poles of GH is s = 0, -2, -1 + j, -1 – j, therefore the center of asymptotes:
The angles of asymptotes:
7‐10


jω
jω
σ × σ
7‐11


7-6 (a) Angles of departure and arrival.
K > 0: −θ 1 − θ 2 − θ 3 + θ 4 = −180o
−θ 1 − 90 − 45 + 90 = −180
o o o o
θ 1 = 135
o
K < 0: −θ 1 − 90o − 45o + 90o = 0o
θ 1 = −45
o
(b) Angles of departure and arrival.
−θ 1 − θ 2 − θ 3 + θ 4 = −180
o
K > 0:
−θ 1 − 135 − 90 + 90 = 0
o o o o
K < 0:
θ 1 = −135
o
7‐12


(c) Angle of departure:
−θ 1 − θ 2 − θ 3 + θ 4 = −180
o
K > 0:
−θ 1 − 135 − 90 − 45 = −180
o o o o
θ 1 = −90
o
(d) Angle of departure
−θ 1 − θ 2 − θ 3 − θ 4 = −180
o
K > 0:
−θ 1 − 135 − 135 − 90 = −180

o o o o
θ 1 = −180
o
7‐13


(e) Angle of arrival
θ 1 + θ 6 − θ 2 − θ 3 − θ 4 − θ 5 = −360
o
K < 0:
θ 1 + 90 − 135 − 135 − 45 − 26.565 = −360

o o o o o o
θ 1 = −108.435
o
7‐14


7-7) a) ‫ ס‬G(s)H(s) =
we know that
therefore
As a result, θ D = ‫ ס‬G(s)H’(s) – 180o = 180 + , when -180o = 180o
b) Similarly:
‫ ס‬G(s)H(s) =
Therefore:
As a result, θ =180 –
7-8) zeros: and poles:
Departure angles from:
7‐15


Arrival angles at
7-9) (a) (b)
(c) (d)
7‐16


7‐17


7‐18


7-10) The breaking points are on the real axis of and must satisfy
If and α is a breakaway point, then
Finding α where K is maximum or minimum , therefore
or
5 4 3 2
7-11) (a) Breakaway-point Equation: 2 s + 20 s + 74 s + 110 s + 48 s = 0
Breakaway Points: −0.7275, − 2.3887
6 5 4 3 2
(b) Breakaway-point Equation: 3 s + 22 s + 65 s + 100 s + 86 s + 44 s + 12 = 0
Breakaway Points: −1, − 2.5
6 5 4 3 2
(c) Breakaway-point Equation: 3 s + 54 s + 347.5 s + 925 s + 867.2 s − 781.25 s − 1953 = 0
7‐19


Breakaway Points: − 2.5, 109

.
6 5 4 3 2
(d) Breakaway-point Equation: − s − 8 s − 19 s + 8 s + 94 s + 120 s + 48 = 0
Breakaway Points: −0.6428, 2.1208
7-12) (a)
K ( s + 8)
G( s) H ( s) =
s( s + 5)( s + 6)

o o o o
Asymptotes: K > 0: 90 and 270 K < 0: 0 and 180
0 − 5 − 6 − ( −8 )
σ1 = = −15
.
3−1
Breakaway‐point Equation:
3 2
2 s + 35 s + 176 s + 240 = 0
Breakaway Points: −2.2178, − 5.5724, − 9.7098
Root Locus Diagram:
7‐20


7‐12 (b)
K
G( s) H ( s) =
s( s + 1)( s + 3)( s + 4 )
o o o o o o o o
Asymptotes: K > 0: 45 , 135 , 225 , 315 K < 0: 0 , 90 , 180 , 270
0 − 1− 3 − 4
σ1 = = −2
4
3 2
Breakaway‐point Equation: 4 s + 24 s + 38 s + 12 = 0
Breakaway Points: −0.4189, − 2, − 3.5811
Root Locus Diagram:
7‐21


7‐12 (c)
K( s + 4)
G( s) H ( s) = 2 2

s ( s + 2)

Asymptotes: K > 0: 60 o , 180 o , 300 o K < 0:

o o o
0 , 120 , 240
0 + 0 − 2 − 2 − ( −4 )
σ1 = = 0
4 −1
4 3 2
3 s + 24 s + 52 s + 32 s = 0
Breakaway Points: 0, − 1085
. , − 2, − 4.915
Root Locus Diagram:

7‐22


7‐12 (d)
K ( s + 2)
G( s) H ( s) = 2

s( s + 2 s + 2 )
Asymptotes: K > 0: 90 o , 270 o

o
K < 0: 0 , 180
o
0 − 1 − j − 1 − j − ( −2 )
σ1 = = 0
3−1
3 2
Breakaway‐point Equation: 2s + 8s + 8s + 4 = 0
Breakaway Points: −2.8393 The other two solutions are not breakaway points.

7‐23


Root Locus Diagram

7‐12 (e)
K ( s + 5)
G(s) H (s) =
(
s s + 2s + 2
2
)
Asymptotes: K > 0: 90 o , 270 o K < 0:
o o
0 , 180
0 − 1 − j − 1 − j − ( −5)
σ 1 = = 15
.
3−1
3 2
2 s + 17 s + 20 s + 10 = 0
Breakaway Points: −7.2091 The other two solutions are not breakaway points.
7‐24


7‐12 (f)
K
G(s) H (s) =
(
s ( s + 4 ) s + 2s + 2
2
)
o o o o o o o o
Asymptotes: K > 0: 45 , 135 , 225 , 315 K < 0: 0 , 90 , 180 , 270
0 − 1− j − 1+ j − 4
σ1 = = −15
.
4
3 2
Breakaway Point: −3.0922 The other solutons are not breakaway points.

7‐25


7‐12 (g)
2
K( s + 4)
G( s) H ( s) = 2 2

s ( s + 8)
Asymptotes: K > 0: 90 o , 270 o o

K < 0: 0 , 180
o
Intesect of Asymptotes:
0 + 0 − 8 − 8 − ( −4 ) − ( −4 )
σ1 =
4−2

5 4 3 2
Breakaway‐point Equation: s + 20 s + 160 s + 640 s + 1040 s = 0
Breakaway Points: 0, −4, −8, −4 − j4, −4 + j4
7‐26


7‐12 (h)
K
G( s) H ( s) = 2 2

s ( s + 8)

Asymptotes: K > 0: 45o , 135o , 225o , 315o K < 0: 0 o , 90 o , 180 o , 270 o
−8 − 8
σ1 = = −4
4
3 2
Breakaway‐point Equation: s + 12 s + 32 s = 0
Breakaway Point: 0, −4, −8
7‐27


7‐12 (i)
G(s) H (s) =
(
K s + 8s + 20
2
)
s ( s + 8)
2 2
Asymptotes: K > 0: 90 o , 270 o K < 0: 0 o , 180 o
−8 − 8 − ( −4 ) − ( −4 )
σ1 = = −4
4−2
5 4 3 2
Breakaway Points: −4, − 8, − 4 + j 4.9, − 4 − j 4.9
7‐28


(j)
2
Ks
G(s) H (s) =
(s 2
−4 )

Since the number of finite poles and zeros of G ( s ) H ( s ) are the same, there are no asymptotes.
Breakaway‐point Equation: 8 s = 0
Breakaway Points: s = 0
7‐29


7‐12 (k)
G(s) H (s) =
K s −4 ( 2
)
(s 2
+1 )( s 2
+4 )
Asymptotes: K > 0: 90 o , 270 o K < 0:
o o
0 , 180
−2 + 2
Intersect of Asymptotes: σ1 = = 0
4−2
6 4 2
Breakaway‐point Equation: s − 8 s − 24 s = 0
Breakaway Points: 0, 3.2132, − 3.2132, j15246

. , − j15246
.
7‐30


7‐12 (l)
K ( s − 1)
2
G(s) H (s) =
(s 2
+1 )( s 2
+4 )
o o o o
Asymptotes: K > 0: 90 , 270 K < 0: 0 , 180
−1 + 1
σ1 = = 0
4−2
5 3
Breakaway‐point Equation: s − 2s − 9s = 0
Breakaway Points: −2.07, 2.07, − j1.47, j1.47
7‐31


(m)
K ( s + 1)( s + 2)( s + 3)
G( s) H ( s) = 3

s ( s − 1)

7‐32


o o
Asymptotes: K > 0: 180 K < 0: 0
6 5 4 3 2
Breakaway‐point Equation: s + 12 s + 27 s + 2 s − 18 s = 0
Breakaway Points: −1.21, −2.4, −9.07, 0.683, 0, 0
7‐33


(n)
K ( s + 5)( s + 40)
G( s) H ( s) = 3

s ( s + 250)( s + 1000)

Asymptotes: K > 0: 60 o , 180 o , 300 o K < 0:

o o
0 , 120 , 240
o
Intersect of asymptotes:
0 + 0 + 0 − 250 − 1000 − ( −5) − ( −40 )

σ1 = = −40167
.
5− 2
6 5 8 4 10 3 11 2
Breakaway‐point Equation: 3750 s + 335000 s + 5.247 × 10 s + 2.9375 × 10 s + 1875
. × 10 s = 0
Breakaway Points: −7.288, −712.2, 0, 0
7‐34


7‐12 (o)
K ( s − 1)
G( s) H ( s) =
s( s + 1)( s + 2 )
Asymptotes: K > 0: 90 o , 270 o K < 0:

o o
0 , 180

7‐35


−1 − 2 − 1
σ1 = = −2
3−1
3
Breakaway‐point Equation: s − 3s − 1 = 0
Breakaway Points; −0.3473, −1.532, 1.879
7‐36


Root Locus
7-13(a) MATLAB code: 10
num=[1 7]; 8
den=conv([1 0],[1 5]);
6
den=conv(den,[1 6]);
mysys=tf(num,den) 4
rlocus(mysys); 2
Imaginary Axis
0
-2
-4
-6
-8
-10
-8 -7 -6 -5 -4 -3 -2 -1 0 1
Real Axis
Root Locus
10
7-13(b) MATLAB code: 4
num=[0 1]; 2
Imaginary Axis
den=conv([1 0],[1 1]);

0
den=conv(den,[1 4]); -2
mysys=tf(num,den)
-4
rlocus(mysys);
-6
-8
-10
-12 -10 -8 -6 -4 -2 0 2 4 6 8
Real Axis
7‐37


Root Locus
15
10
Imaginary Axis
7-13(c) MATLAB code: 0
num=[1 4];
-5
den=conv([1 0],[1 0]);
mysys=tf(num,den)
rlocus(mysys); -15
-12 -10 -8 -6 -4 -2 0 2 4 6 8
Real Axis
Root Locus
10
2
Imaginary Axis
-2
7-13(d) MATLAB code:
-4
num=[1 2];
-6
den=conv([1 0],[1
(1+j)]); -8
den=conv(den,[1 (1-
-10
j)]); -2.5 -2 -1.5 -1 -0.5 0 0.5
Real Axis
mysys=tf(num,den)
rlocus(mysys);
7‐38


Root Locus
25
20
15
10
5
Imaginary Axis
0
-5
-10
-15
7-13(e) MATLAB code: -20
num=[1 5]; -25

-6 -5 -4 -3 -2 -1 0 1 2
den=conv([1 0],[1 Real Axis
(1+j)]);
den=conv(den,[1 (1-
j)]);
mysys=tf(num,den)
rlocus(mysys);
7‐39


Root Locus
5
7-13(f) MATLAB code: 4
num=conv([1 4],[1 4]); 3

den=conv([1 0],[1 0]);
2
den=conv(den,[1 8]); 1
Imaginary Axis
mysys=tf(num,den) 0
rlocus(mysys);
-1
-2
-3
-4
-5
-6 -5 -4 -3 -2 -1 0 1 2
Real Axis
Root Locus
10
8
7-13(g) MATLAB code:
6
num=conv([1 4],[1 4]);
den=conv([1 0],[1 0]); 4
Imaginary Axis
0
mysys=tf(num,den)
rlocus(mysys); -2
-4
-6
-8
-10
-9 -8 -7 -6 -5 -4 -3 -2 -1 0 1
Real Axis
7‐40


Root Locus
25
20
15
7-13(h) MATLAB code: 10
num=[0 1]; 5
Imaginary Axis
den=conv([1 0],[1 0]);
mysys=tf(num,den)
-10
rlocus(mysys);
-15
-20
-25
-30 -25 -20 -15 -10 -5 0 5 10 15 20
Real Axis
Root Locus
15
10
5
Imaginary Axis
7-13(i) MATLAB code:

0
num=conv([1 4-2j],[1
4+2j])
-5
den=conv([1 0],[1 0]);
mysys=tf(num,den)
rlocus(mysys);
-15
-9 -8 -7 -6 -5 -4 -3 -2 -1 0 1
Real Axis
7‐41


Root Locus
0.25
0.2
0.15
0.1
0.05
Imaginary Axis 0
7-13(j) MATLAB code: -0.05
num=conv([1 0],[1 0]); -0.1

den=conv([1 2],[1 -2]);
-0.15
mysys=tf(num,den)
rlocus(mysys); -0.2
-0.25
-2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5
Real Axis
7-13(k) MATLAB code:

num=conv([1 2],[1 -2]);
den=conv([1 -j],[1 j]);
den=conv(den,[1 -2j]);
den=conv(den,[1 2j]);
mysys=tf(num,den)
7‐42


rlocus(mysys);
Root Locus
5
Imaginary Axis
0
-1
-2
-3
-4
-5
-2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5 2 2.5
Real Axis
7-13(l) MATLAB code:

num=conv([1 1],[1 -1]); Root Locus
5
den=conv([1 -j],[1 j]);
den=conv(den,[1 -2j]); 4
den=conv(den,[1 2j]); 3
mysys=tf(num,den)
rlocus(mysys); 2
1
Imaginary Axis
-1
-2
-3
-4
-5
-1.5 -1 -0.5 0 0.5 1 1.5
Real Axis
7‐43


Root Locus
7-13(m) MATLAB code: 6
num=conv([1 1],[1 2]);

4
num=conv(num,[1 3]);
den=conv([1 0],[1 0]);
den=conv(den,[1 -1]);
Imaginary Axis
mysys=tf(num,den) 0
rlocus(mysys);
-2
-4
-6
-18 -16 -14 -12 -10 -8 -6 -4 -2 0 2
Real Axis
Root Locus
1500
1000
7-13(n) MATLAB code:
num=conv([1 5],[1 40]); 500
den=conv([1 0],[1 0]);
Imaginary Axis
0
mysys=tf(num,den) -500
rlocus(mysys);
-1000
-1500
-1400 -1200 -1000 -800 -600 -400 -200 0 200 400 600
Real Axis
7‐44


Root Locus
10
Imaginary Axis
7-13(o) MATLAB code: 0
num=conv([1 5],[1 40]); -2

den=conv([1 0],[1 0]);
-4
-8
mysys=tf(num,den)
rlocus(mysys); -10
-2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5
Real Axis
7-14) (a) Q( s ) = s + 5 ( )
P ( s ) = s s + 3s + 2 = s ( s + 1)( s + 2)
2
Asymptotes: K > 0: 90 o , 270 o K < 0:

o
0 , 180
o
−1 − 2 − ( −5)
σ1 = = 1
3−1
3 2
Breakaway‐point Equation: s + 9 s + 15 s + 5 = 0
Breakaway Points: −0.4475, −1.609, −6.9434
7‐45


7‐14 (b) Q( s ) = s + 3 ε 2
ϕ
P( s) = s s + s + 2
7‐46


o o o o
Asymptotes: K > 0: 90 , 270 K < 0: 0 , 180
−1 − ( −3)
σ1 = = 1
3−1

3 2
Breakaway‐point Equation: s + 5s + 3s + 3 = 0
Breakaway Points: −4.4798 The other solutions are not breakaway points.
7‐47


2
7‐14 (c) Q( s ) = 5s P ( s ) = s + 10
o o
Asymptotes: K > 0: 180 K < 0: 0
7‐48


2
Breakaway‐point Equation: 5 s − 50 = 0
7‐49


7‐14 (d) (
Q( s ) = s s + s + 2
2
) P ( s ) = s + 3s + s + 5s + 10
4 3 2

o o
Asymptotes: K > 0: 180 K < 0: 0
6 5 4 3 2
Breakaway‐point Equation: s + 2 s + 8 s + 2 s − 33 s − 20 s − 20 = 0
Breakaway Points: −2, 1.784. The other solutions are not breakaway points.
7‐50


7‐14 (e) ( )
Q( s ) = s − 1 ( s + 2 )
2
(
P( s) = s s + 2s + 2
2
)
7‐51


Since Q( s ) and P ( s ) are of the same order, there are no asymptotes.

3 2
Breakaway Points: −1.3848
7‐52


7‐14 (f) Q ( s ) = ( s + 1)( s + 4) ( )
P(s) = s s − 2
2
7‐53


o o
Asymptotes: K > 0: 180 K < 0: 0
4 3 2
Breakaway‐point equations: s + 10 s + 14 s − 8 = 0

7‐54


7‐14 (g) Q( s ) = s + 4 s + 5
2
P( s ) = s
2
(s 2
)
+ 8s + 16
Asymptotes: K > 0: 90 o , 270 o o o

K < 0: 0 , 180
−8 − ( −4 )
σ1 = = −2
4−2
5 4 3 2
Breakaway Points: 0, − 2, − 4, − 2 + j 2.45, − 2 − j 2.45
7‐55


7‐14 (h) (
Q( s ) = s − 2
2
) ( s + 4) (
P( s) = s s + 2s + 2
2
)

Since Q ( s ) and P ( s ) are of the same order, there are no asymptotes.
7‐56


Breakaway Points: −2, 6.95
7‐57


7‐14 (i) Q( s ) = ( s + 2 )( s + 0.5) ε ϕ
2
P( s) = s s − 1
Asymptotes: K > 0: 180 o K < 0:

o
0
4 3 2
Breakaway‐point Equation: s + 5s + 4 s − 1 = 0

Breakaway Points: −4.0205, 0.40245 The other solutions are not breakaway points.
7‐58


7‐59


7‐14 (j) Q( s) = 2s + 5 P(s) = s
2
(s 2
)
+ 2s + 1 = s
2
( s + 1)
2
Asymptotes: K > 0: 60 o , 180 o , 300 o o o

K < 0: 0 , 120 , 240
o
Intersect of Asymptotes;
0 + 0 − 1 − 1 − ( −2.5) 0.5
σ1 = = = 0.167
4 −1 3
4 3 2
Breakaway‐point Equation: 6 s + 28 s + 32 s + 10 s = 0
Breakaway Points: 0, −0.5316, −1, −3.135
7‐60


7-15 ) MATLAB code:

clear all;
close all;
s = tf('s')
%a)
num_GH_a=(s+5);
den_GH_a=(s^3+3*s^2+2*s);
GH_a=num_GH_a/den_GH_a;
figure(1);
rlocus(GH_a)
7‐61


%b)
num_GH_b=(s+3);
den_GH_b=(s^3+s^2+2*s);
GH_b=num_GH_b/den_GH_b;
figure(2);
rlocus(GH_b)
%c)
num_GH_c= 5*s^2;
den_GH_c=(s^3+10);
GH_c=num_GH_c/den_GH_c;
figure(3);
rlocus(GH_c)
%d)
num_GH_d=(s^3+s^2+2);
den_GH_d=(s^4+3*s^3+s^2+15);
GH_d=num_GH_d/den_GH_d;
figure(4);
rlocus(GH_d)
%e)
num_GH_e=(s^2-1)*(s+2);
den_GH_e=(s^3+2*s^2+2*s);
GH_e=num_GH_e/den_GH_e;
figure(5);
rlocus(GH_e)
%f)
num_GH_f=(s+4)*(s+1);
den_GH_f=(s^3-2*s);
GH_f=num_GH_f/den_GH_f;
figure(6);
rlocus(GH_f)
%g)
num_GH_g=(s^2+4*s+5);
den_GH_g=(s^4+6*s^3+9*s^2);
GH_g=num_GH_g/den_GH_g;
figure(7);
rlocus(GH_g)
%h)
num_GH_h=(s^2-2)*(s+4);
den_GH_h=(s^3+2*s^2+2*s);
GH_h=num_GH_h/den_GH_h;
7‐62


figure(8);
rlocus(GH_h)
%i)
num_GH_i=(s+2)*(s+0.5);
den_GH_i=(s^3-s);
GH_i=num_GH_i/den_GH_i;
figure(9);
rlocus(GH_i)
%j)
num_GH_j=(2*s+5);
den_GH_j=(s^4+2*s^3+2*s^2);
GH_j=num_GH_j/den_GH_j;
figure(10);
rlocus(GH_j)
%k)
num_GH_k=1;
den_GH_k=(s^5+2*s^4+3*s^3+2*s^2+s);
GH_k=num_GH_k/den_GH_k;
figure(11);
rlocus(GH_k)
Root Locus diagram – 7-15(a):
Root Locus
20
15
10
5
Imaginary Axis
-5
-10
-15
-20
-6 -5 -4 -3 -2 -1 0 1
Real Axis
7‐63


Root Locus diagram – 7-15(b):
Root Locus
15
10
5
Imaginary Axis
-5
-10
-15
-3.5 -3 -2.5 -2 -1.5 -1 -0.5 0 0.5 1
Real Axis
Root Locus diagram – 7-15(c):
7‐64


Root Locus
2
1.5
0.5
Imaginary Axis
-0.5
-1
-1.5
-2
-5 -4 -3 -2 -1 0 1 2
Real Axis
Root Locus diagram – 7-15(d):
Root Locus
1.5
0.5
Imaginary Axis
-0.5
-1
-1.5
-6 -5 -4 -3 -2 -1 0 1 2
Real Axis
Root Locus diagram – 7-15(e):
7‐65


Root Locus
1.5
0.5
Imaginary Axis
-0.5
-1
-1.5
-2.5 -2 -1.5 -1 -0.5 0 0.5 1 1.5
Real Axis
Root Locus diagram – 7-15(f):

poles:
Asymptotes angle:
Therefore,
Departure angle from:
7‐66


Root Locus
5
1
Imaginary Axis
-1
-2
-3
-4
-5
-18 -16 -14 -12 -10 -8 -6 -4 -2 0 2
Real Axis
Root Locus diagram – 7-15(g):

Poles: and zeroes:
Asymptotes angles:
7‐67


Root Locus
15
10
5
Imaginary Axis
-5
-10
-15
-3.5 -3 -2.5 -2 -1.5 -1 -0.5 0 0.5
Real Axis
Root Locus diagram – 7-15(h):

Poles: and zeros:
The break away points:
which means:
or
7‐68


Root Locus
1.5
0.5
Imaginary Axis
-0.5
-1
-1.5
-5 -4 -3 -2 -1 0 1 2
Real Axis
Root Locus diagram – 7-15(i):

Poles:
breaking points:
which means :
Asymptotes angles:
or θ =
7‐69


Root Locus
2.5
1.5
0.5
Imaginary Axis
-0.5
-1
-1.5
-2
-2.5
-10 -8 -6 -4 -2 0 2
Real Axis
Root Locus diagram – 7-15(j):
Root Locus
10
2
Imaginary Axis
-2
-4
-6
-8
-10
-10 -5 0 5
Real Axis
Root Locus diagram – 7-15(k):
7‐70


Root Locus
6
2
Imaginary Axis
-2
-4
-6
-6 -5 -4 -3 -2 -1 0 1 2 3 4
Real Axis
7-16) (a) Asymptotes: K > 0: 45o , 135o , 225o , 315o
−2 − 2 − 5 − 6 − ( −4 )
σ1 = = −2.75
5−1
5 4 3 2
Breakaway‐point Equation: 4 s + 65 s + 396 s + 1100 s + 1312 s + 480 = 0

Breakaway Points: −0.6325, −5.511 (on the RL)
When ζ = 0.707, K = 13.07
7‐71


7‐16 (b) Asymptotes: K > 0: 45o , 135o , 225o , 315o
7‐72


0 − 2 − 5 − 10
σ1 = = −4.25
4
3 2
Breakaway‐point Equation: 4 s + 51s + 160 s + 100 = 0
When ζ = 0.707, K = 61.5
7‐73


7‐16 (c) Asymptotes: K > 0: 180 o
4 3 2
Breakaway‐point Equation: s + 4 s + 10 s + 300 s + 500 = 0
Breakaway Points: −1.727 (on the RL)
When ζ = 0.707, K = 9.65
7‐74


7‐75


o o
7‐16 (d) K > 0: 90 , 270
−2 − 2 − 5 − 6
σ1 = = −7.5
4−2

When ζ = 0.707, K = 8.4
7‐76


7-17) MATLAB code:

clear all;
close all;
s = tf('s')
%a)
num_G_a=(s+3);
den_G_a=s*(s^2+4*s+4)*(s+5)*(s+6);
G_a=num_G_a/den_G_a;
figure(1);
rlocus(G_a)
%b)
7‐77


num_G_b= 1;
den_G_b=s*(s+2)*(s+4)*(s+10);
G_b=num_G_b/den_G_b;
figure(2);
rlocus(G_b)
%c)
num_G_c=(s^2+2*s+8);
den_G_c=s*(s+5)*(s+10);
G_c=num_G_c/den_G_c;
figure(3);
rlocus(G_c)
%d)
num_G_d=(s^2+4);
den_G_d=(s+2)^2*(s+5)*(s+6);
G_d=num_G_d/den_G_d;
figure(4);
rlocus(G_d)
%e)
num_G_e=(s+10);
den_G_e=s^2*(s+2.5)*(s^2+2*s+2);
G_e=num_G_e/den_G_e;
figure(5);
rlocus(G_e)
%f)
num_G_f=1;
den_G_f=(s+1)*(s^2+4*s+5);
G_f=num_G_f/den_G_f;
figure(6);
rlocus(G_f)
%g)
num_G_g=(s+2);
den_G_g=(s+1)*(s^2+6*s+10);
G_g=num_G_g/den_G_g;
figure(7);
rlocus(G_g)
%h)
num_G_h=(s+3)*(s+2);
den_G_h=s*(s+1);
G_h=num_G_h/den_G_h;
figure(8);
rlocus(G_h)
7‐78


%i)
num_G_i=1;
den_G_i=s*(s^2+4*s+5);
G_i=num_G_i/den_G_i;
figure(9);
rlocus(G_i)

By using “Data Cursor” tab on the figure window and clicking on the root locus diagram, gain and
damping values can be observed. Damping of ~0.707 can be observed on intersection of the root locus
diagram with two lines originating from (0,0) by angles of ArcCos(0.707) from the real axis. These
intersection points are shown for part (a) where the corresponding gain is 19. In the other figures for
section (b) to (i), similar points have been picked by the “Data Cursor”, and the gains are reported here.
Root Locus
15
10
System: G_a
Gain: 19
Pole: -0.584 + 0.589i
5 Damping: 0.704
Overshoot (%): 4.43
Imaginary Axis
Frequency (rad/sec): 0.829
System: G_a
Gain: 19
-5 Pole: -0.584 - 0.589i
Damping: 0.704
Overshoot (%): 4.43
-10
-15
-20 -15 -10 -5 0 5 10 15
Real Axis
Root Locus diagram – 7-17(b): (K = 45.5 @ damping = ~0.0707)
7‐79


Root Locus
25
20
15
10
5
Imaginary Axis
-5
-10
-15
-20
-25
-30 -25 -20 -15 -10 -5 0 5 10 15 20
Real Axis
Root Locus diagram – 7-17(c): (K = 12.8 @ damping = ~0.0707)
Root Locus
3
1
Imaginary Axis
-1
-2
-3
-18 -16 -14 -12 -10 -8 -6 -4 -2 0 2
Real Axis
Root Locus diagram – 7-17(d): (K = 8.3 @ damping = ~0.0707)
7‐80


Root Locus
30
20
10
Imaginary Axis
-10
-20
-30
-8 -7 -6 -5 -4 -3 -2 -1 0 1
Real Axis
Root Locus diagram – 7-17(e): (K = 0 @ damping = 0.0707)
Root Locus
20
15
10
5
Imaginary Axis
-5
-10
-15
-20
-15 -10 -5 0 5 10 15 20
Real Axis
Root Locus diagram – 7-17(f): (K = 2.33 @ damping = ~0.0707)
7‐81


Root Locus
5
1
Imaginary Axis
-1
-2
-3
-4
-5
-7 -6 -5 -4 -3 -2 -1 0 1
Real Axis
Root Locus diagram – 7-17(g): (K = 7.03 @ damping = ~0.0707)
Root Locus
10
2
Imaginary Axis
-2
-4
-6
-8
-10
-3.5 -3 -2.5 -2 -1.5 -1 -0.5 0 0.5
Real Axis
Root Locus diagram – 7-17(h): (no solution exists for damping =0.0707)
7‐82


Root Locus
1
0.8
0.6
0.4
0.2
Imaginary Axis
-0.2
-0.4
-0.6
-0.8
-1
-3.5 -3 -2.5 -2 -1.5 -1 -0.5 0 0.5
Real Axis
Root Locus diagram – 7-17(i): (K = 2.93 @ damping = ~0.0707)
Root Locus
4
1
Imaginary Axis
-1
-2
-3
-4
-6 -5 -4 -3 -2 -1 0 1
Real Axis
o o o
7-18) (a) Asymptotes: K > 0: 60 , 180 , 300
7‐83


0 − 10 − 20
σ1 = = −10
3
2
Breakaway‐point Equation: 3 s + 60 s + 200 = 0 Breakaway Point: (RL) −4.2265, K = 384.9

o o o o
(b) Asymptotes: K > 0: 45 , 135 , 225 , 315
0 − 1− 3 − 5
σ1 = = −2.25
4
3 2
Breakaway Points: (RL) −0.4258 K = 2.879, −4.2537 K = 12.95
7‐84


c) Zeros: and poles:
Angle of asymptotes:
The breakaway points: Î
Then and
d) Poles:
breakaway points:
e) Zeros: and poles:
7‐85


breakaway points: Î
f) Poles:
Angles of asymptotes:
breakaway point:
7-19) MATLAB code:

clear all;
close all;
s = tf('s')
%a)
num_G_a=1;
den_G_a=s*(s+10)*(s+20);
figure(1);
rlocus(G_a)
%b)
num_G_b= 1;
den_G_b=s*(s+1)*(s+3)*(s+5);
figure(2);
rlocus(G_b)
%c)
num_G_c=(s-0.5);
den_G_c=(s-1)^2;
figure(3);
rlocus(G_c)
%d)
7‐86


num_G_d=1;
den_G_d=(s+0.5)*(s-1.5);
figure(4);
rlocus(G_d)
%e)
num_G_e=(s+1/3)*(s+1);
den_G_e=s*(s+1/2)*(s-1);
figure(5);
rlocus(G_e)
%f)
num_G_f=1;
den_G_f=s*(s^2+6*s+25);
G_f=num_G_f/den_G_f;
figure(6);
rlocus(G_f)
Root Locus
60
40
System: G_a
Gain: 385
20 Pole: -4.23 - 1.45e-007i
Damping: 1
Overshoot (%): 0
Imaginary Axis

0
-20
-40
-60
-80 -60 -40 -20 0 20 40
Real Axis
7‐87


Root Locus
15
10
System: G_b
Gain: 2.88
5 Pole: -0.426 - 1.02e-008i
Damping: 1
Overshoot (%): 0
Imaginary Axis

0
System: G_b
Gain: 13
Pole: -4.25 + 0.0278i
-5 Damping: 1
Overshoot (%): 0
-10
-15
-15 -10 -5 0 5 10
Real Axis
Root Locus
0.8
0.6
System: G_c
0.4
Gain: 2
Pole: 0
0.2 Damping: -1
Overshoot (%): 0
Imaginary Axis
Frequency (rad/sec): 0
0
-0.2
-0.4
-0.6
-0.8
-0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1 1.2
Real Axis
7‐88


Root Locus
1.5
1
System: G_d
Gain: 1
0.5 Pole: 0.5
Damping: -1
Overshoot (%): 0
Imaginary Axis

0
-0.5
-1
-1.5
-1 -0.5 0 0.5 1 1.5 2
Real Axis
Root Locus
1.5
1
System: G_e System: G_e
Gain: 5.34 Gain: 0.211
0.5 Pole: -2.24 - 3.95e-008i Pole: 0.383
Damping: 1 Damping: -1
Overshoot (%): 0 Overshoot (%): 0
Imaginary Axis
Frequency (rad/sec): 2.24 Frequency (rad/sec): 0.383

0
-0.5
-1
-1.5
-5 -4 -3 -2 -1 0 1 2
Real Axis
Root Locus diagram – 7-19(f): (No breakaway points)
7‐89


Root Locus
25
20
15
10
5
Imaginary Axis
-5
-10
-15
-20
-25
-30 -25 -20 -15 -10 -5 0 5 10 15
Real Axis
7‐90


7-20)
7‐91


7-21) MATLAB code:

clear all;
close all;
s = tf('s')
%a)
n=1;
num_G_a= 1;
den_G_a=(s+4)^n;
figure(n);
rlocus(G_a)
%b)
n=2;
num_G_b= 1;
den_G_b=(s+4)^n;
figure(n);
rlocus(G_b)
%c)
n=3;
num_G_c= 1;
den_G_c=(s+4)^n;
figure(n);
rlocus(G_c)
%d)
n=4;
num_G_d= 1;
den_G_d=(s+4)^n;
figure(n);
rlocus(G_d)
%e)
n=5;
num_G_e= 1;
den_G_e=(s+4)^n;
figure(n);
rlocus(G_e)
7‐92


Root Locus
0.8
0.6
0.4
0.2
Imaginary Axis
-0.2
-0.4
-0.6
-0.8
-16 -14 -12 -10 -8 -6 -4 -2 0 2
Real Axis
Root Locus
2.5
1.5
0.5
Imaginary Axis
-0.5
-1
-1.5
-2
-2.5
-4.5 -4 -3.5 -3 -2.5 -2 -1.5 -1 -0.5 0 0.5
Real Axis
7‐93


Root Locus
4
1
Imaginary Axis
-1
-2
-3
-4
-9 -8 -7 -6 -5 -4 -3 -2 -1 0 1
Real Axis
Root Locus
1.5
0.5
Imaginary Axis
-0.5
-1
-1.5
-6 -5 -4 -3 -2 -1 0 1
Real Axis
7‐94


Root Locus
5
1
Imaginary Axis
-1
-2
-3
-4
-5
-9 -8 -7 -6 -5 -4 -3 -2 -1 0 1
Real Axis
7‐95


7-22) P ( s ) = s 3 + 25 s 2 + 2 s + 100 Q ( s ) = 100 s
o o
Asymptotes: Kt > 0: 90 , 270
−25 − 0
σ1 = = −12.5
3−1
3 2
Breakaway‐point Equation: s + 12.5 s − 50 = 0
Breakaway Points: (RL) −2.2037, −12.162
7-23) MATLAB code:
7‐96


s = tf('s')
num_G= 100;
den_G=s^3+25*s+2*s+100;
G=num_G/den_G;
figure(1);
rlocus(G)
Root Locus diagram – 7-23:
Root Locus
30
20
10
Imaginary Axis
-10
-20
-30
-30 -25 -20 -15 -10 -5 0 5 10 15 20
Real Axis
7-24) Characteristic equation: s 3 + 5s 2 + Kt s + K = 0
(a) K t = 0 : P(s) = s
2
( s + 5) Q(s) = 1

o o o
Asymptotes: K > 0: 60 , 180 , 300
−5 − 0
σ1 = = −1667
.
3
7‐97


2
Breakaway‐point Equation: 3 s + 10 s = 0 Breakaway Points: 0, −3.333
7‐98


3 2
7‐24 (b) P ( s ) = s + 5 s + 10 = 0 Q( s ) = s
−5 − 0
σ1 = = 0
2 −1
3
Breakaway‐point Equation: 2 s + 5 s − 10 = 0
There are no breakaway points on RL.
7‐99


7-25)
By collapsing the two loops, and finding the overall close loop transfer function, the characteristic
equation (denominator of closed loop transfer function) can be found as:
s 3 + 5s 2 + K t s + K
1 + GH =
s 2 ( s + 5) + K t s
For part (a): Root locus diagram, part (a):

Kt =0. Therefore, assuming 15
Root Locus
Den(GH)= s 3 + 5s 2 and
10
Num(GH) =1, we can use rlocus
command to construct the root 5
locus diagram.
Imaginary Axis
For part (b): -5
K =10. Therefore, assuming

-10
Den(GH)= s 3 + 5 s 2 + 10 and
Num(GH) = s , we can use rlocus -15
-20 -15 -10 -5 0 5 10
command to construct the root Real Axis
locus diagram. Root locus diagram, part (b):

Root Locus
15
MATLAB code (7-25): 10
s = tf('s')
5
%a)
Imaginary Axis
num_G_a= 1;
den_G_a=s^3+5*s^2; 0
GH_a=num_G_a/den_G_a;
figure(1); -5
rlocus(GH_a)
-10
%b)
num_G_b= s;
den_G_b=s^3+5*s^2+10; -15
-6 -5 -4 -3 -2 -1 0 1
GH_b=num_G_b/den_G_b; Real Axis
figure(2);
rlocus(GH_b)
7‐100


7-26) P ( s ) = s 2 + 116.84 s + 1843 2

Q ( s ) = 2.05 s ( s + 5)
o
Asymptotes: J L = 0: 180
4 3 2
Breakaway‐point Equation: −2.05 s − 479 s − 12532 s − 37782 s = 0
Breakaway Points: (RL) 0, −204.18
7-27) MATLAB code:
7‐101


s = tf('s')
num_G = (2.05*s^3 + 10.25*s^2);
den_G = (s^2 + 116.84*s + 1843);
G = num_G/den_G;
figure(1);
rlocus(G)
Root locus diagram:
Root Locus
150
100
50
Imaginary Axis
-50
-100
-150
-400 -350 -300 -250 -200 -150 -100 -50 0 50
Real Axis
7‐102


7-28) (a) P( s ) = s ( s 2 − 1) Q ( s ) = ( s + 5)( s + 3)
o
Asymptotes: K > 0: 180

4 3 2
Breakaway‐point Equation: s + 16 s + 46 s − 15 = 0
Breakaway Points: (RL) 0.5239, −12.254
7‐103


7‐28 (b) P( s ) = s ( s 2 + 10 s + 29 ) Q ( s ) = 10( s + 3)
0 − 10 − ( −3)
σ1 = = −3.5
3−1
3 2
There are no breakaway points on the RL.
7‐104


7-29)
Root locus diagram, part (a):
MATLAB code (7-29):
s = tf('s')
%a)
num_G_a = (s+5)*(s+3);
den_G_a = s*(s^2 - 1);
G_a = num_G_a/den_G_a;
figure(1);
rlocus(G_a)
7‐105


K=10;
%b) Root Locus
8
num_G_b = (3*K+K*s);
den_G_b =
6
(s^3+K*s^2+K*3*s-s);
G_b = num_G_b/den_G_b; 4
figure(2);
rlocus(G_b) 2
Imaginary Axis
0
-2
-4
-6
-8
-25 -20 -15 -10 -5 0 5
Real Axis
Root locus diagram, part (b):
Root Locus
20
15
10
5
Imaginary Axis
-5
-10
-15
-20
-6 -5 -4 -3 -2 -1 0 1
Real Axis
7-30) Poles: zeros:
7‐106


Angles of asymptotes:
breakaway points:
Ö Î
MATLAB code:
s = tf('s')
num_G=(s+0.4);
den_G=s^2*(s+3.6);
G=num_G/den_G;
figure(1);
rlocus(G)
Root locus diagram:
Root Locus
5
1
Imaginary Axis
-1
-2
-3
-4
-5
-4 -3.5 -3 -2.5 -2 -1.5 -1 -0.5 0 0.5
Real Axis

7‐31 (a) P ( s ) = s( s + 12.5)( s + 1) Q( s ) = 83.333
7‐107


o o o
Asymptotes: N > 0: 60 , 180 , 300
0 − 12.5 − 1
σ1 = = −4.5
3

2
Breakaway‐point Equation: 3s + 27 s _12.5 = 0
Breakaway Point: (RL) −0.4896

2 2
7‐31 (b) P ( s ) = s + 12.5 s + 833.333 Q ( s ) = 0.02 s ( s + 12.5)
7‐108


o
A > 0: 180
4 3 2
Breakaway‐point Equation: 0.02 s + 0.5 s + 53.125 s + 416.67 s = 0
Breakaway Points: (RL) 0
7‐31 c) P ( s ) = s 3 + 12.5 s 2 + 1666.67 = ( s + 17.78 )( s − 2.64 + j 9.3)( s − 2.64 − j 9.3)
7‐109


Q( s ) = 0.02 s( s + 12.5)

o
Asymptotes: Ko > 0: 180
4 3 2
Breakaway‐point Equation: 0.02 s + 0.5 s + 3.125 s − 66.67 s − 416.67 = 0
7‐110


7-32) MATLAB code:
s = tf('s')
%a)
A=50;
K0=50;
num_G_a = 250;
den_G_a = 0.06*s*(s + 12.5)*(A*s+K0);
figure(1);
rlocus(G_a)
%b)
N=10;
K0=50;
num_G_b = 0.06*s*(s+12.5)*s
den_G_b = K0*(0.06*s*(s+12.5))+250*N;
G_b = num_G_b/den_G_b;
figure(2);
rlocus(G_b)
%c)
A=50;
N=20;
num_G_c = 0.06*s*(s+12.5);
den_G_c = 0.06*s*(s+12.5)*A*s+250*N;
G_c = num_G_c/den_G_c;
figure(3);
rlocus(G_c)
7‐111


Root Locus
30
20
10
Imaginary Axis
-10
-20
-30
-40 -30 -20 -10 0 10 20
Real Axis
Root Locus
30
20
10
Imaginary Axis
-10
-20
-30
-25 -20 -15 -10 -5 0 5
Real Axis
Root locus diagram, part (c):
7‐112


Root Locus
10
2
Imaginary Axis
-2
-4
-6
-8
-10
-35 -30 -25 -20 -15 -10 -5 0 5
Real Axis
7-33) (a) A = Ko = 100: P ( s ) = s( s + 12.5)( s + 1) Q( s ) = 4167

.
o o o
Asymptotes: N > 0: 60 180 300
0 − 1 − 12.5
σ1 = = −4.5
3
2
Breakaway‐point Equation: 3 s + 27 s + 12.5 = 0
Breakaway Points: (RL) −0.4896
7‐113


7‐33 (b) P ( s ) = s 2 + 12.5 + 1666.67 = ( s + 6.25 + j 40.34 )( s + 6.25 − j 40.34 )
Q ( s ) = 0.02 s 2 ( s + 12.5)
7‐114


Asymptotes: A > 0: 180 o
4 3 2
Breakaway‐point Equation: 0.02 s + 0.5 s + 103.13 s + 833.33 s = 0
Breakaway Points: (RL) 0
7‐115


7‐33 (c) P ( s ) = s 3 + 12.5 s 2 + 833.33 = ( s + 15.83)( s − 1.663 + j 7.063)( s − 1.663 − j 7.063)
Q( s ) = 0.01s( s + 12.5)
Asymptotes: Ko > 0: 180o
4 3 2
Breakaway‐point Equation: 0.01s + 0.15 s + 1.5625 s − 16.67 s − 104.17 = 0
7‐116


7-34) MATLAB code:
s = tf('s')
%a)
A=100;
K0=100;
num_G_a = 250;
den_G_a = 0.06*s*(s + 12.5)*(A*s+K0);
figure(1);
rlocus(G_a)
%b)
N=20;
K0=50;
7‐117


num_G_b = 0.06*s*(s+12.5)*s
den_G_b = K0*(0.06*s*(s+12.5))+250*N;
G_b = num_G_b/den_G_b;
figure(2);
rlocus(G_b)
%c)
A=100;
N=20;
num_G_c = 0.06*s*(s+12.5);
den_G_c = 0.06*s*(s+12.5)*A*s+250*N;
G_c = num_G_c/den_G_c;
figure(3);
rlocus(G_c)

Root Locus
30
20
10
Imaginary Axis
-10
-20
-30
-40 -30 -20 -10 0 10 20
Real Axis
7‐118


Root Locus
50
40
30
20
10
Imaginary Axis
-10
-20
-30
-40
-50
-30 -25 -20 -15 -10 -5 0 5 10
Real Axis
Root locus diagram, part (c):
Root Locus
8
2
Imaginary Axis
-2
-4
-6
-8
-30 -25 -20 -15 -10 -5 0 5
Real Axis
7-35) a) zeros: , poles:
7‐119


Breakaway points:
Ö Î
b) There is no closed loop pole in the right half s-plane; therefore the system is stable for all K>0
c) MATLAB code:
num_G=25*(s+2)^2;
den_G=(s^2+4)*(s+5)^2;
G_a=num_G/den_G;
figure(1);
rlocus(G_a)
Root locus diagram:

Root Locus
10
2
Imaginary Axis
-2
-4
-6
-8
-10
-6 -5 -4 -3 -2 -1 0 1
Real Axis
7‐120


2
7-36) (a) P ( s ) = s ( s + 1)( s + 5) Q ( s ) = 1
o o o o
Asymptotes: K > 0: 45 , 135 , 225 , 315
0 + 0 − 1− 5
σ1 = = −15
.
4
3 2
Breakaway‐point Equation: 4 s + 18 s + 10 s = 0 Breakaway point: (RL) 0, −3.851
2
(b) P ( s ) = s ( s + 1)( s + 5) Q ( s ) = 5 s + 1
o o o
Asymptotes: K > 0: 60 , 180 , 300
0 + 0 − 1 − 5 − ( −0.2) 5.8
σ1 = =− = −193
.
4 −1 3
4 3 2
Breakaway‐point Equation: 15 s + 64 s + 43 s + 10 s = 0
Breakaway Points: (RL) −3.5026
7‐121


7-37) Root locus diagram, part (a):

MATLAB code (7-37):
Root Locus
s = tf('s') 15
%a)
num_GH_a= 1;
10
den_GH_a=s^2*(s+1)*(s+
5);
GH_a=num_GH_a/den_GH_a 5
;
Imaginary Axis
figure(1); 0
rlocus(GH_a)
-5
%b)
num_GH_b= (5*s+1);
den_GH_b=s^2*(s+1)*(s+ -10
5);
GH_b=num_GH_b/den_GH_b
-15
; -15 -10 -5 0 5 10 15
figure(2); Real Axis
rlocus(GH_b)
7‐122


Root Locus
20
15
10
Imaginary Axis
0
-5
-10
-15
-20
-25 -20 -15 -10 -5 0 5 10
Real Axis
7-38) a) can be approximated by ( easy way to verify is to compare both funtions’ Taylor
series expansions)
Therefore:
Zeros: and poles:

Angle of asymptotes :
Breakaway points:
Which means: Î
b) Î
S2 1 2+2k
7‐123


S 3-k 0
S0 (3-k)(2+2k)

As a result:
Since K must be positive, the range of stability is then
c) In this problem, e −Ts term is a time delay. Therefore, MATLB PADE command is used for pade
approximation, where brings e −Ts term to the polynomial form of degree N.
s = tf('s')
T=1
N=1;
num_GH= pade(exp(-1*T*s),N);
den_GH=(s+1);
GH=num_GH/den_GH;
figure(5);
rlocus(GH)
Root locus diagram:
7‐124


Root Locus
4
1
Imaginary Axis
-1
-2
-3
-4
-4 -2 0 2 4 6 8 10 12
Real Axis
7-39)
(a) P ( s ) = s 2 ( s + 1)( s + 5) + 10 = ( s + 4.893)( s + 1896
. )( s − 0.394 + j 0.96 )( s − 0.394 + j 0.96 )
Q( s ) = 10 s
o o o
Asymptotes: Td > 0: 60 , 180 , 300
−4.893 − 1896
. + 0.3944 + 0.3944
Intersection of Asymptotes: σ1 = = −2
4 −1
There are no breakaway points on the RL.
7‐125


(b) MATLAB code:

s = tf('s')
num_GH= 10*s;
den_GH=s^2*(s+1)*(s+5)+10;
GH=num_GH/den_GH;
figure(1);
rlocus(GH)
Root locus diagram:
7‐126


Root Locus
15
10
5
Imaginary Axis
-5
-10
-15
-20 -15 -10 -5 0 5 10
Real Axis
7-40) (a) K = 1: P ( s ) = s 3 ( s + 117.23)( s + 4882.8 ) Q ( s ) = 1010( s + 1.5948 )( s + 114.41)( s + 4884 )
o o
Asymptotes: K L > 0: 90 , 270
−117.23 − 4882.8 + 15948

. + 114.41 + 4884
σ1 = = −0.126
5− 3
Breakaway Point: (RL) 0
7‐127


3
7‐40 (b) K = 1000: P ( s ) = s ( s + 117.23)( s + 4882.8 )
3 2 5
Q ( s ) = 1010( s + 5000 s + 5.6673 × 10 s + 891089110 )

= 1010( s + 4921.6 )( s + 39.18 + j 423.7 )( s + 39.18 − 423.7 )
o o
Asymptotes: K L > 0: 90 , 270
−117.23 − 4882.8 + 49216

. + 39.18 + 39.18
σ1 = = −0.033
5− 3
7 7 6 10 5 13 4 16 3 18 2
2020 s + 2.02 × 10 s + 5.279 × 10 s + 1.5977 × 10 s + 18655
. × 10 s + 1.54455 × 10 s = 0
Breakaway points: (RL) 0, −87.576
7‐128


7-41) MATLAB code:

s = tf('s')
Ki=9;
Kb=0.636;
Ra=5;
La=.001;
Ks=1;
n=.1;
Jm=0.001;
Jl=0.001;
Bm=0;
%a)
7‐129


K=1;
num_G_a=((n^2*La*Jl+La*Jm)*s^3+(n^2*Ra*Jl+Ra*Jm+Bm*La)*s^2+Ra*Bm*s+Ki
*Kb*s+n*Ks*K*Ki);
den_G_a=((La*Jm*Jl)*s^5+(Jl*Ra*Jm+Jl*Bm*La)*s^4+(Ki*Kb*Jl+Ra*Bm*Jl)*s
^3);
figure(1);
rlocus(G_a)
%b)
K=1000;
num_G_b=((n^2*La*Jl+La*Jm)*s^3+(n^2*Ra*Jl+Ra*Jm+Bm*La)*s^2+Ra*Bm*s+Ki
*Kb*s+n*Ks*K*Ki);
den_G_b=((La*Jm*Jl)*s^5+(Jl*Ra*Jm+Jl*Bm*La)*s^4+(Ki*Kb*Jl+Ra*Bm*Jl)*s
^3);
figure(2);
rlocus(G_b)

Root Locus
1000
800
600
400
200
Imaginary Axis
-200
-400
-600
-800
-1000
-3500 -3000 -2500 -2000 -1500 -1000 -500 0 500
Real Axis
7‐130


Root Locus
1000
800
600
400
200
Imaginary Axis
-200
-400
-600
-800
-1000
-4000 -3500 -3000 -2500 -2000 -1500 -1000 -500 0 500
Real Axis
7-42
(a) Characteristic Equation: s + 5000 s + 572,400 s + 900,000 + J L 10 s + 50,000 s
3 2
( 3 2
)= 0
3 2 2
P ( s ) = s + 5000 s + 572, 400 s + 900,000 = ( s + 1.5945)( s + 115.6 )( s + 4882.8 ) Q ( s ) = 10 s ( s + 5000 )
Since the pole at −5000 is very close to the zero at −4882.8, P ( s ) and Q( s ) can be approximated as:
2
P ( s ) ≅ ( s + 1.5945)( s + 115.6 ) Q ( s ) ≅ 10.24 s
2
Breakaway‐point Equation: 1200 s + 3775 s = 0 Breakaway Points: (RL): 0, −3.146
7‐131


(b) MATLAB code:

s = tf('s')
K=1;
Jm=0.001;
La=0.001;
n=0.1;
Ra=5;
Ki=9;
Bm=0;
Kb=0.0636;
Ks=1;
num_G_a = (n^2*La*s^3+n^2*Ra*s^2);
den_G_a = (La*Jm*s^3+(Ra*Jm+Bm*La)*s^2+(Ra*Bm+Ki*Kb)*s+n*Ki*Ks*K);
figure(1);
rlocus(G_a)
Root locus diagram:
Root Locus
300
200
100
Imaginary Axis
-100
-200
-300
-6000 -5000 -4000 -3000 -2000 -1000 0 1000
Real Axis
7‐132


7-43) (a) α = 12: P ( s ) = s 2 ( s + 12 ) Q ( s ) = s + 1
o o o o
Asymptotes: K > 0: 90 , 270 K < 0: 0 , 180
0 + 0 − 12 − ( −1)
σ1 = = −5.5
3−1
3 2
Breakaway‐point Equation: 2 s + 15 s + 24 s = 0 Breakaway Points: 0, −2.314, −5.186
2
7‐43 (b) α = 4: P( s) = s ( s + 4) Q ( s ) = s + 1
o o o o
Asymptotes: K > 0: 90 , 270 K < 0: 0 , 180
0 + 0 − 4 − ( −10
σ1 = = −15
.
3−1
3 2
Breakaway‐point Equation: 2 s + 7 s + 8 s = 0 Breakaway Points: K > 0 0. None for K < 0.

7‐133


2
2 α +3 (α + 3) − 16α
(c) Breakaway‐point Equation: 2 s + (α + 3) s + 2 s = 0 Solutions: s = − ± , s = 0
4 4
For one nonzero breakaway point, the quantity under the square‐root sign must equal zero.
2
Thus, α − 10α + 9 = 0, α = 1 or α = 9. The answer is α = 9. The α = 1 solution represents pole‐zero
cancellation in the equivalent G ( s ). When α = 9, the nonzero breakaway point is at s = −3. σ 1 = −4.
7‐134


7-44)
For part (c), after finding the expression for:
dk − 3 − α ± (α − 1)(α − 9)
= ,
ds 4
there is one acceptable value of alpha that makes the square root zero ( α = 9 ). Zero square root means
one answer to the breakaway point instead of 2 answers as a result of ± sign. α = 1 is not acceptable
dk 0
since it results in s = −1@ = 0 and then k = .
ds 0
MATLAB code:
s = tf('s')
%(a)
alpha=12
num_GH= s+1;
den_GH=s^3+alpha*s^2;
GH=num_GH/den_GH;
figure(1);
rlocus(GH)
%(b)
alpha=4
num_GH= s+1;
GH=num_GH/den_GH;
figure(2);
rlocus(GH)
%(c)
alpha=9
num_GH= s+1;
GH=num_GH/den_GH;
figure(3);
rlocus(GH)
7‐135


Root Locus
10
2
Imaginary Axis
-2
-4
-6
-8
-10
-14 -12 -10 -8 -6 -4 -2 0 2
Real Axis
Root Locus
10
2
Imaginary Axis
-2
-4
-6
-8
-10
-4.5 -4 -3.5 -3 -2.5 -2 -1.5 -1 -0.5 0 0.5
Real Axis
7‐136


Root locus diagram, part (c): (alpha=9 resulting in 1 breakaway point)
Root Locus
8
2
Imaginary Axis
-2
-4
-6
-8
-10 -8 -6 -4 -2 0 2
Real Axis
7-45) (a) P ( s ) = s 2 ( s + 3) Q( s ) = s + α
3
Breakaway‐point Equation: 2 s + 3(1 + α ) s + 6α = 0
The roots of the breakaway‐point equation are:
2
−3(1 + α ) 9(1 + α ) − 48α
s= ±
4 4
2
For no breakaway point other than at s = 0 , set 9(1 + α ) − 48α < 0 or ‐ 0.333 < α < 3
Root Locus Diagram with No Breakaway Point other than at s = 0.
7‐137


7‐45 (b) One breakaway point other than at s = 0: α = 0.333, Breakaway point at s = −1.
7‐138


7‐139


7‐45 (d) Two breakaway points: α > 3 :
7-46) First we can rearrange the system as:
H 1 (s) e − st
where
Now designing a controller is similar to the designing a controller for any unity feedback system.
7‐140


7-47) Let the angle of the vector drawn from the zero at s = j12 to a point s1 on the root locuss near the zero
be θ . Let
θ 1 = angle of the vector drawn from the pole at j10 to s1 .

θ 2 = angle of the vector drawn from the pole at 0 to s1 .

θ 3 = angle of the vector drawn from the pole at − j10 to s1 .
θ 4 = angle of the vector drawn from the zero at − j12 to s1 .
Then the angle conditions on the root loci are:
θ = θ 1 − θ 2 − θ 3 + θ 4 = odd multiples of 180

o

o
θ 1 = θ 2 = θ 3 = θ 4 = 90 θ =0
o
Thus,

The root loci shown in (b) are the correct ones.
Answers to True and False Review Questions:
6. (F) 7. (T) 8. (T) 9. (F) 10. (F) 11. (T) 12. (T) 13. (T) 14. (T)
7‐141


Chapter 8
6.54
8‐1 (a) K = 5 ω n = 5 = 2.24 rad / sec ζ= = 1.46 Mr = 1 ω r = 0 rad / sec
4.48
6.54 1
(b) K = 21.39 ω n = 2139
. = 4.62 rad / sec ζ = = 0.707 M r = =1
9.24 2
2ζ 1 − ζ
2
ω r = ω n 1 − ς = 3.27 rad / sec
6.54
(c) K = 100 ω n = 10 rad / sec ζ = = 0.327 M r = 1618
. ω r = 9.45 rad / sec
20
8-2 Bode diagram (a) – k=5: data points from top to bottom indicate
MATLAB code: bandwidth BW, resonance peak Mr, and resonant frequency ωr.
% Question 8-2, System: CL_a
clear all; Frequency (rad/sec): 0.855
Bode Diagram Magnitude (dB): -3
close all; 0
s = tf('s') -10
System: CL_a
Magnitude (dB)
-20 Frequency (rad/sec): 2.25

%a) Magnitude (dB): -9.4
num_G_a= 5; -30
den_G_a=s*(s+6.54); -40
-50
CL_a=G_a/(1+G_a)
0
BW = bandwidth(CL_a)
bode(CL_a)
-45
Phase (deg)
%b)
figure(2); -90
System: CL_a
num_G_b=21.38; Frequency (rad/sec): 2.25
den_G_b=s*(s+6.54); -135 Phase (deg): -90.2
-1 0
10 10
G_b=num_G_b/den_G_b; Frequency (rad/sec)
CL_b=G_b/(1+G_b)
BW = bandwidth(CL_b) Bode diagram (b) – k=21.38: data points from top to bottom
bode(CL_b)
indicate bandwidth BW, resonance peak Mr, and resonant frequency
%c) ωr.
figure(3);
num_G_c=100;
den_G_c=s*(s+6.54);
8‐1


CL_c=G_c/(1+G_c) System: CL_b

BW = bandwidth(CL_c) Magnitude (dB): -3
Bode Diagram
0
bode(CL_c)
System: CL_b
Magnitude (dB)
Magnitude (dB): -3.03
-40
-60
-80
-100
0
-45
Phase (deg)
-90
System: CL_b
Phase (deg): -89.9
-180
0 1 2
10 10 10
Frequency (rad/sec)
Bode diagram (c) – k=100: data points from top to bottom indicate
resonance peak Mr, bandwidth BW, and resonant frequency ωr.
Bode Diagram
20
0
System: CL_c
Magnitude (dB)
Frequency (rad/sec): 9.99 System: CL_c

Magnitude (dB): 3.62
-40
-60
-80
0
-45
Phase (deg)
-90
System: CL_c
-135 Frequency (rad/sec): 10
Phase (deg): -90
-180
-1 0 1 2 3
10 10 10 10 10
Frequency (rad/sec)
8‐2


8-3) If is the input, then
where
and
Therefore:
As a result:
8-4 (a) Mr = 2.944 ( 9.38 dB) ω r = 3 rad / sec BW = 4.495 rad / sec
(b) M r = 15.34 ( 23.71 dB) ω r = 4 rad / sec BW = 6.223 rad / sec
(c) M r = 4.17 (12.4 dB) ω r = 6.25 rad / sec BW = 9.18 rad / sec
(d) M r = 1 ( 0 dB) ω r = 0 rad / sec BW = 0.46 rad / sec
(e) M r = 157
. ( 3.918 dB) ω r = 0.82 rad / sec BW = 1.12 rad / sec
8‐3


(f) Mr = ∞ ( unstable) ω r = 15
. rad / sec BW = 2.44 rad / sec
(g) M r = 3.09 ( 9.8 dB) ω r = 1.25 rad / sec BW = 2.07 rad / sec
(h) M r = 4.12 (12.3 dB) ω r = 3.5 rad / sec BW = 5.16 rad / sec
8‐5)
Maximum overshoot = 0.1 Thus, ζ = 0.59
2
1 1 − 0.416ζ + 2.917ζ
Mr = = 105
. tr = = 0.1 sec
2ζ 1 − ζ
2 ωn
Thus, minimum ω n = 17.7 rad / sec Maximum M r = 105

.
((1 − 2ζ )+ )
1/ 2
Minimum BW = ω n 4ζ − 4ζ + 2 = 20.56 rad/sec

2 4 2

8‐6)
− πζ
2
1−ζ
Maximum overshoot = 0.2 Thus, 0.2 = e ζ = 0.456
2
1 1 − 0.416ζ + 2.917ζ
Mr = = 1.232 t r = = 0.2 Thus, minimum ω n = 14.168 rad/sec
2ζ 1 − ζ
2 ωn

(( )+ )
1/ 2
Maximum M r = 1.232 Minimum BW = 1 − 2ζ 4ζ − 4ζ + 2 = 18.7 rad/sec

2 4 2

− πζ
2
1−ζ
8-7) Maximum overshoot = 0.3 Thus, 0.3 = e ζ = 0.358

2
1 1 − 0.416ζ + 2.917ζ
Mr = = 1.496 t r = = 0.2 Thus, minimum ω n = 6.1246 rad/sec
2ζ 1 − ζ
2 ωn
8‐4


(( )+ )
1/ 2
Maximum M r = 1.496 Minimum BW = 1 − 2ζ 4ζ − 4ζ + 2 = 1.4106 rad/sec

2 4 2

8-8) (a)
At the gain crossover:
Therefore:
at ω = 1.5 Ö
(b)
MATLAB code:
%solving for k:
syms kc
omega=1.5
sol=eval(solve('0.25*kc^2=0.7079^2*((-0.25*omega^3+omega)^2+(-
0.375*omega^2+0.5*kc)^2)',kc))
%ploting bode with K=1.0370
s = tf('s')
K=1.0370;
num_G_a= 0.5*K;
den_G_a=s*(0.25*s^2+0.375*s+1);
CL_a = G_a/(1+G_a)
bode(CL_a);
Bode diagram: data point shows -3dB point at 1.5 rad/sec frequency which is the closed loop
bandwidth
8‐5


Bode Diagram
0
System: CL_a
-20
Magnitude (dB) -40
-60
-80
-100
-120
0
-90
Phase (deg)
-180
-270
-2 -1 0 1 2
10 10 10 10 10
Frequency (rad/sec)
8-9) θ=
α = 90 – θ = 63o
Therefore:
As a result:
8‐6


Therefore:
To change the crossover frequency requires adding gain as:
(b) MATLAB code:

s = tf('s')
%(b)
K = 0.95*2;
num_G_a = 0.5*K;
den_G_a = s*(0.25*s^2+0.375*s+1);
CL_a = G_a/(1+G_a)
bode(CL_a);
figure(2);
sisotool
Peak mag = 2.22 can be converted to dB units by: 20*Log(2.22)= 6.9271 dB
8‐7


By using sisotool and importing the loop transfer function, the overall gain (0.5K) was changed until the
magnitude of the resonance in Bode was about 6.9 dB. At 0.5K=~0.95 or K=1.9, this resonance peak
was achieved as can be seen in the BODE diagram of the following figure:
Root Locus Editor for Open Loop 1 (OL1) Open-Loop Bode Editor for Open Loop 1 (OL1)
6 20
4 0
2 -20
0 -40
-2 -60
-4 -80
G.M.: 4.48 dB
-6 -100 Freq: 2 rad/sec
-8 -6 -4 -2 0 2 4
Stable loop
Bode Editor for Closed Loop 1 (CL1) -120
-90
6
4 -135
2
-180
0
-225
-180
P.M.: 59.1 deg
Freq: 1.11 rad/sec
-360 -270
-2 -1 0 1 2 -1 0 1 2
10 10 10 10 10 10 10 10 10
Frequency (rad/sec) Frequency (rad/sec)
8-10)
− πζ
1 1−ζ
2
M r = 1.4 = Thus, ζ = 0.387 Maximum overshoot = e = 0.2675 (26.75%)
2
2ζ 1 − ζ
2 3
ω r = 3 rad / sec = ω n 1 − 2ζ = 0.8367ω n rad/sec ω n = = 3.586 r ad/sec
0.8367

π π
t max = = = 0.95 sec At ω = 0, M = 0.9.
2 2
ω n 1− ζ 3.586 1 − ( 0.387 )
8‐8


This indicates that the steady‐state value of the unit‐step response is 0.9.
8-11) a) The closed loop transfer function is:
as ,which means ξ = 0.387
According to the transfer function: ξ ω n = 0.1 Ö ω n = 0.129 rad/s

As ωn2 = 0.1K ; then, K = 10 ωn2 = 0.1669
b)
As K = 0.1664, then
which means
Accordingly PM = 42o
As , then
8‐9


8-12)
T BW (rad/sec) Mr
________________________________________________________________
0 1.14 1.54
0.5 1.17 1.09
1.0 1.26 1.00
2.0 1.63 1.09
3.0 1.96 1.29
4.0 2.26 1.46
5.0 2.52 1.63
_________________________________________________________________
8‐13)
T BW (rad/sec) Mr
_________________________________________________________________
0 1.14 1.54
0.5 1.00 2.32
1.0 0.90 2.65
2.0 0.74 2.91
3.0 0.63 3.18
4.0 0.55 3.37
5.0 0.50 3.62
_________________________________________________________________
8-14) The Routh array is:
8‐10


S3 0.25 1
S2 0.375 0.5K
S1 1-1/3 0
S0 0.5K
Therefore:
As , if GH is rearranged as:
then
which gives
where
therefore,
(c)
MATLAB code: Bode diagram:
s = tf('s')
%c)
K = 1.03697;
num_G_a = 0.5*K;
den_G_a =
s*(0.25*s^2+0.375*s+1
);
%create closed-loop
system
G_a =
num_G_a/den_G_a;
8‐11


CL_a = G_a/(1+G_a)
bode(CL_a); Bode Diagram
0
System: CL_a
-20
Magnitude (dB)
Notes: -40
-60
1 - BW is verified by finding - -80

3dB point at Freq = 1.5 rad/sec -100
in the Bode graph at calculated -120
k. 0
2- By comparison to diagram of Phase (deg)

-90
typical 2nd order poles with

different damping ratios, -180
damping ratio is approximated

as: -270
-2 -1 0 1 2
ξ = ~ .707 10 10 10 10 10
Frequency (rad/sec)
8‐12


8‐15 (a)
20
L( s ) = Pω = 1, P = 0
s(1 + 0.1s )(1 + 0.5s )
When ω = 0: ∠L( jω ) = −90o L( jω ) = ∞ When ω = ∞: ∠L( jω ) = −270

o
L( j ω ) = 0
L ( jω ) =
20
=
20 ⎡⎣ −0.6ω − jω 1 − 0.05ω
2
( 2
)⎤⎦ Setting Im L( jω ) = 0
2
(
−0.6ω + jω 1 − 0.05ω
2
) 0.36ω + ω
4 2
(1 − 0.05ω )
2 2
2
1 − 0.05ω = 0 Thus, ω = ±4.47 rad / sec L( j 4.47 ) = −1667
.
360
o
Φ11 = 270 = ( Z − 0.5 Pω − P ) 180 = ( Z − 0.5 ) 180 Thus, Z = = 2

o o o

180
o
The closed‐loop system is unstable. The characteristic equation has two roots in the right‐half
s‐plane.
MATLAB code:
s = tf('s')
%a)
figure(1);
num_G_a= 20;
den_G_a=s*(0.1*s+1)*(0.5*s+1);
nyquist(G_a)

Nyquist Plot of L( jω ):
8‐13


(b)
10
L( s ) = Based on the analysis conducted in part (a), the intersect of the negative
s(1 + 0.1s )(1 + 0.5s )

real axis by the L( jω ) plot is at −0.8333, and the corresponding ω is 4.47 rad/sec.

o
δ ι o
Φ 11 = −90 = Z − 0.5 Pω − P 180 = 180 Z − 90
o
Thus, Z = 0. The closed‐loop system is stable.

MATLAB code:
s = tf('s')
%b)
figure(1);
num_G_a= 10;
den_G_a=s*(0.1*s+1)*(0.5*s+1);
nyquist(G_a)

8‐14


(c)
100(1 + s )
L( s ) = Pω = 1, P = 0.
s(1 + 0.1s )(1 + 0.2 s )(1 + 0.5s )
When ω = 0: ∠L( j 0) = −90o L( j 0) = ∞ When ω = ∞: ∠L( j∞ ) = −270 o L( j∞ ) = 0
When ω = ∞: ∠L( jω ) = −270 L( j ω ) = 0 When ω = ∞: ∠L( jω ) = −270 L( j ω ) = 0

o o

L ( jω ) =
100(1 + jω )
=
(
100(1 + jω ) ⎡⎣ 0.01ω − 0.8ω
4 2
) − jω (1 − 0.17ω )⎤⎦
2
( 0.01ω 4
− 0.8ω
2
) + jω (1 − 0.17ω ) 2
( 0.01ω 4
− 0.8ω
2
) + ω (1 − 0.17ω )
2 2 2 2
4 2 2 4 2
Setting Im L( jω ) = 0 0.01ω − 0.8ω − 1 + 0.17ω = 0 ω − 63ω − 100 = 0
2
Thus, ω = 64.55 ω = ±8.03 rad/sec
⎛ 100 ⎣⎡( 0.01ω 4 − 0.8ω 2 ) + ω 2 (1 − 0.17ω 2 ) ⎦⎤ ⎞

L( j8.03) = ⎜ ⎟ = −10
⎜ ( 0.01ω 2 − 0.8ω 2 )2 + ω 2 (1 − 0.17ω 2 )2 ⎟
⎝ ⎠ ω =8.03
8‐15


Φ 11 = 270 = ( Z − 0.5 Pω − P ) 180 = ( Z − 0.5 ) 180 Thus, Z = 2 The closed‐loop system is

o o o

unstable.
The characteristic equation has two roots in the right‐half s‐plane.
MATLAB code:
s = tf('s')
%c)
figure(1);
num_G_a= 100*(s+1);
den_G_a=s*(0.1*s+1)*(0.2*s+1)*(0.5*s+1);
nyquist(G_a)


8‐16


(d)
10
L( s ) = 2
Pω = 2 P = 0
s (1 + 0.2 s )(1 + 0.5s )
When ω = 0: ∠L( jω ) = −180o L( jω ) = ∞ When ω = ∞: ∠L( jω ) = −360

o
L( jω ) = 0
L ( jω ) =
10
=
(
10 0.1ω − ω + j 0.7ω
4 2 3
)
( 0.1ω 4
−ω
2
) − j 0.7ω 3
( 0.1ω 4
−ω
2
)
2
+ 0.49ω
6
Setting Im L( jω ) = 0, ω = ∞. The Nyquist plot of L( jω ) does not intersect the real axis except at the
origin where ω = ∞.
Φ 11 = ( Z − 0.5 Pω − P ) 180 = ( Z − 1) 180 Thus, Z = 2.

o o

The closed‐loop system is unstable. The characteristic equation has two roots in the right‐half s‐plane.
MATLAB code:
s = tf('s')
%d)
figure(1);
num_G_a= 10;
den_G_a=s^2* (0.2*s+1)*(0.5*s+1);
nyquist(G_a)

8‐17


8‐15 (e)
3( s + 2)
L( s ) = Pω = 1 P = 2
(
s s + 3s + 1
3
)

When ω = 0: ∠L( j 0) = −90o L( j 0 ) = ∞ When ω = ∞: ∠L( j∞ ) = −270

o
L( j∞ ) = 0
L ( jω ) =
3( jω + 2)
=
3( jω + 2) ⎡⎣ ω − 3ω( 4 2
) − jω ⎤⎦ Setting Im L( jω ) = 0,
(ω 4
− 3ω
2
) + jω (4 4
− 3ω
2
)
2
+ω
2
4 2 2
ω − 3ω − 2 = 0 or ω = 3.56 ω = ±189
. rad/sec. . ) = 3
L( j189
Φ11 = ( Z − 0.5 Pω − P ) 180 = ( Z − 2.5 ) 180 = −90 Thus, Z = 2

o o o

MATLAB code:
8‐18


s = tf('s')
%e)
figure(1);
num_G_a= 3*(s+2);
den_G_a=s*(s^3+3*s+1);
nyquist(G_a)

8‐15 (f)
0.1
L( s ) = Pω = 1 P = 0
(
s ( s + 1) s + s + 1
2
)

When ω = 0: ∠L( j 0) = −90 When ω = ∞: ∠L( j∞ ) = −360

o o
L( j 0 ) = ∞ L( j∞ ) = 0
L ( jω ) =
0.1
=
(
0.1 ⎡⎣ ω − 2ω
4 2
) − jω (1 − 2ω )⎤⎦ 2
Settiing Im L( jω ) = 0
(ω 4
− 2ω
2
) + jω (1 − 2ω ) 2
(ω 4
− 2ω
2
) + ω (1 − 2ω )
2 2 2 2
2
ω = ∞ or ω = 0.5 ω = ±0.707 rad / sec L( j 0.707 ) = −0.1333
Φ11 = ( Z − 0.5 Pω − P ) 180 = ( Z − 0.5 ) 180 = −90 Thus, Z = 0 The closed‐loop system is stable.

o o o
8‐19


MATLAB code:
s = tf('s')
%f)
figure(1);
num_G_a= 0.1;
den_G_a=s*(s+1)*(s^2+s+1);
nyquist(G_a)

8‐15 (g)
100
L( s ) = Pω = 3 P = 0
(
s ( s + 1) s + 2
2
)
When ω = 0: ∠L( j 0) = −90 When ω = ∞: ∠L( j∞ ) = −360
o o
L( j 0 ) = ∞ L( j∞ ) = 0
The phase of L( jω ) is discontinuous at ω = 1.414 rad/sec.
o

o
(
Φ11 = 35.27 + 270 − 215.27
o o
) = 90
o
Φ11 = ( Z − 1.5 ) 180 = 90 Thus, P11 =
o o 360
o
= 2
180
8‐20


MATLAB code:
s = tf('s')
%g)
figure(1);
num_G_a= 100;
den_G_a=s*(s+1)*(s^2+2);
nyquist(G_a)

8‐15 (h)
10( s + 10)
L( s ) = Pω = 1 P = 0
s( s + 1)( s + 100)

When ω = 0: ∠L( j 0) = −90o L( j 0 ) = ∞ When ω = ∞: ∠L( j∞ ) = −180

o
L( j∞ ) = 0
L ( jω ) =
10( jω + 10)
=
(
10( jω + 10) ⎡⎣ −101ω − jω 100 − ω
2 2
)⎤⎦
−101ω + jω (100 − ω ) (100 − ω )
2 2 2
10201ω + ω
4 2 2
8‐21


Setting Im L( jω ) = 0, ω = 0 is the only solution. Thus, the Nyquist plot of L( jω ) does not intersect
the real axis, except at the origin.

Φ 11 = ( Z − 0.5 Pω − P ) 180 = ( Z − 0.5 ) 180 = −90 Thus, Z = 0.

o o o

The closed‐loop system is stable.
MATLAB code:
s = tf('s')
%h)
figure(1);
num_G_a= 10*(s+10);
den_G_a=s*(s+1)*(s+100);
nyquist(G_a)

8‐22



8-16
MATLAB code:
s = tf('s')
%a)
figure(1);
num_G_a= 1;
den_G_a=s*(s+2)*(s+10);
nyquist(G_a)
%b)
figure(2);
num_G_b= 1*(s+1);
den_G_b=s*(s+2)*(s+5)*(s+15);
nyquist(G_b)
%c)
figure(3);
num_G_c= 1;
den_G_c=s^2*(s+2)*(s+10);
nyquist(G_c)
%d)
figure(4);
num_G_d= 1;
den_G_d=(s+2)^2*(s+5);
8‐23


nyquist(G_d)
%e)
figure(5);
num_G_e= 1*(s+5)*(s+1);
den_G_e=(s+50)*(s+2)^3;
nyquist(G_e)
Nyquist graph, part(a):
Nyquist Diagram
1.5
0.5
Imaginary Axis
-0.5
-1
-1.5
-1 -0.9 -0.8 -0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0
Real Axis
Nyquist graph, part(b):
8‐24


Nyquist Diagram
4
1
Imaginary Axis
-1
-2
-3
-4
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4
Real Axis
Nyquist graph, part(c):
Nyquist Diagram
0.5
0.4
0.3
0.2
0.1
Imaginary Axis
-0.1
-0.2
-0.3
-0.4
-0.5
-10 -8 -6 -4 -2 0 2
Real Axis
Nyquist graph, part(d):
8‐25


Nyquist Diagram
0.04
0.03
0.02
0.01
Imaginary Axis
-0.01
-0.02
-0.03
-0.04
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4
Real Axis
Nyquist graph, part (e):
Nyquist Diagram
0.01
0.008
0.006
0.004
0.002
Imaginary Axis
-0.002
-0.004
-0.006
-0.008
-0.01
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4
Real Axis
8‐26


8-17 (a)
K
G ( s ) = 2
Pω = 0 P = 0
( s + 5)

o K
∠G ( 0) = 180 ( K < 0) G ( j 0) =
o
∠G ( j 0 ) = 0 ( K > 0 )
25
G ( j∞ ) = 0
o o
G ( j∞ ) = −180 (K > 0) ∠G ( j∞ ) = 0 (K < 0)
For stability, Z = 0.
Φ 11 = − ( 0.5 Pω + P ) 180 = 0

o o

o
0 < K < ∞ Φ11 = 0 Stable
o
K < −25 Φ11 = 180 Unstable
o
−25 < K < 0 Φ11 = 0 Stable

The system is stable for −25 < K < ∞ .
8‐17 (b)
K
G ( s ) = 3
Pω = 0 P = 0
( s + 5)
o K
∠G ( 0) = 180 ( K < 0) G ( j 0) =
o
∠G ( j 0 ) = 0 ( K > 0 )
125
∠G ( j∞ ) = 270 (K < 0) G ( j∞ ) = 0
o o
G ( j∞ ) = −270 (K > 0)
8‐27


Φ 11 = − ( 0.5 Pω + P ) 180 = 0

o o

o
0 < K < 1000 Φ11 = 0 Stable
o
K > 1000 Φ11 = 360 Unstable
o
K < −125 Φ11 = 180 Unstable
o
−125 < K < 0 Φ11 = 0 Stable
The system is stable for −125 < K < 0.
8‐17 (c)
K
G ( s ) = 4
Pω = P = 0
( s + 5)
o K
∠G ( 0 ) = 180 ( K < 0) G ( j 0) =
o
∠G ( j 0 ) = 0 ( K > 0 )
625
∠G ( j∞ ) = 180 (K < 0) G ( j∞ ) = 0

o o
G ( j∞ ) = 0 (K > 0)
8‐28


Φ 11 = − ( 0.5 Pω + P ) 180 = 0

o o

o
0 < K < 2500 Φ11 = 0 Stable
o
K > 2500 Φ11 = 360 Unstable
o
K < −625 Φ11 = 180 Unstable

o
−625 < K < 0 Φ11 = 0 Stable
The system is stable for −625 < K < 2500.

8-18) The characteristic equation:
or
if , the cross real axis at s = 0.1. Ö For stability
if K < 0, the Nyquist cross the real axis at .So, for stability,
therefore, the rage of stability for the system is
MATLAB code:
s = tf('s')
K=1
G= K/(s^2+2*s+2);
H=1/(s+1);
GH=G*H;
sisotool
8‐29


K=10.15
G2= K/(s^2+2*s+2);
H2=1/(s+1);
GH2=G2*H2;
nyquist(GH2)
xlim[(-1.5,.5)]
ylim[(-1,1)]
After generating the feed-forward (G) and feedback (H) transfer functions in the MATLAB code, these
transfer functions are imported to sisotool. Nyquist diagram is added to the results of sisotool. The
overall gain of the transfer function is changed until Nyquist diagram passes through -1+0j point. Higher
values of K resulted in unstable Nyquist diagram. Therefore K<10.15 determines the range of stability
for the closed loop system.
Nyquist at margin of stability:
Nyquist Diagram
4
1
Imaginary Axis
-1
-2
-3
-4
-2 -1 0 1 2 3 4 5 6
Real Axis
8-19)
( )
s s + 2s + s + 1 + K s + s + 1 = 0
3 2
( 2
)
8‐30


Leq ( s ) =
(
K s + s +1
2
) Pω = 1 P = 0 Leq ( j 0 ) = ∞∠ − 90
o
Leq ( j∞ ) = 0∠180
o

(
s s + 2s + s + 1
3 2
)
Leq ( jω ) =
(
K ⎡⎣ 1 − ω ) + jω ⎤⎦ = K ⎡⎣− (ω
2 6
+ω
4
) − jω (ω − 2ω + 1)⎤⎦
4 2
(ω 4
−ω
2
) + jω (1 − 2ω ) 2
(ω 4
−ω
2
) + ω (1 − 2ω )
2 2 2 2
Setting Im Leq ( jω ) = 0

4 2
ω − 2ω + 1 = 0
Thus, ω = ±1 rad/sec are the real solutions.

Leq ( j1) = − K
For stability,
o o

Φ11 = − ( 0.5Pω + P )180 = −90
When K = 1 the system is marginally stable.

o
K > 0 Φ11 = −90 Stable
o
K < 0 Φ11 = +90 Unstable

8‐31


Routh Tabulation
4
s 1 K +1 K
3
s 2 K +1
2 K +1
s K K > −1
2
2 2
1 K − 2K + 1 ( K − 1)
s =
K +1 K +1
0
s K K > 0
1 2
When K = 1 the coefficients of the s row are all zero. The auxiliary equation is s + 1 = 0 The solutions
are ω = ±1 rad/sec. Thus the Nyquist plot of Leq ( jω ) intersects the −1 point when K = 1, when ω = ±1
rad/sec. The system is stable for 0 < K < ∞ , except at K = 1.
8-20) Solution is similar to the previous problem. Let’s use Matlab as an alternative approach
MATLAB code:
s = tf('s')
figure(1);
K=8.09
num_GH= K;
den_GH=(s^3+3*s^2+3*s+1);
GH=num_GH/den_GH;
nyquist(GH)
xlim([-1.5,.5])
ylim([-1,1])
sisotool;
8‐32


After generating the loop transfer function and analyzing Nyquist in MATLAB sisotool, it was found
that for values of K higher than ~8.09, the closed loop system is unstable. Following is the Nyquist
diagram at margin of stability.
Part(a), Nyquist at margin of stability:
Nyquist Diagram
1
0.8
0.6
0.4
0.2
Imaginary Axis
-0.2
-0.4
-0.6
-0.8
-1
-1.5 -1 -0.5 0 0.5
Real Axis
Part(b), Verification by Routh-Hurwitz criterion:

Using Routh criterion, the coefficient table is as follows:
S3 1 3
S2 3 K+1
S1 (8-K)/3 0
S0 K+1 0
The system is stable if the content of the 1st column is positive:

(8-K)/3>0 Æ K<8
8‐33


K+1>0 Æ K>-1
which is consistent with the results of the Nyquist diagrams.
8-21)
Parabolic error constant K a = lim s G ( s ) = lim10 ( K P + K D s ) = 10 K P = 100 Thus K P = 10

2

s →0 s →0

2
Characteristic Equation: s + 10 K D s + 100 = 0

10 KD s
Geq ( s ) = 2
Pω = 2 P = 0
s + 100
For stability,
o o
Φ11 = − ( 0.5 Pω + P )180 = −180

The system is stable for 0 < K D < ∞ .

8‐34


8‐22 (a) The characteristic equation is 1 + G ( s ) − G ( s ) − 2 G ( s ) 2 = 1 − 2 G ( s ) 2 = 0

2
2 −2 K
Geq ( s ) = −2 G ( s ) = 2 2
Pω = 0 P = 0
( s + 4 ) ( s + 5)
Geq ( jω ) =
−2 K
2
=
2
(
−2 K ⎡⎣ 400 − 120ω + ω
2 2
) − jω ( 360 − 18ω )⎤⎦
2
( 400 − 120ω 2
+ω
4
) + jω ( 360 − 18ω ) 2
( 400 − 120ω 2
+ω
2
) + ω ( 360 − 18ω )
2 2 2
2
K
Geq ( j 0) = ∠180
o
Geq ( j∞ ) = 0∠180
o
Setting Im Geq ( jω ) = 0
200
2
K
ω = 0 and ω = ±4.47 rad / sec Geq ( j 4.47 ) =
800
For stability,
Φ 11 = − ( 0.5 Pω + P ) 180 = 0

o o

2
The system is stable for K < 200
or K < 200

8‐35


8‐22 (b)
4 3 2 2
Characteristic Equation: s + 18 s + 121s + 360 s + 400 − 2 K = 0
Routh Tabulation
4 2
s 1 121 400 − 2 K
3
s 18 360
2 2
s 101 400 − 2 K
2
1 29160 - 36 K 2
s 29160 + 36 K > 0
101
s 0 400 − 2 K 2 K 2 < 200
Thus for stability, K < 200
8‐23 (a)
Nyquist Plot
83.33N
G ( s ) =
s( s + 2)( s + 11767
. )
For stability, N < 3.89
Thus N < 3 since N must be an integer.

8‐36


(b)
2500
G ( s) =
s( 0.06 s + 0.706)( As + 100)
Characteristic Equation: 0.06 As 3 + ( 6 + 0.706 A) s 2 + 70.6 s + 2500 = 0

2
As ( 0.06 s + 0.706 )
Geq ( s ) = 2
Since Geq ( s ) has more zeros than poles, we should sketch the Nyquist
6 s + 70.6 s + 2500
plot of 1/ Geq ( s ) for stability study.

1
=
( 2500 − 6ω ) + j 70.6ω = ⎡⎣( 2500 − 6ω ) + j 70.6ω ⎤⎦ ( −0.706ω + j 0.06ω )
2 2 2 3
Geq ( jω ) A ( −0.706ω − j 0.06ω ) A ( 0.498ω + 0.0036ω )

2 3 4 6
⎡ 1 ⎤
⎥ = 0
o o
1 / Geq ( j 0 ) = ∞∠ − 180 1 / Geq ( j∞ ) = 0∠ − 90 Setting Im ⎢
⎣ Geq ( jω ) ⎦
2 1 −4.23
100.156 − 0.36ω = 0 ω = ±16.68 rad/sec =
Geq ( j16.68) A
For stability,
Φ 11 = − ( 0.5 Pω + P ) 180 = −180

o o

o
For A > 4.23 Φ11 = 180 Unstable

o
For 0 < A < 4.23 Φ11 = −180 Stable
The system is stable for 0 < A < 4.23.
(c)
8‐37


2500
G ( s ) =
s( 0.06 s + 0.706 )( 50 s + Ko )
Characteristic Equation: s( 0.06 s + 0.706)(50 s + Ko ) + 2500 = 0
Ko s( 0.06 s + 0.706)
Geq ( s ) = Pω = 0 P = 0 Geq ( j 0 ) = 0∠ 90
o o
3 2
Geq ( j∞ ) = 0∠ − 90
3s + 35.3s + 2500
Geq ( jω ) =
K o ( −0.06ω + 0.706 jω )
3
=
( )(
K o −0.06ω + 0.706 jω ⎡⎣ 2500 − 35.5ω
2 2
) + j3ω 3
⎤⎦

( 2500 − 35.3ω ) − j3ω
2 3
( 2500 − 35.3ω ) 2 2
+ 9ω
6
4 2 2
Setting Im Geq ( jω ) = 0 ω + 138.45ω − 9805.55 = 0 ω = 516
. ω = ±7.18 rad/sec

Geq ( j 7.18 ) = −0.004 K o
For stability, Φ 11 = − ( 0.5 Pω + P ) 180 = 0

o o

For stability, 0 < Ko < 217.4

8‐38


8‐24 (a) Kt = 0:
Y (s) 10000 K 10000 K

G(s) = = =
E ( s) (
s s + 10 s + 10000 K t
2
) s ( s + 10)
2

The (−1, j0) point is enclosed for all
values of K. The system is unstable
for all values of K.
(b) Kt = 0.01:
G(s) =
10000 K
G ( jω ) =
10000 K ⎣⎡ −10ω − jω 100 − ω
2
( 2
)⎦⎤
(
s s + 10 s + 100
2
) 100ω + ω
4 2
(100 − ω ) 2 2
Setting Im G ( jω ) = 0 ω 2 = 100
ω = ±10 rad / sec G ( j10 ) = −10 K
The system is stable for 0 < K < 0.1

8‐39


(c) Kt = 0.1:
G( s) =
10000 K
G ( jω ) =
10000 K ⎡⎣ −10ω − jω 1000 − ω
2
( 2
)⎤⎦
(
s s + 10 s + 1000
2
) 100ω + ω
4 2
(1000 − ω ) 2 2
Setting Im G ( jω ) = 0 ω 2 = 100 ω = ±31.6 rad/sec . ) = −K

G ( j 316

For stability, 0 < K < 1

8‐25) The characteristic equation for K = 10 is:
3 2
s + 10 s + 10,000 Kt s + 100,000 = 0
10,000 Kt s
Geq ( s ) = 3 2
Pω = 0 P = 2
s + 10 s + 100,000
Geq ( jω ) =
10, 000 K t jω
=
(
10, 000 K t ⎣⎡ −ω + jω 10, 000 − 10ω
4 2
)⎦⎤ Setting Im G ( jω ) = 0
100, 000 − 10ω − jω (10, 000 − 10ω )
2 3 2 eq
+ω
2 6
8‐40


2
ω = 0, ω = 10,000
ω = ±100 rad / sec Geq ( j100) = − Kt
For stability,
Φ 11 = − ( 0.5 Pω + P ) 180 = −360

o o

The system is stable for Kt > 0.
8-26)
a) Ö
b) Ö
c) Ö
MATLAB code:
s = tf('s')
figure(1);
J=1;
B=1;
K=1;
Kf=0
G1= K/(J*s+B);
CL1=G1/(1+G1*Kf);
H2 = 1;
G1G2 = CL1/s;
L_TF=G1G2*H2;
nyquist(L_TF)
sisotool
8‐41


Part (a), Kf=0: by plotting the Nyquist diagram in sisotool and varying the gain, it was observed that
all values of gain (K) will result in a stable system. Location of poles in root locus diagram of the second
figure will also verify that.
Nyquist Diagram
20
15
10
5
Imaginary Axis
-5
-10
-15
-20
-1 -0.9 -0.8 -0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0
Real Axis
8‐42


1 40
20
0.5
0
0
-20
-0.5
-40
G.M.: Inf
-1 -60
Freq: Inf
-1 -0.8 -0.6 -0.4 -0.2 0
Stable loop
50
-90
-50
-100 -135
0
-90
P.M.: 54.9 deg
Freq: 0.704 rad/sec
-180 -180
-2 -1 0 1 2 -2 -1 0 1 2
10 10 10 10 10 10 10 10 10 10
Part (b), Kf=0.1: The result and approach is similar to part (a), a sample of Nyquist diagram is
presented for his case as follows:
Nyquist Diagram
15
10
5
Imaginary Axis
-5
-10
-15
-1 -0.9 -0.8 -0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0
Real Axis
8‐43


Part (c), Kf=0.2: The result and approach is similar to part (a), a sample of Nyquist diagram is
presented for his case as follows:
Nyquist Diagram
15
10
5
Imaginary Axis
-5
-10
-15
-1 -0.9 -0.8 -0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0
Real Axis
8-27)
MATLAB code:
s = tf('s')
figure(1);
J=1;
B=1;
K=10;
Kf=0.2
G1= K/(J*s+B);
CL1=G1/(1+G1*Kf);
H2 = 1;
G1G2 = CL1/s;
L_TF=G1G2*H2;
nyquist(L_TF)
After assigning K=10, different values of Kf has been used in the range of 0.01<K<104. The Nyquist
diagrams shows the stability of the closed loop system for all 0<K<∞. A sample of Nyquist diagram is
plotted as follows:
8‐44


Nyquist Diagram
25
20
15
10
5
Imaginary Axis
-5
-10
-15
-20
-25
-1.4 -1.2 -1 -0.8 -0.6 -0.4 -0.2 0
Real Axis
8-28) a) K > 2 Ö system is stable
b) 0 < K < 1 and -2 < K < 0 Ö -2 < K < 1 Ösystem is stable
MATLAB code:
s = tf('s')
%a)
figure(1);
K=1
num_GH_a= K*(s+1);
den_GH_a=(s-1)^2;
nyquist(GH_a)
%b)
figure(2);
K=1
num_GH_b= K*(s-1);
den_GH_b=(s+1)^2;
nyquist(GH_b)
8‐45


sisotool
Part(a): Using MATLAB sisotool, the transfer function gain can be iteratively changed in order to
obtain different phase margins. By changing the gain so that PM=0 (margin of stability), K>~2 resulted
in stable Nyquist diagram for part(a). Following two figures illustrate the sisotool and Nyquist results at
margin of stability for part (a).
2 10
1 5
0
0
-5
-1
-10 G.M.: -0.00581 dB

-2 Freq: 1.73 rad/sec
-4 -3 -2 -1 0 1
Stable loop
100
-90
P.M.: 0.0688 deg
50
0 -180
-50 -225
180
-270
0
-315
-180 -360
-2 -1 0 1 2 0
10 10 10 10 10 10
8‐46


Nyquist Diagram
2
1.5
Imaginary Axis 0.5
-0.5
-1
-1.5
-2
-1.5 -1 -0.5 0 0.5 1 1.5 2 2.5
Real Axis
Part(b): Similar methodology applied as in part (a). K<1 results in closed loop stability. Following are
sisotool and Nyquist results at margin of stability (K=1):
1 0
0.5
-5
0
-10
-0.5
-15
G.M.: -0.00236 dB
-1 Freq: 0 rad/sec
-5 -4 -3 -2 -1 0 1 2
Unstable loop
80
180
60
40
135
20
0 90
-20
-40 45
90
45 0
0
-45 P.M.: -4.01 deg
-45
Freq: 0.0233 rad/sec
-90 -90
-6 -4 -2 0 2 -2 -1 0 1
10 10 10 10 10 10 10 10 10
8‐47


Nyquist Diagram
1
0.8
0.6
0.4
0.2
Imaginary Axis
-0.2
-0.4
-0.6
-0.8
-1
-1.2 -1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6
Real Axis
8-29) (a)
− Td s 100
Let G ( s ) = G1 ( s )e Then G1 ( s ) =
(
s s + 10 s + 100
2
)
100 100
Let =1 or =1
−10ω + jω 100 − ω
2
( 2
) ⎡100ω 4 + ω 2 (100 − ω 2 )2 ⎤
1/ 2
⎣ ⎦
(100 − ω )
2
Thus 100ω + ω = 10, 000 ω − 100ω + 10, 000ω − 10, 000 = 0
4 2 2 6 4 2

The real solution for ω are ω = ±1 rad/sec.
8‐48


⎡ 100 − ω 2 ⎤
−1
∠G1 ( j1) = − tan ⎢ = 264.23
o
⎥
⎣ −10ω ⎦ ω =1
Equating ωTd =
( 264.23 o
− 180 π
o
)
ω =1
180
84.23π
= = 1.47 rad
180
Thus the maximum time delay for stability
is
Td = 1.47 sec.
(b) Td = 1 sec.
−s − jω
100 Ke 100 Ke
G( s) = G ( jω ) =
(
s s + 10 s + 100
2
) (
−10ω + jω 100 − ω
2 2
)
At the intersect on the negative real axis, ω = 1.42 rad/sec.

G ( j1.42) = −0.7107 K .
The system is stable for
0 < K < 1.407

8‐49


8‐30 (a) K = 0.1
− Td s
10e − Td s
G ( s ) = = G1 ( s )e
(
s s + 10 s + 100
2
)
10 10
Let =1 or = 1
−10ω + jω 100 − ω
2
( 2
) ⎡100ω 4 + ω 2 (100 − ω 2 )2 ⎤
1/ 2
⎣ ⎦
ω − 100ω + 10,000ω − 100 = 0 The real solutions for ω is ω = ±0.1 rad/sec.

6 4 2
Thus
⎡ 100 − ω 2 ⎤ −1
∠G1 ( j 0.1) = − tan ⎢ = 269.43
o
⎥
⎣ −10ω ⎦ ω = 0.1
Equate ωTd =
( 269.43 o
− 180 π
o
) = 1.56 rad We have Td = 15.6 sec.
ω = 0.1
180
o
We have the maximum time delay

for stability is 15.6 sec.

8‐30 (b) Td = 0.1 sec.
−0.1 s −0.1 jω
100 Ke 100 Ke
G( s) = G ( jω ) =
(
s s + 10 s + 100
2
) (
−10ω + jω 100 − ω
2 2
)
At the intersect on the negative real axis,
ω = 6.76 rad/sec. G ( j 6.76 ) = −0.1706 K
8‐50


The system is stable for
0 < K < 5.86

8-31)
MATLAB code:
s = tf('s')
%a)
figure(1);
K=1
num_GH_a= K;
den_GH_a=s*(s+1)*(s+1);
nyquist(GH_a)
%b)
figure(2);
K=20
num_GH_b= K;
den_GH_b=s*(s+1)*(s+1);
nyquist(GH_b)
sisotool;
8‐51


By using sisotool and importing the loop transfer function, different values of K has been tested which
resulted in a stable system when K<2, and unstable system for K>2. Following diagrams correspond to
margin of stability:
3 30
2 20
10
1
0
0
-10
-1
-20
-2
-30
G.M.: -0.282 dB
-3 Freq: 1 rad/sec
-4 -3 -2 -1 0 1 -40
Unstable loop
100
-90
0
-135
-100
-200 -180
0
-180
-225
-360 P.M.: -0.925 deg
Freq: 1.02 rad/sec
-540 -270
-1 0 1 2 0
10 10 10 10 10
Part(a): small K resulted in stable system as shown below for K=0.219:
8‐52


3 10
2 0
-10
1
-20
0
-30
-1
-40
-2
-50
G.M.: 19.2 dB
-3 Freq: 1 rad/sec
-4 -3 -2 -1 0 1 -60
Stable loop
0
-90
-50
-135
-100
-150 -180
0
-225
-180
P.M.: 66.3 deg
Freq: 0.21 rad/sec
-360 -270
-2 -1 0 1 2 0
10 10 10 10 10 10
Nyquist Diagram
10
2
Imaginary Axis
-2
-4
-6
-8
-10
-1 -0.9 -0.8 -0.7 -0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0
Real Axis
Part(b): Large K resulted in unstable system as shown below for K=20:
8‐53


3 50
40
2
30
1
20
0 10
-1 0
-10
-2
-20
G.M.: -20 dB
-3 Freq: 1 rad/sec
-4 -3 -2 -1 0 1 -30
Unstable loop
-90
0
-50 -135
-100 -180
0
-90
-180 -225
-270 P.M.: -47.8 deg
-450 -270
-1 0 1 2 0
10 10 10 10 10
Nyquist Diagram
30
20
10
Imaginary Axis
-10
-20
-30
-15 -10 -5 0 5
Real Axis
8‐54


The system is stable for small value of K, since there is no encirclement of the s = -1
The system is unstable for large value of K, since the locus encirclement the twice in
CCW; which means two poles are in the right half s-plane.
8-32) (a) The transfer function (gain) for the sensor‐amplifier combination is 10 V/0.1 in = 100 V/in. The velocity
of flow of the solution is
3
10 in / sec
v= = 100 in/sec
0.1 in
The time delay between the valve and the sensor is Td = D / v sec. The loop transfer function is
− Td s
100 Ke
G( s) = 2

s + 10 s + 100
8‐55


8‐56


8-33)
(a) The transfer function (gain) for the sensor‐amplifier combination is 1 V/0.1 in = 10 V/in. The velocity
of flow of the solutions is
3
10 in / sec
v= = 100 in / sec
0.1 in
The time delay between the valve and sensor is Td = D / v sec. The loop transfer function is
− Td s
10 Ke
G( s) = 2

s + 10 s + 100
(b) K = 10:
− jω Td
100e
G ( jω ) =
− Td s
G ( s ) = G1 ( s )e
(100 − ω ) + j10ω2
100
(100 − ω )
2
=1 + 100ω = 10, 000
2 2
Setting
( 100 − ω
2
) + j10ω
4 2
Thus, ω − 100ω = 0 Real solutions: ω = 0, ω = ±10 rad / sec
∠G1 ( j10) = − tan ⎜

−1 ⎛ 10ω ⎞ = −90
2 ⎟
o

⎝ 100 − ω ⎠ ω =10
Thus,
90 π π
o
10Td = o
= rad
180 2
Thus,
π
Td = = 0.157 sec
20

Maximum D = vTd = 100 × 0.157 = 15.7 in
8‐57


(c) D = 10 in.
−0.1s
D 10 10 Ke
Td = = = 0.1 sec G( s) = 2

v 100 s + 10 s + 100
The Nyquist plot of G ( jω ) intersects the negative real axis at ω = 12.09 rad/sec. G ( j ) = −0.0773K
For stability, the
maximum value of K is 12.94 .

8-34)
The system (GH) has zero poles in the right of s plane: P=0.
According to Nyquist criteria (Z=N+P), to ensure the stability which means the number of right poles of
1+ GH=0 should be zero (Z=0), we need N clockwise encirclements of Nyquist diagram about -1+0j
point. That is N=-P or in other words, we need P counter-clockwise encirclement about -1+0j. In this
case, we need 0 CCW encirclements.
8-34(a) According to Nyquist diagrams, this happens when K<-1. The three Nyquist diagrams are
plotted with K=-10, K=-1, K=10 as examples:
MATLAB code:
s = tf('s')
%a)
figure(1);
K=-10
num_G_a = K ;
8‐58


den_G_a =(s+1);
num_H_a = (s+2);
den_H_a = (s^2+2*s+2);
H_a = num_H_a/den_H_a;
OL_a = G_a*H_a
nyquist(OL_a)
Case 1) Nyquist graph, K=-10: margin of stability K<-1 unstable
figure(2);
K=-1
num_G_a = K ;
den_G_a =(s+1);
num_H_a = (s+2);
den_H_a = (s^2+2*s+2);
%CL_a = G_a/(1 + G_a*H_a);
OL_a = G_a*H_a
nyquist(OL_a)
8‐59


Case 2) Nyquist graph, K=-1: marginally unstable
figure(3);
K=10
num_G_a = K ;
den_G_a =(s+1);
num_H_a = (s+2);
den_H_a = (s^2+2*s+2);
%CL_a = G_a/(1 + G_a*H_a);
OL_a = G_a*H_a
nyquist(OL_a)
8‐60


Case 3) Nyquist graph, K=10: stable case, -1<K no CCW encirclement about -1+0j point
8‐61


8-34 (b)
For K<-1 (unstable), there will be 1 real pole in the right hand side of s-plane for the closed loop system,
by running the following code.
K=-10
num_G_a = K ;
den_G_a =(s+1);
num_H_a = (s+2);
den_H_a = (s^2+2*s+2);
OL_a = G_a*H_a
nyquist(OL_a)
CL=1/(1+OL_a)
pole(CL)
K =
-10
Transfer function:
-10 s - 20
---------------------
s^3 + 3 s^2 + 4 s + 2
Transfer function:
s^3 + 3 s^2 + 4 s + 2
----------------------
s^3 + 3 s^2 - 6 s - 18
ans =
2.4495
-3.0000
-2.4495
For K=-1 (marginally unstable), there will be 2 negative complex conjugate poles and a pole at zero for
the closed loop system, by running the following code.
K=-1
num_G_a = K ;
den_G_a =(s+1);
num_H_a = (s+2);
den_H_a = (s^2+2*s+2);
8‐62


OL_a = G_a*H_a
nyquist(OL_a)
CL=1/(1+OL_a)
pole(CL)
K=
-1
Transfer function:
-s - 2
---------------------
s^3 + 3 s^2 + 4 s + 2
Transfer function:
s^3 + 3 s^2 + 4 s + 2
---------------------
s^3 + 3 s^2 + 3 s
ans =
0
-1.5000 + 0.8660i
-1.5000 - 0.8660i
Note: you may also wish to use MATLAB sisotool.

See alternative solution to 8-38.
8‐63


8-34(c) The Characteristic Equation is: s3 + 3 s2 + (4+K) s + 2+2K

Using Routh criterions, the coefficient table is as follows:
s3 1 4+K
s2 3 2K+2
s1 K+10 0
s0 2K+2 0
The system is stable if the content of the 1st column is positive:

K+10>0 Æ K>-10
2K+2>0 Æ K>-1
which is consistent with the results of the Nyquist diagrams. For K>-1 system is stable.
8-35) (a) M r = 2.06, ω r = 9.33 rad / sec, BW = 15.2 rad / sec
(b)
1 4 2
M r = = 2.06 ζ − ζ + 0.0589 = 0 The solution for ζ < 0.707 is ζ = 0.25.
2
2ζ 1 − ζ
2 9.33
ω r 1 − 2ζ = 9.33 rad / sec Thus ω n = = 9.974 rad / sec
0.9354
ωn
2
99.48 19.94
GL ( s ) = = = BW = 15.21 rad/sec
s ( s + 2ζω n ) s ( s + 4.987) s (1 + 0.2005 s )
8‐64


8-36) Assuming a unity feedback loop (H=1), G(s)H(s)=G(s)

(a)
5
G ( s ) =
s(1 + 0.5s )(1 + 0.1s )

8‐65


8‐36 (b)

10
G ( s ) =
s(1 + 0.5s )(1 + 0.1s )
(c)
500
G ( s ) =
( s + 1.2)( s + 4 )( s + 10 )

8‐66


(d)
10( s + 1)
G ( s ) =
s( s + 2 )( s + 10 )
8‐36 (e)
0.5
G ( s ) =
(
s s + s +1
2
)

8‐67


(f)

−s
100e
G ( s ) =
(
s s + 10 s + 50
2
)

8‐68


(g)

−s
100e
G ( s ) =
(
s s + 10 s + 100
2
)

8‐36 (h)
10( s + 5)
G ( s ) =
(
s s + 5s + 5
2
)

(a)
8‐69


MATLAB code:
s = tf('s')
num_G_a= 5;
den_G_a=s*(0.5*s+1)*(0.1*s+1);
G_a=num_G_a/den_G_a
margin(G_a)
Bode diagram:
(b)
MATLAB code:
s = tf('s')
num_G_a= 10;
den_G_a=s*(1+0.5*s)*(1+0.1*s);
G_a=num_G_a/den_G_a
margin(G_a)
8‐70


(c)
MATLAB code:
s = tf('s')
num_G_a= 500;
den_G_a=s*(s+1.2)*(s+4)*(s+10);
G_a=num_G_a/den_G_a
margin(G_a)
8‐71


(d)
MATLAB code:
s = tf('s')
num_G_a= 10*(s+1);
den_G_a=s*(s+2)*(s+10);
G_a=num_G_a/den_G_a
margin(G_a)
8‐72


(e)
MATLAB code:
s = tf('s')
num_G_a= 0.5;
den_G_a=s*(s^2+s+1);
G_a=num_G_a/den_G_a
margin(G_a)
8‐73


(f)
MATLAB code:
s = tf('s')
num_G_a= 100*exp(-s);
den_G_a=s*(s^2+10*s+50);
G_a=num_G_a/den_G_a
margin(G_a)
8‐74


(g)
MATLAB code:
s = tf('s')
num_G_a= 100*exp(-s);
den_G_a=s*(s^2+10*s+100);
G_a=num_G_a/den_G_a
margin(G_a)
8‐75


(h)
MATLAB code:
s = tf('s')
num_G_a= 10*(s+5);
den_G_a=s*(s^2+5*s+5);
G_a=num_G_a/den_G_a
margin(G_a)
8‐76


8‐77


8-37)
MATLAB code:
s = tf('s')
num_GH_a= 25*(s+1);
den_GH_a=s*(s+2)*(s^2+2*s+16);
GH_a=num_GH_a/den_GH_a
margin(GH_a)
8‐78


8-38) MATLAB code:

s = tf('s')
num_G_a= 25*(s+1);
den_G_a=s*(s+2)*(s^2+2*s+16);
G_a=num_G_a/den_G_a
margin(G_a)
Bode diagram: PM=101 deg, GM=3.91 dB @ 4.22 rad/sec
Bode Diagram
Gm = 3.91 dB (at 4.22 rad/sec) , Pm = 101 deg (at 1.08 rad/sec)
20
0
Magnitude (dB)
-20
-40
-60
-80
-100
-45
-90
Phase (deg)
-135
-180
-225
-270
-1 0 1 2
10 10 10 10
Frequency (rad/sec)
8-38 Alternative solution

MATLAB code:
s = tf('s')
%a)
figure(1);
num_G_a = 1 ;
den_G_a =(s+1);
num_H_a = (s+2);
den_H_a = (s^2+2*s+2);
8‐79


CL_a = G_a/(1 + G_a*H_a);
margin(CL_a)
sisotool

Bode diagram: for k=1
Bode Diagram
Gm = Inf , Pm = Inf
0
-10
Magnitude (dB)
-20
-30
-40
0
Phase (deg)
-45
-90
-1 0 1 2
10 10 10 10
Frequency (rad/sec)

Using MATLAB sisotool, the transfer function gain can be iteratively changed in order to obtain different phase
margins. By changing the gain K between very small and very big numbers, it was found that the closed loop
system are stable (positive PM) for every positive K in this system.
K=0.034
8‐80


10 -20
5 -40
-60
0
-80
-5
-100 G.M.: Inf

-10 Freq: Inf
-2 -1.5 -1 -0.5 0
Stable loop
0
0
-50 -45
-100 -90
0
-135
-45
P.M.: Inf
Freq: NaN
-90 -180
-2 -1 0 1 2 -1 0 1 2
10 10 10 10 10 10 10 10 10
K=59.9
10 40
5 20
0
0
-20
-5
-40 G.M.: Inf

-10 Freq: Inf
-2 -1.5 -1 -0.5 0
Stable loop
40
0
20
-45
0
-20 -90
90
-135
0
P.M.: 7.67 deg
Freq: 7.89 rad/sec
-90 -180
-1 0 1 2 -1 0 1 2
10 10 10 10 10 10 10 10
8‐81


In order to test the negative range of K, ‐1 was multiplied to the loop transfer function through the following
code, and sisotool was used again.
figure(1);
num_G_a = -1 ;
den_G_a =(s+1);
num_H_a = (s+2);
den_H_a = (s^2+2*s+2);
CL_a = G_a/(1 + G_a*H_a);
margin(CL_a)
sisotool
at K=‐1, margin of stability is observed as PM~=0:
K= ‐1
1 10
0.5 -10
-20
0 -30
-40
-0.5
-50
-60
G.M.: -0.0609 dB
-1 Freq: 0 rad/sec
-4 -2 0 2 4 -70
Unstable loop
40 180
20
0 135
-20
-40 90
135
90
45
45 P.M.: -11.8 deg
Freq: 0.137 rad/sec
0 0
-4 -2 0 2 -1 0 1 2
10 10 10 10 10 10 10 10
The system is stable for K>‐1 as follows: K= ‐0.6
8‐82


1 0
0.5 -20
-40
0
-60
-0.5
-80 G.M.: 4.37 dB

-1 Freq: 0 rad/sec
-6 -4 -2 0 2 4
Stable loop
50
180
0 135
-50 90
180
45
90
P.M.: Inf
Freq: NaN
0 0
-2 -1 0 1 2 -1 0 1 2
10 10 10 10 10 10 10 10 10

And the system is unstable for K< ‐1: K= ‐3
1 20
0.5 0
-20
0
-40
-0.5
-60 G.M.: -9.59 dB

-1 Freq: 0 rad/sec
-6 -4 -2 0 2 4
Unstable loop
20
180
0
135
-20
-40 90
180
45
90
P.M.: -129 deg
Freq: 1.84 rad/sec
0 0
-2 -1 0 1 2 -1 0 1 2
10 10 10 10 10 10 10 10 10

*Combining the individual ranges for K, the system will be stable in the range of K> ‐1
8‐83


8‐39 See sample MATLAB code in Part e. The MATLAB codes are identical to problem 8‐36.
(a)
K
G ( s ) =
s(1 + 0.1s )(1 + 0.5s )
The Bode plot is done with K = 1.
GM = 21.58 dB For GM = 20 dB,
K must be reduced by −1.58 dB.
Thus K = 0.8337
o o
PM = 60.42 . For PM = 45
K should be increased by 5.6 dB.
Or, K = 1.91
(b)
K ( s + 1)
G( s) = The
s(1 + 0.1s )(1 + 0.2 s)(1 + 0.5s )
Bode plot is done with K = 1.
GM = 19.98 dB. For GM = 20 dB,
K ≅ 1.
o o
PM = 86.9 . For PM = 45

K should be increased by 8.9 dB.
Or, K = 2.79.
8‐84


8‐39 (c) See the top plot
K
G ( s ) = 3

( s + 3)
GM = 46.69 dB
PM = infinity.
For GM = 20 dB K can be
increased by 26.69 dB or K = 21.6.
For PM = 45 deg. K can be
increased by 28.71 dB, or
K = 27.26.

(d) See the middle plot
K
G( s) = 4

( s + 3)
GM = 50.21 dB
PM = infinity.
For GM = 20 dB K can be
increased by 30.21 dB or K = 32.4
For PM = 45 deg. K can be
increased by 38.24 dB, or
K = 81.66
(e) See the bottom plot

8‐85


−s
Ke MATLAB code:
G ( s ) =
(
s 1 + 0.1s + 0.01s
2
) s = tf('s')
num_G_a= exp(-s);
GM=2.97 dB; PM = 26.58 deg
den_G_a=s*(0.01*s^2+0.1*s+1);
For GM = 20 dB K must be G_a=num_G_a/den_G_a
margin(G_a)
decreased by −17.03 dB or
K = 0.141.
For PM = 45 deg. K must be
decreased by −2.92 dB or K = 0.71.
8‐86


8‐39 (f)
K (1 + 0.5s )
G ( s ) =
(
s s + s +1
2
)
GM = 6.26 dB
PM = 22.24 deg
For GM = 20 dB K must be decreased by −13.74 dB or
K = 0.2055.
For PM = 45 deg K must be decreased by −3.55 dB or
K = 0.665.
8‐40 (a)
10 K
G ( s ) =
s(1 + 0.1s )(1 + 0.5s )
The gain‐phase plot is done with
K = 1.
GM = 1.58 dB
PM = 3.95 deg.

For GM = 10 dB, K must be decreased by −8.42
dB or K = 0.38.
For PM = 45 deg, K must be decreased by −14 dB, or K = 0.2.
For M r = 1.2 , K must be decreased to 0.16.
8‐87


Sample MATLAB code:
s = tf('s')
num_G_a= 10;
den_G_a=s*(1+0.1*s)*(0.5*s+1);
G_a=num_G_a/den_G_a
nichols(G_a)

(b)
8‐88


5K ( s + 1)
G( s) =
s(1 + 0.1s )(1 + 0.2 s)(1 + 0.5s )
The Gain‐phase plot is done with
K = 1.
GM = 6 dB
PM = 22.31 deg.
For GM = 10 dB, K must be decreased by −4

dB or K = 0.631.
For PM = 45 deg, K must be
decrease by −5 dB.
For M r = 1.2 , K must be decreased to 0.48.
8‐40 (c)
10 K
G ( s ) =
(
s 1 + 0.1s + 0.01s
2
)
The gain‐phase plot is done for
K = 1.
GM = 0 dB M r = ∞
PM = 0 deg
For GM = 10 dB, K must be decreased by −10
dB or K = 0.316.
For PM = 45 deg, K must be decreased by −5.3
dB, or
K = 0.543.
For M r = 1.2, K must be decreased to 0.2213.
(d)
8‐89


−s
Ke
G ( s ) =
(
s 1 + 0.1s + 0.01s
2
)
The gain‐phase plot is done for
K = 1.
GM = 2.97 dB M r = 3.09
PM = 26.58 deg
For GM = 10 dB, K must be decreased by −7.03
dB, K = 0.445.

For PM = 45 deg, K must be decreased by
−2.92 dB, or
K = 0.71.
For M r = 1.2, K = 0.61.
8‐90


8-41
MATLAB code:
s = tf('s')
%a)
num_GH_a= 1*(s+1)*(s+2);
den_GH_a=s^2*(s+3)*(s^2+2*s+25);
CL_a = GH_a/(1+GH_a)
figure(1);
bode(CL_a)
%b)
figure(2);
rlocus(GH_a)
%c)
num_GH_c= 53*(s+1)*(s+2);
den_GH_c=s^2*(s+3)*(s^2+2*s+25);
GH_c=num_GH_c/den_GH_c;
figure(3);
nyquist(GH_c)
xlim([-2 1])
ylim([-1.5 1.5])
%d)
num_GH_d= (s+1)*(s+2);
den_GH_d=s^2*(s+3)*(s^2+2*s+25);
GH_d=num_GH_d/den_GH_d;
CL_d = GH_d/(1+GH_d)
figure(4);
margin(CL_d)
sisotool
Part (a), Bode diagram:
8‐91


Bode Diagram
50
0
Magnitude (dB)
-50
-100
-150
0
-90
Phase (deg)
-180
-270
-2 -1 0 1 2
10 10 10 10 10
Frequency (rad/sec)
Part (b), Root locus diagram:
Root Locus
30
20
10
Imaginary Axis
-10
-20
-30
-40 -30 -20 -10 0 10 20
Real Axis
Part (c), Gain and frequency that instability occurs: Gain=53, Freq = 4.98 rad/sec, as seen in the data
point in the figure:
8‐92


Nyquist Diagram
1.5
System: GH_a
0.5
Real: -1
Imag: 0.0174
Imaginary Axis
0
-0.5
-1
-1.5
-2 -1.5 -1 -0.5 0 0.5 1
Real Axis
Part (d), Gain and

frequency that
instability occurs:
Gain=0.127 at PM
=20 deg:
By running sisotool
command in
MATLAB, the
transfer functions
are imported and the
gain is iteratively
changed until the
phase margin of
PM=20 deg is
achieved. The Part (e): The corresponding gain margin is GM = 22.7 dB is seen in the
corresponding Gain figure
is K= 0.127.
8‐93


8‐42 (a) Gain crossover frequency = 2.09 rad/sec PM = 115.85 deg
Phase crossover frequency = 20.31 rad/sec GM = 21.13 dB
(b) Gain crossover frequency = 6.63 rad/sec PM = 72.08 deg
(c) Gain crossover frequency = 19.1 rad/sec PM = 4.07 deg
(d) For GM = 40 dB, reduce gain by (40 − 21.13) dB = 18.7 dB, or gain = 0.116 × nominal value.

(e) For PM = 45 deg, the magntude curve reads −10 dB. This means that the loop gain can be increased by
10 dB from the nominal value. Or gain = 3.16 × nominal value.

(f) The system is type 1, since the slope of G ( jω ) is −20 dB/decade as ω → 0.
(g) GM = 12.7 dB. PM = 109.85 deg.
(h) The gain crossover frequency is 2.09 rad/sec. The phase margin is 115.85 deg.
Set
115.85 π
o
ωTd = 2.09Td = o
= 2.022 rad
180
Thus, the maximum time delay is Td = 0.9674 sec.
8‐94


8‐43 (a) The gain is increased to four times its nominal value. The magnitude curve is raised by 12.04 dB.
Gain crossover frequency = 10 rad/sec PM = 46 deg
(b) The GM that corresponds to the nominal gain is 21.13 dB. To change the GM to 20 dB we need to
increase the gain by 1.13 dB, or 1.139 times the nominal gain.
(c) The GM is 21.13 dB. The forward‐path gain for stability is 21.13 dB, or 11.39.
(d) The PM for the nominal gain is 115.85 deg. For PM = 60 deg, the gain crossover frequency must be
moved to approximately 8.5 rad/sec, at which point the gain is −10 dB. Thus, the gain must be increased
by 10 dB, or by a factor of 3.162.
(e) With the gain at twice its nominal value, the system is stable. Since the system is type 1, the steady‐state
error due to a step input is 0.
(f) With the gain at 20 times its nominal value, the system is unstable. Thus the steady‐state error would be
infinite.
(g) With a pure time delay of 0.1 sec, the magnitude curve is not changed, but the the phase curve is subject
to a negative phase of − 0.1ω rad. The PM is
PM = 115.85 − 0.1 × gain crossover frequency = 115.85 − 0.209 = 115.64 deg
The new phase crossover frequency is approximately 9 rad/sec, where the original phase curve is
reduced by −0.9 rad or −51.5 deg. The magnitude of the gain curve at this frequency is −10 dB.
Thus, the gain margin is 10 dB.
(h) When the gain is set at 10 times its nominal value, the magnitude curve is raised by 20 dB. The new
gain crossover frequency is approximately 17 rad/sec. The phase at this frequency is −30 deg.
Thus, setting
30 π
o
ωTd = 17Td = o
= 0.5236 Thus Td = 0.0308 sec.
180

8‐95


8-44
MATLAB code:
s = tf('s')
%using pade cammand for PADE approximation of exponential term
num_G_a= pade((80*exp(-0.1*s)),2);
den_G_a=s*(s+4)*(s+10);
CL_a = G_a/(1+G_a)
OL_a = G_a*1;
%(a)
figure(1)
nyquist(OL_a)
%(b) and (c)

figure(2);
margin(CL_a)
Part (a), Nyquist diagram:
Nyquist Diagram
8
2
Imaginary Axis
-2
-4
-6
-8
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4
Real Axis
Part (b) and (c), Bode diagram:

Using Margin command, the gain and phase margins are obtained as GM = 6.3 dB, PM = 42.6 deg:
8‐96


Bode Diagram
Gm = 6.3 dB (at 4.02 rad/sec) , Pm = 42.6 deg (at 2.9 rad/sec)
50
0
Magnitude (dB)
-50
-100
-150
360
180
Phase (deg)
-180
-360
-1 0 1 2 3
10 10 10 10 10
Frequency (rad/sec)
8‐97


8‐45 (a) Bode Plot:
For stability: 166 (44.4 dB) < K < 7079 (77 dB) Phase crossover frequencies: 7 rad/sec and 85 rad/sec
8‐98


Nyquist Plot:

8‐99


8‐45 (b) Root Loci.
8‐100


8‐46 (a) Nyquist Plot

Bode Plot

8‐101


Root Locus
8‐102


8-47
MATLAB code:
%a)
k=1
num_GH= k*(s+1)*(s+5);
den_GH=s*(s+0.1)*(s+8)*(s+20)*(s+50);
GH=num_GH/den_GH;
CL = GH/(1+GH)
figure(1);
bode(CL)
figure(2);
OL = GH;
nyquist(GH)
xlim([-1.5 0.5]);
ylim([-1 1]);
sisotool
Part (a), Bode diagram:
8‐103


Bode Diagram
0
-50
Magnitude (dB)
-100
-150
-200
0
-90
Phase (deg)
-180
-270
-4 -3 -2 -1 0 1 2 3
10 10 10 10 10 10 10 10
Frequency (rad/sec)
Part (a), Nyquist diagram:
Nyquist Diagram
1
0.8
0.6
0.4
0.2
Imaginary Axis
-0.2
-0.4
-0.6
-0.8
-1
-1.5 -1 -0.5 0 0.5
Real Axis
Part (a), range of K for stability: By running sisotool command in MATLAB, the transfer functions
are imported and the gain is iteratively changed until the phase margin of PM=0 deg is achieved (where
K = ) which is the margin of stability. The stable rang for K is K>8.16x104:
8‐104


Part (b), Root-locus diagram, K and ω at the points where the root loci cross the jω-axis:
As can be seen in the figure at K=8.13x104 and ω =33.8 rad/sec, the poles cross the jω axis. Both of
these values are consistent with the results of part(a) from sisotool.
8‐105


Root Locus
150
System: CL
100 Gain: 8.13e+004
Pole: 0.0788 + 33.8i
Damping: -0.00233
Overshoot (%): 101
50 Frequency (rad/sec): 33.8
Imaginary Axis
-50 System: CL
Gain: 7.99e+004
Pole: -0.045 - 33.5i
Damping: 0.00134
-100 Overshoot (%): 99.6
-150
-200 -150 -100 -50 0 50 100
Real Axis
8‐106


8‐48 Bode Diagram
When K = 1, GM = 68.75 dB, PM = 90 deg. The critical value of K for stability is 2738.
8‐107


8‐49 (a) Forward‐path transfer function:
Θ L ( s) K a Ki ( Bs + K )
G( s) = = K aG p ( s ) =
E ( s) Δo
where
(
Δ o = 0.12 s ( s + 0.0325 ) s + 2.5675 s + 6667
2
)

(
= s 0.12 s + 0.312 s + 80.05 s + 26
3 2
)
43.33( s + 500)
G ( s ) =
(
s s + 2.6 s + 667.12 s + 216.67
3 2
)
(b) Bode Diagram:
8‐108


Gain crossover frequency = 5.85 rad/sec PM = 2.65 deg.
8‐109


8‐49 (c) Closed‐loop Frequency Response:
M r = 17.72, ω r = 5.75 rad / sec, BW = 9.53 rad / sec
8‐110


8-50
(a)
(b)
(c) to (e)
MATLAB code:
s = tf('s')
m = 0.11;
r = 0.015;
d = 0.03;
g = 9.8;
L = 1.0;
J = 9.99*10^-6
K=1
num_GH= K*m*g*d;
den_GH=L*(J/r^2+m)*s^2;
GH=num_GH/den_GH;
CL = GH/(1+GH)
%c)
figure(1);
nyquist(GH)
xlim([-1.5 0.5]);
ylim([-1 1]);
%d)
figure(2);
margin(CL)
Part (c): since the system is a double integrator (1/s2), the phase is always -180 deg, and the system is
always marginally stable for any K, leading to a complicated control problem.
8‐111


Nyquist Diagram
1
0.8
0.6
0.4
0.2
Imaginary Axis
-0.2
-0.4
-0.6
-0.8
-1
-1.5 -1 -0.5 0 0.5
Real Axis
Part (d), Bode diagram:

As explained in section (c), since the system is always marginally stable, GM=inf and PM = 0, as can be
seen by MATLAB MARGIN command, resulting in the following figure:
8‐112


Bode Diagram
Gm = Inf dB (at Inf rad/sec) , Pm = 0 deg (at 0.647 rad/sec)
150
System: CL
100 Frequency (rad/sec): 0.458
Magnitude (dB): 143
Magnitude (dB)
50
0
System: CL
-100
-180
-225
Phase (deg)
-270
-315
-360
-2 -1 0 1
10 10 10 10
Frequency (rad/sec)
Part (e), Mr,= 143 dB, ωr = 0.458 rad/sec, and BW= 0.712 rad/sec as can be seen in the data points in
the above figure.
8-51 (a) When K = 1, the gain crossover frequency is 8 rad/sec.
(b) When K = 1, the phase crossover frequency is 20 rad/sec.
(c) When K = 1, GM = 10 dB.
(d) When K = 1, PM = 57 deg.
(e) When K = 1, M r = 1.2.
(f) When K = 1, ω r = 3 rad/sec.
(g) When K = 1, BW = 15 rad/sec.
(h) When K = −10 dB (0.316), GM = 20 dB
(i) When K = 10 dB (3.16), the system is marginally stable. The frequency of oscillation is 20 rad/sec.
(j) The system is type 1, since the gain‐phase plot of G ( jω ) approaches infinity at −90 deg. Thus, the
steady‐state error due to a unit‐step input is zero.
8‐113


8-52 When K = 5 dB, the gain‐phase plot of G ( jω ) is raised by 5 dB.
(a) The gain crossover frequency is ~10 rad/sec.
(b) The phase crossover frequency is ~20 rad/sec.
(c) GM = 5 dB.
(d) PM = ~34.5 deg.
(e) When K = 5, Mr= ~ 2 (smallest circle tangent to an M circle).
(f) ω r = 15 rad/sec
(g) BW = 30 rad/sec
(h) When K = −30 dB, the GM is 40 dB (shift the graph of K=1, 30 dbs down).
When K = 10 dB, the gain‐phase plot of G ( jω ) is raised by 10 dB.
(a) The gain crossover frequency is 20 rad/sec.
(b) The phase crossover frequency is 20 rad/sec.
(c) GM = 0 dB.
(d) PM = 0 deg.
(e) When K = 10, Mr= ~ 1.1 (smallest circle tangent to an M circle).
(f) ωr = 5 rad/sec
(g) BW = ~40 rad/sec
(h) When K = −30 dB, the GM is 40 dB (shift the graph of K=1, 30 dbs down).
8-53
Since the function has exponential term, PADE command has been used to obtain the transfer function.
80e −0.1s
G ( s) H ( s) =
s ( s + 4)( s + 10)
MATLAB code:
s = tf('s')
%a)
8‐114


num_GH= pade(80*exp(-0.1*s),2);
den_GH=s*(s+4)*(s+10);
GH=num_GH/den_GH;
CL = GH/(1+GH)
BW = bandwidth(CL)
bode(CL)
%b)
figure(2);
nichols(GH)
Part(a), Nicholas diagram:
Nichols Chart
40
20
-20
Open-Loop Gain (dB)
-40
-60
-80
-100
-120
-140
-270 -180 -90 0 90 180 270
Open-Loop Phase (deg)
Part(b), Bode diagram:
8‐115


Bode Diagram
50
0
Magnitude (dB)
-50
-100
-150
360
180
Phase (deg)
-180
-360
-1 0 1 2 3
10 10 10 10 10
Frequency (rad/sec)
G
8- 54) Note: GCL =
1+ G
To draw the Bode and polar plots use the closed loop transfer function, GCL, and find BW. Use G to
obtain the gain-phase plots and Gm and Pm. Use the Bode plot to graphically obtain Mr.
Sample MATLAB code:
s = tf('s')
%a)
num_G= 1+0.1*s;
den_G=s*(s+1)*(0.01*s+1);
G=num_G/den_G
figure(1)
nyquist(G)
figure(2)
margin(G)
GCL = G/(1+G)
BW = bandwidth(GCL)
figure(3)
bode(GCL)
8‐116


Transfer function:
0.1 s + 1
-----------------------
0.01 s^3 + 1.01 s^2 + s
Transfer function:
0.001 s^4 + 0.111 s^3 + 1.11 s^2 + s
--------------------------------------------------------------
0.0001 s^6 + 0.0202 s^5 + 1.041 s^4 + 2.131 s^3 + 2.11 s^2 + s
BW =
1.2235
8‐117


8‐118


8-55) (a) The phase margin with K = 1 and Td = 0 sec is approximately 57 deg. For a PM of 40 deg, the time delay
produces a phase lag of −17 deg. The gain crossover frequency is 8 rad/sec.
Thus,
17 π
o
ωTd = 17 = Thus ω = 8 rad / sec
o
o
= 0.2967 rad / sec
180
0.2967
Td = = 0.0371 sec
8
(b) With K = 1, for marginal stability, the time delay must produce a phase lag of −57 deg.
Thus, at ω = 8 rad/sec,
57 π
o
0.9948
ωTd = 57 =
o
o
= 0.9948 rad Td = = 0.1244 sec
180 8
8‐119


8‐56 (a) The phase margin with K = 5 dB and Td = 0 is approximately 34.5 deg. For a PM of 30 deg, the time
delay must produce a phase lag of −4.5 deg. The gain crossover frequency is 10 rad/sec. Thus,
4.5 π
o
0.0785
ωTd = 4.5 =
o
o
= 0.0785 rad Thus Td = = 0.00785 sec
180 10
(b) With K = 5 dB, for marginal stability, the time delay must produce a phase lag of −34.5 deg.
Thus at ω = 10 rad/sec,
34.5 π
o
0.602
ωTd = 34.5 =
o
o
= 0.602 rad Thus Td = = 0.0602 sec
180 10
8‐57) For a GM of 5 dB, the time delay must produce a phase lag of −34.5 deg at ω = 10 rad/sec. Thus,
34.5 π
o
0.602
ωTd = 34.5 =
o
o
= 0.602 rad Thus Td = = 0.0602 sec
180 10

−2 s
Y ( s) e
G( s) = =
E ( s) (1 + 10 s )(1 + 25s )

From the Bode diagram, phase crossover frequency = 0.21 rad/sec GM = 21.55 dB
8‐120


gain crossover frequency = 0 rad/sec PM = infinite
(b)
1
G(s) =
(1 + 10 s )(1 + 25s ) (1 + 2 s + 2 s 2
)

From the Bode diagram, phase crossover frequency = 0.26 rad/sec GM = 25 dB

(c)
1− s
G(s) =
(1 + s )(1 + 10 s )(1 + 2 s )

Sample MATLAB code

s = tf('s')
%a)
num_G=exp(-2*s);
den_G=(10*s+1)*(25*s+1);
G=num_G/den_G
figure(1)
margin(G)

8‐121


8‐58 (continued) Bode diagrams for all three parts.
8‐122


−s
e
G( s) =
(1 + 10 s )(1 + 25s )
(b)
1
G( s) =
(1 + 10s )(1 + 25s ) (1 + s + 0.5s 2
)
(c)
(1 − 0.5s )
G( s) =
(1 + 10 s )(1 + 25s )(1 + 0.5 s )
8‐123


Plots 8‐59 (a‐c)
8‐124


8‐60 Sensitivity Plot:
8‐125


ω max = 5.75 rad/sec

M
SG = 17.15
max
8-61)
(a)
(b)
(c)&(d)
MATLAB code:
s = tf('s')
%c)
K = 1
num_GH= K*(1.151*s+0.1774);
den_GH=(s^3+0.739*s^2+0.921*s);
GH=num_GH/den_GH;
CL = GH/(1+GH)
sisotool
%(d)
figure(1)
margin(CL)
Part (c), range of K for stability:

Sisotool Result shows that by changing K between 0 and inf., all the roots of closed loop system remain
in the left hand side plane and PM remains positive. Therefore, the system is stable for all positive K.
8‐126


4
Root
x 10 Locus Editor for Open Loop 1 (OL1) Open-Loop Bode Editor for Open Loop 1 (OL1)
3 200
2
180
1
160
0
-1
140
-2
120 G.M.: Inf
-3 Freq: Inf
-0.4 -0.3 -0.2 -0.1 0
Stable loop
Bode Editor for Closed Loop 1 (CL1) 100
100
0
0 -45
-100 -90
0
-135
-90
P.M.: 0.00149 deg
Freq: 2.26e+004 rad/sec
-180 -180
-2 0 2 4 6 -2 -1 0 1 2
10 10 10 10 10 10 10 10 10 10
Part (d), Bode, GM & PM for K=1:
8‐127


Bode Diagram
Gm = Inf dB (at Inf rad/sec) , Pm = 61.3 deg (at 1.59 rad/sec)
20
Magnitude (dB) 0
-20
-40
-60
-80
0
-45
Phase (deg)
-90
-135
-180
-2 -1 0 1 2
10 10 10 10 10
Frequency (rad/sec)
8‐128


Chapter 9
9‐1 Forward‐path Transfer Function:
M ( s) K
G( s) = = 3 2

1− M ( s) s + ( 20 + a ) s + ( 200 + 20a ) s + 200a − K
K 200 a
K v = lim sG ( s ) = = = 5 Thus a = 10 K = 2000
s→ 0 200 + 20a 200 + 20 a
G ( s ) =
2000
Gc ( s ) =
G(s)
=
(
20 s + 10 s + 100
2
)
(
s s + 30 s + 400
2
) Gp (s) (s 2
+ 30 s + 400 )

The maximum overshoot of the unit‐step response is 0 percent.
M ( s) K
G( s) = = 3 2

1− M ( s) s + ( 20 + a ) s + ( 200 + 20a ) s + 200a − K
K 200 a
K v = lim sG ( s ) = = = 9 Thus a = 90 K = 18000
s→ 0 200 + 20a 200 + 20 a
9‐1


G ( s ) =
18000
Gc ( s ) =
G(s)
=
(
180 s + 10 s + 100
2
)
(
s s + 110 s + 2000
2
) G p ( s) ( s + 110s + 2000 )
2
The maximum overshoot of the unit‐step response is 4.3 percent.
From the expression for the ramp‐error constant, we see that as a or K goes to infinity, Kv approaches 10.
Thus the maximum value of Kv that can be realized is 10. The difficulties with very large values of K and
a are that a high‐gain amplifier is needed and unrealistic circuit parameters are needed for the controller.
9‐3) The close loop transfer function is:

Comparing with second order system:
and 2
exp 0.254 Ö ξ = 0.4
3 Ö ωn = 1.14
Ö τ = 1.09
Ö K = 1.42
9-4) The forward path transfer function of the system is:
24
G ( s) H ( s) =
s ( s + 1)( s + 6)
1. The steady state error is less than to π/10 when the input is a ramp with a slope of 2π rad/sec
9‐2


2 2
lim lim
10 24 4K
s ( s + 1)( s + 6)
As a result K 0.2
2. The phase margin is between 40 to 50 degrees
In this part of the solution, MATLAB sisotool can be very helpful. More detailed instructions
on using MATLAB sisotool is presented in the solutions for this particular problem. Similar
guidelines could be used for similar questions of this chapter.
SISOTOOL quick instructions:
Once opening the sisotool by typing sisotool in MATLAB the command window (or in the “m” code), the
following window pops up:
Where you can insert transfer functions for C, G, H, and F, or you can leave some of them as default value (1), by
clicking on “System Data”. Once you substitute transfer functions, you will see a graph including the root‐locus
diagram, a closed loop Bode diagram, and an open loop Bode diagram indicating the Gain Margin and Phase
Margin as well.
** You can drag the open‐loop bode magnitude diagram up and down to see the effect of gain change on all of
the graphs. Sisotool updates all these graph instantly. You can also drag the poles and zeros on the root locus
diagram to observe the effect on the other diagrams.
9‐3


You can also open “Analysis Plots” tab to add other graphs such as Nyquist diagram as shown in the following
figure:
9‐4


In this particular question, you need to add a zero to include the effect of K d . You can add a zero by using the
“Compensator Editor” tab, as shown in the following graph. The last thing you need to do for this problem, is to
drag the location of zero and gain in the following diagram (or edit these locations by assigning C gain and Zero
location in the “Compensator Editor” tab), so it satisfies the PM of 40 to 50 deg; while gain of C is kept above
0.2 ( K p ≥ 0.2 , from part (a)).
In the following snapshot of “Compensator Editor”, C gain or K p is set to 1, and the zero location is set to ‐2.7,
resulting in 44.2 [deg] phase margin, presented in the following figure.
9‐5


60
2000
40
1000
20
Imag Axis
Magnitude (dB)
0
0
-1000
-20
-2000
-40 G.M.: Inf
-6 -5 -4 -3 -2 -1 0 Freq: Inf
Stable loop
Bode Editor for Closed Loop 1 (CL1)
-60
20
-90
Magnitude (dB)
-20
-40 -120
Phase (deg)
-60
0
Phase (deg)
-45 -150
-90
P.M.: 44.2 deg
-135
Freq: 2.07 rad/sec
-180 -180
-1 0 1 2 -2 -1 0 1 2
10 10 10 10 10 10 10 10 10
Final answers: K p = 1, K d = 0.3704
9‐6


Preliminary MATLAB code for 9‐4:
%solving for k:
syms kc
omega=1.5
sol=eval(solve('0.25*kc^2=0.7079^2*((-0.25*omega^3+omega)^2+(-
0.375*omega^2+0.5*kc)^2)',kc))
%ploting bode with K=1.0370
s = tf('s')
K=1.0370;
num_G_a= 0.5*K;
den_G_a=s*(0.25*s^2+0.375*s+1);
CL_a = G_a/(1+G_a)
bode(CL_a);
sisotool;
Alternatively we can use ACSYS.
9‐7


K p = 1, K d = 0.3704
9‐8


9‐9


9-5) (a) Ramp‐error Constant:
1000 ( K P + K D s ) 1000 K P
K v = lim s = = 100 K P = 1000 Thus K P = 10
s →0 s ( s + 10) 10
s + (10 + 1000 K D ) s + 1000 K P = 0

2
ω n = 1000 K P = 10000 = 100 rad/sec 2ζω n = 10 + 1000 K D = 2 × 0.5 × 100 = 100
90
Thus KD = = 0.09
1000
9‐10


9‐5 (b) For K v = 1000 and ζ = 0.707 , and from part (a), ω n = 100 rad/sec,
131.4
2ζω n = 10 + 1000 K D = 2 × 0.707 × 100 = 141.4 Thus KD = = 0.1314
1000
(c) For K v = 1000 and ζ = 1.0 , and from part (a), ω n = 100 rad/sec,
190
2ζω n = 10 + 1000 K D = 2 × 1 × 100 = 200 Thus KD = = 0.19
1000
9-6) The ramp‐error constant:
1000 ( K P + K D s )
K v = lim s = 100 K P = 10, 000 Thus K P = 100
s →0 s ( s + 10)
1000 (100 + K D s )
The forward‐path transfer function is: G(s) =
s ( s + 10)
KD PM (deg) GM Mr BW (rad/sec) Max overshoot (%)
0 1.814 ∞ 13.5 493 46.6
0.2 36.58 ∞ 1.817 525 41.1
0.4 62.52 ∞ 1.291 615 22
0.6 75.9 ∞ 1.226 753 13.3
0.8 81.92 ∞ 1.092 916 8.8
1.0 84.88 ∞ 1.06 1090 6.2
The phase margin increases and the maximum overshoot decreases monotonically as K D increases.
9‐11


Sample MATLAB CODE for time frequency responses

clear all
KD=1.0;
num = [-100/KD];
den = [0 -10];
G=zpk(num,den,1000);
figure(1)
margin(G)
M=feedback(G,1)
figure(2)
step(M);
figure(3)
bode(M)

Zero/pole/gain:
1000 (s+100)
‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐
(s+111.3) (s+898.7)
Phase margin is Pm=180-95=85 Deg
Gm=∞
9‐12


Use the cursor to obtain the PO and tr values.
KD increase results in the minimum overshoot.
Bandwidth is 1090 rads/s.
9‐13


Use ACSYS to find the same results:
9‐14


9-7)
PD controller design
The open‐loop transfer function of a system is:
1
G(s) H (s) =
(2 s + 1)( s + 1)(0.5s + 1)
The solution is very similar to 9‐4. The transfer functions are inserted into sisotool, where another real zero is
added to represent the effect of K d . That is C ( s ) = K p + K d s = K p (1 + K d s / K p ) , which is called the
compensator transfer function in sisotool. The place of real zero is Z = − K p / K d , and the gain of the
compensator is equal to K p , as noted in the following sisotool window:
9‐15


By fixing the gain to 9, and starting to change the zero location, PM can be adjusted to above 25 [deg] as
required by the question. The current setting has a zero at ‐3.5, which resulted in 30 [deg] phase margin and
33.1 dB gain margin as seen in the following diagrams.
The design requires K p = 9 and K d = − K p / Z = −9 / − 3.6 = 2.5
15 20
10
10
5
0
0
-10
-5
-20
-10
G.M.: 33.1 dB
-15 -30
Freq: 10.8 rad/sec
-4 -3 -2 -1 0 1
Stable loop
50
0
0
-45
-50
-90
-100
0 -135
-90
-180
-180 P.M.: 29.8 deg
Freq: 1.77 rad/sec
-270 -225
-1 0 1 2 -2 -1 0 1 2
10 10 10 10 10 10 10 10 10
s = tf('s')
Kp = 1
num_GH= Kp*1;
den_GH=(2*s+1)*(s+1)*(0.5*s+1);
9‐16


GH=num_GH/den_GH;
CL = GH/(1+GH)
Sisotool
9-8: PD controller design: The open-loop transfer function of a system is:
60
G ( s) H ( s) =
s (0.4 s + 1)( s + 1)( s + 6)
(a) Design a PD controller to satisfy the following specifications:
(i) Kv =10
(ii) the phase margin is 45 degrees.
60
lim lim 10 10
0.4 1 1 6
As a result: 1
The rest of the procedure is similar to 9‐7:

9‐17


The transfer functions are inserted into sisotool, where another real zero is added to represent the effect of K d
. That is C ( s ) = K p + K d s = K p (1 + K d s / K p ) , which is called the compensator transfer function in sisotool.
The place of real zero is Z = − K p / K d , and the gain of the compensator is equal to K p , as noted in the
sisotool window:
K p = 1 fixed to 1, and zero location was changed in the entire real axis. However, 2 of the closed loop poles
remained in the right hand side of S plane in the root locus diagram, indicating instability for all Kd values.
Solution for K v = 10 with PD controller and PM=45 [deg] does not exist. Unstable close loop poles are
indicated in the root locus diagram of the following figure:
9‐18


9‐8)
s = tf('s')
Kp = 1
num_GH= Kp*60;
den_GH=s*(0.4*s+1)*(s+1)*(s+6);
GH=num_GH/den_GH;
CL = GH/(1+GH)
sisotool
9‐19


4500 K ( K D + K P s )
G ( s ) = Gc ( s )G p ( s ) =
s ( s + 361.2)
4500 KK P
Ramp Error Constant: K v = lim sG ( s ) = = 12.458 KK P
s→ 0 361.2

1 0.0802
e ss = = ≤ 0.001 Thus KK P ≥ 80.2 Let K P = 1 and K = 80.2
Kv KK P

Attributes of Unit‐step Response:
KD tr (sec) ts (sec) Max Overshoot (%)
0 0.00221 0.0166 37.1
0.0005 0.00242 0.00812 21.5
0.0010 0.00245 0.00775 12.2
0.0015 0.0024 0.0065 6.4
0.0016 0.00239 0.00597 5.6
0.0017 0.00238 0.00287 4.8
0.0018 0.00236 0.0029 4.0
0.0020 0.00233 0.00283 2.8
Select K D ≥ 0.0017
(b) BW must be less than 850 rad/sec.
9‐20


KD GM PM (deg) Mr BW (rad/sec)
0.0005 ∞ 48.45 1.276 827
0.0010 ∞ 62.04 1.105 812
0.0015 ∞ 73.5 1.033 827
0.0016 ∞ 75.46 1.025 834
0.0017 ∞ 77.33 1.018 842
0.00175 ∞ 78.22 1.015 847
0.0018 ∞ 79.07 1.012 852
Select K D ≅ 0.00175 . A larger K D would yield a BW larger than 850 rad/sec.
9‐21


9-10)
The forward‐path Transfer Function: N = 20
200 ( K P + K D s )
G(s) =
s ( s + 1)( s + 10)

To stabilize the system, we can reduce the forward‐path gain. Since the system is type 1, reducing the
gain does not affect the steady‐state liquid level to a step input. Let K P = 0.05
200 ( 0.05 + K D s )
G(s) =
s ( s + 1)( s + 10)
Unit‐step Response Attributes:
KD ts (sec) Max Overshoot (%)
0.01 5.159 12.7
0.02 4.57 7.1
0.03 2.35 3.2
0.04 2.526 0.8
0.05 2.721 0
0.06 3.039 0
0.10 4.317 0
When K D = 0.05 the rise time is 2.721 sec, and the step response has no oversho
9‐22


9-11)
(a) For e ss = 1,
200 ( K P + K D s )
K v = lim sG ( s ) = lim s = 20 K P = 1 Thus K P = 0.05
s →0 s →0 s ( s + 1)( s + 10)

200 ( 0.05 + K D s )
G ( s ) =
s ( s + 1)( s + 10)

Attributes of Frequency Response:
KD PM (deg) GM (deg) Mr BW (rad/sec)
0 47.4 20.83 1.24 1.32
0.01 56.11 ∞ 1.09 1.24
0.02 64.25 ∞ 1.02 1.18
0.05 84.32 ∞ 1.00 1.12
0.09 93.80 ∞ 1.00 1.42
0.10 93.49 ∞ 1.00 1.59
0.11 92.71 ∞ 1.00 1.80
0.20 81.49 ∞ 1.00 4.66
0.30 71.42 ∞ 1.00 7.79
0.50 58.55 ∞ 1.03 12.36
For maximum phase margin, the value of K D is 0.09. PM = 93.80 deg. GM = ∞, M r = 1,
and BW = 1.42 rad/sec.
9‐23


(b) Sensitivity Plots:

M
The PD control reduces the peak value of the sensitivity function SG ( jω )
9‐24


9-12)
PD controller design: The open loop transfer function of a system is:
100
G ( s) H ( s) =
s (0.1s + 1)(0.02 s + 1)
Design the PD controller so that the phase margin is greater than 50 degrees and the BW is greater than
20 rad/sec.
The transfer functions are generated and imported in sisotool as in 9-4:
MATLAB code:
s = tf('s')
Kp = 1
num_GH= Kp*100;
den_GH=s*(0.1*s+1)*(0.02*s+1);
GH=num_GH/den_GH;
CL = GH/(1+GH)
sisotool
Following similar steps in 9‐4, the loop transfer functions are inserted into sisotool. Another real zero is added to
represent the effect of K d . That is C ( s ) = K p + K d s = K p (1 + K d s / K p ) , which is called the compensator
transfer function in sisotool. The place of real zero is Z = − K p / K d , and the gain of the compensator is equal
to K p . The zero location and K p gain were changed interactively in sisotool until the desired PM (59.5 [deg])
and BW is achieved. Following figures shows this PM at cross over frequency of 38.6 rad/sec, which insures BW
9‐25


of higher than 38.6 rad/sec (as the bandwidth is @ ‐3dB rather than 0 DB, i.e. bandwidth occurs at higher
frequency compared to cross over frequency).
Final possible answer: K p = 0.25 and K d = − K p / Z = −0.25 / − 5 = 0.05
9‐26


9‐27


9-13)
Lead compensator design:
1000 K
G(s) H (s) =
s (0.2 s + 1)(0.005s + 1)
Design a compensator such that the steady state error to the unit step input is less than 0.01 and the closed loop
damping ratio ζ>0.4.
The transfer functions are generated and imported in sisotool as in 9‐4:
1
e ss = < 0.01 ⇒ 1000 K p > 101 ⇒ Therefore, K p is selected as 150: ( K p =0.11)
1 + 1000 K p
To achieve the required damping ratio, the poles of the closed loop system are placed with an angle of less than
ArcCos(ζ=0.4), in the root locus diagram of sisotool. This is done by iteratively change the location of poles and
zeros of a lead compensator and setting K p =0.11. The pole and zero (which perform as a lead compensator
when the pole is further away from zero to the left) are inserted in sisotool as explained in 9‐4. The lead
compensator will introduce some phase lead at lower frequencies about the zero location which improves the
closed loop response in terms of damping and phase margin. Following is the chosen location for lead
compensator pole and zero:
Pole @ ‐100 rad/sec
Zero @ ‐1.08 rad/sec
Which resulted in smallest angles of dominant pole locations (the ones closer to imaginary axis) with the real
axis. This small angle means higher damping of the poles as ζ=ArcCos(pole’s angle with real axis).
9‐28


50
2
0
1
-50
0
-100
-1
-150
-2
G.M.: 17.3 dB
-200
Freq: 3.64 rad/sec
-2 -1 0 1
Stable loop
-90
0
-135
-100
-180
-200
-225
0
-90 -270
-180
-315 P.M.: 56.7 deg
-270
Freq: 0.975 rad/sec
-360 -360
-2 0 2 4 -2 0 2 4
10 10 10 10 10 10 10 10

9‐29


In this particular case, closed loop complex poles can be observed in the shown root locus diagram at about ‐
0.8±1.2j. This corresponds to damping of about:
cos(atan(1.2/0.8))=0.554 Æ ζ 0.55
s = tf('s')
Kp = 1
num_GH= Kp*60;
den_GH=s*(0.4*s+1)*(s+1)*(s+6);
GH=num_GH/den_GH;
CL = GH/(1+GH)
figure(1)
margin(CL)
sisotool
9‐30


9-14)
PD controller designed for a maximum overshoot and a maximum steady state error
1
0.005
lim
Therefore:
1 1 4
≤ 0.005 ⇒ K p > ⇒ Kp >
250 K p 0.005 * 250 5
Let K p = 1 , then:
πξ
M p = exp( − ) < 0.20 ⇒ ξ > 0.45
2
1− ξ

Let ζ 0.6; then:
250 1 1250 1

0.2 250 1 250 1250 5 1250
Accordingly, 1250, or 35.35. therefore,

1250 5 2 Ö 1250 2 0.6 0.35
which gives: 0.034

The design characteristics can be observed in the diagram below:
9‐31


1 110
100
0.5
90
0
80
-0.5
70
G.M.: Inf
-1 60
Freq: NaN
-4 -3 -2 -1 0
Stable loop
Bode Editor for Closed Loop 1 (CL1) x 104 50
0
0
P.M.: 90 deg
-10 Freq: 3.68e+004 rad/sec
-20 -45
-30
0
-90
-45
-90 -135
-2 0 2 4 6 -4 -2 0 2
10 10 10 10 10 10 10 10 10
9-15)
Lead compensator controller design
The bode diagram of the system without lead compensator is shown below:
9‐32


Bode Diagram
20
Magnitude (dB) 0
-20
-40
-60
-80
0
-45
Phase (deg)
-90
-135
-180
-1 0 1 2 3
10 10 10 10 10
Frequency (rad/sec)

Indicating a PM of 32.7 [deg]. To reach 45 [deg] phase margin, additional 12.3 [deg] phase lead is
needed. At ω = 17.3 rad/s crossover frequency, if Φ in = 12.3 then
1 sin Φ
r 0.6488
1 sin Φ
As ω = 17.3 , then τ = 0.0718
√
Since the gain is lowered by 1.2838
.
A gain compensator with gain of 0.7789 is required, where,
1

1
9‐33


9-16) If r = 0.1 Ö Φ sin 55
tan tan Ö 17.3 tan 17.3 tan 1.73 12.3
then from trial and error we found τ = 0.014088 and required gain would be 9.7185
9-17) K lim sG s G s 100K 100 Ö K 1
First plot the bode diagram of uncompensated system when K = 1
Bode diagram for Loop T.F. is included for K=1
MATLAB CODE:
s = tf('s')
Kp = 1
num_GH= Kp*100;
den_GH=s*(0.2*s+1)*(0.05*s+1);
GH=num_GH/den_GH
lag_tf=(s/2+1)/(s/0.2+1)
lead_tf=(s/4+1)/(s/50+1)
LL=lag_tf*lead_tf
OL=GH*LL
CL =OL/(1+OL);
figure(1)
margin(GH)
figure(2)
margin(OL)
figure(3)
bode(CL)
grid on;
9‐34


Bode Diagram
Gm = -12 dB (at 10 rad/sec) , Pm = -28.3 deg (at 18.8 rad/sec)
100
50
Magnitude (dB)
-50
-100
-150
-90
-135
Phase (deg)
-180
-225
-270
-1 0 1 2 3
10 10 10 10 10
Frequency (rad/sec)
The bode diagram with K=1 shows -28 deg PM at 18.8 rad/sec.
According to the requirements the gain must be greater than or 250 for 0.2 / and
.
must be less than or 0.01 for 200 /
In order to achieve above requirements, a lead-lag compensator will be appropriate.
Using a lag compensator will allow lower gain at frequencies less then ω1 and using a lead
compensator will allow to increase phase margin
For the lag compensator, 1/10 is chosen to boost the low frequency amplitude
1 /2 1
1 /0.2 1
In order to introduce some phase lead to obtain the require PM, a lead compensator is also
designed as:
1 1
0.0112
√10 28.3√10
9‐35


Where is overlaid with the crossover frequency (28.3 rad/sec) for applying the maximum
phase lead at this frequency. The Lead compensator T.F. will be as follows:
1 /5 1
1 /50 1
Resulting in the following Bode diagram for the compensated system, showing 44 deg PM:
To obtain a slightly higher PM, lead compensator zero was re-tuned, where the zero is pulled
closer to imaginary axis from -5 to -4:
1 /4 1
1 /50 1
This resulted in a higher PM as shown in the following bode diagram of loop transfer function:
Bode Diagram
Gm = 14.5 dB (at 30.7 rad/sec) , Pm = 45 deg (at 10.6 rad/sec)
100
50
Magnitude (dB)
-50
-100
-90
-135
Phase (deg)
-180
-225
-270
-2 -1 0 1 2 3
10 10 10 10 10 10
Frequency (rad/sec)
Correspondingly, the Bode diagram of closed loop system can be shown as:
9‐36


Bode Diagram
20
Magnitude (dB) -20
-40
-60
-80
-100
0
-90
Phase (deg)
-180
-270
-1 0 1 2 3
10 10 10 10 10
Frequency (rad/sec)
9-18) See Chapter 5 solutions for MATLAB codes for this problem.
100 ⎜ K P +
⎛ KI ⎞
⎟ 100 ( K P s + K I )
G ( s ) =
⎝ s ⎠
For K v = 10, K v = lim sG ( s ) = lim s = K I = 10
s + 10 s + 100
2
s →0 s →0
(
s s + 10 s + 100
2
)

Thus K I = 10.
(b) Let the complex roots of the characteristic equation be written as s = −σ + j15 and s = −σ − j15.
The quadratic portion of the characteristic equation is
2
(
s + 2σ s + σ + 225 = 0
2
)

9‐37


s + 10 s + (100 + 100 K P ) s + 1000 = 0

3 2
The characteristic equation of the system is
The quadratic equation must satisfy the characteristic equation. Using long division and solve for zero
remainder condition.
s + (10 − 2σ )
s + 2σ s + σ + 225 s + 10 s + (100 + 100 K P ) s + 1000

2 2 3 2

s + 2σ s + σ + 225 s
3 2
( 2
)
(
(10 − 2σ ) s + 100 K P − σ − 125 s + 1000
2
) 2

(10 − 2σ ) s + ( 20σ − 4σ ) s + (10 − 2σ ) ( s + 225 )
2 2 2
(100 K P )
+ 3σ − 20σ − 125 s + 2σ − 10σ + 450σ − 1250
2 3 2
3 2
For zero remainder, 2σ − 10σ + 450σ − 1250 = 0 (1)
2
and 100 K P + 3σ − 20σ − 125 = 0 (2)
The real solution of Eq. (1) is σ = 2.8555 . From Eq. (2),
2
125 + 20σ − 3σ
KP = = 15765
.
100
The characteristic equation roots are: s = −2.8555 + j15, − 2.8555 − j15, and s = −10 + 2σ = −4.289
(c) Root Contours:
9‐38


100 K P s 100 K P s
Geq ( s ) = =
s + 10 s + 100 s + 1000
3 2
( s + 10 ) ( s 2
+ 100 )

Root Contours:
9‐39


9-19)

⎛
100 ⎜ K P +
KI ⎞
⎟ 100 ( K P s + K I )
G ( s ) =
⎝ s ⎠ For K = 10, K v = lim sG ( s ) = lim s = K I = 10
s + 10 s + 100
2 v
s →0 s →0
(
s s + 10 s + 100
2
)

Thus the forward‐path transfer function becomes
10 (1 + 0.1K P s )
G(s) =
(
s 1 + 0.1s + 0.01s
2
)

Attributes of the Frequency Response:

KP PM (deg) GM (dB) Mr BW (rad/sec)
0.1 5.51 1.21 10.05 14.19
0.5 22.59 6.38 2.24 15.81
0.6 25.44 8.25 1.94 16.11
0.7 27.70 10.77 1.74 16.38
0.8 29.40 14.15 1.88 16.62
9‐40


0.9 30.56 20.10 1.97 17.33
1.0 31.25 ∞ 2.00 18.01
1.5 31.19 ∞ 1.81 20.43
1.1 31.51 ∞ 2.00 18.59
1.2 31.40 ∞ 1.97 19.08
When K P = 11
. and K I = 10, K v = 10, the phase margin is 31.51 deg., and is maximum.
The corresponding roots of the characteristic equation roots are:
−5.4, − 2.3 + j13.41, and − 2.3 − j13.41
Referring these roots to the root contours in Problem 10‐8(c), the complex roots corresponds to
a relative damping ratio that is near optimal.
(b) Sensitivity Function:
9‐41


In the present case, the system with the PI controller has a higher maximum value for the sensitivity
function.
9‐42


9-20)
100 ( K P s + K I )
G(s) =
(
s s + 10 s + 100
2
)
For K v = 100,
100 ( K P s + K I )
K v = lim sG ( s ) = lim s = K I = 100 Thus K I = 100.
s →0 s →0
(
s s + 10 s + 100
2
)
s + 10 s + (100 + 100 K P ) s + 100 K I = 0
3 2
(b) The characteristic equation is
Routh Tabulation:
3
s 1 100 + 100 K P
2
s 10 10,000
For stability, 100 K P − 900 > 0 Thus KP > 9
1
s 100 K P − 900 0
0
s 10,000
Root Contours:

100 K P s 100 K P s
Geq ( s ) = 3 2
=
s + 10 s + 100 s + 10,000 ( s + 23.65)( s − 6.825 + j19.4)( s − 6.825 − j19.4)
9‐43


(c) K I = 100
100 ( K P s + 100 )
G(s) =
(
s s + 10 s + 100
2
)
The following maximum overshoots of the system are computed for various values of K P .
KP 15 20 22 24 25 26 30 40 100 1000
ymax 1.794 1.779 1.7788 1.7785 1.7756 1.779 1.782 1.795 1.844 1.859
When KP = 25, minimum ymax = 1.7756
9‐44


9-21)
100 ( K P s + K I ) 100 K I
G(s) = For Kv = = 10, K I = 10
(
s s + 10 s + 100
2
) 100
(b) Characteristic Equation: (
s3 + 10 s 2 + 100 K + 1 s + 1000 = 0
P )
Routh Tabulation:

3
s 1 100 + 100 K P
2
s 10 1000
For stability, KP > 0
1
s 100 K P 0
0
s 1000
Root Contours:
100 K P s
Geq ( s ) = 3 2

s + 10 s + 100 s + 1000

9‐45


(c) The maximum overshoots of the system for different values of K P ranging from 0.5 to 20 are
computed and tabulated below.
KP 0.5 1.0 1.6 1.7 1.8 1.9 2.0 3.0 5.0 10 20
ymax 1.393 1.275 1.2317 1.2416 1.2424 1.2441 1.246 1.28 1.372 1.514 1.642

When KP = 1.7, maximum ymax = 1.2416
K
9-22) K lim sG s G s 20 Ö 20 Ö K 5
let K = 6 and targeted PM = 45o . To include some integral action, Ki is set to 1.
First , let’s take a look at uncompensated system:
The open loop bode shows as PM of -25.2 @ 4.35 rad/sec:
9‐46


Bode Diagram
Gm = -12.6 dB (at 2.2 rad/sec) , Pm = -25.2 deg (at 4.35 rad/sec)
100
50
Magnitude (dB)
-50
-100
-150
-135
-180
Phase (deg)
-225
-270
-2 -1 0 1 2 3
10 10 10 10 10 10
Frequency (rad/sec)
To achieve the PM of 45 deg, we need to add a phase lead of (45 – (-25.2)=70.2). By try and error, 2
compensators (a double lead compensator) each with phase lead of 55 deg was found suitable.
Considering the change in cross over frequency after applying the lead filters, overall, a PM of 52 deg
was obtained as seen in the bode diagram of compensated loop:
Double Lead filter design:
1 sin Φ 1 sin 55 1 1
α 10.0590, 0.0210
1 sin Φ 1 sin 55 √10.059 15√10.0590
The maximum phase lead of the compensators are placed at 15 rad/sec, which resulted in a larger PM
(=52.5 deg) compared to applying this phase lead at original cross over frequency of 4.35 rad/sec. This
was due to the shape of phase diagram affected by integral action (i.e. phase starts at -180 @ ω = 0
rad/sec.
The gain crossover frequency is ω 6.92 rad/sec. Bode diagram of compensated loop transfer
function can be observed in the following figure, showing a PM pf 52.5 deg:
9‐47


Bode Diagram
100
Magnitude (dB) 50
-50
-100
-90
-135
Phase (deg)
-180
-225
-270
-2 -1 0 1 2 3
10 10 10 10 10 10
Frequency (rad/sec)
Bode Diagram
20
0
Magnitude (dB)
-20
-40
-60
-80
-100
0
-90
Phase (deg)
-180
-270
-1 0 1 2 3
10 10 10 10 10
Frequency (rad/sec)
9‐48


MATLAB code:
s = tf('s')
Kp = 6;
Ki = 1;
num_GH= 24*(Kp+Ki/s);
den_GH=s*(s+1)*(s+6);
GH=num_GH/den_GH;
%lead design
PL=55
CRover=15
alpha=(1+sin(PL/180*pi))/(1-sin(PL/180*pi))
T=1/alpha^0.5/CRover
lead=(1+T*alpha*s)/(1+T*s)
LT=GH*lead*lead %double lead compensation
CL = LT/(1+LT);
figure(1)
Margin(GH)
figure(2)
Margin(LT)
figure(3)
Bode(CL)
grid on;
9-23)
1
e 0.05
K
40 K s K I
K lim s G s G s lim KI 20
s 2 s 20
Let’s consider 21
As gain crossover frequency is ω 1 Ö |G G jω | 1 Ö 1.25
Let’s see if the PM is in the required range. The bode of the loop transfer function shows a PM
of -57 deg at 3.31.
9‐49


Bode Diagram
Gm = -Inf dB (at 0 rad/sec) , Pm = -57.1 deg (at 3.31 rad/sec)
100
50
Magnitude (dB)
-50
-100
-180
Phase (deg)
-225
-270
-1 0 1 2
10 10 10 10
Frequency (rad/sec)
By try and error, a double lead compensator, each with phase lead of 53 deg was found suitable.
Considering the change in cross over frequency after applying the lead filters, overall, a PM of 35.4 deg
1 sin Φ 1 sin 53 1 1
α 8.9322, 0.0446
1 sin Φ 1 sin 53 √8.9322 7.5√8.9322
The maximum phase lead of the compensators are placed at 7.5 rad/sec, resulting in a larger PM (= 52.5
deg) compared to applying this phase lead at original cross over frequency of 3.31 rad/sec. This was due
to the shape of phase diagram affected by integral action (i.e. phase starts at -180 @ ω = 0 rad/sec.
Then the gain crossover frequency is ω 6.8 rad/sec. Bode diagram of compensated loop can be
observed in the following figure, showing a PM pf 35.4 deg:
9‐50


MATLAB code: Bode Diagram

s = tf('s') 100
Kp = 1.25;
Ki = 21; 50
Magnitude (dB)
den_GH=s*(s+2)*(s+20); 0
GH=num_GH/den_GH;
CL = GH/(1+GH);
-50
%lead design
PL=53
-100
CRover=7.5
alpha=(1+sin(PL/180*pi))/(1
-150
-sin(PL/180*pi)) -135
LT=GH*lead*lead -180
Phase (deg)
CL = LT/(1+LT);
figure(1) -225
Margin(GH)
figure(2)
Margin(LT) -270
-1 0 1 2 3
figure(3) 10 10 10 10 10
Bode(CL) Frequency (rad/sec)
grid on;
Bode Diagram
50
0
Magnitude (dB)
-50
-100
-150
0
-90
Phase (deg)
-180
-270
-1 0 1 2 3
10 10 10 10 10
Frequency (rad/sec)
9‐51


9-24)
To satisfy the unity DC gain (G(0)/(1+Kp*G(0)) = 1), Kp should be equal to 1: Kp=1
In order to add some integral action, 0.2 was chosen as the integral gain.
First, the bode plot of the Loop transfer function is obtained demonstrating a PM of -23.8 deg at 3
rad/sec cross over frequency:
Bode Diagram
Gm = -9.18 dB (at 1.82 rad/sec) , Pm = -23.8 deg (at 3 rad/sec)
100
50
Magnitude (dB)
-50
-100
-135
-180
Phase (deg)
-225
-270
-2 -1 0 1 2
10 10 10 10 10
Frequency (rad/sec)
First, the bode plot of the Loop transfer function is obtained demonstrating a PM of -23.8 deg at 3
rad/sec cross over frequency:
By try and error, a double lead compensator, each with phase lead of 48 deg was found suitable.
Considering the change in cross over frequency after applying the lead filters, overall, a PM of 41.9 deg
1 sin Φ 1 sin 48 1 1
α 6.7865, 0.0427
1 sin Φ 1 sin 48 √6.7865 9√8.9322
9‐52


The maximum phase lead of the compensators are placed at 9 rad/sec, resulting in a larger PM (= 41.9
deg) compared to applying this phase lead at original cross over frequency of 3 rad/sec. This was due to
the shape of phase diagram affected by integral action (i.e. phase starts at -180 @ ω = 0 rad/sec.
Then the gain crossover frequency is ω 4.35 rad/sec. Bode diagram of compensated loop can be
observed in the following figure, showing a PM pf 35.4 deg:
Bode Diagram
100
50
Magnitude (dB)
-50
-100
-150
-90
-135
Phase (deg)
-180
-225
-270
-2 -1 0 1 2 3
10 10 10 10 10 10
Frequency (rad/sec)
MATLAB code:
s = tf('s')
Kp=1;
Ki=0.2;
den_GH=s*(5*s+7)*(s+3);
GH=num_GH/den_GH;
%lead design
PL=48
CRover=9
alpha=(1+sin(PL/180*pi))/(1-sin(PL/180*pi))
LT=GH*lead*lead
CL = LT/(1+LT);
9‐53


figure(1)
Margin(GH)
figure(2)
Margin(LT)
figure(3)
Margin(CL)
Bode Diagram
50
0
Magnitude (dB)
System: CL
-50
-100
-150
0
-90
Phase (deg)
-180
-270
-1 0 1 2 3
10 10 10 10 10
Frequency (rad/sec)
The Bandwidth can be obtained from -3dB in magnitude diagram of the Bode plot. He above data point
in the figure shows BW = 7.44 rad/sec
9‐54


9-25) 2 , therefore, 2
From bode plot of uncompensated loop, we have PM = -111 at ω = 0.363 rad/s:
Bode Diagram
Gm = -25.9 dB (at 0.0576 rad/sec) , Pm = -111 deg (at 0.363 rad/sec)
100
50
Magnitude (dB)
-50
-100
270
180
Phase (deg)
90
0
-3 -2 -1 0 1
10 10 10 10 10
Frequency (rad/sec)
K
A PI controller can be expressed as PI K sK . The effect is similar to adding a Zero at . Let’s
K
place this zero at 71/500 to cancel the Phase lag originating from the unstable zero of G at +71/500:
2353 K (71 − 500 s )
( G (s) = ).
71s ( 40 s + 13)(5000 s + 181)
The compensator can be expressed as:

K
PI K s K s 71/500 , where K can be adjusted in sisotool as the overall gain of the
K
loop, until the required PM is achieved. At K 37, PM=50 deg as seen in the following sisotool
results:
9‐55


10 50
0
5
-50
0 -100
-150
-5
-200
G.M.: 10.5 dB
-10 -250 Freq: 0.0343 rad/sec
-10 -5 0 5 10
Stable loop
200
270
0
180
-200
90
-400
360 0
0 -90
P.M.: 50.5 deg
Freq: 0.0137 rad/sec
-360 -180
-4 -2 0 2 4 -4 -2 0 2 4
10 10 10 10 10 10 10 10 10 10
Considering the slow dynamic of the ship, and the RHS zero, the crossover frequency is relatively low.
The root locus diagram can be seen as:
Zoom in
9‐56


Root Locus
0.06
0.04
0.02
Imaginary Axis
-0.02
-0.04
-0.06
-0.08
-0.6 -0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2
Real Axis
Zoom out
Root Locus
0.6
0.4
0.2
Imaginary Axis
-0.2
-0.4
-0.6
-0.8
-8 -7 -6 -5 -4 -3 -2 -1 0
Real Axis
MATAB Code:
s = tf('s')
Kp=1;
Ki=1;
num_G= 2353*2*(71-500*s)
den_G=71*s*(40*s+13)*(5000*s+181)*(71+500*s)^2;
G=num_G/den_G;
%PI design
9‐57


Kp=1
Ki=71/500
PI=Kp+Ki*s
figure(100)
Margin(G)
figure(101)
rlocus(G*PI)
sisotool
9-26) a) Transfer functions G and H are generated in MATLAB and imported into sisotool:
MATLAB Code:
s = tf('s')
num_G= 2*10^5;
den_G=s*(s+20)*(s^2+50*s+10000);
G=num_G/den_G;
num_c1= 0.05*s+1;
den_c1=0.01*s+1;
c1=num_c1/den_c1;
num_c2= s/0.316+1;
den_c2=s/3.16+1;
c2=num_c2/den_c2;
sisotool
(a) The gain was changed until the cross over frequency matches 31.6 rad/sec as a requirement. At
K=30.3, the desired cross over frequency of 31.6 rad/sec happens as can be seen in the following
sisotool results:
9‐58


300 50
200
0
100
0
-50
-100
-200
-100
G.M.: 5.26 dB
-300 -200 -100 0 100 200
Stable loop
50
-90
-50 -180
-100
0
-270
-180
P.M.: 62.5 deg
Freq: 31.6 rad/sec
-360 -360
0 1 2 3 0 1 2 3 4
10 10 10 10 10 10 10 10 10
9‐59


. K
b) K lim sG s G s ÆK Lim s K 30.3 , then
.
K 30.3
c) Again, K lim sG s G s ÆK Lim s r. To have

K 100 , the overall gain of the PI controller should be equal to 100 (r = 100).
d & e) In this part, the PI pole is asked to be placed at -3.16 rad/sec and the crossover frequency needs to
be at 31.6 rad/sec. The zero and the gain of the PI controller needs to be designed.
r (τs + 1)
Considering the structure of the PI controller given in the question, H ( s ) = , the corresponding
(rτs + 1)
pole is set to -3.16 in sisotool. The place of the zero and the overall gain is iteratively changed in the
MATLAB sisotool to achieve the crossover frequency of 31.6 rad/sec.
With a zero at -0.06321 rad/sec and overall gain of K=270, required crossover frequency (31.6 rad/sec)
and PM of 21.9 deg is obtained as shown in the following sisotool results:
300 150
200 100
100 50
0 0
-100 -50
-200 -100
G.M.: 6.61 dB
-300 -150 Freq: 53.2 rad/sec
-300 -200 -100 0 100 200
Stable loop
100
-90
-100 -180
-200
0
-270
-180
P.M.: 21.9 deg
Freq: 31.6 rad/sec
-360 -360
-2 0 2 4 -4 -2 0 2 4
10 10 10 10 10 10 10 10 10
9‐60


e) the presented sisotool figure shows the compensated bode diagram and 21.9 deg of PM.
9‐61


9-27)
KDs + KPs + KI ⎛ KI 2 ⎞
2
KI
Gc ( s ) = K P + K D s + = = (1 + K D 1 s ) ⎜ K P 2 + ⎟
s s ⎝ s ⎠
where
K P = K P 2 + K D1 K I 2 K D = K D1 K P 2 KI = KI 2
100 (1 + K D1 s )( K P 2 s + K I 2 )
G ( s ) = Gc ( s )G p ( s ) = K v = lim sG ( s ) = K I 2 = 100
(
s s + 10 s + 100
2
) s→ 0
Thus
K I = K I 2 = 100

Consider only the PI controller, (with K D1 = 0 )
Forward‐path Transfer Function: Characteristic Equation:
100 ( K P 2 s + 100 )
G ( s ) = s + 10 s + (100 + 100 K P 2 ) s + 10, 000 = 0
3 2

(
s s + 10 s + 100
2
)

For stability, K P 2 > 9 . Select K P 2 = 10 for fast rise time.
1000 (1 + K D1 s )( s + 10 )
G(s) =
(
s s + 10 s + 100
2
)
9‐62


When K D1 = 0.2, the rise time and
overshoot requirements are satisfied.
K D = K D1 K P 2 = 0.2 × 10 = 2
K P = K P 2 + K D1 K I 2 = 10 + 0.2 × 100 = 30

100
Gc ( s ) = 30 + 2 s +
s

9‐63


9-28)
Process Transfer Function:
−0.2 s
Y (s) e 1
Gp (s) = = ≅
U (s) 1 + 0.25s (1 + 0.25s ) (1 + 0.2 s + 0.02 s 2
)
(a) PI Controller:
KI
KP +
s 200 ( K P s + K I )
G ( s ) = Gc ( s )G p ( s ) ≅ =
(1 + 0.25s ) ( 1 + 0.2 s + 0.02 s
2
)
s ( s + 4 ) s + 10 s + 50
2
( )
200 K I
For K v = 2, K v = lim sG ( s ) = = KI = 2 Thus K I = 2
s→ 0 4 × 50
200 ( 2 + K P s )
Thus G(s) =
(
s ( s + 4 ) s + 10 s + 50
2
)
The following values of the attributes of the unit‐step response are computed for the system with various
values for K P .
Max overshoot ts ts

KP
(%) (sec) (sec)

0.1 19.5 0.61 2.08
0.2 13.8 0.617 1.966
0.3 8.8 0.615 1.733
0.4 4.6 0.606 0.898
0.5 1.0 0.5905 0.878
0.6 0 0.568 0.851
0.7 0 0.541 1.464
0.8 0 0.5078 1.603
1.0 0 0.44 1.618
9‐64


9‐65


9‐66


9‐67


9-29)
9‐68


9-30) If there is no disturbance then
0.9 2
2
Let’s consider PID controller as:
1 1 1/
If τD is sufficiently smaller than τI then the τI has minor effect in PID controller. Let’s examine
PID controller when τD is varied.
0.5 180
100 90
If 10 65 0.5
20 1
so let’s consider τD 10, then 0.1

D
As τI to be large enough with respect to τD, let τI 20 τD 200 Ö 0.005

I
Now we have to determine the value of K so that the gain at the crossover frequency remains at 1.
If K=1 then |G s G s | . 20. Therefore, 20, or K=0.05
9-31) Let
1 1
1 1
where 0.01. therefore, 10 1 100
which gives 10 . Now 10,then PM 40 . As a result, 91o phase lead is

required to achieve PM 45
The crossover frequency is 12.3 rad/sec as can be seen in the uncompensated bode diagram. The
lag compensator must positionω 5rad/sec, where its gain is 17.5 dB. Therefore the ratio of
lag compensator can be chosen for this purpose as 3 10
9‐69


Bode Diagram
40
20
Magnitude (dB)
-20
-40
0
-45
Phase (deg)
-90
-135
-180
-2 -1 0 1 2
10 10 10 10 10
Frequency (rad/sec)
Now the gain, which is obtained from combination of lead and lag compensator, is
1 5 1 5
·
1 5 1 5
or
1 5 1 5
20 20
1 5 1 5
where 1 for α1 < 1. Since it is required that the final gain is increased by 17.5 dB,
let’s choose .
On the other hand, the corner frequency is 0.5 Î 2. Therefore,
1 5
20 23.5
1 5
9‐70


1 5
20 23.5 17.5 6
1 5
As a result, the actual phase reduction is 5.33
The required phase lead is 45 5.33 3 53.33, where

53.33
By trial and error, we can find 0.068 and 0.35. Therefore 147 where
9‐71


9-32)
9‐72


Unit‐step Response.

The unit‐step response shows a maximum overshoot of 26%. Although the relative damping ratio of the
complex roots is 0.707, the real pole of the third‐order system transfer function is at −0.667 which adds
to the overshoot.
(c)
0.00667 (1 + K D1 s )( K P 2 s + K I 2 )
G ( s ) = Gc ( s )G p ( s ) =
(
s s + 0.00667
2
)
For K v = 100, K I 2 = K I = 100. Let us select K P 2 = 50. Then
0.00667 (1 + K D1 s )( 50 s + 100 )
G(s) =
(
s s + 0.00667
2
)
For a small overshoot, K D1 must be relatively large. When K D1 = 100, the maximum overshoot is
approximately 4.5%. Thus,
9‐73


K P = K P 2 + K D1 K I 2 = 50 + 100 × 100 = 10050
K D = K D1 K P 2 = 100 × 50 = 5000
K I = 100
3 2
System Characteristic Equation: s + 33.35 s + 67.04 s + 0.667 = 0
Roots: −0.01, −2.138, −31.2

9‐74


9-33)(a)
Z ( s) 1 1 0.00667
G p ( s) = = 2
= 2
= 2

F ( s) Ms + K s 150 s + 1 s + 0.00667
The transfer function G p ( s ) has poles on the jω axis. The natural undamped frequency is
ω n = 0.0816 rad/sec.
(b) PID Controller:
G ( s ) = Gc ( s )G p ( s ) =
(
0.00667 K D s + K P s + K I
2
)
(
s s + 0.00667
2
)
K v = lim sG ( s ) = K I = 100 Thus K I = 100
s→ 0
s + 0.00667 K D s + 0.00667 (1 + K P ) s + 0.00667 K I = 0

3 2
2
For ζ = 1 and ω n = 1 rad/sec, the second‐order term of the characteristic equation is s + 2 s + 1.
Dividing the characteristic equation by the seond‐order term.
s + ( 0.00667 K D − 2 )
s + 2 s + 1 s + 0.00667 K D s + ( 0.00667 + 0.00667 K P ) s + 0.00667 K I

2 3 2

s + 2s + s
3 2
( 0.00667 K D
− 2 ) s + ( 0.00667 K P − 0.99333 ) s + 0.00667 K I
2
( 0.00667 K − 2 ) s + ( 0.01334 K D − 4 ) s + 0.00667 K D − 2

2
D

( 0.00667 K P
− 0.01334 K D + 3.00667 ) s + 0.00667 K I − 0.00667 K D + 2
For zero remainder,
0.00667 K P − 0.01334 K D + 3.00667 = 0 (1)
−0.00667 K D + 0.00667 K I + 2 = 0 (2)
From Eq. (2),
0.00667 K D = 0.00667 K I + 2 = 2.667 Thus K D = 399.85
From Eq. (1),
9‐75


0.00667 K P = 0.01334 K D − 3.00667 = 2.3273 Thus K P = 348.93
G(s) =
(
0.00667 399.85s + 348.93s + 100
2
)
(
s s + 0.00667
2
)
s + 2.667 s + 2.334 s + 0.667 = ( s + 1) ( s + 0.667 ) = 0

3 2 2

Roots: −1, −1, −0.667

The maximum overshoot is 20%.
9‐76


9-34) a) As , therefore, or
b)
According to the second order system:
1.8 1.8
0.36
5
0.6
1
Let's first add a PD controller with G s 1 K D s, and find KD which satisfy the maximum
overshoot requirement.
After writing the closed-loop transfer function including the PD controller, the characteristic equation
(denominator of the closed loop T.F.) is:
Therefore:
1000 0.36 129.6
Also, 2 2 2 1000 0.6 0.36 50 382 Nsec/m
KI
Now let’s add a PI controller with G s K , and find Kp and KI by using following table.
KI
Now, for the PI part, K I should be selected so that the additional pole at does not interfere
K
with the system dynamics. This pole is usually placed at least 1 decade lower (frequency wise)
than the slowest existing poles of the system. In this case, since 0.05, let’s have
KI
0.005, resulting in K I K 0.005 129.6 0.005 0.648
K
The step response is obtained through the following MATLAB code, showing the rise time of
less than 5 sec, and almost no overshoot
9‐77


Step Response
1
0.99
0.98
0.97
0.96
Amplitude
0.95
0.94
0.93
0.92
0.91
0.9
0 5 10 15 20 25 30 35 40 45 50
Time (sec)
MATLAB code:
s = tf('s')
Kp = 129.6
Kd = 382
Ki = 0.648
num_GH= (Kp*Kd*s)*(1+Ki/s);
den_GH=(1000*s+50);
GH=num_GH/den_GH;
CL = GH/(1+GH)
figure(1)
step(CL)
xlim([-1 50])
ylim([0.9 1])
9‐78


9-35)(a) Process Transfer Function: Forward‐path Transfer Function
4 4 ( KP + KD s)
G p ( s) = 2
G ( s ) = Gc ( s )G p ( s ) = 2

s s
2 2
Characteristic Equation: s + 4 K D s + 4 K P = s + 1.414 s + 1 = 0 for ζ = 0.707, ω n = 1 rad/sec
K P = 0.25 and K D = 0.3535

Maximum overshoot = 20.8%
(b) Select a relatively large value for K D and a small value for K P so that the closed‐loop poles are real.
The closed‐loop zero at s = − K P / K D is very close to one of the closed‐loop poles, and the system
dynamics are governed by the other closed‐loop poles. Let K D = 10 and use small values of K P .
The following results show that the value of K P is not critical as long as it is small.

9‐79


Kp Max overshoot tr ts
(%) (sec) (sec)
0.1 0 0.0549 0.0707
0.05 0 0.0549 0.0707
0.2 0 0.0549 0.0707
(c) For BW ≤ 40 rad/sec and M r = 1, we can again select K D = 10 and a small value for K P .
The following frequency‐domain results substantiate the design.

Kp PM Mr BW
(deg) (rad/sec)
0.1 89.99 1 40
0.05 89.99 1 40
0.2 89.99 1 40
9‐80


9-36) (a) Forward‐path Transfer Function: Characteristic Equation:
10, 000 ( K P + K D s ) 3 2
G ( s ) = Gc ( s )G p ( s ) = s + 10 s + 10,000 K D s + 10,000 K P = 0
s
2
( s + 10 )
Routh Tabulation:
3
s 1 10,000 K D
2
s 10 10,000 K P
The system is stable for K P > 0 and K D > 0.1K P
1
s 10,000 K D − 1000 K P 0
0
s 10,000 K P
(b) Root Locus Diagram:
10, 000 K P
G ( s ) =
s
2
( s + 10 )
9‐81


Root Contours:
0 ≤ K D < ∞ , K P = 0.001, 0.002, 0.005, 0.01.
10,000 K D s
Geq ( s ) = 3 2

s + 10 s + 10,000 K P

(c) The root contours show that for small values of K P the design is insensitive to the variation of K P . This

means that if we choose K P to be between 0.001 and 0.005, the value of K D can be chosen to be 0.005
for a relative damping ratio of 0.707. Let K P = 0.001 and K D = 0.005. Gc ( s ) = 0.001 + 0.005 s.
The forward‐path transfer function becomes

9‐82


10 (1 + 5 s )
G(s) =
s
2
( s + 10 )
Since the zero of G ( s ) is at s = −0.2, which is very close to s = 0, G(s) can be approximated as:
50
G(s) ≅
s ( s + 10 )
For the second‐order system, ζ = 0.707. Using Eq. (7‐104), the rise time is obtained as
2
1 − 0.4167ζ + 2.917ζ
tr = = 0.306 sec
ωn
(d) Frequency‐domain Characteristics:
10 (1 + 5 s )
G(s) =
s
2
( s + 10 )
PM GM Mr BW
(deg) (dB) (rad/sec)
63 ∞ 1.041 7.156
9‐83


9-37) This problem is extensively discussed in Chapters 5 and 6. Use the transfer function (5-123) for
the open-loop system, and a series PID compensator in a unity feedback system.
Gc
Reduce to:
Gc G
Θin _ Va Θ
Where:
Θ(s) Kt
G (s) = =
Va ( s ) s ( La Js + ( La B + Ra J ) s + Ra B + K t K b )
2
KI KDs + K ps + KI
2
Gc ( s ) = K p + K D s + =
s s
With
the rotor inertia (J) = 0.01 kg.m2/s2
damping ratio of the mechanical system (B) = 0.1 Nms
back-emf constant (Kb) = 0.01 Nm/Amp
torque constant (Kt) = 0.01 Nm/Amp
armature resistance (Ra) = 1 Ω
armature inductance (La) = 0.5 H
Starting systematically, set KI=KD=0. Assume a small electric time constant (or small inductance) and
simplify to Equation (5-126):
KK m K s
Θm ( s ) Ra
= (5-125)
Θin ( s ) ⎧ ⎛ KK ⎞ KK K ⎫
(τ e s + 1) ⎪⎨J m s 2 + ⎜ B + b m ⎟ s + m s ⎪⎬
⎪⎩ ⎝ Ra ⎠ Ra ⎪⎭
9‐84


Where Ks is the sensor gain, and, as before, τe = (La/Ra) may be neglected for small La.
KK t K s
Θm ( s ) Ra J
=
Θin ( s ) ⎛ R B + Kt Kb ⎞ KK t K s
s2 + ⎜ a ⎟ s+
⎝ Ra J ⎠ Ra J
KK t K s
Θm ( s ) Ra J
=
Θin ( s ) ⎛ R B + Kt Kb ⎞ KK t K s
s2 + ⎜ a ⎟ s+ (5-126)
⎝ Ra J ⎠ Ra J
K p 0.1
Θm ( s ) 0.01
=
Θin ( s ) ⎛ 0.1 + (0.01)(0.01) ⎞ K p 0.1
s2 + ⎜ ⎟ s+
⎝ 0.01 ⎠ 0.01
Θm ( s ) 10 K p
=
Θin ( s ) s 2 + 12 s + 10 K p
Where Ks=0.
3 .2
Using ts ≅ ; for a less than 2 sec settling time ζωn ≤ 1.6
ζωn
For a PO of 4.3, ζ=0.707, resulting in ωn=2.26.
Then a standard 2nd order prototype system that will have the desired response, with zero steady state
error, takes the following form
Θm ( s ) 5.12
= 2
Θin ( s ) s + 3.2 s + 5.12
For obvious reasons
Θm ( s ) 10 K p 5.12
= 2 ≠ 2
Θin ( s ) s + 12 s + 10 K p s + 3.2 s + 5.12
Let’s add a PD controller
Gc ( s ) = K p + K D s
9‐85


Θm ( s ) 10 ( K D s + K p ) 5.12
= 2 ≠ 2
Θin ( s ) s + (12 + 10 K D ) s + 10 K p s + 3.2 s + 5.12
(12 + 10 K D ) = 5.12
10 K p = 5.12
K D = −0.688
K p = 0.512
Although the two systems are not the same because one has a zero, we chose the controller gain values
by matching the two characteristic equations – as an initial approximation. The resulting zero in the right
hand plane is troubling.
Lets find the response of the system through ACSYS:
Looking at the TF poles, it seems prudent to design the controller by placing its zero farther to LHS of
the s-plane. Set z=-Kp/KD=-20 and vary KD to find the root locus or the response.
9‐86


Done!
9-38) The same as 9-37
9‐87


9-39)
⎡ 0 1 0 0⎤ ⎡ s −1 0 0⎤
⎢ 25.92 0 0 0⎥ ⎢ −25.92 s 0 0⎥
A =⎢ ⎥ sI − A = ⎢ ⎥
∗ ∗

⎢ 0 0 0 1⎥ ⎢ 0 0 s −1⎥
⎢ ⎥ ⎢ ⎥
⎣ −2.36 0 0 0⎦ ⎣ 2.36 0 0 s⎦
s 0 0 −25.92 0 0
Δ = sI − A = s 0 s −1 +
∗
0 s −1 = s
2
(s 2
− 25.92 )
0 0 s 2.36 0 s
⎡ s3 s
2
0 ⎤ 0
⎢ 2 3 ⎥
1 ⎢ 25.92 s s 0 0
( sI − A ) ⎥
∗ −1
=
Δ ⎢ −2.36 s −2.36 s − 25.92 s s − 25.92 ⎥
3 2
⎢ ⎥
⎣ −2.36 s −2.36 s s − 25.92 s ⎦
2 3
0
⎡ 0 ⎤ ⎡ −0.0732 s 2 ⎤
⎢ −0.0732 ⎥ 1 ⎢ −0.0732 s 3 ⎥
( s I − A )
∗ −1
B = ( s I − A ) ⎢ 0 ⎥ = Δ ⎢⎢
∗ −1
⎥
⎢ ⎥ 0.0976 s − 2.357 ⎥
2
⎢ ⎥ ⎢ ⎥
⎣ 0.0976 ⎦ ⎣ 0.0976 s − 2.357 s ⎦
3

Y ( s ) = D ( sI − A
0.0976 s − 24.15 ( 2
)
) B = [0 (
0 ] sI − A )
∗ −1 ∗ −1
0 1 B=
s
2
(s 2
− 25.92 )
Characteristic Equation: s + 0.0976 s + ( 0.0976 K P − 25.92 ) s − 2.357 K D s − 2.357 K P = 0
4 3 2

3 1
The system cannot be stabilized by the PD controller, since the s and the s terms involve K D which
require opposite signs for K D .
9‐88


9-40)
Let us first attempt to compensate the system with a PI controller.
KI 100 ( K P s + K I )
Gc ( s ) = K P + Then G ( s ) = Gc ( s )G p ( s ) =
s (
s s + 10 s + 100
2
)
Since the system with the PI controller is now a type 1 system, the steady‐state error of the system due to a
step input will be zero as long as the values of K P and K I are chosen so that the system is stable.
Let us choose the ramp‐error constant K v = 100 . Then, K I = 100. The following frequency‐domain
performance characteristics are obtained with K I = 100 and various value of K P ranging from 10 to 100.

KP PM GM Mr BW
10 1.60 ∞ 29.70 50.13
20 6.76 ∞ 7.62 69.90
30 7.15 ∞ 7.41 85.40
40 6.90 ∞ 8.28 98.50
50 6.56 ∞ 8.45 106.56
100 5.18 ∞ 11.04 160.00
The maximum phase margin that can be achieved with the PI controller is only 7.15 deg when K P = 30.

Thus, the overshoot requirement cannot be satisfied with the PI controller alone.
Next, we try a PID controller.
9‐89


KI (1 + K s )( K s + KI 2 ) (1 + K s )( K s + 100 )
Gc ( s ) = K P + K D s + = D1 P2
= D1 P2

s s s
Based on the PI‐controller design, let us select K P 2 = 30. Then the forward‐path transfer function
becomes
100 ( 30 s + 100 )(1 + K D1 s )

G(s) =
(
s s + 10 s + 100
2
)
The following attributes of the frequency‐domain performance of the system with the PID controller
are obtained for various values of K D1 ranging from 0.05 to 0.4.

KD1 PM GM Mr BW
0.05 85.0 ∞ 1.04 164.3
0.10 89.4 ∞ 1.00 303.8
0.20 90.2 ∞ 1.00 598.6
0.30 90.2 ∞ 1.00 897.0
0.40 90.2 ∞ 1.00 1201.0
We see that for values of K D1 greater than 0.2, the phase margin no longer increases, but the
bandwidth increases with the increase in K D1. Thus we choose

K D1 = 0.2, K I = K I 2 = 100, K D = K D1 K P 2 = 0.2 × 30 = 6 ,
K P = K P 2 + K D1 K I 2 = 30 + 0.2 × 100 = 50
100
The transfer function of the PID controller is Gc ( s ) = 50 + 6 s +
s
The unit‐step response is show below. The maximum overshoot is zero, and the rise time is 0.0172 sec.
9‐90


9‐91


9-41)
3.6 9 1 1
3.6 9
Let’s consider the characteristic equation like:
3.6 9 2
for 2% settling time.
Therefore we can choose 0.5 and 4 where 2 3. Now, we can choose pole p
far enough from pole dominant of second order. Let p = 10, then the characteristic equation
would be:
36 9 14 56 160
where 160 , 0.29, and 0.065
160 0.065 0.29 1
Verify using MATLAB
9‐92


9-42)
(a)
4 4 (1 + aTs ) 4 aTs
Gp (s) = G ( s ) = Gc ( s )G p ( s ) = Geq ( s ) =
s (1 + Ts )
3 2
Ts + s + 4
2 2
s
Root Contours: (T is fixed and a varies)

Select a small value for T and a large value for a. Let T = 0.02 and a = 100.
1+ 2s 400 ( s + 0.5 )
Gc ( s ) = G(s) =
1 + 0.02 s s
2
( s + 50 )
3 2
The characteristic equation is s + 50 s + 400 s + 200 = 0
The roots are: −0.5355, −9.3, −40.17
The system transfer function is
Y (s) 400 ( s + 0.5 )

=
R(s) ( s + 0.5355 )( s + 9.3 )( s + 40.17 )
9‐93


Since the zero at −0.5 is very close to the pole at −0.5355, the system can be approximated by a second‐
order system,
Y (s) 373.48
=
R(s) ( s + 9.3 )( s + 40.17 )
The unit‐step response is shown below. The attributes of the response are:
Maximum overshoot = 5% t s = 0.6225 sec t r = 0.2173 sec
The following attributes of the frequency‐domain performance are obtained for the system with the
phase‐lead controller.
PM = 77.4 deg GM = infinite Mr = 1.05 BW = 9.976 rad/sec
(b) The Bode plot of the uncompensated forward‐path transfer function is shown below. The diagram
shows that the uncompensated system is marginally stable. The phase of G ( jω ) is −180 deg at all
9‐94


frequencies. For the phase‐lead controller we need to place ω m at the new gain crossover frequency
to realize the desired phase margin which has a theoretical maximum of 90 deg.
For a desired phase margin of 80 deg,
o
1 + sin 80
a= o
= 130
1 − sin 80
The gain of the controller is 20 log10 a = 42 dB. The new gain crossover frequency is at
42
G ( jω ) = − = −21 dB
2
4 2
Or G ( jω ) = 2
= 0.0877 Thus ω = 45.61 ω = 6.75 rad/sec
ω
1
= aω m = 130 × 6.75 = 77 Thus T = 0.013
T
1
= 0.592 Thus aT = 169
.
aT
1 + aTs 1 + 1.702 s 4 (1 + 1.702 s )

Gc ( s ) = = G (s) =
1 + Ts 1 + 0.0131s s (1 + 0.0131s )
2
Bode Plot.
9‐95


9‐96


⎛
1000a ⎜ s +
1 ⎞
1000 (1 + aTs )
⎟
G ( s ) = Gc ( s )GP ( s ) = =
⎝aT ⎠

s ( s + 10 )(1 + Ts ) ⎛ 1⎞
s ( s + 10 ) ⎜ s + ⎟
⎝ T⎠
Set 1/aT = 10 so that the pole of G ( s ) at s = −10 is cancelled. The characteristic equation of the system
becomes
2 1
s + s + 1000 a = 0
T
1
ω n = 1000 a 2ζω n = = 2 1000 a Thus a = 40 and T = 0.0025
T
Controller Transfer Function: Forward‐path Transfer Function:

1 + 0.01s 40, 000

Gc ( s ) = G ( s ) =
1 + 0.0025 s s ( s + 400 )
The attributes of the unit‐step response of the compensated system are:
Maximum overshoot = 0 t r = 0.0168 sec t s = 0.02367 sec
(b) Frequency‐domain Design

The Bode plot of the uncompensated forward‐path transfer function is made below.
1000
G(s) = The attributes of the system are PM = 17.96 deg, GM = infinite.
s ( s + 10 )
Mr = 3.117, and BW = 48.53 rad/sec.
9‐97


To realize a phase margin of 75 deg, we need more than 57 deg of additional phase. Let us add an
additional 10 deg for safety. Thus, the value of φ m for the phase‐lead controller is chosen to be
67 deg. The value of a is calculated from
o
1 + sin 67
a= o
= 24.16
1 − sin 67
The gain of the controller is 20 log10 a = 20 log10 24.16 = 27.66 dB. The new gain crossover frequency
is at
' 27.66
G ( jω m ) = − = −13.83 dB
2

'
From the Bode plot ω m is found to be 70 rad/sec. Thus,
1
= aT = 24.16 × 70 = 344 or T = 0.0029 aT = 0.0702
T

1 + aTs 1 + 0.0702 s
Thus Gc ( s ) = =
1 + Ts 1 + 0.0029 s
The compensated system has the following frequency‐domain attributes:
PM = 75.19 deg GM = infinite Mr = 1.024 BW = 91.85 rad/sec
The attributes of the unit‐step response are:
Rise time tr = 0.02278 sec Settling time ts = 0.02828 sec Maximum overshoot = 3.3%
9‐98


9‐99


9-44) Also see Chapter 6 for solution to this problem.
Mathematical Model:
Draw free body diagrams (Assume both xc and xw are positive and are measured from
equilibrium). Refer to Chapter 4 problems for derivation details.
kc ( xc − xw ) cc ( x&c − x&w )
mw mc

k w ( xw − y ) cw ( x&w − y& ) kc ( xc − xw ) cc ( x&c − x&w )
⎡ mw xw ⎤ ⎡cw + cc
0 ⎤ ⎡ && −cc ⎤ ⎡ x&w ⎤ ⎡ kw + kc −kc ⎤ ⎡ xw ⎤ ⎡cw 0 ⎤ ⎡ y& ⎤ ⎡ kw 0⎤ ⎡ y ⎤
⎢0 ⎥ ⎢ + + = +
⎣ xc ⎥⎦ ⎢⎣ cc
mc ⎦ ⎣ && cc ⎥⎦ ⎢⎣ x&c ⎥⎦ ⎢⎣ kc kc ⎥⎦ ⎢⎣ xc ⎥⎦ ⎢⎣ 0 0 ⎥⎦ ⎢⎣ 0 ⎥⎦ ⎢⎣ 0 0 ⎥⎦ ⎢⎣ 0 ⎥⎦
To Solve we need to simplify, since the problem is very difficult.
Assume the wheel is very stiff; hence k w = ∞ , which implies xw=y. Then
xc + cc x&c + kc xc = cc y& + kc y
mc &&
or
xc + 2ζωn x&c + ωn2 xc = 2ζωn y& + ωn2 y
&&
Placing an actuator between the two masses (ignore actuator dynamics for simplicity), and use a
PD control: the control force is (mc is added to make the final equation look simpler):
F = mc K D ( x&c − x&w ) + mc K P ( xc − xw )
where
xw = y, x&w = y&
The transfer function of the system is:
9‐100


Xc ( 2ζωn + K D ) s + (ωn2 + K P )
= 2
Y s + ( 2ζωn + K D ) s + ( 2ζωn + K D )
The rest is a standard PD controller design.
9‐101


9-45) (a) Forward‐path Transfer Function: (N = 10)
200 (1 + aTs )
G ( s ) = Gc ( s )G p ( s ) =
s ( s + 1)( s + 10 )(1 + Ts )
Starting with a = 1000, we vary T first to stabilize the system. The following time‐domain attributes
are obtained by varying the value of T.

T Max Overshoot tr ts
(%)
0.0001 59.4 0.370 5.205
0.0002 41.5 0.293 2.911
0.0003 29.9 0.315 1.83
0.0004 22.7 0.282 1.178
0.0005 18.5 0.254 1.013
0.0006 16.3 0.230 0.844
0.0007 15.4 0.210 0.699
0.0008 15.4 0.192 0.620
0.0009 15.5 0.182 0.533
0.0010 16.7 0.163 0.525
The maximum overshoot is at a minimum when T = 0.0007 or T = 0.0008. The maximum overshoot
is 15.4%.
9‐102


Unit‐step Response. (T = 0.0008 sec a = 1000)

(b) Frequency‐domain Design.
Similar to the design in part (a), we set a = 1000, and vary the value of T between 0.0001 and 0.001.
The attributes of the frequency‐domain characteristics are given below.
PM GM Mr BW
T (deg) (dB) (rad/sec)
0.0001 17.95 60.00 3.194 4.849
0.0002 31.99 63.53 1.854 5.285
0.0003 42.77 58.62 1.448 5.941
0.0004 49.78 54.53 1.272 6.821
0.0005 53.39 51.16 1.183 7.817
0.0006 54.69 48.32 1.138 8.869
0.0007 54.62 45.87 1.121 9.913
0.0008 53.83 43.72 1.125 10.92
0.0009 52.68 41.81 1.140 11.88
0.0010 51.38 40.09 1.162 12.79
The phase margin is at a maximum of 54.69 deg when T = 0.0006. The performance worsens if the value of a is less than
1000.
9‐103


9‐46 (a) Bode Plot.
The attributes of the frequency response are:
PM = 4.07 deg GM = 1.34 dB Mr = 23.24 BW = 4.4 rad/sec
(b) Single‐stage Phase‐lead Controller.
6 (1 + aTs )
G(s) =
s (1 + 0.2 s )(1 + 0.5 s )(1 + Ts )
We first set a = 1000, and vary T. The following attributes of the frequency‐domain characteristics
are obtained.

T PM Mr
(deg)
0.0050 17.77 3.21
0.0010 43.70 1.34
0.0007 47.53 1.24
0.0006 48.27 1.22
0.0005 48.06 1.23
0.0004 46.01 1.29
0.0002 32.08 1.81
0.0001 19.57 2.97
The phase margin is maximum at 48.27 deg when T = 0.0006.
9‐104


Next, we set T = 0.0006 and reduce a from 1000. We can show that the phase margin is not very
sensitive to the variation of a when a is near 1000. The optimal value of a is around 980, and the
corresponding phase margin is 48.34 deg.
With a = 980 and T = 0.0006, the attributes of the unit‐step response are:
Maximum overshoot = 18.8% tr = 0.262 sec ts = 0.851 sec
(c) Two‐stage Phase‐lead Controller. (a = 980, T = 0.0006)
G(s) =
(
6 (1 + 0.588s ) 1 + bT s
2 )
( )

s (1 + 0.2 s )(1 + 0.5s )(1 + 0.0006 s ) 1 + T s
2
Again, let b = 1000, and vary T2 . The following results are obtained in the frequency domain.
PM Mr

T2
(deg)

0.0010 93.81 1.00
0.0009 94.89 1.00
0.0008 96.02 1.00
0.0007 97.21 1.00
0.0006 98.43 1.00
0.0005 99.61 1.00
0.0004 100.40 1.00
0.0003 99.34 1.00
0.0002 91.98 1.00
0.0001 73.86 1.00
Reducing the value of b from 1000 reduces the phase margin. Thus, the maximum phase margin of
100.4 deg is obtained with b = 1000 and T2 = 0.0004. The transfer function of the two‐stage phase‐
lead controller is
9‐105


(1 + 0.588s )(1 + 0.4 s )

Gc ( s ) =
(1 + 0.0006 s )(1 + 0.0004 s )

(c) Unit‐step Responses.

9‐106


9-47) Also see derivations in 4-9.
L
x(t)
F
L
M
θ
Here is an alternative representation including friction (damping) μ. In this case the angle θ is
measured differently.
Let’s find the dynamic model of the system:
1)
2)
Let . If Φ is small enough then 1 and , therefore
which gives:
Φ s
F s
Ignoring friction 0.
Φ s
F s
where
Ignoring actuator dynamics (DC motor equations), we can incorporate feedback control using a
series PD compensator and unity feedback. Hence,
9‐107


KP+KDs
R _ F Φ
F ( s) = K p ( R ( s ) − Φ ) − K D s ( R ( s ) − Φ )
The system transfer function is:
Φ A( K p + KDs)
= 2
R (s + K D s + A ( K p − B )
Control is achieved by ensuring stability (Kp>B)
Use Routh Hurwitz to establish stability first. Use Acsys to do that as

demonstrated in this chapter problems. Also Chapter 2 has many examples.
Use MATLAB to simulate response:

clear all
Kp=10;
Kd=5;
A=10;
B=8;
num = [A*Kd A*Kp];
G=tf(num,den)
step(G)
Transfer function:
50 s + 100
--------------
s^2 + 5 s + 20
9‐108


Adjust parameters to achieve desired response. Use THE PROCEDURE in Example 5-11-1.
You may look at the root locus of the forward path transfer function to get a better perspective.
Φ A ( K p + K D s ) AK D ( z + s )
= =
E s 2 − AB s 2 − AB
fix z and vary K D .
clear all
z=100;
Kd=0.01;
A=10;
B=8;
num = [A*Kd A*Kd*z];
den =[1 0 -(A*B)];
G=tf(num,den)
rlocus(G)
Transfer function:
0.1 s + 10
----------
s^2 – 80
9‐109


For z=10, a large KD=0.805 results in:
clear all
Kd=0.805;
Kp=10*Kd;
A=10;
B=8;
num = [A*Kd A*Kp];
G=tf(num,den)
pole(G)
zero(G)
step(G)
Transfer function:
8.05 s + 80.5
-------------------
s^2 + 0.805 s + 0.5
ans =
9‐110


-0.4025 + 0.5814i
-0.4025 - 0.5814i
ans =
-10
Looking at dominant poles we expect to see an oscillatory response with overshoot

close to desired values.
For a better design, and to meet rise time criterion, use Example 5-
11-1 and Chapter 9 PD design examples.
9‐111


9-48) (a) The loop transfer function of the system is
−6
10 K p K a K e 1 + R2 Cs 68.76 1 + R2 × 10 s
G(s) H (s) = =
Ns (1 + 0.05 s ) R1Cs s (1 + 0.05 s ) 2s

3 2 −4
The characteristic equation is s + 20 s + 6.876 × 10 R2 s + 687.6 = 0
For root locus plot with R2 as the variable parameter, we have

−4 −4
6.876 × 10 R2 s 6.876 × 10 R2 s
Geq ( s ) = =
s + 20 s + 687.6
3 2
( s + 21.5 )( s − 0.745 + j 5.61)( s − 0.745 − j 5.61)
Root Locus Plot.

9‐112


5
When R2 = 2.65 × 10 , the roots are at −6.02 ± j 7.08 , and the relative damping ratio is 0.65 which is
maximum. The unit‐step response is plotted at the end together with those of parts (b) and (c).
(b) Phase‐lead Controller.
68.76 (1 + aTs )
G(s) H (s) =
s (1 + 0.05 s )(1 + Ts )

Characteristic Equation: Ts + (1 + 20T ) s + ( 20 + 1375.2 aT ) s + 1375.2 = 0

3 2

With T = 0.01, the characteristic equation becomes
3 2
s + 120 s + ( 2000 + 1375.2a ) s + 137520 = 0
The last equation is conditioned for a root contour plot with a as the variable parameter.
Thus
1375.2as
Geq ( s ) = 3 2

s + 120 s + 2000 s + 137,520

From the root contour plot on the next page we see that when a = 3.4 the characteristic equation roots
are at −39.2, − 40.4 + j 43.3, and − 40.4 − j 43.3, and the relative damping ratio is maximum and is
0.682.

Root Contour Plot (a varies).
9‐113


Unit‐step Responses.
9‐114


(c) Frequency‐domain Design of Phase‐lead Controller.
For a phase margin of 60 deg, a = 4.373 and T = 0.00923. The transfer function of the controller is
1 + aTs 1 + 0.04036 s
Gc ( s ) = =
1 + Ts 1 + 0.00923 s
9‐49 (a) Time‐domain Design of Phase‐lag Controller.
Process Transfer Function:
200
Gp (s) =
s ( s + 1)( s + 10 )

For the uncompensated system, the two complex characteristic equation roots are at s = −0.475 + j 0.471
and −0.475 − j 0.471 which correspond to a relative damping ratio of 0.707, when the forward path gain
is 4.5 (as against 200). Thus, the value of a of the phase‐lag controller is chosen to be
4.5
a= = 0.0225 Select T = 1000 which is a large number.
200
Then
1 + aTs 1 + 22.5 s 4.5 ( s + 0.0889) )

Gc ( s ) = = G ( s ) = Gc ( s )G p ( s ) =
1 + Ts 1 + 1000 s s ( s + 1)( s + 10 )( s + 0.001)


9‐115


Maximum overshoot = 13.6 tr = 3.238 sec ts = 18.86 sec
Bode Plot (with phase‐lag controller, a = 0.0225, T = 1000)
9‐116


PM = 59 deg. GM = 27.34 dB Mr = 1.1 BW = 0.6414 rad/sec
(b) Frequency‐domain Design of Phase‐lag Controller.
For PM = 60 deg, we choose a = 0.02178 and T = 1130.55. The transfer function of the phase‐lag
controller is
1 + 24.62 s
Gc ( s ) = GM = 27.66 dB Mr = 1.093 BW = 0.619 rad/sec
1 + 1130.55s

9‐117


Unit‐step Response. Max overshoot = 12.6%, tr = 3.297 sec ts = 18.18 sec

9‐50 (a) Time‐domain Design of Phase‐lead Controller
Forward‐path Transfer Function.
K (1 + aTs ) K
G ( s ) = Gc ( s )G p ( s ) = K v = lim sG ( s ) = = 10 Thus K = 250
s ( s + 5 ) (1 + Ts )
2
s→ 0 25
With K = 250, the system without compensation is marginally stable. For a > 1, select a small value for
T and a large value for a. Let a = 1000. The following results are obtained for various values of T
ranging from 0.0001 to 0.001. When T = 0.0004, the maximum overshoot is near minimum at 23%.

T Max Overshoot tr ts
(%) (sec) (sec)
0.0010 33.5 0.0905 0.808
0.0005 23.8 0.1295 0.6869
0.0004 23.0 0.1471 0.7711
9‐118


0.0003 24.4 0.1689 0.8765
0.0002 30.6 0.1981 1.096
0.0001 47.8 0.2326 2.399
As it turns out a = 1000 is near optimal. A higher or lower value for a will give larger overshoot.
(b) Frequency‐domain Design of Phase‐lead Controller
250 (1 + aTs )
G(s) =
s
2
( s + 5 ) (1 + Ts )
2
Setting a = 1000, and varying T, the following attributes are obtained.
9‐119


T PM Mr BW
(deg) (rad/sec)
0.00050 41.15 1.418 16.05
0.00040 42.85 1.369 14.15
0.00035 43.30 1.355 13.16
0.00030 43.10 1.361 12.12
0.00020 38.60 1.513 10.04
When a = 1000, the best value of T for a maximum phase margin is 0.00035, and PM = 43.3 deg.
As it turns out varying the value of a from 1000 does not improve the phase margin. Thus the
transfer function of the controller is
1 + aTs 1 + 0.35 s 250 (1 + 0.35 s )

Gc ( s ) = = and G(s) =
s ( s + 5 ) (1 + 0.00035s )
2
1 + Ts 1 + 0.00035 s
(c) Time‐domain Design of Phase‐lag Controller
Without compensation, the relative damping is critical when K = 18.5. Then, the value of a is
chosen to be
18.5
a = = 0.074
250
We can use this value of a as a reference, and conduct the design around this point. The value of
T is preferrably to be large. However, if T is too large, rise and settling times will suffer.
The following performance attributes of the unit‐step response are obtained for various values of
a and T.

9‐120


a T Max Overshoot tr ts
(%)
0.105 500 2.6 1.272 1.82
0.100 500 2.9 1.348 1.82
0.095 500 2.6 1.422 1.82
0.090 500 2.5 1.522 2.02
0.090 600 2.1 1.532 2.02
0.090 700 1.9 1.538 2.02
0.090 800 1.7 1.543 2.02
0.090 1000 1.4 1.550 2.22
0.090 2000 0.9 1.560 2.22
0.090 3000 0.7 1.566 2.22
As seen from the results, when a = 0.09 and for T ≥ 2000, the maximum overshoot is less
than 1% and the settling time is less than 2.5 sec. We choose T = 2000 and a = 0.09.
The corresponding frequency‐domain characteristics are:
PM = 69.84 deg GM = 20.9 dB Mr = 1.004 BW = 1.363 rad/sec
(d) Frequency‐domain Design of Phase‐lag Controller
9‐121


250 (1 + aTs )
G ( s ) = a <1
s ( s + 5 ) (1 + Ts )
2
The Bode plot of the uncompensated system is shown below. Let us add a safety factor by
requiring that the desired phase margin is 75 degrees. We see that a phase margin of 75 degrees
can be realized if the gain crossover is moved to 0.64 rad/sec. The magnitude of G ( jω ) at this
frequency is 23.7 dB. Thus the phase‐lag controller must provide an attenuation of −23.7 dB at
the new gain crossover frequency. Setting
20 log10 a = −23.7 dB we have a = 0.065

We can set the value of 1/aT to be at least one decade below 0.64 rad/sec, or 0.064 rad/sec. Thus,
we get T = 236. Let us choose T = 300. The transfer function of the phase‐lag controller becomes

1 + aTs 1 + 19.5s
Gc ( s ) = =
1 + Ts 1 + 300 s
9‐122


9‐123


The attributes of the frequency response of the compensated system are:
PM = 71 deg GM = 23.6 dB Mr = 1.065 BW = 0.937 rad/sec
Maximum overshoot = 6% tr = 2.437 sec ts = 11.11 sec
Comparing with the phase‐lag controller designed in part (a) which has a = 0.09 and T = 2000,
the time response attributes are:
Maximum overshoot = 0.9% tr = 1.56 sec ts = 2.22 sec
The main difference is in the large value of T used in part (c) which resulted in less overshoot,
rise and settling times.
9‐124


9-51)
9-52) Forward‐path Transfer Function (No compensation)
6.087 × 10
7
G(s) = Gp (s) =
(
s s + 423.42 s + 2.6667 × 10 s + 4.2342 × 10
3 2 6 8
)

The uncompensated system has a maximum overshoot of 14.6%. The unit‐step response is shown
below.
(a) Phase‐lead Controller
1 + aTs
Gc ( s ) = (a > 1)
1 + Ts
By selecting a small value for T, the value of a becomes the critical design parameter in this case.
If a is too small, the overshoot will be excessive. If the value of a is too large, the oscillation in
the step response will be objectionable. By trial and error, the best value of a is selected to be 6,
and T = 0.001. The following performance attributes are obtained for the unit‐step response.
However, the step response still has oscillations due to the compliance in the motor shaft. The unit‐
step response of the phase‐lead compensated system is shown below, together with that of the
uncompensated system.
(b) Phase‐lead and Second‐order Controller
9‐125


The poles of the process G p ( s ) are at −161.3, −131+ j1614.7 and −131 − j1614.7. The second‐
2
order term is s + 262 s + 2,624, 417.1. Let the second‐order controller transfer function be
2
s + 262 s + 2,624,417.1
Gc1 ( s ) = 2 2

s + 2ζ pω n s + ω n

The value of ω n is set to 2,624, 417.1 = 1620 rad/sec, so that the steady‐state error is not affected.

Let the two poles of Gc1 ( s ) be at s = −1620 and − 1620. Then, ζ p = 405 .
2
s + 262 s + 2,624,417.1
Gc1 ( s ) = 2

s + 3240 s + 2,624,417.1
6.087 × 10
10
(1 + 0.006 s )
G ( s ) = Gc ( s )Gc1 ( s )G p ( s ) =
( )
s ( s + 161.3 ) s + 3240 s + 2, 624, 417.1 (1 + 0.001s )
2

The unit‐step response is shown below, and the attributes are:
Maximum overshoot = 0.2 tr = 0.01012 sec ts = 0.01414 sec
The step response does not have any ripples.
Unit‐step Responses
9‐126


9‐53 (a) System Equations.
dω m
ea = Ra ia + Kbω m Tm = Ki ia Tm = J m + Bmω m + K L (θ m − θ L ) + BL (ω m − ω L )
dt
dω L
K L (θ m − θ L ) + BL (ω m − ω L ) = J L
dt
State Equations in Vector‐matrix Form:
9‐127


⎡ dθ L ⎤
⎢ dt ⎥ ⎡ 0 1 0 0 ⎤
⎢ ⎥ ⎢ K BL KL BL ⎥ ⎡θ ⎤ ⎡ 0 ⎤
⎢ dω L ⎥ ⎢ − L − ⎥⎢ L ⎥ ⎢ 0 ⎥
⎢ dt ⎥ ⎢ J L JL JL JL ⎥ ⎢ω L ⎥ ⎢ ⎥
⎢ dθ ⎥ = ⎢ 0 ⎥ + ⎢ 0 ⎥ ea
⎥ ⎢⎢ m ⎥⎥ ⎢ K ⎥
0 0 1 θ
⎢ m⎥ ⎢
B + Bl K i K b ⎥ ⎣ω m ⎦ ⎢
a ⎥
⎢ dt ⎥ ⎢ K L BL KL
⎢ dω m ⎥ ⎢ J − − m − ⎥ ⎢
⎣ a m ⎥⎦
R J
⎢ ⎥ ⎣ m Jm Jm Jm J m Ra ⎦
⎣ dt ⎦
State Diagram:

Transfer Functions:

Ωm (s)
=
( )
K i s + BL s + K L / Ra
2

J m J L s + ( K e J L + BL J L + BL J m ) s + ( J m K L + J L K L + K e BL ) s + K L K e
2 2
Ea ( s )
Ω L (s) K i ( BL s + K L ) / Ra
=
J m J L s + ( K e J L + B L J L + B L J m ) s + ( J m K L + J L K L + K e BL ) s + K L K e
3 2
Ea ( s )
9‐128


Ωm (s)
=
(
133.33 s + 10 s + 3000
2
) =
(
133.33 s + 10 s + 3000
2
)
Ea ( s ) s + 318.15 s + 60694.13s + 58240
3 2
( s + 0.9644 )( s + 158.59 + j187.71)( s + 158.59 − j187.71)
Ω L (s) 1333.33 ( s + 300 )

=
Ea ( s ) ( s + 0.9644 )( s + 158.59 + j187.71)( s + 158.59 − j187.71)

(b) Design of PI Controller.
⎛ KI ⎞
Ω L (s)
1333.33 K P ⎜ s + ⎟ ( s + 300 )
G ( s ) = =
⎝ KP ⎠

E (s) s ( s + 0.9644 ) ( s + 317.186 s + 60388.23
2
)

1333.33 × 300 K I
K v = lim sG ( s ) = = 6.87 K I = 100 Thus K I = 14.56
s→ 0 0.9644 × 60388.23
With K I = 14.56 , we study the effects of varying K P . The following results are obtained.
KP tr ts Max Overshoot
(sec) (sec) (%)
20 0.00932 0.02778 4.2
18 0.01041 0.01263 0.7
17 0.01113 0.01515 0
16 0.01184 0.01515 0
15 0.01303 0.01768 0
10 0.02756 0.04040 0.6
With K I = 14.56 and K P ranging from 15 to 17, the design specifications are satisfied.
9‐129



(c) Frequency‐domain Design of PI Controller (KI = 14.56)
1333.33 ( K P s + 14.56 )( s + 300 )

G(s) =
(
s s + 318.15s + 60694.13s + 58240
3 2
)

9‐130


The following results are obtained by setting K I = 14.56 and varying the value of K P .

KP PM GM Mr BW Max Overshoot tr ts
(deg) (dB) (rad/sec) (%) (sec) (sec)
20 65.93 ∞ 1.000 266.1 4.2 0.00932 0.02778
18 69.76 ∞ 1.000 243 0.7 0.01041 0.01263
17 71.54 ∞ 1.000 229 0 0.01113 0.01515
16 73.26 ∞ 1.000 211.6 0 0.01184 0.01515
15 74.89 ∞ 1.000 190.3 0 0.01313 0.01768
10 81.11 ∞ 1.005 84.92 0.6 0.0294 0.0404
8 82.66 ∞ 1.012 63.33 1.3 0.04848 0.03492
7 83.14 ∞ 1.017 54.19 1.9 0.03952 0.05253
6 83.29 ∞ 1.025 45.81 2.7 0.04697 0.0606
5 82.88 ∞ 1.038 38.12 4.1 0.05457 0.0606
From these results we see that the phase margin is at a maximum of 83.29 degrees when K P = 6 .
However, the maximum overshoot of the unit‐step response is 2.7%, and M r is slightly greater than
one. In part (b), the optimal value of K P from the standpoint of minimum value of the maximum
overshoot is between 15 and 17. Thus, the phase margin criterion is not a good indicator in the present
case.

9‐131


9‐54 (a) Forward‐path Transfer Function
G p ( s ) =
K Θm (s)
=
(
100 K s + 10 s + 100
2
=
) ( 10, 000 s + 10 s + 100
2
)
Tm ( s ) (
s s + 20 s
3 2
+ 2100 s + 10, 000 ) s ( s + 4.937 ) ( s
2
+ 15.06 s + 2025.6 )

The unit‐step response is plotted as shown below. The attributes of the response are:
(b) Design of the Second‐order Notch Controller
The complex zeros of the notch controller are to cancel the complex poles of the process transfer
function. Thus
Gc ( s ) =
2
s + 15.06 s + 2025.6
and G ( s ) = G ( s )G ( s ) =
(
10, 000 s + 10 s + 100
2
)
2
s ( s + 4.937 ) ( s )
c p
s + 90ζ p s + 2025.6 + 90ζ p s + 2025.6
2
The following results are obtained for the unit‐step response when various values of ζ p are used.
The maximum overshoot is at a minimum of 4.1% when ζ p = 1.222 . The unit‐step response is
plotted below, along with that of the uncompensated system.
ζ p 2ζω n Max Overshoot
(%)
2.444 200 7.3
2.333 210 6.9
2.222 200 6.5
9‐132


1.667 150 4.9
1.333 120 4.3
1.222 110 4.1
1.111 100 5.8
1.000 90 9.8

(c) Frequency‐domain Design of the Notch Controller
The forward‐path transfer function of the uncompensated system is
9‐133


G ( s) =
(
10000 s + 10 s + 100
2
)
s ( s + 4.937 ) ( s + 15.06 s + 2025.6 )
2

The Bode plot of G ( jω ) is constructed in the following. We see that the peak value of G ( jω ) is
approximately 22 dB. Thus, the notch controller should provide an attentuation of −22 dB or 0.0794
at the resonant frequency of 45 rad/sec. Using Eq. (10‐155), we have
ζz 0.167
Gc ( j 45) = = = 0.0794 Thus ζ p = 2.1024
ζp ζp

Notch Controller Transfer Function Forward‐path Transfer Function
Gc ( s ) =
2
s + 15.06 s + 2025.6
G ( s ) =
(
10, 000 s + 10 s + 100
2
)
s ( s + 4.937 ) ( s )
2
s + 189.216 s + 2025.6 + 189.22 s + 2025.6
2
Bode Plots
9‐134


Attributes of the frequency response: PM = 80.37 deg GM = infinite M r = 1097
. BW = 66.4 rad/sec
Attributes of the frequency response of the system designed in part (b):
PM = 59.64 deg GM = infinite M r = 1048

. BW = 126.5 rad/sec

9‐135


9‐55 (a) Process Transfer Function

500 ( s + 10 )
Gp ( s) =
(
s s + 10 s + 1000
2
)

The Bode plot is constructed below. The frequency‐domain attributes of the uncompensated
system are:
PM = 30 deg GM = infinite M r = 186

. and BW = 3.95 rad/sec

The unit‐step response is oscillatory.
(b) Design of the Notch Controller
For the uncompensated process, the complex poles have the following constants:

ω n = 1000 = 316
. rad / sec 2ζω n = 10 Thus ζ = 0.158

The transfer function of the notch controller is
2 2
s + 2ζ zω n s + ω n
Gc ( s ) = 2 2

s + 2ζ p s + ω n

For the zeros of Gc ( s ) to cancel the complex poles of G p ( s ) , ζ z = ζ = 0.158 .
From the Bode plot, we see that to bring down the peak resonance of G ( jω n ) in order
to smooth out the magnitude curve, the notch controller should provide approximately
9‐136


−26 dB of attenuation. Thus, using Eq. (10‐155),
−26
ζz 0.158
= 10 20
= 0.05 Thus ζp = = 31525
.
ζp 0.05

The transfer function of the notch controller is
2
s + 10 s + 1000 500 ( s + 10 )
Gc ( s ) = G ( s ) = Gc ( s )G p ( s ) =
2
s + 199.08 s + 1000 (
s s + 199.08s + 1000
2
)

The attributes of the compensated system are:
PM = 72.38 deg GM = infinite M r = 1 BW = 5.44 rad/sec
Maximum overshoot = 3.4% t r = 0.3868 sec t s = 0.4848 sec

Bode Plots
9‐137


Step Responses
9‐138


(c) Time‐domain design of the Notch Controller
With ζ z = 0.158 and ω n = 316

. , the forward‐path transfer function of the compensated system is
500 ( s + 10 )
G ( s ) = Gc ( s )G p ( s ) =
(
s s + 63.2ζ p s + 1000
2
)

The following attributes of the unit‐step response are obtained by varying the value of ζ p .

ζ p 2ζω n Max Overshoot tr ts
(%) (sec) (sec)
1.582 100 0 0.4292 0.5859
1.741 110 0 0.4172 0.5657
1.899 120 0 0.4074 0.5455
2.057 130 0 0.3998 0.5253
2.215 140 0.2 0.3941 0.5152
2.500 158.25 0.9 0.3879 0.4840
3.318 209.7 4.1 0.3884 0.4848
When ζ p = 2.5 the maximum overshoot is 0.9%, the rise time is 0.3879 sec and the setting
time is 0.4840 sec. These performance attributes are within the required specifications.

9‐139


9‐56 Let the transfer function of the controller be
Gc ( s ) =
(
20, 000 s + 10 s + 50
2
)
( s + 1000 ) 2

Then, the forward‐path transfer function becomes
G ( s ) = Gc ( s )G p ( s ) =
(
20, 000 K s + 10 s + 50
2
)
(
s s + 10 s + 100
2
) ( s + 1000 ) 2
6
10 K
For Gcf ( s ) = 1, K v = lim sG ( s ) = 8
= 50 Thus the nominal K = 5000
s→ 0 10
For ± 20% variation in K, K min = 4000 and K max = 6000. To cancel the complex closed‐loop poles,
we let
50 ( s + 1)
Gcf ( s ) = where the (s + 1) term is added to reduce the rise time.
s + 10 s + 50
2

Closed‐loop Transfer Function:
10 K ( s + 1)
6
Y (s)
=
R(s) (
s s + 10 s + 100
2
)( s + 1000 ) + 20, 000 K ( s
2 2
+ 10 s + 50 )


5 4 3 7 2 8 9
K = 4000: s + 2010 s + 1,020,100 s + 9.02 × 10 s + 9 × 10 + 4 × 10 = 0
Roots: −97.7, − 648.9, − 1252.7, − 5.35 + j 4.6635, − 5.35 − j 4.6635
Max overshoot ≅ 6.7% Rise time < 0.04 sec
9‐140


5 4 3 8 2 9 9
K = 5000: s + 2010 s + 1,020100 s + 11
. × 10 s + 11
. × 10 s + 5 × 10 = 0
Roots: −132.46, 587.44, − 1279.6, − 5.272 + j 4.7353, − 5.272 − j 4.7353
Max overshoot ≅ 4% Rise time < 0.04 sec

5 4 3 8 2 9 9
K = 6000 s + 2010 s + 1,020,100 s + 13
. × 10 s + 13
. × 10 s + 6 × 10 = 0
Roots: −176.77, − 519.37, − 1303.4, − 5.223 + j 4.7818, − 5.223 − j 4.7818
Max overshoot ≅ 2.5% Rise time < 0.04 sec
Thus all the required specifications stay within the required tolerances when the value of K varies by
plus and minus 20%.
Unit‐step Responses

9‐141


9‐57 Let the transfer function of the controller be
Gc ( s ) =
(
200 s + 10 s + 50
2
)
( s + 100 ) 2
G ( s ) = Gc ( s )G p ( s ) =
(
200, 000 K s + 10 s + 50
2
)
s ( s + a )( s + 100 )
2

For a = 10,
7
10 K
K v = lim sG ( s ) = 5
= 100 K = 100 Thus K = 1
s→ 0 10
Characteristic Equations: (K = 1)
4 3 5 2 6 7
a = 10: s + 210 s + 2.12 × 10 s + 2.1 × 10 s + 10 = 0
Roots: −4.978 + j 4.78, − 4.978 − j 4.78, − 100 + j 447.16 − 100 − j 447.16

4 3 5 2 6 7
a = 8: s + 208 s + 2.116 × 10 s + 2.08 × 10 s + 10 = 0
Roots: −4.939 + j 4.828, − 4.939 − j 4.828, − 99.06 + j 446.97, − 99.06 − j 446.97
4 3 5 6 7
a = 12: s + 212 s + 2.124 × 10 + 2.12 × 10 s + 10 = 0
Roots: −5.017 + j 4.73, − 5.017 − j 4.73, − 100.98 + j 447.36, − 100.98 − j 447.36
9‐142


Unit‐step Responses: All three responses for a = 8, a = 10, and 12 are similar.
Y ( s) K K
G (s) = = K v = lim sG ( s ) = =1
E (s) s ( s + 1)( s + 10 ) + KK t s s →0 10 + KK t
Characteristic Equation: s + 11s + (10 + KK t ) s + K = s + 11s + Ks + K = 0

3 2 3 2

K ( s + 1)
For root loci, Geq ( s ) =
s
2
( s + 11)
9‐143


Root Locus Plot (K varies)
The root loci show that a relative damping ratio of 0.707 can be realized by two values of K. K = 22
and 59.3. As stipulated by the problem, we select K = 59.3.
9‐144


10 K 10 K K
G(s) = K v = lim sG ( s ) = = = 1 Thus Kt = K − 1
s ( s + 1)( s + 10 ) + 10 K t s s→ 0 10 + 10 Kt 1 + Kt
Characteristic Equation: s ( s + 1)( s + 10 ) + 10 K t + 10 K = s + 11s + 10 Ks + 10 K = 0

3 2

When K = 5.93 and Kt = K − 1 = 4.93 , the characteristic equation becomes
3 2
s + 11s + 10.046 s + 4.6 = 0

The roots are: −10.046, − 0.47723 + j 0.47976, − 0.47723 − j 0.47976
K (1 + aTs ) 1
G(s) = K v = lim sG ( s ) = = 100 Thus Kt = 0.01
(
s ( (1 + Ts ) s + 10 s + KK t
2
) s→ 0 Kt
Let T = 0.01 and a = 100. The characteristic equation of the system is written:

2
(
s + 110 s + 1000 s + K 0.001s + 101s + 100 = 0
4 3 2
)

To construct the root contours as K varies, we form the following equivalent forward‐path transfer
function:
Geq ( s ) =
(
0.001K s + 101, 000 s + 100, 000
2
) = 0.001K ( s + 1)( s + 50499 )
s
2
( s + 10 )( s + 100 ) s
2
( s + 10 )( s + 100 )
9‐145


From the root contour diagram we see that two sets of solutions exist for a damping ratio of 0.707.
These are:
K = 20: Complex roots: −1158

. + j1155
. , − 1158
. − j1155
.
K = 44.6: Complex roots: −4.0957 + j 4.0957, − 4.0957 − j 4.0957

The unit‐step responses of the system for K = 20 and 44.6 are shown below.
Unit‐step Responses:
9‐146


K s KK1 K i N
G(s) =
s ⎡⎣ J t La s + ( Ra J t + La Bt + K1 K 2 J t ) s + Ra Bt + K1 K 2 Bt + K b K i + KK1 K i K t ⎤⎦
2
1.5 × 10 K
7
G ( s) =
(
s s + 3408.33s + 1, 204, 000 + 1.5 × 10 KK t
2 8
)

15 K
Ramp Error Constant: K v = lim sG ( s ) = = 100
s→ 0 1.204 + 150 KKt

Thus 1.204 + 150 KKt = 0.15 K
9‐147


1.5 × 10 K
7
G(s) =
(
s s + 3408.33s + 150, 000 K
2
)
3 7
Characteristic Equation: s + 3408.33 s + 150,000 Ks + 15
. × 10 K = 0
When K = 38.667 the roots of the characteristic equation are at
−0.1065, − 1651
. + j165
. , − 1651
. − j165
. (ζ ≅ 0.707 for the complex roots)
5.8 × 10
8
G(s) =
(
s s + 3408.33s + 5.8 × 10
2 6
)

9‐148


Unit‐step response attributes: Maximum overshoot = 0 Rise time = 0.0208 sec Settling time = 0.0283 sec
9‐62 (a) Disturburnce‐to‐Output Transfer Function
Y (s) 2 (1 + 0.1s )
== Gc ( s ) = 1
TL ( s ) r =0
s (1 + 0.01s )(1 + 0.1s ) + 20 K
For TL ( s ) = 1 / s
1
lim y ( t ) = lim sY ( s ) = ≤ 0.01 Thus K ≥ 10
t →∞ s→ 0 10 K
(b) Performance of Uncompensated System. K = 10, Gc ( s ) = 1
9‐149


200
G(s) =
s (1 + 0.01s )(1 + 0.1s )
The Bode diagram of G ( jω ) is shown below. The system is unstable. The attributes of the
frequency response are: PM = −9.65 deg GM = −5.19 dB.
(c) Single‐stage Phase‐lead Controller Design
To realize a phase margin of 30 degrees, a = 14 and T = 0.00348.
1 + aTs 1 + 0.0487 s
Gc ( s ) = =
1 + Ts 1 + 0.00348 s
The Bode diagram of the phase‐lead compensated system is shown below. The performance
attributes are: PM = 30 deg GM = 10.66 dB M r = 1.95 BW = 131.6 rad/sec.
(d) Two‐stage Phase‐lead Controller Design
200 (1 + 0.0487 s )
Starting with the forward‐path transfer function G ( s ) =
s (1 + 0.1s )(1 + 0.01s )(1 + 0.00348 s )
The problem becomes that of designing a single‐stage phase‐lead controller. For a phase margin
or 55 degrees, a = 7.385 and T = 0.00263. The transfer function of the second‐stage controller is

1 + aTs 1 + 0.01845 s
Gc1 ( s ) = =
1 + Ts 1 + 0.00263 s
9‐150


200 (1 + 0.0487 s )(1 + 0.01845 s )

Thus G(s) =
s (1 + 0.1s )(1 + 0.01s )(1 + 0.00348 s )(1 + 0.00263s )
The Bode diagram is shown on the following page. The following frequency‐response attributes
are obtained:
PM = 55 deg GM = 12.94 dB M r = 111

. BW = 256.57 rad/sec

Bode Plot [parts (b), (c), and (d)]
9‐151


9‐152


9‐63 (a) Two‐stage Phase‐lead Controller Design.
The uncompensated system is unstable. PM = −43.25 deg and GM = −18.66 dB.
With a single‐stage phase‐lead controller, the maximum phase margin that can be realized affectively
is 12 degrees. Setting the desired PM at 11 deg, we have the parameters of the single‐stage phase‐
lead controller as a = 128.2 and T1 = 0.00472 . The transfer function of the single‐stage controller
1 + aT1 s 1 + 0.6057 s
is Gc1 ( s ) = =
1 + T1 s 1 + 0.00472 s
Starting with the single‐stage‐controller compensated system, the second stage of the phase‐lead
controller is designed to realize a phase margin of 60 degrees. The parameters of the second‐stage
controller are: b = 16.1 and T2 = 0.0066 . Thus,
1 + bT2 s 1 + 0.106 s
Gc 2 ( s ) = =
1 + T2 s 1 + 0.0066 s
1 + 0.6057 s 1 + 0.106 s
Gc ( s ) = Gc1 ( s )Gc 2 ( s ) =
1 + 0.00472 s 1 + 0.0066 s
1, 236, 598.6 ( s + 1.651)( s + 9.39 )

G ( s ) = Gc1 ( s )Gc 2 ( s )G p ( s ) =
s ( s + 2)( s + 5 )( s + 211.86 )( s + 151.5 )
Attributes of the frequency response of the compensated system are:
GM = 19.1 dB PM = 60 deg M r = 1.08 BW = 65.11 rad/sec
The unit‐step response is plotted below. The time‐response attributes are:
Maximum overshoot = 10.2% t s = 0.1212 sec t r = 0.037 sec
9‐153


(b) Single‐stage Phase‐lag Controller Design.
With a single‐stage phase‐lag controller, for a phase margin of 60 degrees, a = 0.0108 and T = 1483.8.
The controller transfer function is
1 + 16.08 s
Gc ( s ) =
1 + 1483.8 s
The forward‐path transfer function is
G ( s ) = Gc ( s )G p ( s ) =
β γ
6.5 s + 0.0662
sβ
s + 2γ
βs + 5γβs + 0.000674γ

From the Bode plot, the following frequency‐response attributes are obtained:
PM = 60 deg GM = 20.27 dB M r = 1.09 BW = 1.07 rad/sec

The unit‐step response has a long rise time and settling time. The attributes are:
Maximum overshoot = 12.5% t s = 12.6 sec t r = 2.126 sec
(c) Lead‐lag Controller Design.
For the lead‐lag controller, we first design the phase‐lag portion for a 40‐degree phase margin.
The result is a = 0.0238 and T1 = 350 . The transfer function of the controller is
1 + 8.333 s
Gc1 ( s ) =
1 + 350 s
The phase‐lead portion is designed to yield a total phase margin of 60 degrees. The result is
b = 4.8 and T2 = 0.2245. The transfer function of the phase‐lead controller is
1 + 1076
. s
Gc 2 ( s ) =
1 + 0.2245s
9‐154


The forward‐path transfer function of the lead‐lag compensated system is
68.63 ( s + 0.12 )( s + 0.929 )

G(s) =
s ( s + 2 )( s + 5 )( s + 0.00286 )( s + 4.454 )
Frequency‐response attributes: PM = 60 deg GM = 13.07 dB M r = 1.05 BW = 3.83 rad/sec
Unit‐step response attributes: Maximum overshoot = 5.9% t s = 1512
. sec t r = 0.7882 sec
Unit‐step Responses.

9‐64 (a) The uncompensated system has the following frequency‐domain attributes:
PM = 3.87 deg GM = 1 dB M r = 7.73 BW = 4.35 rad/sec
The Bode plot of G p ( jω ) shows that the phase curve drops off sharply, so that the phase‐lead
controller would not be very effective. Consider a single‐stage phase‐lag controller. The phase
9‐155


margin of 60 degrees is realized if the gain crossover is moved from 2.8 rad/sec to 0.8 rad/sec.
The attenuation of the phase‐lag controller at high frequencies is approximately −15 dB.
Choosing an attenuation of −17.5 dB, we calculate the value of a from
20 log10 a = −17.5 dB Thus a = 0.1334
The upper corner frequency of the phase‐lag controller is chosen to be at 1/aT = 0.064 rad/sec.
Thus, 1/T = 0.00854 or T = 117.13. The transfer function of the phase‐lag controller is
1 + 15.63 s
Gc ( s ) =
1 + 117.13
The forward‐path transfer function is
5 (1 + 15.63s )(1 − 0.05 s )

G ( s ) = Gc ( s ) G p ( s ) =
s (1 + 0.1s )(1 + 0.5 s )(1 + 117.13s )(1 + 0.05 s )
From the Bode plot of G ( jω ) , the following frequency‐domain attributes are obtained:
PM = 60 deg GM = 18.2 dB M r = 1.08 BW = 1.13 rad/sec

The unit‐step response attributes are:
maximum overshoot = 10.7% t s = 10.1 sec t r = 2.186 sec

Bode Plots
9‐156


9‐157


9‐64 (b) Using the exact expression of the time delay, the same design holds. The time and frequency domain
attributes are not much affected.
9‐65 (a) Uncompensated System.
10
Forward‐path Transfer Function: G(s) =
(1 + s )(1 + 10 s )(1 + 2 s )(1 + 5s )
The Bode plot of G ( jω ) is shown below.
The performance attributes are: PM = −10.64 deg GM = −2.26 dB
The uncompensated system is unstable.
(b) PI Controller Design.
10 ( K p s + K I )
Forward‐path Transfer Function: G (s) =
s (1 + s )(1 + 10 s )(1 + 2 s )(1 + 5 s )
Ramp‐error Constant: K v = lim sG ( s ) = 10 K I = 0.1 Thus K I = 0.01

s→ 0
0.1 (1 + 100 K P s )
G ( s) =
s (1 + s )(1 + 10 s )(1 + 2 s )(1 + 5 s )
The following frequency‐domain attributes are obtained for various values of K P .
KP PM GM Mr BW
0.01 24.5 5.92 2.54 0.13
0.02 28.24 7.43 2.15 0.13
0.05 38.84 11.76 1.52 0.14
9‐158


0.10 50.63 12.80 1.17 0.17
0.12 52.87 12.23 1.13 0.18
0.15 53.28 11.22 1.14 0.21
0.16 52.83 10.88 1.16 0.22
0.17 51.75 10.38 1.18 0.23
0.20 49.08 9.58 1.29 0.25
The phase margin is maximum at 53.28 degrees when K P = 0.15.
The forward‐path transfer function of the compensated system is
0.1 (1 + 15 s )
G ( s) =
s (1 + s )(1 + 10 s )(1 + 5 s )(1 + 2 s )
The attributes of the frequency response are:
PM = 53.28 deg GM = 11.22 dB M r = 114

. BW = 0.21 rad/sec
Maximum overshoot = 14.1% t r = 10.68 sec t s = 48 sec

Bode Plots
9‐159


9‐160


9‐65 (c) Time‐domain Design of PI Controller.
By setting K I = 0.01 and varying K P we found that the value of K P that minimizes the maximum
overshoot of the unit‐step response is also 0.15. Thus, the unit‐step response obtained in part (b)
is still applicable for this case.
9‐66 Closed‐loop System Transfer Function.
Y (s) 1
=
s + ( 4 + k3 ) s + ( 3 + k 2 + k3 ) s + k1
3 2
R(s)
For zero steady‐state error to a step input, k1 = 1. For the complex roots to be located at −1 +j and −1 − j,
2
we divide the characteristic polynomial by s + 2 s + 2 and solve for zero remainder.
s + ( 2 + k2 )
s + 2 s + 2 s + ( 4 + k3 ) s + ( 3 + k 2 + k3 ) s + 1
2 3 2

s + + 2s
3 2
2s
(2 + k ) s
3
2
+ (1 + k 2 + k 3 ) s + 1
(2 + k ) s
3
2
+ ( 4 + 2 k3 ) s + 4 + 2 k3

( -3+k 2
− k3 ) s − 3 − 2k3
For zero remainder, − 3 − 2 k3 = 0 Thus k3 = −15

.
−3 + k 2 − k3 = 0 Thus k 2 = 15
.
The third root is at −0.5. Not all the roots can be arbitrarily assigned, due to the requirement on the
steady‐state error.

9‐161


9‐67 (a) Open‐loop Transfer Function.
X 1 (s) k3
G ( s) = =
s ⎡⎣ s + ( 4 + k 2 ) s + 3 + k1 + k 2 ⎤⎦
2
E (s)
Since the system is type 1, the steady‐state error due to a step input is zero for all values of k1 , k 2 , and k3
that correspond to a stable system. The characteristic equation of the closed‐loop system is
s + ( 4 + k 2 ) s + ( 3 + k1 + k 2 ) s + k 3 = 0
3 2

For the roots to be at −1 + j, −1 − j, and −10, the equation should be:
3 2
s + 12 s + 22 s + 20 = 0
Equating like coefficients of the last two equations, we have
4 + k 2 = 12 Thus k2 = 8
3 + k1 + k 2 = 22 Thus k1 = 11
k3 = 20 Thus k3 = 20
(b) Open‐loop Transfer Function.
Y (s) Gc ( s ) 20 20 ( s + 1)( s + 3 )
= = Thus Gc ( s ) =
E ( s) ( s + 1)( s + 3 ) (
s s + 12 s + 22
2
) (
s s + 12 s + 22
2
)

9‐162


Chapter 10_________________________________________________________________________
dy
10-1) (a) State variables: x1 = y, x 2 =
dt
State equations: Output equation:
⎡ dx1 ⎤
⎢ dt ⎥ ⎡ 0 1 ⎤ ⎡ x1 ⎤ ⎡ 0 ⎤ ⎡ x1 ⎤
⎢ ⎥=⎢ ⎥ ⎢ x ⎥ + ⎢5 ⎥ r y = [1 0] ⎢ ⎥ = x1
⎢ 2⎥
dx ⎣ − 1 − 4 ⎦⎣ 2⎦ ⎣ ⎦ ⎣ x2 ⎦
⎢⎣ dt ⎥⎦
2
dy d y
(b) State variables: x1 = y, x 2 = , x3 = 2

dt dt
⎡ dx1 ⎤
⎢ dt ⎥
⎢ ⎥ ⎡0 1 0 ⎤ ⎡ x1 ⎤ ⎡ 0 ⎤ ⎡ x1 ⎤
⎢ ⎢
⎥= 0 1 ⎥ ⎢ x2 ⎥ + ⎢ 0 ⎥ r y = [1 0 0 ] ⎢ x2 ⎥ = x1
dx 2
0
⎢ dt ⎥ ⎢ ⎥⎢ ⎥ ⎢ ⎥ ⎢ ⎥
⎢ dx ⎥ ⎣ ⎢ −1 −2.5 − 1.5 ⎥⎦ ⎢⎣ x3 ⎥⎦ ⎢⎣ 0.5⎥⎦ ⎢⎣ x3 ⎥⎦
⎢ 3⎥
⎣⎢ dt ⎦⎥
2
dx1 dy d y
∫
t
(c) State variables: x1 = y (τ ) dτ , x2 = , x3 = , x4 = 2

0
dt dt dt
⎡ x&1 ⎤ ⎡ 0 1 0 0 ⎤ ⎡ x1 ⎤ ⎡ 0 ⎤ ⎡ x1 ⎤
⎢ x& ⎥ ⎢ 0 0 1 0 ⎥ ⎢ x ⎥ ⎢ 0 ⎥ ⎢x ⎥
⎢ ⎥ = ⎢ ⎥ ⎢ 2 ⎥ + ⎢ ⎥ r y = [1 0 0 0 ] ⎢ ⎥ = x1
2 2

⎢ x&3 ⎥ ⎢ 0 0 0 1 ⎥ ⎢ x3 ⎥ ⎢ 0 ⎥ ⎢ x3 ⎥
⎢& ⎥ ⎢ ⎥⎢ ⎥ ⎢ ⎥ ⎢ ⎥
⎣ x4 ⎦ ⎣ −1 −1 −3 −5⎦ ⎣ x4 ⎦ ⎣1 ⎦ ⎣ x4 ⎦
2 3
dy d y d y
(d) State variables: x1 = y, x 2 = , x3 = 2
, x4 = 3

dt dt dt
⎡ x&1 ⎤ ⎡ 0 1 0 0 ⎤ ⎡ x1 ⎤
⎡0 ⎤ ⎡ x1 ⎤
⎢ x& ⎥ ⎢ 0 0 1 0 ⎥ ⎢ x ⎥ ⎢0 ⎥ ⎢x ⎥
⎢ ⎥ = ⎢ ⎥⎢ 2⎥+ ⎢ ⎥r y = [1 0 0 0 ] ⎢ ⎥ = x1
2 2

⎢ x&3 ⎥ ⎢ 0 0 0 1 ⎥ ⎢ x3 ⎥ ⎢ 0 ⎥ ⎢ x3 ⎥
⎢& ⎥ ⎢ ⎥⎢ ⎥ ⎢ ⎥ ⎢ ⎥
⎣ x4 ⎦ ⎣ −1 −2.5 0 −1.5 ⎦ ⎣ x4 ⎦ ⎣ 1 ⎦ ⎣ x4 ⎦
10‐1


10-2) a)
Ö 3 2 3
Ö 3
Let
3 3
Then Ö
2 3 2 3
If and , then
3
2 3
or
3 1 1
2 0 3
1 0
b)
Ö 6 11 6 6
Ö 6
Let X s , therefore 11
and Let , then 6 6 . As a result:
6
11
6 6
or
6
11
6 6
If , , then
6 1 0 0
11 0 1 0 u
6 0 0 6
10‐2


1 0 0
c)
Ö 7 12 2
Ö 7
Let , then 12 2 . As a result:
7
12 2
Let and , then
7 1 1
12 0 2
1 0
d)
Ö 4 39 108 11
Ö 4
Let X s , then
39
Now, let , therefore
sX s 39Y s X s U s
11 35 250
sX s 108Y s U s
s s
Let X s U s , then 11
Let X s , or 35
Let , then 250 . If , then:
4
39
108
11
35 36
250
10‐3


or
4 1 0 0 0 0 0
39 0 1 0 0 0 1
108 0 0 1 0 0 0
0 0 0 0 1 0 11
0 0 0 0 0 1 35
0 0 0 0 0 0 250
10-3) (a) Alternatively use equations (10-225), (10-232) and (10-233)
3
3 2
The state variables are defined as
x1 ( t ) = y ( t )
dy ( t )
x2 ( t ) =
dt
Then the state equations are represented by the vector‐matrix equation
dx ( t )
= Ax ( t ) + Bu ( t )
dt
where x(t) is the 2 × 1 state vector, u(t) the scalar input, and
⎡0 1⎤ ⎡ 3⎤
A=⎢ ⎥ B=⎢ ⎥
⎣ −2 −3⎦ ⎣1⎦ (Also see section 2‐3‐3 or 10‐6)
C = [1 0] D=0
G ( s ) = C ( sI − A ) B + D
−1
10‐4


MATLAB
>> clear all
>> syms s
>> A=[0,1;-2,-3]
A=
0 1
-2 -3
>> B=[0;1]
B=
>> C=[3,1]
C=
3 1
>> s*eye(2)-A
ans =
[ s, -1]
[ 2, s+3]
>> inv(ans)
ans =
[ (s+3)/(s^2+3*s+2), 1/(s^2+3*s+2)]
[ -2/(s^2+3*s+2), s/(s^2+3*s+2)]
>> C*ans*B
ans =
3/(s^2+3*s+2)+s/(s^2+3*s+2)
10‐5


Use ACSYS as demonstrate in section 10-19-2
1) Activate MATLAB
2) Go to the folder containing ACSYS
3) Type in Acsys
4) Click the “Transfer Function Symbolic” pushbutton
5) Enter the transfer function
6) Use the “State Space” option as shown below:
You get the next window. Enter the A,B,C, and D values.
10‐6


-------------------------------------------------------------
State Space Analysis
-------------------------------------------------------------
Inputs:
A=| 0 1| B=|0|
|-2 -3| |1|
C=|3 1| D=|0|
State Space Representation:
Dx = | 0 1|x + |0|u
|-2 -3| |1|
10‐7


y = |3 1|x + |0|u
Determinant of (s*I-A):
s +3s+2
Characteristic Equation of the Transfer Function:
s +3s+2
The eigen values of A and poles of the Transfer Function are:
-1
-2
Inverse of (s*I-A) is:
[ s+3 1 ]
[ ------------ ------------]
[ 2 2 ]
[s +3s+2 s + 3 s + 2]
[ ]
[ 2 s ]
[- ------------ ------------]
[ 2 2 ]
[ s +3s+2 s + 3 s + 2]
State transition matrix (phi) of A:
[ 2 exp(-t) - exp(-2 t) exp(-t) - exp(-2 t) ]
[ ]
[-2 exp(-t) + 2 exp(-2 t) -exp(-t) + 2 exp(-2 t)]
Transfer function between u(t)and y(t) is:

10‐8


s+3
------------
s +3s+2
No Initial Conditions Specified
States (X) in Laplace Domain:
[ 1 ]
[---------------]
[(s + 2) (s + 1)]
[ ]
[ s ]
[---------------]
[(s + 2) (s + 1)]
Inverse Laplace x(t):
[ exp(-t) - exp(-2 t) ]
[ ]
[-exp(-t) + 2 exp(-2 t)]
Output Y(s):
s+3
---------------
(s + 2) (s + 1)
Inverse Laplace y(t):
2 exp(-t) - exp(-2 t)
10‐9


Use the same procedure for parts b, c and d.
10-4) a) x Î 2 ,Î 2
Î Î
.
Î 0.5 Î 0.5
Î Î
As a result:
2 0 0 1 1
0.5 0 0 0 0
0 1 0 0 0
1 0 1 0 0
1 1 1 0
b) Î 2 Î 2
Î 4 3 Î 4 3 2 3
Î 6 Î
As a result:
2 0 0 1
2 3 0 1
1 1 6 0
1 0 1
c) Î 5 Î 5
Î 2 Î 2
Î 4 Î 4
10‐10


Î 2 Î 2
As a result:
5 0 0 0 1
1 2 1 0 1 u
0 1 4 0 0
0 0 2 0 0
0 1 0 1
10-5) We shall first show that
2
I A 1 A
Φ ( s ) = ( sI − A ) =
−1
+ 2
+ 2
+L
s s 2! s
We multiply both sides of the equation by ( sI − A ) , and we get I = I. Taking the inverse Laplace transform
on both sides of the equation gives the desired relationship for φ ( t ) .
10-6)
(a) USE MATLAB
Amat=[0 1;-2 -1]
[mA,nA]=size(Amat);
rankA=rank(Amat);
disp(' Characteristic Polynomial: ')
chareq=poly(Amat);
[mchareq,nchareq]=size(chareq);
syms s;
poly2sym(chareq,s)
[evecss,eigss]=eig(Amat);
disp(' Eigenvalues of A = Diagonal Canonical Form of A is:');
Abar=eigss,
disp('Eigen Vectors are ')
T=evecss
% state transition matrix
ilaplace(inv([s 0;0 s]-Amat))
Results in MATLAB COMMAND LINE

Amat =
0 1
10‐11


‐2 ‐1
Characteristic Polynomial:
ans =
s^2+s+2
Eigenvalues of A = Diagonal Canonical Form of A is:
Abar =
‐0.5000 + 1.3229i 0
0 ‐0.5000 ‐ 1.3229i
Eigen Vectors are
T =
‐0.2041 ‐ 0.5401i ‐0.2041 + 0.5401i
0.8165 0.8165
phi=ilaplace(inv([s 0;0 s]‐Amat))
phi =
[ 1/7*exp(‐1/2*t)*(7*cos(1/2*7^(1/2)*t)+7^(1/2)*sin(1/2*7^(1/2)*t)), 2/7*7^(1/2)*exp(‐
1/2*t)*sin(1/2*7^(1/2)*t)]
[‐4/7*7^(1/2)*exp(‐1/2*t)*sin(1/2*7^(1/2)*t), 1/7*exp(‐1/2*t)*(7*cos(1/2*7^(1/2)*t)‐
7^(1/2)*sin(1/2*7^(1/2)*t))]
% use vpa to convert to digital format. Use digit(#) to adjust level of precision if necessary.
vpa(phi)
ans =
[ .1428571*exp( .5000000*t)*(7.*cos(1.322876*t)+2.645751*sin(1.322876*t)),
.7559289*exp(‐.5000000*t)*sin(1.322876*t)]
[ ‐1.511858*exp(‐.5000000*t)*sin(1.322876*t),
10‐12


.1428571*exp(‐.5000000*t)*(7.*cos(1.322876*t)‐2.645751*sin(1.322876*t))]
ANALYTICAL SOLUTION:
2
Characteristic equation: Δ( s ) = sI − A = s + s + 2 = 0
Eigenvalues: s = −0.5 + j1323

. , − 0.5 − j1323
.
State transition matrix:
⎡ cos1.323t + 0.378 sin 1.323t 0.756 sin 1.323t ⎤ −0.5 t

φ (t ) = ⎢
−1.069 sin (1.323t − 69.3 ) ⎥⎦
e
−1.512 sin 1.323t
o
⎣
Alternatively
USE ACSYS as illustrated in section 10-19-1

1) Activate MATLAB
3) Type in Acsys
4) Click the “State Space” pushbutton
5) Enter the A,B,C, and D values. Note C must be entered here and must have the same number of
columns as A. We us [1,1] arbitrarily as it will not affect the eigenvalues.
6) Use the “Calculate/Display” menu and find the eigenvalues.
10‐13


From MATLAB Command Window:
The A matrix is:
Amat =
0 1
‐2 ‐1
10‐14


ans =
s^2+s+2
Abar =
‐0.5000 + 1.3229i 0
0 ‐0.5000 ‐ 1.3229i
Eigen Vectors are
T =
‐0.2041 ‐ 0.5401i ‐0.2041 + 0.5401i
0.8165 0.8165
THE REST ARE SAME AS PART A.
2
(b) Characteristic equation: Δ( s ) = sI − A = s + 5s + 4 = 0 Eigenvalues: s = −4, − 1
⎡ 1.333e − t − 0.333e −4 t −t
0.333e − 0.333e ⎤−4 t
φ (t ) = ⎢ −t −4 t −t ⎥ −4 t
⎣ −1.333e − 1.333e −0.333e + 1.333e ⎦
(c) Characteristic equation: Δ ( s ) = ( s + 3 ) = 0 Eigenvalues: s = −3, − 3

2
⎡ e −3t 0 ⎤
φ (t ) = ⎢ −3 t ⎥
⎣0 e ⎦
Eigenvalues: s = −3, 3
2
(d) Characteristic equation: Δ( s ) = s − 9 = 0
⎡ e3t 0 ⎤
φ (t ) = ⎢ −3 t ⎥
⎣0 e ⎦
10‐15


Eigenvalues: s = j 2, − j 2
2
(e) Characteristic equation: Δ( s ) = s + 4 = 0
⎡ cos 2t sin 2t ⎤
φ (t ) = ⎢ ⎥
⎣ − sin 2t cos 2t ⎦
Eigenvalues: s = −1, − 2, − 2
3 2
(f) Characteristic equation: Δ( s ) = s + 5 s + 8 s + 4 = 0
⎡e − t 0 0 ⎤
⎢ ⎥
φ (t ) = 0
−2 t −2 t

⎢ e te
⎥
⎢⎣ 0 0 e ⎥⎦
−2 t
(g) Characteristic equation: Δ( s ) = s + 15 s + 75 s + 125 = 0 Eigenvalues: s = −5, − 5, − 5

3 2
⎡ e −5 t te
−5 t
0 ⎤
⎢ ⎥
φ (t ) = 0
−5 t −5 t

⎢ e te
⎥
⎢⎣ 0 0 e ⎥⎦
−5 t
State transition equation: x(t ) = φ (t ) x(t ) + ∫ φ (t − τ )Br (τ ) dτ

t
φ (t ) for each part is given in Problem 5‐3.
0
10-7)In MATLAB USE ilaplace to find L−1 ⎡⎣( sI − A ) BR ( s ) ⎤⎦ see previous problem for codes.
−1
(a)
⎧ 1 ⎡ s + 1 1 ⎤ ⎡ 0 1 ⎤ ⎡1⎤ 1 ⎫
∫ φ (t − τ )Br (τ )dτ = L ⎡⎣( sI − A ) BR ( s ) ⎤⎦ = L ⎨
t −1
−1 −1
⎢ ⎥⎢ ⎥⎢ ⎥ ⎬
0
⎩ Δ ( s ) ⎣ −2 s ⎦ ⎣1 0 ⎦ ⎣1⎦ s ⎭
⎡ s+2 ⎤
⎢ s ( s2 + s + 2) ⎥
⎥=⎡
1 + 0.378 sin 1.323t − cos1.323t ⎤
=L ⎢
−1
t≥0
⎢ s−2 ⎥ ⎢⎣ −1 + 1.134 sin 1.323t + cos1.323t ⎥⎦
⎢ s s2 + s + 2 ⎥
⎣ ( )⎦
10‐16


(b)
⎧ 1 ⎡ s + 5 1 ⎤ ⎡1⎤ ⎡1⎤ 1 ⎫
t −1
−1 −1
⎢ ⎥⎢ ⎥⎢ ⎥ ⎬
0
⎩ Δ ( s ) ⎣ −4 s ⎦ ⎣1⎦ ⎣1⎦ s ⎭
⎡ s+6 ⎤ ⎡ 1.5 − 1.67 + 0.167 ⎤
⎢ s ( s + 1)( s + 2) ⎥ ⎢ ⎥ ⎡ 1.5 − 1.67e − t + 0.167e −4 t ⎤
=L ⎢
−1
⎥ = L−1 ⎢ s s + 1 s + 4 ⎥ = ⎢ ⎥ t≥0
⎢ s−4 ⎥ ⎢ −1 + 1.67 − 0.667 ⎥ ⎣ −1 + 1.67e − 0.667e − 4t ⎦
−t
⎢⎣ s ( s + 1)( s + 4) ⎥⎦ ⎢⎣ s s + 1 s + 4 ⎥⎦
(c)
⎧⎡ 1 0
⎤ ⎫
⎪⎢
−1 ⎪ s + 3
⎥ ⎡ 0 ⎤ 1 ⎪⎪
∫0 φ (t − τ )Br (τ )dτ = L ⎡⎣( sI − A ) BR( s) ⎤⎦ = L ⎨ ⎢
t −1
−1
⎥ ⎬
⎪⎢ 0 1 ⎥ ⎢⎣ 1 ⎥⎦ s ⎪
⎪⎩⎣⎢ s + 3 ⎦⎥ ⎭⎪
⎡ 0 ⎤
=L ⎢
−1 ⎥=⎡ 0 ⎤
t≥0
⎢
1 ⎢ −3 t ⎥
⎥ 0.333 (1 − e ) ⎦
⎢⎣ s ( s + 3 ) ⎥⎦ ⎣
(d)
⎧⎡ 1 0
⎤ ⎫
⎪⎢
−1 ⎪ s − 3
⎥ ⎡ 0 ⎤ 1 ⎪⎪
∫0 ⎡ ( ) ⎤
t −1
φ τ τ τ L L
−1
− = ⎣ − ⎦ = ⎨⎢ ⎥ ⎬
1 ⎥ ⎢⎣ 1 ⎥⎦ s ⎪
( t ) B r ( ) d s I A B R ( s )
⎪⎢ 0
⎪⎢
⎩⎣ s + 3 ⎦⎥ ⎭⎪
⎡ 0 ⎤
=L ⎢
−1 ⎥=⎡ 0 ⎤
t≥0
⎥ ⎣ 0.333 (1 − e −3t ) ⎥⎦
⎢
1
⎢
⎣⎢ s ( s + 3 ) ⎦⎥
(e)
⎧⎡ 1 2
⎤ ⎫
⎪⎢ 2
−1 ⎪ s + 4
⎥ ⎡ 0 ⎤ 1 ⎪⎪
∫0 ⎡ ( ) ⎤
t −1
φ τ τ τ
−1
− = L ⎣ − ⎦ = L ⎨⎢ ⎥ ⎬
s ⎥ ⎢⎣ 1 ⎥⎦ s ⎪
( t ) B r ( ) d s I A B R ( s )
⎪ ⎢ −2
⎪⎢ s 2 + 4
⎩⎣ s + 4 ⎦⎥
2
⎪⎭

⎡ 2 ⎤
⎢ s ⎥ ⎡ 2 ⎤
=L ⎢ ⎥=⎢
−1
t ≥0
⎥
⎢ 1 ⎥ ⎣ 0.5 sin 2t ⎦
⎢⎣ ( s 2 + 4 ) ⎥⎦
(f)
10‐17


⎧⎡ 1 ⎤ ⎫
⎪⎢ 0 0 ⎥ ⎪
⎪⎢ s + 1 ⎥ ⎡0 ⎤ ⎪
−1 ⎪ ⎢ 1 1 ⎥ ⎢1 ⎥ 1 ⎪
∫0 φ (t − τ )Br (τ )dτ = L ⎡⎣( sI − A ) BR( s ) ⎤⎦ = L ⎨ ⎢ 0
t −1
−1
2 ⎥ ⎬
⎪⎢ s+2 ( s + 2) ⎥ ⎢ ⎥ s ⎪
⎪⎢ ⎣⎢ 0 ⎦⎥ ⎪
1 ⎥
⎪ 0 ⎪
⎩⎣⎢ ⎥
0
s+2 ⎦ ⎭
⎡ 0 ⎤
⎢ ⎥ ⎡ 0 ⎤
=L ⎢
−1 1 ⎢ −2 t ⎥
⎥ = 0.5 (1 − e ) t ≥ 0
⎢ s ( s + 2) ⎥ ⎢ ⎥
⎢ ⎥ ⎣⎢ 0 ⎦⎥
⎣ 0 ⎦
(g)
⎧⎡ 1 1
0
⎤ ⎫
⎪⎢ s + 5 ( s + 5) 2 ⎥ ⎪
⎪⎢ ⎥ ⎡0⎤ ⎪
−1 ⎪⎢ 1 1 ⎥ ⎢ ⎥ 1⎪
∫0 φ (t − τ )Br (τ ) dτ = L ⎡⎣( sI − A ) BR ( s ) ⎤⎦ = L ⎨ ⎢ 0
t −1
−1
2 ⎥
0 ⎬
⎪⎢ s+5 ( s + 5) ⎥ ⎢ ⎥ s ⎪
⎢⎣1 ⎥⎦
⎪⎢ 1 ⎥ ⎪
⎪⎢ 0 0 ⎥ ⎪
⎩⎣ s+5 ⎦ ⎭

⎡ ⎤ ⎡ ⎤
⎢ ⎥ ⎢ ⎥
⎢ 0
⎥ ⎢
0
⎥ ⎡ 0 ⎤
−1
⎢ 1 ⎥ −1 ⎢ 0.04 0.04 0.2 ⎥ ⎢ −5 t −5 t ⎥
=L ⎢ =L
2 ⎥ ⎢ s − s + 5 − ( s + 5 )2 ⎥ = ⎢ 0.04 − 0.04e − 0.2te ⎥ u s (t )
⎢ s ( s + 5) ⎥ ⎢ ⎥ ⎢⎣ 0.2 − 0.2e
−5 t
⎥⎦
⎢ 1 ⎥ ⎢ 0.2 0.2 ⎥
⎢ ⎥ ⎢⎣ − ⎥⎦
⎢⎣ s ( s + 5 ) ⎥⎦ s s+5
10-8) State transition equation: x(t ) = φ (t ) x(t ) + ∫0 φ (t − τ )Br (τ ) dτ φ (t ) for each part is given in Problem 5‐3.

t
(a)
10‐18


⎧ 1 ⎡ s + 1 1 ⎤ ⎡ 0 1 ⎤ ⎡1⎤ 1 ⎫
t −1
−1 −1
⎢ ⎥⎢ ⎥⎢ ⎥ ⎬
0
⎩ Δ ( s ) ⎣ −2 s ⎦ ⎣1 0 ⎦ ⎣1⎦ s ⎭
⎡ s+2 ⎤
⎢ s ( s + s + 2) ⎥
2
⎥=⎡
1 + 0.378 sin 1.323t − cos1.323t ⎤
=L ⎢
−1
t≥0
⎢ s−2 ⎥ ⎣ −1 + 1.134 sin1.323t + cos1.323t ⎥⎦
⎢
⎢ s s2 + s + 2 ⎥
⎣ ( )⎦
(b)
⎧ 1 ⎡ s + 5 1 ⎤ ⎡1⎤ ⎡1⎤ 1 ⎫
∫ φ (t − τ )Br (τ ) dτ = L ⎡⎣( sI − A ) BR ( s ) ⎤⎦ = L ⎨
t −1
−1 −1
⎢ ⎥⎢ ⎥⎢ ⎥ ⎬
0
⎩ Δ ( s ) ⎣ −4 s ⎦ ⎣1⎦ ⎣1⎦ s ⎭
⎡ s+6 ⎤ ⎡ 1.5 − 1.67 + 0.167 ⎤
⎢ s ( s + 1)( s + 2) ⎥ ⎢ s s + 1 s + 4 ⎥ ⎡ 1.5 − 1.67e − t + 0.167e −4 t ⎤
=L ⎢ ⎥=L ⎢
−1 −1
⎥=⎢ ⎥ t≥0
⎢ s−4 ⎥ ⎢ −1 + 1.67 − 0.667 ⎥ ⎣ −1 + 1.67e − 0.667e − 4t ⎦
−t
⎢⎣ s ( s + 1)( s + 4) ⎥⎦ ⎣⎢ s s + 1 s + 4 ⎦⎥
(c)
⎧⎡ 1 0
⎤ ⎫
⎪⎢
−1 ⎪ s + 3
⎥ ⎡ 0 ⎤ 1 ⎪⎪
∫0 ⎡ ( ) ⎤
t −1
φ τ τ τ L L
−1
− = ⎣ − ⎦ = ⎨⎢ ⎥ ⎬
1 ⎥ ⎢⎣ 1 ⎥⎦ s ⎪
( t ) B r ( ) d s I A B R ( s )
⎪⎢ 0
⎪⎢
⎩⎣ s + 3 ⎦⎥ ⎭⎪
⎡ 0 ⎤
=L ⎢
−1 ⎥=⎡ 0 ⎤
t≥0
⎢
1 ⎢ −3 t ⎥
⎥ ⎣ 0.333 (1 − e ) ⎦
⎣⎢ s ( s + 3 ) ⎦⎥
(d)
⎧⎡ 1 0
⎤ ⎫
⎪⎢
−1 ⎪ s − 3
⎥ ⎡ 0 ⎤ 1 ⎪⎪
∫0 φ (t − τ )Br (τ )dτ = L ⎣⎡( sI − A ) BR( s ) ⎦⎤ = L ⎨ ⎢
t −1
−1
⎥ ⎬
⎪⎢ 0 1 ⎥ ⎢⎣ 1 ⎥⎦ s ⎪
⎪⎢
⎩⎣ s + 3 ⎦⎥ ⎭⎪
⎡ 0 ⎤
=L ⎢
−1 ⎥=⎡ 0 ⎤
t≥0
⎢
1 ⎢ −3 t ⎥
⎥ 0.333 (1 − e ) ⎦
⎢⎣ s ( s + 3 ) ⎥⎦ ⎣
(e)
10‐19


⎧⎡ 1 2
⎤ ⎫
⎪⎢ 2
−1 ⎪ s + 4
⎥ ⎡ 0 ⎤ 1 ⎪⎪
∫0 ⎡ ( ) ⎤
t −1
φ τ τ τ
−1
− = L ⎣ − ⎦ = L ⎨⎢ ⎥ ⎬
s ⎥ ⎢⎣ 1 ⎥⎦ s ⎪
( t ) B r ( ) d s I A B R ( s )
⎪ ⎢ −2
⎪⎢ s 2 + 4
⎩⎣ s + 4 ⎦⎥
2
⎪⎭

⎡ 2 ⎤
⎢ s ⎥ ⎡ 2 ⎤
=L ⎢ ⎥=⎢
−1
t ≥0
⎥
⎢ 1 ⎥ ⎣ 0.5 sin 2t ⎦
⎢⎣ ( s + 4 ) ⎥⎦
2
(f)
⎧⎡ 1 ⎤ ⎫
⎪⎢ 0 0 ⎥ ⎪
⎪⎢ s + 1 ⎥ ⎡0 ⎤ ⎪
−1 ⎪ ⎢ 1 1 ⎥ ⎢1 ⎥ 1 ⎪
∫0 φ (t − τ )Br (τ )dτ = L ⎡⎣( sI − A ) BR( s ) ⎤⎦ = L ⎨ ⎢ 0
t −1
−1
2 ⎥ ⎬
⎪⎢ s+2 ( s + 2) ⎥ ⎢ ⎥ s ⎪
⎢⎣ 0 ⎥⎦
⎪⎢ 1 ⎥ ⎪
⎪ 0 ⎪
⎩⎣⎢ ⎥
0
s+2 ⎦ ⎭
⎡ 0 ⎤
⎢ ⎥ ⎡ 0 ⎤
=L
−1
⎢ 1
⎥ ⎢ −2 t ⎥
= 0.5 (1 − e ) t ≥ 0
⎢ s ( s + 2) ⎥ ⎢ ⎥
⎢ ⎥ ⎣⎢ 0 ⎥⎦
⎣ 0 ⎦
(g)
⎧⎡ 1 1
0
⎤ ⎫
⎪⎢ s + 5 ( s + 5) 2 ⎥ ⎪
⎪⎢ ⎥ ⎡0⎤ ⎪
−1 ⎪⎢ 1 1 ⎥ ⎢ ⎥ 1⎪
∫0 φ (t − τ )Br (τ ) dτ = L ⎡⎣( sI − A ) BR ( s ) ⎤⎦ = L ⎨ ⎢ 0
t −1
−1
2 ⎥
0 ⎬
⎪⎢ s+5 ( s + 5) ⎥ ⎢ ⎥ s ⎪
⎪⎢ ⎣⎢1 ⎦⎥ ⎪
1 ⎥
⎪⎢ 0 0 ⎥ ⎪
⎩⎣ s+5 ⎦ ⎭

⎡ ⎤ ⎡ ⎤
⎢ ⎥ ⎢ ⎥
⎢ 0 ⎥ ⎢
0
⎥ ⎡ 0 ⎤
−1
⎢ 1 ⎥ −1 ⎢ 0.04 0.04 0.2 ⎥ ⎢ −5 t −5 t ⎥
=L ⎢ =L

( + ) 2 ⎥ ⎢ s − s + 5 − ( s + 5 )2 ⎥ = ⎢ 0.04 − 0.04e − 0.2te ⎥ u s (t )
⎢ s s 5 ⎥ ⎢ ⎥ ⎣⎢ ⎦⎥
−5 t
0.2 − 0.2e
⎢ 1 ⎥ ⎢ 0.2 0.2 ⎥
⎢ ⎥ ⎢⎣ − ⎥⎦
⎣⎢ s ( s + 5 ) ⎦⎥ s s+5
10‐20


10-9) (a) Not a state transition matrix, since φ ( 0) ≠ I (identity matrix).
(b) Not a state transition matrix, since φ ( 0) ≠ I (identity matrix).
(c) φ ( t ) is a state transition matrix, since φ ( 0 ) = I and
−1
⎡ 1 ⎤
0 ⎡ 1 0⎤
[φ (t ) ] = ⎢ − t
−1
−t ⎥
=⎢ ⎥ = φ ( −t )
⎣1 − e e ⎦ ⎣1 − e ⎦
t t
e
(d) φ ( t ) is a state transition matrix, since φ ( 0) = I , and
⎡e2t −te
2t
t e / 2⎤
2 2t
[φ (t ) ]−1 = ⎢⎢ 0 e
2t
−te
2t ⎥
⎥ = φ ( −t )
⎢⎣ 0 0 e
2t
⎥⎦
1 3 1
10-10) a) Î and
2 3 2
Therefore:
2
2 2 2
If 0 0, then 0.5 0.5
b)
0.5
. 1 1
. .
0.5 0.5 0.5
. .
2 0.5 5 0.5
If x 0 0, then
.
0 1 0.5 0.5 – 0.5 0.5
2 2 .
0.5
.
5
.
0.5 0.5 1
and
x .
y t 1 0 x x e sin 0.5t
10-11) (a) (1) Eigenvalues of A: 2.325, − 0.3376 + j 0.5623, − 0.3376 − j 0.5623

10‐21


(2) Transfer function relation:
⎡ s 2 + 3s + 2 1 ⎤ ⎡0⎤
−1
⎡ s −1 0 ⎤ ⎡0 ⎤ s+3 ⎡1⎤
X( s ) = ( sI − A ) BU ( s ) =
−1 1 ⎢
0 s −1 ⎥ ⎢ 0 ⎥ U ( s ) = 1 ⎢ −1 ⎥ ⎢ ⎥
s ( s + 3) s 0 U (s) =
1 ⎢ ⎥
s U (s)
Δ( s) ⎢ ⎥ ⎢ ⎥ Δ(s) ⎢ ⎥⎢ ⎥ Δ(s) ⎢ 2 ⎥
⎣⎢ 1 2 s + 3⎦⎥ ⎣⎢ 1 ⎦⎥ ⎢⎣ − s −2 s − 1 s ⎥⎦ ⎣⎢ 1 ⎦⎥
2
⎣⎢ s ⎦⎥
3 2
Δ( s ) = s + 3 s + 2 s + 1
(3) Output transfer function:
⎡1⎤
Y ( s) 1 ⎢ ⎥ 1
= C( s ) ( sI − A ) B = [1 0]
−1
0 s = 3
Δ ( s ) ⎢ 2 ⎥ s + 3s + 2 s + 1
2
U (s)
⎣⎢ s ⎦⎥

7) Activate MATLAB
9) Type in Acsys
12) Use the “Calculate/Display” menu and find the eigenvalues and other State space calculations.
10‐22


The A matrix is:
Amat =
0 1 0
0 0 1
‐1 ‐2 ‐3
10‐23


ans =
s^3+3*s^2+2*s+1
Abar =
‐2.3247 0 0
0 ‐0.3376 + 0.5623i 0
0 0 ‐0.3376 ‐ 0.5623i
Eigen Vectors are
T =
0.1676 0.7868 0.7868
‐0.3896 ‐0.2657 + 0.4424i ‐0.2657 ‐ 0.4424i
0.9056 ‐0.1591 ‐ 0.2988i ‐0.1591 + 0.2988i
State‐Space Model is:

a =
x1 x2 x3
x1 0 1 0
x2 0 0 1
x3 ‐1 ‐2 ‐3

b =
u1
x1 0
x2 0
x3 1

10‐24


c =
x1 x2 x3
y1 1 0 0

d =
u1
y1 0

Continuous‐time model.
ans =
s^3+3*s^2+2*s+1

Equivalent Transfer Function Model is:
Transfer function:
1.776e‐015 s^2 + 6.661e‐016 s + 1
‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐
s^3 + 3 s^2 + 2 s + 1

Pole, Zero Form:

Zero/pole/gain:
1.7764e‐015 (s^2 + 0.375s + 5.629e014)
‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐
(s+2.325) (s^2 + 0.6753s + 0.4302)
The numerator is basically equal to 1
10‐25


Use the same procedure for other parts.
(b) (1) Eigenvalues of A: −1, − 1.
⎡ 1 ⎤
1 ⎡s + 1 1 ⎤ ⎡0 ⎤ ⎢ ( s + 1)2 ⎥
X ( s ) = ( sI − A ) BU ( s ) =
−1
= ⎢ ⎥ U (s) Δ( s) = s + 2s + 1
2
Δ ( s ) ⎢⎣ 0 s + 1⎥⎦ ⎢⎣1 ⎥⎦
U ( s )
⎢ 1 ⎥
⎢ ( s + 1) ⎥
⎣ ⎦
⎡ 1 ⎤
Y ( s) ⎢ ( s + 1)2 ⎥ 1 1 s+2
= C( s ) ( sI − A ) B = [1 1] ⎢ ⎥=
−1
+ =
⎢ 1 ⎥ ( s + 1) s + 1 ( s + 1)
2 2
U (s)
⎢⎣ s + 1 ⎥⎦
(c) (1) Eigenvalues of A: 0, − 1, − 1.
⎡ s 2 + 2s = 1 s+2 1 ⎤ ⎡0 ⎤ ⎡1⎤
⎢ ⎥⎢ ⎥ 1 ⎢ ⎥
( )
1
X ( s ) = ( s I − A ) BU ( s ) = s ( s + 2) ) s 0 U ( s ) =
−1
Δ(s ) = s s + 2s + 1
2
0 s U (s)
Δ( s) ⎢ ⎥⎢ ⎥ Δ( s) ⎢ 2 ⎥
⎢⎣ 0 −s s ⎥⎦ ⎣⎢1 ⎦⎥
2
⎣⎢ s ⎦⎥
⎡1⎤
s +1
= C( s ) ( sI − A ) B = [1 1 0 ] ⎢ s ⎥ =
Y (s) −1 1
=
U (s) ⎢ 2 ⎥ s ( s + 1) s ( s + 1)
2
⎣⎢ s ⎦⎥

2
dy dx1 dx 2 d y dx 2 dx3
10-12) We write = + = x 2 + x3 = + = − x1 − 2 x 2 − 2 x3 + u
dt dt dt dt dt dt
10‐26


⎡ dx1 ⎤
⎢ dt ⎥
⎢ ⎥ ⎡ 0 1 0 ⎤ ⎡ x1 ⎤ ⎡ 0 ⎤
d x ⎢ dy ⎥ ⎢
= = 0 1 1 ⎥ ⎢ x2 ⎥ + ⎢ 0 ⎥ u (1)
dt ⎢ dt ⎥ ⎢ ⎥⎢ ⎥ ⎢ ⎥
⎢ 2 ⎥ ⎢⎣ −1 −2 −2 ⎥⎦ ⎢⎣ x3 ⎥⎦ ⎢⎣1 ⎥⎦
⎢d y ⎥
⎣⎢ dt 2 ⎦⎥
⎡ x1 ⎤ ⎡1 0 0 ⎤ ⎡ 1 0 0⎤
x = ⎢ y ⎥ = ⎢ 1 1 0 ⎥ x x = ⎢ −1 1 0 ⎥ x (2)
⎢ ⎥ ⎢ ⎥ ⎢ ⎥
⎣⎢ y& ⎦⎥ ⎣⎢ 0 1 1 ⎥⎦ ⎣⎢ 1 −1 1 ⎥⎦
Substitute Eq. (2) into Eq. (1), we have
⎡ −1 1 0 ⎤ ⎡0 ⎤
= A 1 x + B 1u = 0 0 1 x = ⎢ 0 ⎥ u
⎢ ⎥
dx

dt ⎢ ⎥ ⎢ ⎥
⎣⎢ −1 0 −2 ⎦⎥ ⎢⎣1 ⎦⎥
10-13) For MATLAB Codes see 10‐15
(a)
s −2 0
3 2 3 2
sI − A = −1 s − 2 0 = s − 3 s + 2 = s + a2 s + a1 s + a0 a0 = 2, a1 = 0, a2 = −3
1 0 s −1
⎡ a1 a2 1⎤ ⎡ 0 −3 1 ⎤ ⎡0 2 4 ⎤
M = ⎢ a2 1 0 = ⎢ −3 1 0 ⎥
⎥ S = ⎡⎣ B AB A B ⎤⎦ = ⎢ 1 2 6 ⎥
2

⎢ ⎥ ⎢ ⎥ ⎢ ⎥
⎣⎢ 1 0 0 ⎦⎥ ⎣⎢ 1 0 0 ⎥⎦ ⎣⎢1 1 −1⎦⎥
⎡ −2 2 0 ⎤
P = SM = ⎢ 0 −1 1 ⎥
⎢ ⎥
⎣⎢ −4 −2 1 ⎥⎦

10‐27


(b)
s −2 0
3 2 3 2
sI − A = −1 s − 1 0 = s − 3s + 2 = s + a2 s + a1 s + a0 a0 = 2, a1 = 0, a2 = −3
1 −1 s −1
⎡ a1 a2 1⎤ ⎡ 0 −3 1 ⎤ ⎡1 2 6 ⎤
M = ⎢ a2 1 0 = ⎢ −3 1 0 ⎥
⎥ S = ⎡⎣ B AB A B ⎤⎦ = ⎢1 3 8 ⎥
2
⎢ ⎥ ⎢ ⎥ ⎢ ⎥
⎣⎢ 1 0 0 ⎦⎥ ⎣⎢ 1 0 0 ⎥⎦ ⎣⎢ 0 0 1 ⎥⎦
⎡ 0 −1 1 ⎤
P = SM = ⎢ −1 0 1 ⎥
⎢ ⎥
⎣⎢ 1 0 0 ⎥⎦
(c)
s+2 −1 0
3 2 3 2
sI − A = 0 s+2 0 = s + 7 s + 16 s + 12 = s + a2 s + a1 s + a0 a0 = 12, a1 = 16, a2 = 7
1 2 s+3
⎡ a1 a2 1⎤ ⎡16 7 1 ⎤ ⎡1 −1 0 ⎤
M = ⎢ a2 0 = ⎢ 7 1 0⎥
⎥ S = ⎡⎣ B AB A B ⎤⎦ = ⎢1 −2 4 ⎥
2
1
⎢ ⎥ ⎢ ⎥ ⎢ ⎥
⎢⎣ 1 0 0 ⎥⎦ ⎢⎣ 1 0 0 ⎥⎦ ⎢⎣1 −6 23⎥⎦
⎡ 9 6 1⎤
P = SM = ⎢ 6 5 1⎥
⎢ ⎥
⎢⎣ −3 1 1⎥⎦
(d)
s +1 −1 0
3 2 3 2
sI − A = 0 s =1 −1 = s + 3 s + 3s + 1 = s + a2 s + a1 s + a0 a0 = 1, a1 = 3, a2 = 3
0 0 s +1
10‐28


⎡ a1 a2 1⎤ ⎡3 3 1 ⎤ ⎡ 0 1 −1⎤
M = ⎢ a2 1 0 ⎥ = ⎢3 1 0 ⎥ S = ⎡⎣ B AB A B ⎤⎦ = ⎢ 1 0 −1⎥
2
⎢ ⎥ ⎢ ⎥ ⎢ ⎥
⎣⎢ 1 0 0 ⎦⎥ ⎣⎢1 0 0 ⎥⎦ ⎣⎢1 −1 1 ⎥⎦
⎡2 1 0⎤
P = SM = ⎢ 2 3 1 ⎥
⎢ ⎥
⎣⎢ 1 2 1 ⎥⎦
(e)
s −1 −1
2 2
sI − A = = s + 2 s − 1 = s + a1 s + a0 a0 = −1, a1 = 2
2 s+3
⎡ a1 1⎤ ⎡2 1⎤ ⎡0 1⎤
M=⎢ =⎢ S = [B AB ] = ⎢
⎣1 0 ⎦ ⎣ 1 0 ⎥⎦
⎥
⎣1 −3⎥⎦
⎡1 0⎤
P = SM = ⎢
⎣ −1 1 ⎥⎦
10-14) For MATLAB codes see 10‐15
(a) From Problem 10‐13(a),
⎡ 0 −3 1 ⎤
M = ⎢ −3 1 0 ⎥
⎢ ⎥
⎢⎣ 1 0 0 ⎥⎦
Then,
⎡ C ⎤ ⎡ 1 0 1⎤ ⎡ 0.5 1 3 ⎤
V = ⎢ CA ⎥ = ⎢ −1 2 1⎥ Q = ( MV ) = ⎢ 0.5 1.5 4 ⎥
−1

⎢ 2⎥ ⎢ ⎥ ⎢ ⎥
⎢⎣CA ⎥⎦ ⎢⎣ 1 2 1⎥⎦ ⎢⎣ −0.5 −1 −2 ⎥⎦
(b) From Problem 10‐13(b),
⎡16 7 1 ⎤
M = ⎢ 7 1 0⎥
⎢ ⎥
⎢⎣ 1 0 0 ⎥⎦
10‐29


⎡ C ⎤ ⎡ 1 0 1⎤ ⎡ 0.2308 0.3077 1.0769 ⎤
V = ⎢ CA ⎥ = ⎢ −1 3 1⎥ Q = ( MV ) = ⎢ 0.1538 1.3846 ⎥
−1
0.5385
⎢ 2⎥ ⎢ ⎥ ⎢ ⎥
⎣⎢CA ⎦⎥ ⎣⎢ 2 5 1⎥⎦ ⎣⎢ −0.2308 −0.3077 −0.0769 ⎦⎥
(c) From Problem 10‐13(c),
⎡ C ⎤ ⎡ 1 0 0⎤
V = ⎢ CA ⎥ = ⎢ −2 1 0 ⎥
⎢ 2⎥ ⎢ ⎥
⎣⎢CA ⎦⎥ ⎣⎢ 4 −4 0 ⎥⎦
Since V is singular, the OCF transformation cannot be conducted.
(d) From Problem 10‐13(d),
⎡3 3 1 ⎤
M = ⎢3 1 0 ⎥
⎢ ⎥
⎣⎢1 0 0 ⎥⎦
Then,

⎡ C ⎤ ⎡1 0 1⎤ ⎡ −1 1 0 ⎤
V = ⎢ CA ⎥ = ⎢ −1 1 −1⎥ Q = ( MV ) = ⎢ 0 1 −2 ⎥
−1

⎢ 2⎥ ⎢ ⎥ ⎢ ⎥
⎣⎢CA ⎦⎥ ⎣⎢ 1 −2 2 ⎥⎦ ⎣⎢ 1 −1 1 ⎥⎦
(e) From Problem 10‐13(e),
⎡2 1⎤ ⎡ C ⎤ ⎡1 0⎤ ⎡0 1⎤
Q = ( MV ) = ⎢
−1
M=⎢ ⎥ Then, V = ⎢ 2⎥
=⎢
⎣1 0⎦ ⎣CA ⎦ ⎣1 1 ⎥⎦ ⎣1 −3⎥⎦
10-15) (a) Eigenvalues of A: 1, 2.7321, −0.7321
⎡ 0 0.5591 0.8255 ⎤
T = [ p1 p2 p 3 ] = ⎢ 0 0.7637 −0.3022 ⎥
⎢ ⎥
⎣⎢1 −0.3228 0.4766 ⎦⎥
10‐30


where p1 , p 2 , and p 3 are the eigenvectors.
(b) Eigenvalues of A: 1, 2.7321, −0.7321
⎡ 0 0.5861 0.7546 ⎤
T = [ p1 p2 p 3 ] = ⎢ 0 0.8007 −0.2762 ⎥
⎢ ⎥
⎣⎢1 0.1239 0.5952 ⎦⎥
where p1 , p 2 , and p 3 are the eigenvalues.
(c) Eigenvalues of A: ‐3, ‐2, ‐2. A nonsingular DF transformation matrix T cannot be found.
(d) Eigenvalues of A: −1, −1, −1
The matrix A is already in Jordan canonical form. Thus, the DF transformation matrix T is the identity
matrix I.
(e) Eigenvalues of A: 0.4142, −2.4142
⎡ 0.8629 −0.2811⎤
T = [p 2 p2 ] = ⎢
⎣ −0.5054 0.9597 ⎥⎦

1) Activate MATLAB
3) Type in Acsys
7) Next use the “Calculate/Display” menu and conduct State space calculations.
8) Next use the “Calculate/Display” menu and conduct Controlability calculations.
NOTE: the above order of calculations MUST BE followed in the order stated, otherwise you will get an
error.
10‐31


SOLVE PART (a)
The A matrix is:
10‐32


Amat =
0 2 0
1 2 0
-1 0 1
ans =
s^3-3*s^2+2
Abar =
1.0000 0 0
0 2.7321 0
0 0 -0.7321
10‐33


Eigen Vectors are
T=
0 0.5591 0.8255
0 0.7637 -0.3022
1.0000 -0.3228 0.4766
State-Space Model is:
a=
x1 x2 x3
x1 0 2 0
x2 1 2 0
x3 -1 0 1
b=
u1
x1 0
x2 1
x3 1
c=
x1 x2 x3
y1 0 1 1
10‐34


d=
u1
y1 0
Continuous-time model.
ans =
s^3-3*s^2+2
Transfer function:
2 s^2 - 3 s - 4
------------------------------
s^3 - 3 s^2 + 8.882e-016 s + 2
Pole, Zero Form:
Zero/pole/gain:
2 (s-2.351) (s+0.8508)
--------------------------
10‐35


(s-2.732) (s-1) (s+0.7321)
The Controllability Matrix [B AB A^2B ...] is =
Smat =
0 2 4
1 2 6
1 1 -1
The system is therefore Controllable, rank of S Matrix is =
rankS =
Mmat =
0 -3 1
-3 1 0
1 0 0
The Controllability Canonical Form (CCF) Transformation matrix is:
10‐36


Ptran =
-2 2 0
0 -1 1
-4 -2 1
The transformed matrices using CCF are:
Abar =
0 1.0000 0.0000
0 -0.0000 1.0000
-2.0000 0.0000 3.0000
Bbar =
Cbar =
-4 -3 2
Dbar =
0
10‐37


10-16) a)
Consider:
—2 2
Therefore:
0 1 0 0
0 0 1 0 0
0
0 0 0 1
1
0 0 2 0
1 0 1 0
U U
As , therefore sY s Y s
U U
Let X s Y s . If y x , then sY s sX s X , or x x . As a
result:
Now consider X , and , then
Therefore:
0 1 0 0 0
2 0 1 0 1 u
0 0 0 1 0
0 0 0 0 1
b)
Consider:
2
10‐38


4 4
Therefore:
0 1 0
4 4 1
1 2
As , therefore 4 2 . As a result:
4 2
21
4 4
10-17) For MATLAB codes see 10‐15
(a)
⎡1 −2 ⎤
S = [B AB ] = ⎢ S is singular.
⎣0 0 ⎥⎦
(b)
⎡ 1 −1 1 ⎤
S = ⎡⎣ B AB A B ⎤⎦ = ⎢ 2 −2 2 ⎥ S is singular.
2
⎢ ⎥
⎢⎣ 3 −3 3 ⎥⎦
(c)
⎡ 2 2 + 2 2⎤
S = [B AB ] = ⎢ ⎥ S is singular.
⎢⎣ 2 2+ 2 ⎥⎦
(d)
⎡ 1 −2 4 ⎤
S = ⎡⎣ B AB A B ⎤⎦ = ⎢ 0 0 0⎥ S is singular.
2
⎢ ⎥
⎣⎢1 −4 14 ⎦⎥
10-18) a, d and e are controllable
b, c, and f are not controllable
10‐39


9) Activate MATLAB
11) Type in Acsys
16) Next use the “Calculate/Display” menu and conduct Controlability calculations.
NOTE: the above order of calculations MUST BE followed in the order stated, otherwise you will get an
error.
10‐40


For part b, the system is not Controllable because [B AB] is singular (rank is less than 2):
The A matrix is:
Amat =
-1 0
0 -2
ans =
s^2+3*s+2
Abar =
-2 0
0 -1
Eigen Vectors are
T=
0 1
1 0
ans =
s^2+3*s+2
Transfer function:
-----
s+1
Pole, Zero Form:

10‐41


Zero/pole/gain:
-----
(s+1)
Smat =
2 -2
0 0 ÍRank is 1, and this is a singular matrix
The system is therefore Not Controllable, rank of S Matrix is =
rankS =
Mmat =
3 1
1 0
Ptran =
4 2
0 0
10-19) a, d, and e are observable
b, c, and f are not observable
Using ACSYS (also see the previous problem for more details):
10‐42


For part b, the system is not observable. Note: you must choose a B matrix arbitrarily.
The A matrix is:
Amat =
-1 0
0 -2
ans =
s^2+3*s+2
10‐43


Abar =
-2 0
0 -1
Eigen Vectors are
T=
0 1
1 0
ans =
s^2+3*s+2
Transfer function:
Pole, Zero Form:
Zero/pole/gain:
The Observability Matrix (transpose:[C CA CA^2 ...]) is =
Vmat =
0 1
0 -2
The System is therefore Not Observable, rank of V Matrix is =
rankV =
1
10‐44


Mmat =
3 1
1 0
10-20) (a) Rewrite the differential equations as:
d θm B d θm K b dθ m
2 2
K Ki dia Ra Ka Ks
2
=− 2
− θm + ia =− − ia + (θ r
−θm )
dt J dt J J dt La dt La La
dθ m
State variables: x1 = θ m , x 2 = , x3 = ia
dt
⎡ dx1 ⎤ ⎡ ⎤
⎡ ⎤
⎢ dt ⎥ ⎢ 0 0 ⎥ ⎢ 0 ⎥
1
⎢ ⎥ ⎢ ⎥ ⎡ x1 ⎤
⎢ −K Ki ⎥ ⎢ ⎥ ⎢ ⎥
⎢ 2 ⎥ =
dx B
− x2 + ⎢ 0 ⎥ θ r y= 1 0 0 x = x1
⎢ dt ⎥ ⎢ J J ⎥⎢ ⎥ ⎢
⎥ ⎢x ⎥ K K ⎥
J
⎢ dx ⎥ ⎢ K K Ra ⎥ ⎣ 3 ⎦ ⎢ a s ⎥
⎢ 3 ⎥ ⎢− a s
Kb
− − ⎣⎢ La ⎦⎥
⎣⎢ dt ⎦⎥ ⎢⎣ La La La ⎥⎦
(b) Forward‐path transfer function:
−1
⎡ ⎤
⎢s ⎡ ⎤
−1 0 ⎥ ⎢0 ⎥
⎢ ⎥
Θm (s) ⎢K B Ki ⎥ ⎢ ⎥ Ki K a
G ( s ) = = [1 0 0 ] s+ − ⎢ 0 ⎥=
⎢J J ⎥
E (s)
⎢
J
⎥ ⎢ K ⎥ Δ o (s)
⎢0 ⎢ a⎥
s+ a⎥
Kb R
⎢⎣ La La ⎥⎦ ⎣⎢ La ⎦⎥
Δ o ( s ) = JLa s + ( BLa + Ra J ) s + ( KLa + K i K b + Ra B ) s + KRa = 0

3 2

Closed‐loop transfer function:

10‐45


−1
⎡ ⎤
⎢ s ⎡ ⎤
−1 0 ⎥ ⎢ 0 ⎥
⎢ ⎥
Θm (s) ⎢ K B Ki ⎥ ⎢ ⎥ K sG(s)
M (s) = = [1 0 0 ] s+ − ⎢ 0 ⎥=
Θr (s) ⎢ J J ⎥
⎢
J
⎥ ⎢ K K ⎥ 1 + K s (s)
⎢ Ka Ks ⎢ a s⎥
s+ a⎥
Kb R
⎢⎣ La ⎥⎦
⎣⎢ La La La ⎦⎥
Ki K a K s
=
JLa s + ( BLa + Ra J ) s + ( KLa + K i K b + Ra B ) s + K i K a K s + KRa
3 2
10-21) (a)
⎡0 1⎤ ⎡ −1 0⎤ ⎡0 −1⎤ ⎡1 0⎤
A=⎢ A =⎢ A =⎢ A =⎢
2 3 4
⎣ −1 0 ⎥⎦ ⎣0 −1⎥⎦ ⎣1 0 ⎥⎦ ⎣0 1 ⎥⎦
(1) Infinite series expansion:
⎡ t2 t4 t
3
t
5
⎤
⎢ 1 − + − L t − + − L⎥
1 2 2 2! 4! 3! 5! ⎡ cos t sin t ⎤
φ (t ) = I + At + A t + L = ⎢ ⎥=⎢ ⎥
⎢ ⎥ ⎣ − sin t cos t ⎦
3 5 2 4
2! t t t t
⎢⎣ −t + 3! − 5! + L 1 − 2! + 4! L ⎥⎦
(2) Inverse Laplace transform:
−1
⎡ s −1⎤ 1 ⎡ s 1⎤ ⎡ cos t sin t ⎤
Φ ( s ) = ( sI − A ) = ⎢
−1
⎥ = 2 φ (t ) = ⎢
⎣1 s ⎦ s + 1 ⎢⎣ −1 s ⎥⎦ ⎣ − sin t cos t ⎥⎦
(b)
⎡ −1 0⎤ ⎡1 0⎤ ⎡ −1 0⎤ ⎡ −1 0 ⎤
A=⎢ A =⎢ A =⎢ A =⎢ ⎥
2 3 4
⎣0 −2 ⎥⎦ ⎣0 4 ⎥⎦ ⎣0 −8 ⎥⎦ ⎣ 0 16 ⎦
⎡ t
2
t
3
t
4
⎤
⎢ 1 − t + − + +L 0 ⎥ ⎡e− t
1 2 2 2! 3! 4! 0 ⎤
φ (t ) = I + At + A t + L = ⎢ ⎥=⎢ −2 t ⎥
⎢ ⎥ ⎣0 e ⎦
2 3
2! 4t 8t
1 − 2t + − +L
⎣⎢ ⎦⎥
0
2! 3!
10‐46


⎡ 1 0
⎤
⎡s + 1 0 ⎤ ⎢ s +1 −1
⎥ ⎡e−t 0 ⎤
Φ ( s ) = ( sI − A ) = ⎢
−1
⎥ =⎢ ⎥ φ (t ) = ⎢ ⎥
⎣ 0 s + 2⎦
−2 t
⎢ 0 1 ⎥ ⎣0 e ⎦
⎢⎣ s + 2 ⎥⎦
(c)
⎡0 1⎤ ⎡1 0⎤ ⎡0 1⎤ ⎡1 0⎤
A=⎢ A =⎢ A =⎢ A =⎢
2 3 4
⎣1 0 ⎥⎦ ⎣0 1 ⎥⎦ ⎣1 0 ⎥⎦ ⎣0 1 ⎥⎦
⎡ t2 t4 t
3
t
5
⎤
⎢ 1 + + + L t + + L ⎥
1 2 2 2! 4! 3! 5! ⎡ e − t + et −e + e ⎤
−t t
φ (t ) = I + At + A t + L = ⎢ ⎥ = ⎢ −t −t ⎥
⎢ ⎥ ⎣ −e + e e +e ⎦
3 5 2 4 t t
2! t t t t
t + + L 1+ + + L
⎢⎣ 3! 5! 2! 4! ⎥⎦
⎡ 0.5 − 0.5 −0.5

+
⎤
0.5
⎡s −1⎤
−1
1 ⎢ s +1 s −1
⎡s 1⎤ s + 1 s − 1⎥
Φ ( s ) = ( sI − A ) =
−1
⎢ −1 = = ⎢ ⎥
⎣ s ⎥⎦ s − 1 ⎢⎣1 s ⎥⎦ ⎢ −0.5 0.5
2
0.5 0.5 ⎥
+ +
⎢⎣ s + 1 s − 1 s + 1 s − 1 ⎥⎦
⎡ e − t + et −e + e ⎤
−t t
φ (t ) = 0.5 ⎢ − t −t ⎥
⎣ −e + e e +e ⎦
t t
10-22) (a) e = K s (θ r − θ y ) ea = e − e s e s = R s ia eu = Kea
eu − eb dθ y d θy
2
ia = eb = K b Tm = K i ia = ( J m + J L )
Ra + Rs
2
dt dt
dθ y
Solve for ia in terms of θ y and , we have
dt
dθ y
KK s (θ r − θ y ) − K b
ia = dt
Rs + Rs + KRs
10‐47


Differential equation:
d θy ⎛ dθ y ⎞
2
K i ia Ki
= = ⎜ − Kb − KK sθ y + KK sθ y ⎟
dt
2
Jm + JL ( J m + J L )( Ra + Rs + KRs ) ⎝ dt ⎠
dθ y
State variables: x1 = θ y , x 2 =
dt
State equations:
⎡ dx1 ⎤ ⎡ 0 1 ⎤ ⎡ 0 ⎤
⎢ dt ⎥ ⎢ ⎥ ⎡ x1 ⎤ ⎢ ⎥θ
⎢ ⎥= − KK s K i − Kb Ki + − KK s K i
⎢ dx2 ⎥ ⎢⎢ ( J + J )( R + R + KR ) ⎥ ⎣⎢ x2 ⎦⎥ ⎢ ⎥ r
( J m + J L )( Ra + Rs + KRs ) ⎦⎥ ⎢⎣ ( J m + J L )( Ra + Rs + KRs ) ⎦⎥
⎣⎢ ⎦⎥ ⎣ m L
dt
a s s
⎡ 0 1 ⎤ ⎡ x1 ⎤ ⎡ 0 ⎤
=⎢ + θr
⎣ −322.58 −80.65 ⎥⎦ ⎢⎣ x2 ⎥⎦ ⎢⎣322.58 ⎥⎦
We can let v ( t ) = 322.58θ r , then the state equations are in the form of CCF.
(b)
−1
⎡ s −1 ⎤ 1 ⎡ s + 80.65 1 ⎤
( sI − A ) = ⎢ −1
⎥ = 2
⎣322.58 s + 80.65 ⎦ s + 80.65 s + 322.58 ⎢⎣ −322.58 s ⎥⎦
⎡ −0.06 − 1.059 −0.014

+
⎤ 0.014
⎢ s + 76.42 s + 4.22 s + 76.42 s + 4.22 ⎥
= ⎢ ⎥
⎢ 4.468 − 4.468 1.0622
−
0.0587 ⎥
⎢⎣ s + 76.42 s + 4.22 s + 76.42 s + 4.22 ⎥⎦
For a unit‐step function input, u s ( t ) = 1 / s.
⎡ 322.2 ⎤ ⎡ 1 0.0584 1.058 ⎤

⎢ ⎥ + −
+ + ⎢ ⎥
⎥ = ⎢ s s + 76.42 s + 4.22 ⎥
1 s ( s 76.42)( s 4.22)
( sI − A )−1 B = ⎢
s ⎢ 322.2 ⎥ ⎢ −4.479 + 4.479 ⎥
⎢⎣ s ( s + 76.42)( s + 4.22) ⎥⎦ ⎢⎣ s + 76.42 s + 4.22 ⎥⎦
⎡ −0.06e −76.42 t − 1.059e −4.22 t −0.014e

−76.42 t
+ 0.01e
−4.22 t
⎤
x (t ) = ⎢ −76.42 t −4.22 t −76.42 t −4.22 t ⎥ x(0)
⎣ 4.468e − 4.468e 1.0622e − 0.0587 e ⎦
⎡1 + 0.0584e −76.42 t − 1.058e −4.22 t ⎤

=⎢ −76.42 t −4.22 t ⎥ t ≥ 0
⎣ −4.479e + 4.479e ⎦
10‐48


2
(c) Characteristic equation: Δ( s ) = s + 80.65 s + 322.58 = 0
(d) From the state equations we see that whenever there is Ra there is (1 + K ) Rs . Thus, the purpose of Rs is

to increase the effective value of Ra by (1 + K ) Rs . This improves the time constant of the system.
10-23) (a) State equations:
⎡ dx1 ⎤ ⎡ 0 1 ⎤ ⎡ 0 ⎤
⎢ dt ⎥ ⎢ ⎥ ⎡ x1 ⎤ ⎢ ⎥θ
⎢ ⎥= − KK s K i − Kb Ki +
⎥ ⎢⎣ x2 ⎥⎦ ⎢
KK s K i
⎢ dx2 ⎥ ⎢⎢ J ( R + R + KR ) J ( R + Rs + KRs ) ⎥⎦
⎥ r
⎢⎣ J ( R + Rs + KRs ) ⎥⎦
⎣⎢ ⎦⎥ ⎣
dt
s s
⎡ 0 1 ⎤ ⎡ x1 ⎤ ⎡ 0 ⎤
= ⎢ + θr
⎣ −818.18 −90.91⎥⎦ ⎢⎣ x2 ⎥⎦ ⎢⎣818.18 ⎥⎦
Let v = 818.18θ r . The equations are in the form of CCF with v as the input.
−1
⎡ s −1 ⎤ 1 ⎡ s + 90.91 1⎤
( sI − A ) = ⎢
−1
(b) ⎥ =
( s + 10.128 )( s + 80.782 ) ⎢⎣ −818.18 s ⎥⎦

⎣818.18 s + 90.91⎦
⎡ 1.143e −10.128 t − 0.142e −80.78 t 0.01415e

−10.128 t
− 0.0141e
−80.78 t
⎤ ⎡ x1 (0) ⎤
x (t ) = ⎢ ⎥ ⎢ x (0) ⎥
⎦⎣ 2 ⎦
−10.128 t −80.78 t −10.128 t −80.78 t
⎣ −11.58e + 0.1433e −0.1433e + 1.143e
⎡ 11.58e −10.128 t
− 11.58e ⎤
−80.78 t
+ ⎢ −10.128 t −80.78 t ⎥

t ≥ 0
⎣1 − 1.1434e + 0.1433e ⎦
2
(c) Characteristic equation: Δ( s ) = s + 90.91s + 818.18 = 0
Eigenvalues: −10.128, −80.782
(d) Same remark as in part (d) of Problem 5‐14.
10-24) If and P diagonalizing A, let consider , therefore or
The solution for is 0 , therefore
0 1
on the other hand
10‐49

A Chaptter 10 Solutio
ons Golnarraghi, Kuo

0 2
From
m equation (1
1) and (2):
1
10-25) Consider and . If , then or
The solution
s for x is 0 , theerefore:
0 (1))
On thhe other hand

d:
0 2
From
m equation (1
1) and (2):
1
10‐26 (a) Fo
orward‐path trransfer functio
on: C
Closed‐loop tr
ransfer functio
on:
Y (s) 5 ( K1 + K 2 s ) Y (s) G (s) 5 ( K1 + K 2 s )

G ( s ) = = M (s) = = =
s [ s ( s + 4)( s + 5)) + 10 ] 1 + G(s) s + 9 s + 20 s + (10 + 5 K 2 ) s + 5 K1
4 3 2
E (s) R(s)
(b) Statte diagram by direct decomp
position:

State equations: Outpu
ut equation:
⎡ x&1 ⎤ ⎡ 0 1 0 0 ⎤ ⎡ x1 ⎤
⎡0 ⎤
⎢ x& ⎥ ⎢ 0 0 1 0 ⎥ ⎢ x ⎥ ⎢0 ⎥
⎢ 2⎥ = ⎢ ⎥⎢ 2⎥+ ⎢ ⎥r y = 5 K1 5K2 0 x
⎢ x&3 ⎥ ⎢ 0 0 0 1 ⎥ ⎢ x3 ⎥ ⎢ 0 ⎥
⎢ & ⎥ ⎢− K − (10
⎥⎢ ⎥ ⎢ ⎥
1 + 5 K 2 ) −220 −9 ⎦ ⎣ x4 ⎦ ⎣ 1 ⎦
⎣ x4 ⎦ ⎣ 5
10‐50


1
(c) Final value: r ( t ) = u s ( t ), R( s ) = .
s
5 ( K1 + K 2 s )
lim y (t ) = lim sY ( s ) = lim =1
s + 9 s + 20 s + (10 + 5 K 2 ) s + 5 K1
4 3 2
t →∞ s→0 s →0
10‐27 In CCF form,
⎡ 0 1 0 0 L 0 ⎤ ⎡0⎤
⎢ 0 0 1 0 L 0 ⎥ ⎢0⎥
⎢ ⎥ ⎢ ⎥
A=⎢ M M M M O M ⎥ B = ⎢M ⎥
⎢ ⎥ ⎢ ⎥
L
⎢ 0 0 0 0 1
⎥ ⎢0⎥
⎢⎣ − a0 − a1 − a2 − a3 L − an −1 ⎥⎦ ⎢⎣1 ⎥⎦
⎡s −1 0 0 L 0 ⎤
⎢0 s −1 0 L 0 ⎥
⎢ ⎥
sI − A = ⎢ M M M M O ⎥ M
⎢ ⎥
−1
⎢0 0 0 0 0
⎥
⎢⎣ a0 a1 a2 a3 L s + an ⎥⎦

n n −1 n− 2
sI − A = s + an−1 s + an− 2 s + L + a1 s + a0
Since B has only one nonzero element which is in the last row, only the last column of adj ( sI − A ) is
going to contribute to adj ( sI − A ) B . The last column of adj ( sI − A ) is obtained from the cofactors of
'
the last row of ( sI − A ) . Thus, the last column of adj ( sI − A ) B is 1
2 n −1
s s L s .

2
dy d y
10‐28 (a) State variables: x1 = y, x 2 = , x3 = 2

dt dt
State equations: x& ( t ) = Ax ( t ) + Br ( t )
10‐51


⎡0 1 0⎤ ⎡0 ⎤
A=⎢0 0 1⎥ B = ⎢0 ⎥
⎢ ⎥ ⎢ ⎥
⎣⎢ −1 −3 −3⎥⎦ ⎣⎢1 ⎦⎥
(b) State transition matrix:
⎡ s 2 + 3s + 3 s + 3 1 ⎤
−1
⎡ s −1 0 ⎤
1 ⎢ ⎥
Φ ( s ) = ( sI − A ) = ⎢ 0 s −1 ⎥ =
−1
−1 s + 3s s
2

⎢ ⎥ Δ(s) ⎢ ⎥
⎢⎣1 3 s + 3⎥⎦ ⎢⎣ − s −3s − 1 s ⎥⎦
2
⎡ 1 1 1 1 2 1 ⎤
⎢ + + + ⎥
⎢ s + 1 ( s + 1) ( s + 1) ( s + 1) ( s + 1) ( s + 1) ⎥
2 3 2 3 3
⎢ −1 1 1 2 s ⎥
= ⎢ + − ⎥
⎢ ( s + 1)3 s +1 ( s + 1) 2
( s + 1) 3
( s + 1)3 ⎥
⎢ ⎥
−s −3
2
⎢ +
2 s
⎥
⎣⎢ ( s + 1)3 ( s + 1) 2
( s + 1) 3
( s + 1)3 ⎦⎥
Δ ( s ) = s + 3s + 3s + 1 = ( s + 1)
3 2 3

⎡(1 + t + t 2 / 2 ) e − t (t + t ) e 2 −t 2
t e /2
−t
⎤
⎢
φ (t ) =
2 −t
⎢ −t e / 2 (1 + t − t ) e 2 −t
( t − t / 2 ) e ⎥⎥
2 −t
⎢⎣ ( −t + t 2 / 2 ) e − t 2
t e
−t
(1 − 2t + t 2 / 2 ) e − t ⎥⎦
(c) Use ACSYS or MATLAB and follow the procedure shown in solution to 10‐3.
10‐52


‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐
‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐
Inputs:
A=| 0 1 0| B=|0|
| 0 0 1| |0|
|‐1 ‐3 ‐3| |1|

C=|1 0 0| D=|0|

| 0 1 0| |0|
Dx = | 0 0 1|x + |0|u
|‐1 ‐3 ‐3| |1|

y = |1 0 0|x + |0|u

Determinant of (s*I‐A):

3 2
s + 3 s + 3 s + 1

3 2
10‐53


s + 3 s + 3 s + 1
‐1
‐1
‐1
Inverse of (s*I‐A) is:

[ 2 ]
[s + 3 s + 3 s + 3 1 ]
[‐‐‐‐‐‐‐‐‐‐‐‐ ‐‐‐‐‐ ‐‐‐‐]
[ %1 %1 %1 ]
[ ]
[ 1 s (s + 3) s ]
[ ‐ ‐‐‐‐ ‐‐‐‐‐‐‐‐‐ ‐‐‐‐]
[ %1 %1 %1 ]
[ ]
[ 2 ]
[ s 3 s + 1 s ]
[ ‐ ‐‐‐‐ ‐ ‐‐‐‐‐‐‐ ‐‐‐‐]
[ %1 %1 %1 ]

3 2
%1 := s + 3 s + 3 s + 1

[ 2 2 2 ]
[1/2 exp(‐t) (2 + 2 t + t ) , (t + t ) exp(‐t) , 1/2 t exp(‐t)]

[ 2 2 2
[‐ 1/2 t exp(‐t) , ‐(‐t ‐ 1 + t ) exp(‐t) , ‐ 1/2 exp(‐t) (‐2 t + t )

]
]

[ 2 2
[1/2 exp(‐t) (‐2 t + t ) , exp(‐t) (‐3 t + t ) ,

2 ]
1/2 exp(‐t) (2 ‐ 4 t + t )]

1
‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐
3 2
s + 3 s + 3 s + 1
No Initial Conditions Specified
10‐54


[ 1 ]
[‐‐‐‐‐‐‐‐]
[ 3]
[(s + 1) ]
[ ]
[ s ]
[‐‐‐‐‐‐‐‐]
[ 3]
[(s + 1) ]
[ ]
[ 2 ]
[ s ]
[‐‐‐‐‐‐‐‐]
[ 3]
[(s + 1) ]

[ 2 ]
[ 1/2 t exp(‐t) ]
[ ]
[ 2 ]
[‐ 1/2 exp(‐t) (‐2 t + t ) ]
[ ]
[ 2 ]
[1/2 exp(‐t) (2 ‐ 4 t + t )]
Output Y(s):

1
‐‐‐‐‐‐‐‐
3
(s + 1)

2
1/2 t exp(‐t)
‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐
‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐
Inputs:
A=| 0 1 0| B=|0|
| 0 0 1| |0|
|‐1 ‐3 ‐3| |1|

C=|1 0 0| D=|0|

10‐55


| 0 1 0| |0|
Dx = | 0 0 1|x + |0|u
|‐1 ‐3 ‐3| |1|

y = |1 0 0|x + |0|u

Determinant of (s*I‐A):

3 2
s + 3 s + 3 s + 1

3 2
s + 3 s + 3 s + 1
‐1
‐1
‐1
Inverse of (s*I‐A) is:

[ 2 ]
[s + 3 s + 3 s + 3 1 ]
[‐‐‐‐‐‐‐‐‐‐‐‐ ‐‐‐‐‐ ‐‐‐‐]
[ %1 %1 %1 ]
[ ]
[ 1 s (s + 3) s ]
[ ‐ ‐‐‐‐ ‐‐‐‐‐‐‐‐‐ ‐‐‐‐]
[ %1 %1 %1 ]
[ ]
[ 2 ]
[ s 3 s + 1 s ]
[ ‐ ‐‐‐‐ ‐ ‐‐‐‐‐‐‐ ‐‐‐‐]
[ %1 %1 %1 ]

3 2
%1 := s + 3 s + 3 s + 1

[ 2 2 2 ]
[1/2 exp(‐t) (2 + 2 t + t ) , (t + t ) exp(‐t) , 1/2 t exp(‐t)]

[ 2 2 2
[‐ 1/2 t exp(‐t) , ‐(‐t ‐ 1 + t ) exp(‐t) , ‐ 1/2 exp(‐t) (‐2 t + t )

]
]

10‐56


[ 2 2
[1/2 exp(‐t) (‐2 t + t ) , exp(‐t) (‐3 t + t ) ,

2 ]
1/2 exp(‐t) (2 ‐ 4 t + t )]

1
‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐
3 2
s + 3 s + 3 s + 1
Initial Conditions:
x(0)= 1
0
0

[ 2 ]
[s + 3 s + 4]
[‐‐‐‐‐‐‐‐‐‐‐‐]
[ 3 ]
[ (s + 1) ]
[ ]
[ s ‐ 1 ]
[ ‐‐‐‐‐‐‐‐ ]
[ 3 ]
[ (s + 1) ]
[ ]
[ s (s ‐ 1) ]
[ ‐‐‐‐‐‐‐‐‐ ]
[ 3 ]
[ (s + 1) ]

[ 2 ]
[ (t + 1 + t ) exp(‐t) ]
[ ]
[ 2 ]
[ ‐(‐t + t ) exp(‐t) ]
[ ]
[ 2 ]
[(‐3 t + 1 + t ) exp(‐t)]
Output Y(s) (with initial conditions):

2
s + 3 s + 4
‐‐‐‐‐‐‐‐‐‐‐‐
10‐57


3
(s + 1)

2
(t + 1 + t ) exp(‐t)

3 2
(d) Characteristic equation: Δ( s ) = s + 3 s + 3 s + 1 = 0
Eigenvalues: −1, −1, −1
dy
10‐29 (a) State variables: x1 = y, x 2 =
dt
State equations:
⎡ dx1 (t ) ⎤
⎢ dt ⎥ ⎡ 0 1 ⎤ ⎡ x1 (t ) ⎤ ⎡ 0 ⎤
⎢ ⎥=⎢ ⎥⎢ ⎥ + ⎢ ⎥ r (t )
⎢ dx2 (t ) ⎥ ⎣ −1 −2 ⎦ ⎣ x2 (t ) ⎦ ⎣1 ⎦
⎢⎣ dt ⎥⎦
⎡ s+2 1 ⎤
⎢ ( s + 1)2
−1
( s + 1) ⎥ ⎡ (1 + t )e − t ⎤
⎡ s −1 ⎤
2 −t
te
Φ ( s ) = ( sI − A ) = ⎢
−1
⎢ ⎥ φ (t ) = ⎢
⎥ = ⎢ −1 ⎥ ⎥
⎣ + ⎦
−t −t
1 s 2 s ⎣ −te (1 − t ) e ⎦
⎢ s +1 2 2 ⎥
⎣( ) ( s + 1) ⎦
Δ ( s ) = ( s + 1) = 0
2
Characteristic equation:
dy
(b) State variables: x1 = y, x 2 = y +
dt
State equations:
2
dx1 dy dx 2 d y dy dy
= = x 2 − y = x 2 − x1 = 2
+ = −y − + r = − x2 + r
dt dt dt dt dt dt
⎡ dx1 ⎤
⎢ dt ⎥ ⎡ −1 2 ⎤ ⎡ x1 ⎤ ⎡ 0 ⎤
⎢ ⎥=⎢ ⎥⎢ ⎥+⎢ ⎥r
⎢ dx2 ⎥ ⎣ 0 −1⎦ ⎣ x2 ⎦ ⎣1 ⎦
⎣⎢ dt ⎦⎥
10‐58


⎡ 1 −2 ⎤
⎡ s + 1 −2 ⎤ ⎢ s + 1 ( s + 1) ⎥
−1 2
⎡e − t ⎤
−te
−t
Φ(s) = ⎢ ⎥ =⎢ ⎥ φ (t ) = ⎢ ⎥
⎣ 0 s + 1⎦ ⎢ 1 ⎥
−t
⎣0 e ⎦
⎢⎣ 0 s +1 ⎦
⎥
(c) Characteristic equation: Δ ( s ) = ( s + 1) = 0 which is the same as in part (a).
2
10‐30 (a) State transition matrix:
⎡s − σ ω ⎤ 1 ⎡s − σ −ω ⎤
sI − A = ⎢
s − σ ⎥⎦
( sI − A ) −1
=
Δ ( s ) ⎢⎣ ω s − σ ⎥⎦
2
(
Δ ( s ) = s − 2σ + σ + ω
2 2
)
⎣ −ω
⎡ cos ω t − sin ω t ⎤
φ (t ) = L ⎡⎣( sI − A ) ⎤⎦ = ⎢
−1 −1 σt
⎥ e
⎣ sin ω t cos ω t ⎦
(b) Eigenvalues of A: σ + jω , σ − jω
10‐31 (a)
−3
Y1 ( s ) s 1
= −1 −2 −3
= 3 2

U1 ( s ) 1+ s + 2s + 3s s + s + 2s + 3
−3
Y2 ( s ) s 1 Y1 ( s )
= −1 −2 −3
= 3 2
=
U 2 ( s) 1+ s + 2s + 3s s + s + 2s + 3 U1 ( s )
(b) State equations [Fig. 5‐21(a)]: x& = A1x + B1u1 Output equation: y1 = C1x
⎡0 1 0⎤ ⎡0 ⎤
A1 = ⎢ 0 0 1⎥ B1 = ⎢ 0 ⎥ C1 = [1 0 0]
⎢ ⎥ ⎢ ⎥
⎢⎣ −3 −2 −1⎥⎦ ⎢⎣1 ⎥⎦
State equations [Fig. 5‐21(b)]: x& = A 2 x + B 2 u2 Output equation: y2 = C2x
⎡ 0 0 −3 ⎤ ⎡1 ⎤
A 2 = ⎢ 1 0 −2 ⎥ B 2 = ⎢0 ⎥ C2 = [ 0 0 1]
⎢ ⎥ ⎢ ⎥
⎢⎣ 0 1 −1⎥⎦ ⎢⎣ 0 ⎥⎦
'
Thus, A 2 = A1
10‐59


1
10‐32 (a) State diagram:

(b) State diagram:
1
10‐33 (a)
−3
Y ( s) 10 s X ( s)
G( s) = = −1 −2 −3
Y ( s ) = 100 X ( s )
U ( s) 1 + 8.5 s + 20.5s + 15s X ( s)
−1 −2 −3
X ( s ) = U ( s ) − 8.5 s X ( s ) − 20.5 s X ( s ) − 15 s X ( s )

State diagraam:

10‐60


Statte equation: x& ( t ) = Ax ( t ) + Bu ( t )
⎡ 0 1 0 ⎤ ⎡0⎤
⎢
A= 0 0 1 ⎥ B = ⎢0⎥ A and B are in
A CCF
⎢ ⎥ ⎢ ⎥
⎢⎣ −15 −20.5 −8.55 ⎥⎦ ⎣⎢1 ⎦⎥
(b)
−3 −4
Y ( s) 10 s + 20 s X ( s)
G ( s ) = = −1 −2

U ( s) 1 + 4.5 s + 3.5 s X ( s)
−3 −4 −1 −2
Y ( s ) = 10 s X ( s ) + 20 s X ( s ) X ( s ) = −4.5 s X ( s ) − 3.5 s X ( s ) + U ( s )
Staate diagram:

Staate equations: x& ( t ) = Ax
A ( t ) + Bu ( t )
⎡0 1 0 ⎤ 0 ⎡0⎤
⎢0 0 1 0 ⎥ ⎢0⎥
A=⎢ ⎥ B=⎢ ⎥ A and B are in
n CCF
⎢0 0 0 1 ⎥ ⎢0⎥
⎢ ⎥ ⎢ ⎥
⎣0 0 −3.5 −4.55 ⎦ ⎣1 ⎦
(c)
−2 −3
Y ( s) 5( s + 1) 5s + 5s X ( s)
G( s) = = = −1 −2

U ( s) s( s + 2 )( s + 10) 1 + 12 s + 20 s X ( s)

−2 −3 −1 −
−2
Y ( s) = 5s X ( s ) + 5s X ( s ) X ( s ) = U ( s ) − 12 s X ( s ) − 20 s X ( s )
10‐61


Staate equations: x& ( t ) = Ax ( t ) + Bu ( t )
⎡0 1 0 ⎤ ⎡0 ⎤
⎢
A= 0 0 1 ⎥ B = ⎢0 ⎥ A and B are in CCF
⎢ ⎥ ⎢ ⎥
⎢⎣ 0 −20 −12 ⎥⎦ ⎢⎣1 ⎥⎦
(d)
−4
Y (s) 1 s X (s)
G (s) = = =
U (s) (
s ( s + 5) s + 2s + 2
2
) −1
1 + 7 s + 12 s + 10 s
−2 −3
X (s)
−4 −1 −2 −3
Y ( s) = s X ( s) X ( s ) = U ( s ) − 7 s X ( s ) − 12 s X ( s ) − 10 s

State diagraam:
State equations: x& ( t ) = Ax
A ( t ) + Bu ( t )
10‐62


⎡0 1 0 0⎤ ⎡0 ⎤
⎢0 0 1 0⎥ ⎢0 ⎥
A=⎢ ⎥ B=⎢ ⎥ A and B arre in CCF
⎢0 0 0 1⎥ ⎢0 ⎥
⎢ ⎥ ⎢ ⎥
⎣ 0 −10 −12 −7 ⎦ ⎣1 ⎦
1
10‐34 (a)
Y ( s) 10 5.71 6.67 0.952
G( s) = = 3 2
= − +
U ( s) s + 8.5 s + 20.5s + 15 s + 15 s+2 s+5

A ( t ) + Bu ( t )
⎡ −1.5 0 0 ⎤ ⎡ 5.71 ⎤
A=⎢ 0 −2 0 ⎥ B = ⎢ −6.67 ⎥
⎢ ⎥ ⎢ ⎥
⎢⎣ 0 0 −5 ⎥⎦ ⎣⎢ 0.952 ⎦⎥
w the input and the output branches are allo
The matrix B is not unique. It depends on how ocated.
(b)
Y ( s) 10( s + 2) −4.55 0.49 4 5.71

G( s) = = 2
= + + + 2

U ( s) s ( s = 1)(( s + 3.5) s s + 3.5 s +1 s
Statte diagram:
10‐63


State equation: x& ( t ) = Ax
A ( t ) + Bu ( t )
⎡0 1 0 ⎤ 0 ⎡0⎤
⎢0 0 0 0 ⎥ ⎢1 ⎥
A=⎢ ⎥ B=⎢ ⎥
⎢0 0 −1 0 ⎥ ⎢1 ⎥
⎢ ⎥ ⎢ ⎥
⎣0 0 0 −3.5 ⎦ ⎣1 ⎦
(b)
Y ( s) 5( s + 1) 2.5 0.313 0.563

G( s) = = = + −
U ( s) s( s + 2)( s + 10) s s+2 s + 10

Sttate equationss: x& ( t ) = Ax
A ( t ) + Bu ( t )
⎡0 0 0 ⎤ ⎡1⎤
A = ⎢ 0 −2 0 ⎥ B = ⎢1⎥
⎢ ⎥ ⎢⎥
⎢⎣ 0 0 −10 ⎥⎦ ⎢⎣1⎦⎥
(d)
Y (s) 1 0.1 0.0118 0.0882 s + 0.235

G ( s) = = = − −
U ( s) (
s ( s + 5) s + 2s + 2
2
) s s+5 s + 2s + 2
2
10‐64


Sttate equationss: x& ( t ) = Ax
A ( t ) + Bu ( t )
⎡0 0 0 0 ⎤ ⎡1 ⎤
⎢ 0 −5 0 0 ⎥ ⎢1 ⎥
A=⎢ ⎥ B=⎢ ⎥
⎢0 0 0 1 ⎥ ⎢0⎥
⎢ ⎥ ⎢ ⎥
⎣ 0 0 −2 − 2 ⎦ ⎣1 ⎦
1
10‐35 (a)
Y ( s) 1
10
G( s) = =
U ( s) ( s + 15)(
. s + 2 )( s + 5)
Statte diagram:

A ( t ) + Bu ( t )
⎡ −5 1 0 ⎤ ⎡0⎤
⎢
A = 0 −2 1 ⎥ B=⎢0⎥
⎢ ⎥ ⎢ ⎥
⎢⎣ 0 0 −1.5 ⎥⎦ ⎣⎢10 ⎦⎥
(b)
10‐65


G(s) =
Y (s)
=
10( s + 2) ⎛ 100 ⎞ ⎛ s + 2 ⎞ ⎛ 1 ⎞
=⎜2 ⎟⎜ ⎟⎜ ⎟

s ( s + 1)( s + 3.5) ⎝ s ⎠ ⎝ s + 1 ⎠ ⎝ s + 3.5 ⎠
2
U (s)
Staate diagram:

A ( t ) + Bu ( t )
⎡0 1 0 ⎤ 0 ⎡0⎤
⎢0 0 1 1 ⎥ ⎢0⎥
A=⎢ ⎥ B=⎢ ⎥
⎢0 0 −1 1 ⎥ ⎢0⎥
⎢ ⎥ ⎢ ⎥
⎣0 0 0 −3.5 ⎦ ⎣10 ⎦
(c)
G ( s) =
Y (s)
=
59 s + 1) ⎛ 5 ⎞⎛ s +1 ⎞⎛ 1 ⎞
=⎜
⎟⎜ ⎟⎜ ⎟
U (s) s ( s + 2)( s + 10) ⎝ s ⎠ ⎝ s + 2 ⎠ ⎝ s + 10 ⎠
Staate diagram:
A ( t ) + Bu ( t )
⎡0 1 0⎤ ⎡0 ⎤
A = ⎢ 0 −10 −1⎥ B = ⎢0 ⎥
⎢ ⎥ ⎢ ⎥
⎢⎣ 0 0 −2 ⎥⎦ ⎢⎣ 5 ⎥⎦
(d)
10‐66


Y (s) 1
G(s) = =
U (s) (
s ( s + 5) s + 2s + 2
2
)
S
State diagram:

A ( t ) + Bu ( t )
⎡0 1 0 0 ⎤ ⎡0⎤
⎢0 0 1 0 ⎥ ⎢0⎥
A=⎢ ⎥ B=⎢ ⎥
⎢ 0 −2 −2 1 ⎥ ⎢0⎥
⎢ ⎥ ⎢ ⎥
⎣ 0 0 0 −5 ⎦ ⎣1 ⎦
1
10‐36 (a)
−3
Y ( s) 10 100 s X ( s)
G( s) = = = −1 −2

E ( s) s( s + 4 )( s + 5) 1+ 9s + 20 s X ( s)

(b)
( Dynamic equations:
⎡ x&1 ⎤ ⎡ 0 1 0 ⎤ ⎡ x1 ⎤ ⎡ 0 ⎤
⎢ x& ⎥ = ⎢ 0 0 1 ⎥ ⎢ x2 ⎥ + ⎢ 0 ⎥ r y = [10
1 0 0] x
⎢ 2⎥ ⎢ ⎥⎢ ⎥ ⎢ ⎥
⎢⎣ x&3 ⎥⎦ ⎢⎣ −10
1 −20 −9 ⎥⎦ ⎢⎣ x3 ⎦⎥ ⎢⎣1 ⎦⎥
10‐67


(c) State transition equation:
⎡ X 1 (s) ⎤ ⎡ s −1 (1 + 9 s −1 + 20 s −2 ) s −2 (1 + 9 s −1 ) s −3 ⎤ ⎡ x1 (0) ⎤ ⎡ s −3 ⎤
⎢ X (s)⎥ = 1 ⎢ −10 s
−3 −1 −2 ⎥
s (1 + 9 s ) s ⎢ x2 (0) ⎥ +
−1 1 ⎢ −2 ⎥ 1
s
⎢ 2
⎥ Δ(s) ⎢ ⎥⎢ ⎥ Δ ( s ) ⎢ −1 ⎥ s
⎣⎢ X 3 ( s ) ⎦⎥ ⎢⎣ −10 s
−2
−20 s
−2
s ⎥⎦ ⎣⎢ x3 (0) ⎦⎥
−1
⎢⎣ s ⎥⎦
⎡1⎤
⎡s 2
+ 9 s + 20 s+9 1 ⎤ ⎡ x1 (0) ⎤ ⎢s⎥
⎢ ⎥ ⎢ ⎥
s ⎢ x2 (0) ⎥ +
1 1
= −10 s ( s + 9)
⎥ Δc (s) ⎢ ⎥
1
Δ c (s) ⎢ ⎥⎢
⎢⎣ −10 s −10 ( 2 s + 1) s ⎥⎦ ⎣⎢ x3 (0) ⎦⎥
2
⎢s⎥
⎢ ⎥
⎣ ⎦
−1 −2 −3 3 2
Δ( s) = 1+ 9s + 20 s + 10 s Δ c ( s ) = s + 9 s + 20 s + 10
⎧ ⎡1.612 0.946 0.114 ⎤ ⎡ −0.706 −1.117 −0.169 ⎤ ⎡ 0.0935 0.171 0.055 ⎤ ⎫

⎪ −5.895 t ⎪
x (t ) = ⎨ ⎢ −1.14 −0.669 −0.081⎥ e ⎢ 4.056 ⎥ e ⎢ ⎥
⎬ x(0)
−0.708 t −2.397 t
+ 1.692 2.678 + −0.551 −1.009 −0.325 e
⎢ ⎥ ⎢ ⎥ ⎢ ⎥
⎪ ⎢ 0.807 0.474 0.057 ⎥ ⎪
⎩⎣ ⎦ ⎣⎢ −4.056 −6.420 −0.972 ⎦⎥ ⎣⎢ 3.249 5.947 1.915 ⎦⎥ ⎭
⎡0.1 − 0.161e −0.708 t + 0.0706e −2.397 t − 0.00935e −5.895 t ⎤

⎢ ⎥
+⎢ 0.114e
−0.708 t
− 0.169e
−2.397 t
+ 0.055e
−5.895 t
⎥ t ≥ 0
⎢⎣ −0.087e −0.708 t + 0.406e −2.397 t − 0.325e −5.895 t ⎥⎦
(d) Output:

(
y (t ) = 10 x1 (t ) = 10 1.612e
−0.708 t
− 0.706e
−2.397 t
+ 0.0935e
−5.897 t
) x (0) + 10 ( 0.946e − 1.117e + 0.1711e ) x (0)
1
−0.708 t −2.397 t −5.895 t
2

(
+ 10 1.141e
−0.708 t
− 0.169e
−2.397 t
+ 0.0550e
−5.895 t
) x (0) + 1 − 1.61e−0.708t + 0.706e−2.397t − 0.0935e−5.895t
3
t≥0
10‐37(a) Closed‐loop transfer function:
Y ( s) 10
= 3 2

R( s ) s + 9 s + 20 s + 10
(b) State diagram:
10‐68


(c) State equation
S s: (d) Statee transition eqquations:
[
[Same answers
s as Problem 5
5‐26(d)]
⎡ x&1 ⎤ ⎡ 0 1 0 ⎤ ⎡ x1 ⎤ ⎡ 0 ⎤
⎢ ⎥
x& 2 = 0 ⎢ 0 1 ⎥ ⎢ x2 ⎥ + ⎢ 0 ⎥ r (e) Outpput: [Same ansswer as Probleem 5‐26(e)]
⎢ ⎥ ⎢ ⎥⎢ ⎥ ⎢ ⎥
⎢⎣ x&3 ⎦⎥ ⎣⎢ −10 −20 −9 ⎦⎥ ⎣⎢ x3 ⎦⎥ ⎣⎢1 ⎦⎥
1
10‐38 (a) Statte diagram:
(b) Statte equations:
⎡ x&1 ⎤ ⎡ −2 20 −1 0 ⎤ ⎡ x1 ⎤ ⎡ 0 −1⎤
⎢ x& ⎥ ⎢ 0 −10 1 0 ⎥ ⎢ x2 ⎥ ⎢ 0 0 ⎥⎡u ⎤
⎢ ⎥ = ⎢ ⎥⎢ ⎥+⎢ ⎥
2
⎢ x&3 ⎥ ⎢ −0.1 0 −20 1 ⎥ ⎢ x3 ⎥ ⎢ 0 0 ⎥ ⎢⎣TD ⎥⎦

⎢& ⎥ ⎢ ⎥⎢ ⎥ ⎢ ⎥
⎣ x4 ⎦ ⎣ 0 0 0 −5 ⎦ ⎣ x4 ⎦ ⎣30 0 ⎦
(c) Tran
nsfer function rrelations:
From
m the system b
block diagram,
1 ⎛ −1 ⎞
−0.2 s
0.3 30e U (s) 90U ( s )
Y ( s ) = ⎜ TD ( s ) + TD ( s ) + + ⎟
Δ( s) ⎝ s + 2 ( s + 2)( s + 20) ( s + 2)( s + 5)( s + 20) ( s + 5)( s + 20) ⎠
−0. 2 s −0. 2 s
0.1e ( s + 2)( s + 20) + 0.1e
Δ(
Δ s) = 1+ =
( s + 2 )( s + 20) ( s + 2)( s + 20)
10‐69


−0. 2 s
− ( s + 19.7 ) 30e + 90( s + 2 )U ( s )
Y ( s ) = −0. 2 s
TD ( s ) +
−0. 2 s
( s + 2)( s + 20) + 0.1e ( s + 5) ( s + 2)( s + 20) + 0.1e
−0. 2 s
− ( s + 200) 330e U ( s)
Ω( s ) = −0. 2 s
TD ( s ) +
−0. 2 s
( s + 2 )( s + 20) + 0.1e ( s + 5) ( s + 2 )( s + 20) + 0.1e
−11
1
10‐39 (a) Th
here should no
ot be any incom
ming branches to a state variaable node otheer than the s branch. Thus, we
ould create a new node as shown in the folllowing state diiagram.
sho

−1
(b) State Notice that therre is a loop witth gain −1 after all the s brranches are deleted, so Δ = 2
e equations: N 2.
dx1 17 1 dx 2 15 1 1
= x1 + x2 = x1 − x 2 + r Output equ
uation: y = 6..5 x1 + 0.5 x 2
dt 2 2 dt 2 2 2
1
10‐40 (a) Trransfer functio
on: (b) Charracteristic equation:
Ks + 5 s + 1
( s + 1) ( s )
2
Y (s)
= + 111s + 2 = 0
2

R(s) ( s + 1) ( s 2
+ 11s + 2 )
Roo
ots of characte
eristic equation: −1, −0.185, −‐10..82. These aree not functions of K.
(c) Whe
en K = 1:
2
Y ( s) s + 5s + 1
= 3 2

R( s ) s + 12 s + 133s + 2

State diaggram:
10‐70


(d) Wheen K = 4:
4 s + 5s + 1 ( s + 1)(4 s + 1) 4s + 1
2
Y (s)
= = =
R( s) ( s + 1) ( s 2
+ 11s + 2 ) ( s + 1) ( s 2
+ 11s + 2 ) s + 11s + 2
2
State diagram:

(e)
Ks + 5 s + 1
( s + 1) ( s )
2
Y (s)
= + 111s + 2 = 0
2

R(s) ( s + 1) ( s 2
+ 11s + 2 )
M
MATLAB
ssolve(s^2+11*ss+2)
aans = ‐11/2+1
1/2*113^(1/2)
‐‐11/2‐1/2*1133^(1/2)
>>> vpa(ans)
aans =
‐.20
‐‐10.8
Y (s) Ks 2 + 5s + 1
=
R ( s ) ( s + 1)(
1 s + 0.2)( s + 10.82)
K = 4, 2.1914, 0.4536
P
Pole zero can
ncelation occcurs for the given values of K.

10‐71


1
10‐41 (a)
Y (s) 1 100
Gp (s) = = =
U (s) (1 + 0.5s ) (1 + 0.2 s + 0.02 s 2
) s + 12
1 s + 70 s + 1000
3 2
Statte diagram by ddirect decompposition:

State equations:
⎡ x&1 ⎤ ⎡ 0 1 0 ⎤ ⎡ x1 ⎤ ⎡ 0 ⎤
⎢ x& ⎥ = ⎢ 0 0 1 ⎥ ⎢ x2 ⎥ + ⎢ 0 ⎥ u
⎢ 2⎥ ⎢ ⎥⎢ ⎥ ⎢ ⎥
⎢⎣ x&3 ⎥⎦ ⎢⎣ −100 −70 −12 ⎥⎦ ⎢⎣ x3 ⎥⎦ ⎢⎣1 ⎥⎦
(b) Charracteristic equuation of closedd‐loop system
m: R
Roots of chara
acteristic equation:
−5.88, − 3.06 + j 4.965, − 3.06 − j 4.965

3 2
s + 12 s + 70 s + 200 = 0
1
10‐42 (a)
Y ( s) 1 − 0.066 s −20( s − 15)

1
Gp (s) = ≅ =
U ( s) (1 + 0.5s ) (1 + 0.133s + 0.0067s 2
) s + 22 s + 190 s + 300
3 2
Statte diagram by direct decompposition:
10‐72


State equations:
⎡ x&1 ⎤ ⎡ 0 1 0 ⎤ ⎡ x1 ⎤ ⎡ 0 ⎤
⎢ x& 2 ⎥ = ⎢ 0 0 1 ⎥ ⎢ x2 ⎥ + ⎢ 0 ⎥
⎢ ⎥ ⎢ ⎥⎢ ⎥ ⎢ ⎥
⎣⎢ x&3 ⎦⎥ ⎢⎣ −300 −190
1 −22 ⎥⎦ ⎣⎢ x3 ⎦⎥ ⎣⎢ 1 ⎦⎥
Chaaracteristic equuation of closeed‐loop system
m: R
Roots of chara
acteristic equation:
−12, −5 + j5, −5 −j5
3 2
s + 22 s + 170 s + 600 = 0

1
10‐43 (a) State variables: x1 = ω m and x 2 = ω D
State equations:
dω m Kb Ki + Kb Ra KD KK
Ki dω D KD KD
=− + ωD + e = ωm − ω D
dt JRa J JRa dt JR JR
(b) Statte diagram:
(c) Open‐loop transfe
er function:
Ωm (s) Ki ( J R s + K D )
KK
=
JJ R Ra s + ( K b J R K i + K D Ra J R + K D JRa ) s + K D K b K i
2
E (s)
Clossed‐loop transfer function:
10‐73


Ωm (s) K s KK i ( J R s + K D )
=
Ωr (s) JJ R Ra s + ( K b J R K i + K D Ra J R + K D JRa + K s KK i J R ) s + K D K b K i + K s KK i K D
2
(d) Characteristic equation of closed‐loop system:
Δ ( s ) = JJ R Ra s + ( K D J R K i + K D Ra J R + K D JRa + K s KK i J R ) s + K D K b K i + K s KK i K D = 0
2

2
Δ( s ) = s + 1037 s + 20131.2 = 0
Characteristic equation roots: −19.8, −1017.2
10‐44 (a) State equations: x& ( t ) = Ax ( t ) + Br ( t )
⎡ −b d ⎤ ⎡ −2 1⎤ ⎡0⎤ ⎡0 1⎤
A=⎢ = B=⎢ ⎥ S = [B AB ] = ⎢
⎣c − a ⎥⎦ ⎢⎣ 2 −1⎥⎦ ⎣1 ⎦ ⎣1 −1⎥⎦
Since S is nonsingular, the system is controllable.
(b)
⎡0 d ⎤
S = [B AB ] = ⎢ The system is controllable for d ≠ 0.
⎣1 − a ⎥⎦
10‐45 (a)
⎡1 −1 1⎤
S = ⎡⎣ B AB A B ⎤⎦ = ⎢1 −1 1⎥
2
S is singular. The system is uncontrollable.
⎢ ⎥
⎣⎢1 −1 1⎥⎦
(b)
⎡1 −1 1 ⎤
S = ⎡⎣ B AB A B ⎤⎦ = ⎢1 −2 4 ⎥
2
S is nonsingular. The system is controllable.
⎢ ⎥
⎣⎢1 −3 9 ⎥⎦
10‐46 (a) State equations: x& ( t ) = Ax ( t ) + Bu ( t )
10‐74


⎡ −2 3⎤ ⎡1⎤ ⎡1 1⎤
A = ⎢ B=⎢ ⎥ S = [B AB ] = ⎢ ⎥ S is singular. The system is uncontrollable.
⎣1 0 ⎥⎦ ⎣1⎦ ⎣1 1⎦
Output equation: y= 1 0 x = Cx C= 1 0
⎡1 −2 ⎤
V = ⎡⎣C A C ⎦⎤ = ⎢
' ' '
V is nonsingular. The system is observable.
⎣0 3 ⎥⎦
(b) Transfer function:
Y ( s) s+3 1
= 2
=
U ( s) s + 2s − 3 s −1
Since there is pole‐zero cancellation in the input‐output transfer function, the system is either
uncontrollable or unobservable or both. In this case, the state variables are already defined, and the
system is uncontrollable as found out in part (a).
10‐47 (a) α = 1, 2, or 4 . These values of α will cause pole‐zero cancellation in the transfer function.
(b) The transfer function is expanded by partial fraction expansion,
Y ( s) α −1 α−2 α−4
= − +
R( s ) 3( s + 1) 2( s + 2 ) 6( s + 4 )
By parallel decomposition, the state equations are: x& ( t ) = Ax ( t ) + Br ( t ) , output equation: y ( t ) = Cx( t ).
⎡ −1 0 0 ⎤ ⎡α − 1⎤
A = ⎢ 0 −2 0 ⎥ B = ⎢α − 2 ⎥ D=
⎡1 −
1 1⎤

⎢ ⎥ ⎢ ⎥ ⎢⎣ 3 2 6 ⎥⎦
⎢⎣ 0 0 −4 ⎥⎦ ⎢⎣α − 4 ⎥⎦
The system is uncontrollable for α = 1, or α = 2, or α = 4.
(c) Define the state variables so that
10‐75


⎡1⎤
⎢3⎥
⎡ −1 0 0 ⎤ ⎢ ⎥
−1
A = ⎢ 0 −2 0 ⎥ B=⎢ ⎥ D = [α − 1 α − 2 α − 4]
⎢ ⎥ ⎢2⎥
⎢⎣ 0 0 −4 ⎥⎦ ⎢ −1 ⎥
⎢ ⎥
⎢⎣ 6 ⎥⎦
TThe system is u or α = 1, or α == 2, or α = 4.
unobservable fo
1
10‐48
⎡1 b ⎤
S = [B A ]= ⎢ S = ab − 1 − b ≠ 0
2
ab − 1⎥⎦
AB
⎣b
The boundary of the region of coontrollability iss described by ab − 1 − b 2 = 0.
0
Regions of contro
ollability:

1
10‐49
⎡ b1 b1 + b2 ⎤
S = [B A ]= ⎢ S = 0 when b1b2 − b1b2 − b2 = 0, or b2 = 0
2
AB ⎥
⎣b2 b2 ⎦
Thee system is com ollable when b2 ≠ 0.

mpletely contro
⎡ d1 d2 ⎤
V = ⎡⎣C A C ⎤⎦ = ⎢ V = 0 when d1 ≠ 0.
' ' '

⎣d2 d1 + d 2 ⎥⎦
mpletely observvable when d 2 ≠ 0.
Thee system is com
10‐76


1
10‐50 (a) State equations:
dh 1 K I nN Ko dθ m dω m Ki Kb Ki K a
= (q i
− qo ) = θm − h = ωm =− ωm +
2
ei J = J m + n J L
dt A A A dt d
dt JRa JRa
dθ m
Statte variable: x1 = h, x 2 = θ m , x3 = = ωm
dt
Statte equations: x& = Ax + Bei
⎡ −Ko K I nN ⎤ ⎡ ⎤
⎢ 0 ⎥ ⎡ −1 0.016 0 ⎤ ⎢ 0 ⎥ ⎡ 0 ⎤
⎢ A A
⎥ ⎢ ⎢ ⎥
A=⎢ 0 0 1 ⎥= 0 0 1 ⎥ B=⎢ 0 ⎥=⎢ 0 ⎥
⎢ ⎥ ⎢ ⎥
⎢ K i K b ⎥ ⎢⎣ 0 0 −11..767 ⎥⎦ ⎢ K K ⎥ ⎢8333.33⎥
⎣ ⎦
⎢ 0 0 − ⎥ ⎢ i a
⎥
⎢⎣ JRa ⎦⎥ ⎣⎢ JRa ⎦⎥
Sttate diagram:
(b) Characteristic equ
uation of A:
Ko − K I nN
n
s+ 0
A A
sI − A = −1
⎛
= s⎜s +
Ko⎞ ⎛ s + Ki Kb ⎞
0 s ⎟⎜ ⎟ = s ( s + 1)( s + 11.767)
1
⎝ A ⎠⎝ JRa ⎠
Ki Kb
0 0 s+
JRa
Eige
envalues of A: 0, −1, −11.767.
(c) Conttrollability:
10‐77


⎡ 0 0 133.33 ⎤
S = ⎡⎣ B AB A B ⎤⎦ = ⎢ 0 8333.33 −98058 ⎥ S ≠ 0. The system is controllable.
2

⎢ ⎥
⎣⎢8333.33 −98058 1153848 ⎥⎦
(d) Observability:
(1) C = 1 0 0 :
⎡1 −1 1 ⎤
V = ⎡⎣C ( A ) C ⎤⎦ = ⎢ 0 0.016 −0.016 ⎥ V is nonsingular. The system is observable.
' ' ' ' 2 '
AC
⎢ ⎥
⎣⎢ 0 0 0.016 ⎦⎥

(2) C = 0 1 0 :
⎡0 0 0 ⎤
V = ⎡⎣C ( A ) C ⎤⎦ = ⎢ 1 0 0 ⎥ V is singular. The system is unobservable.
' ' ' ' 2 '
AC
⎢ ⎥
⎣⎢ 0 1 −11.767 ⎦⎥
(3) C = 0 0 1 :
⎡0 0 0 ⎤
V = ⎡⎣C ( A ) C ⎤⎦ = ⎢ 0 0 ⎥ V is singular. The system is unobservable.
' ' ' ' 2 '
AC 0
⎢ ⎥
⎣⎢1 −11.767 138.46 ⎥⎦
∗ 4 2
10‐51 (a) Characteristic equation: Δ( s ) = sI − A = s − 25.92 s = 0
Roots of characteristic equation: −5.0912, 5.0912, 0, 0
(b) Controllability:
⎡ 0 −0.0732 0 −1.8973⎤
⎢ −0.0732 0 −1.8973 0 ⎥
S = ⎡⎣ B
∗
AB
∗ ∗ ∗2
A B
∗
A B ⎤⎦ =
∗3 ∗
⎢ ⎥
⎢ 0 0.0976 0 0.1728 ⎥
⎢ ⎥
⎣ 0.0976 0 0.1728 0 ⎦
∗ ∗
S is nonsingular. Thus, A , B is controllable.
(c) Observability:

10‐78


∗
(1) C = 1 0 0 0
⎡1 0 25.92 0 ⎤
⎢0 1 0 25.92 ⎥
V = ⎡⎣C ( A ) C ⎤⎦ = ⎢ ⎥
∗' ∗' ∗' ∗' 2 ∗' ∗' 3 ∗'
A C (A ) C
⎢0 0 0 0 ⎥
⎢ ⎥
⎣0 0 0 0 ⎦
S is singular. The system is unobservable.
∗
(2) C = 0 1 0 0
⎡ 0 25.92 0 671.85 ⎤
⎢1 0 25.92 0 ⎥
V = ⎡⎣C ( A ) C ⎤⎦ = ⎢ ⎥
∗' ∗' ∗' ∗' ∗' ∗' 3 ∗'
A C (A ) C
2
⎢0 0 0 0 ⎥
⎢ ⎥
⎣0 0 0 0 ⎦
(3) C∗ = 0 0 1 0
⎡0 0 −2.36 0 ⎤
⎢0 0 0 −2.36 ⎥
V = ⎡⎣C ( A ) C ⎤⎦ = ⎢ ⎥
∗' ∗' ∗' ∗' ∗' ∗' 3 ∗'
A C (A ) C
2
⎢1 0 0 0 ⎥
⎢ ⎥
⎣0 1 0 0 ⎦
S is nonsingular. The system is observable.
∗
(4) C = 0 0 0 1
⎡ 0 −2.36 0 −61.17 ⎤
⎢0 0 −2.36 0 ⎥
V = ⎡⎣C
∗'
A C
∗' ∗' ∗'
(A ) C
2 ∗'
( A ) C ⎤⎦ =
∗' 3 ∗'
⎢ ⎥
⎢0 0 0 0 ⎥
⎢ ⎥
⎣1 0 0 0 ⎦
10‐79


1
10‐52 The controllability matrix is
⎡ 0 −1 0 −16 0 3 ⎤
−384
⎢ −1 0 −16 0 −384 0 ⎥
⎢ ⎥
⎢0 0 0 166 0 512 ⎥
S=⎢ ⎥ R
Rank of S is 6. The system is ccontrollable.
0 0 16 0 512 0
⎢ ⎥
⎢0 1 0 0 0 0 ⎥
⎢ ⎥
⎢⎣ 1 0 0 0 0 0 ⎥⎦
1
10‐53 (a) Trransfer functio
on:
Θv ( s) KI H
=
R(s) Jvs
2
(J G
s + KPs + KI + KN
2
)
Statte diagram by direct decomp
position:
Statte equations: x& ( t ) = Ax
A ( t ) + Br ( t )
⎡0 1 0 0 ⎤
⎢0 ⎡0 ⎤
0 1 0 ⎥ ⎢0 ⎥
⎢ ⎥
A = ⎢0 0 0 1 ⎥ B=⎢ ⎥
⎢ ⎥ ⎢0 ⎥
− ( KI + KN ) −KP ⎢ ⎥
⎢0 0 ⎥ ⎣1 ⎦
⎢⎣ JG J G ⎥⎦
(b) Characteristic equ
uation: J v s
2
(J G
s + KP s + KI + KN = 0
2
)

10‐80


1
10‐54 (a) State equations: x& (t ) = Ax
A (t ) + Bu1 (t )
⎡ −3 1⎤ ⎡0 ⎤ ⎡0 1 ⎤
A=⎢ B= S = [B B] = ⎢ 1 −2 ⎥
−2 ⎥⎦ ⎢1 ⎥ AB
⎣0 ⎣ ⎦ ⎣ ⎦
S is nonsinggular. A, B is controllablee.
Output equ
uation: y 2 = Cx C = −1 1
⎡ −1 3 ⎤
V = ⎡⎣C A C ⎦⎤ =
' ' '
⎢ 1 −3 ⎥ V is singu
ular. The system
m is unobservaable.
⎣ ⎦
h feedback, u2 = − kc2 , the sttate equation is: x& (t ) = Ax(t ) + Bu1 (t ) .

(b) With
⎡ −3 − 2 k 1 ⎤ ⎡0 1 ⎤
⎡0 ⎤
A = ⎢ 1+ g 1+ k ⎥ B=⎢ ⎥ S= ⎢ 1+ k ⎥
⎢ ⎥ ⎣1 ⎦ ⎢ ⎥
⎢⎣ 0 −2 ⎥⎦ ⎣ 1 −2 ⎦
S is nonsingular fo
or all finite valu
ues of k. The syystem is contro
ollable.
Statte diagram:

Outtput equation: y2 = Cxx C=
⎡ −1 1 ⎤

⎢⎣ 1 + k 1 + k ⎥⎦
⎡ −1 3 + 2k ⎤
⎢1 + K (1 + k ) ⎥
2
V = ⎡⎣ D' A D ⎤⎦ = ⎢ ⎥
' '

⎢ 1 3 + 2k ⎥
−
⎢⎣ 1 + k (1 + k ) ⎥⎦
2
V is singular for any k. Thee system with feedback is un
nobservable.
10‐81


10‐55 (a)
⎡1 2⎤
S = [B AB ] = ⎢ S is nonsingular. System is controllable.
⎣2 −7 ⎥⎦
⎡1 −1 ⎤
V = ⎡⎣C A C ⎦⎤ = ⎢
' ' '
V is nonsingular. System is observable.
⎣1 −2 ⎥⎦
(b) u = − k1 k 2 x
⎡0 1⎤ ⎡ k1 k2 ⎤ ⎡ − k1 1 − k2 ⎤
A c = A − BK = ⎢ ⎥−⎢ =
⎣ −1 −3 ⎦ ⎣ 2k1 2k 2 ⎥⎦ ⎢⎣ −1 − 2 k1 −3 − 2k 2 ⎥⎦
⎡1 − k1 − 2 k 2 + 2 ⎤
S = [B AcB] = ⎢ S = −11 − 2 k 2 ≠ 0
⎣2 −7 − 2 k1 − 4 k 2 ⎥⎦
11
For controllabillity, k 2 ≠ −
2
⎡ −1 −1 − 3k1 ⎤
V = ⎡⎣C A c C ⎤⎦ = ⎢
' ' '

⎣1 −2 − 3k 2 ⎥⎦
For observability, V = −1 + 3k1 − 3k 2 ≠ 0
10‐56
Same as 10‐21 (a)
10‐57
⎡ 0 ⎤ ⎡ x1 ⎤ ⎡ 0 ⎤
1
=⎢ + θr
⎣ −322.58 −80.65 ⎥⎦ ⎢⎣ x2 ⎥⎦ ⎢⎣322.58 ⎥⎦

10‐82


⎡ 0 1 ⎤
A=⎢ ⎥
⎣ −322.58 −80.65⎦
⎡ x& ⎤ ⎡ 0 1 ⎤ ⎡ x1 ⎤ ⎡ 0 ⎤ ⎡ 0 ⎤
From 10‐22 ⎢ 1 ⎥ = ⎢ ⎥⎢ ⎥+⎢ ⎥ θr B=⎢ ⎥
⎣ x&2 ⎦ ⎣ −322.58 −80.65⎦ ⎣ x2 ⎦ ⎣322.58⎦ ⎣322.58⎦
C = [1 0]
D=0
Use the state space tool of ACSYS
10‐83


The A matrix is:
Amat =
0 1.0000
‐322.5800 ‐80.6500

ans =
s^2+1613/20*s+16129/50

Abar =
‐4.2206 0
0 ‐76.4294
Eigen Vectors are
T =
0.2305 ‐0.0131
‐0.9731 0.9999
10‐84



a =
x1 x2
x1 0 1
x2 ‐322.6 ‐80.65

b =
u1
x1 0
x2 322.6

c =
x1 x2
y1 1 0

d =
u1
y1 0


ans =
s^2+1613/20*s+16129/50

Transfer function:
322.6
‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐
s^2 + 80.65 s + 322.6
Pole, Zero Form:
Zero/pole/gain:
322.58
‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐
(s+76.43) (s+4.221)
10‐85


10‐58
Closed‐loop System Transfer Function.
Y (s) 1
=
s + ( 4 + k3 ) s + ( 3 + k 2 + k3 ) s + k1
3 2
R(s)
For zero steady‐state error to a step input, k1 = 1. For the complex roots to be located at −1 +j and −1 − j,
2
we divide the characteristic polynomial by s + 2 s + 2 and solve for zero remainder.
s + ( 2 + k2 )
s + 2 s + 2 s + ( 4 + k3 ) s + ( 3 + k 2 + k3 ) s + 1
2 3 2

s + + 2s
3 2
2s
(2 + k ) s
3
2
+ (1 + k 2 + k 3 ) s + 1
10‐86


(2 + k ) s
3
2
+ ( 4 + 2 k3 ) s + 4 + 2 k3

( -3+k 2
− k3 ) s − 3 − 2k3
For zero remainder, − 3 − 2 k3 = 0 Thus k3 = −15

.
−3 + k 2 − k3 = 0 Thus k 2 = 15
.
The third root is at −0.5. Not all the roots can be arbitrarily assigned, due to the requirement on the
steady‐state error.
10‐59 (a) Open‐loop Transfer Function.
X 1 (s) k3
G ( s) = =
s ⎡⎣ s + ( 4 + k 2 ) s + 3 + k1 + k 2 ⎤⎦
2
E (s)
Since the system is type 1, the steady‐state error due to a step input is zero for all values of k1 , k 2 , and k3
that correspond to a stable system. The characteristic equation of the closed‐loop system is
s + ( 4 + k 2 ) s + ( 3 + k1 + k 2 ) s + k3 = 0
3 2

For the roots to be at −1 + j, −1 − j, and −10, the equation should be:
3 2
s + 12 s + 22 s + 20 = 0
Equating like coefficients of the last two equations, we have
4 + k 2 = 12 Thus k2 = 8
3 + k1 + k 2 = 22 Thus k1 = 11
k3 = 20 Thus k3 = 20
(b) Open‐loop Transfer Function.
10‐87


Y (s) Gc ( s ) 20 20 ( s + 1)( s + 3 )
= = Thus Gc ( s ) =
E ( s) ( s + 1)( s + 3 ) (
s s + 12 s + 22
2
) (
s s + 12 s + 22
2
)

10‐60 (a)
⎡ 0 1 0 0⎤ ⎡ 0 ⎤
⎢ 25.92 0 0 0⎥ ⎢ −0.0732 ⎥
A = ⎢ ⎥ B =⎢ ⎥
∗ ∗

⎢ 0 0 0 1⎥ ⎢ 0 ⎥
⎢ ⎥ ⎢ ⎥
⎣ −2.36 0 0 0⎦ ⎣ 0.0976 ⎦

The feedback gains, from k1 to k 4 :
−2.4071E+03 −4.3631E+02 −8.4852E+01 −1.0182E+02
The A ∗ − B ∗ K matrix of the closed‐loop system
0.0000E+00 1.0000E+00 0.0000E+00 0.0000E+00
−1.5028E+02 −3.1938E+01 −6.2112E+00 −7.4534E+00
0.0000E+00 0.0000E+00 0.0000E+00 1.0000E+00
2/3258E+02 4.2584E+01 8.2816E+00 9.9379E+00
The B vector
0.0000E+00
−7.3200E−02
0.0000E+00
10‐88


9.7600E−02
Time Responses:
10‐60 (b) The feedback gains, from k1 to k2 :
−9.9238E+03 −1.6872E+03 −1.3576E+03 −8.1458E+02
0.0000E+00 1.0000E+00 0.0000E+00 0.0000E+00
−7.0051E+02 −1.2350E+02 −9.9379E+01 −5.9627E+01
0.0000E+00 0.0000E+00 0.0000E+00 1.0000E+00
9.6621E+02 1.6467E+02 1.3251E+02 7.9503E+01

10‐89


The B vector
0.0000E+00
−7.3200E−02
0.0000E+00
9.7600E−02
Time Responses:
10‐61 The solutions using MATLAB

10‐90


(a) The feedback gains, from k1 to k2 :
−6.4840E+01 −5.6067E+00 2.0341E+01 2.2708E+00
0.0000E+00 1.0000E+00 0.0000E+00 0.0000E+00
−3.0950E+02 −3.6774E+01 1.1463E+02 1.4874E+01
0.0000E+00 0.0000E+00 0.0000E+00 1.0000E+00
−4.6190E+02 −3.6724E+01 1.7043E+02 1.477eE+01
The B vector
0.0000E+00
−6.5500E+00
0.0000E+00
−6.5500E+00
'
(b) Time Responses: Δx(0) = 01
. 0 0 0
10‐91


'
. 0 0 0 , the initial position of Δx1 or Δy1 is preturbed downward
With the initial states Δx(0) = 01
from its stable equilibrium position. The steel ball is initially pulled toward the magnet, so Δx3 = Δy 2 is
negative at first. Finally, the feedback control pulls both bodies back to the equilibrium position.
'
. 0 , the initial position of Δx3 or Δy 2 is preturbed
With the initial states Δx(0) = 0 0 01
downward from its stable equilibrium. For t > 0, the ball is going to be attracted up by the magnet
toward the equilibrium position. The magnet will initially be attracted toward the fixed iron plate, and
then settles to the stable equilibrium position. Since the steel ball has a small mass, it will move more
10‐92


actively.
10‐62 (a) Block Diagram of System.
u = − k1 x1 + k 2 ∫ ( − x1 + w1 ) dt
State Equations of Closed‐loop System:

⎡ dx1 ⎤
⎢ dt ⎥ ⎡ −2 − k1 1 ⎤ ⎡ x1 ⎤ ⎡0 1 ⎤ ⎡ w1 ⎤
⎢ ⎥=⎢ ⎥ ⎢ ⎥ +⎢
⎢ dx2 ⎥ ⎣ − k 2 0 ⎦ ⎣ x2 ⎦ ⎣ k 2 0 ⎥⎦ ⎢⎣ w2 ⎥⎦
⎢⎣ dt ⎥⎦


10‐93


s + 2 + k1 −1
sI − A = = s + ( 2 + k1 ) s + k 2 = 0
2

k2 s

For s = −10, −10, sI − A = s 2 + 20 s + 200 = 0 Thus k1 = 18 and k 2 = 200
200W1 ( s ) s −2 + s −1W2 ( s ) 200W1 ( s) + sW2 ( s)

X ( s) = X 1 ( s) = −1 −1 −2
=
1 + 2s + 18 s + 200 s s 2 + 20 s + 200
1 W2
W1 ( s ) = W2 ( s) = W2 = constant
s s
200 + W2 s
X ( s ) = lim x (t ) = lim sX ( s ) = 1
(
s s + 20 s + 200
2
) t →∞ s →0
10‐62 (b) With PI Controller:
Block Diagram of System:
(K s + K I ) W1 ( s ) + sW2 ( s ) ( 2 s + 200 ) W ( s ) + sW ( s )
Set K P = 2 and K I = 200 . X (s) = P
= 1 2

s + 20 s + 200 s + 20 s + 200
2 2

10‐94


Time Responses:
10‐95


10-63)
10 10
1 2 3 6 11 6
Consider:
10 6 11 6
Therefore:
0 1 0 0
0 0 1 0
6 11 6 10
1 0 0
As a result:
0 1 0 0
0 0 1 0 1 0 0 0
6 11 6 10
Using MATLAB, we’ll find:

15.4 4.5 0.8
10-64)
Inverted Pendulum on a cart
The equations of motion from Problem 4-21 are obtained (by ignoring all the pendulum inertia term):
x − mlθ&& cos θ + mlθ& 2 sin θ = f

( M + m) &&

x cos θ + lθ&&) = 0
ml (− g sin θ − &&
These equations are nonlinear, but they can be linearized. Hence
θ ≈0
cos θ ≈ 1
sin θ ≈ θ
x + mlθ&& = f
( M + m) &&

x + lθ&&) = 0
ml (− gθ − &&
10‐96


Or
⎡( M + m) ml ⎤ ⎡ &&
x⎤ ⎡ f ⎤
⎢ − ml =
ml ⎦ ⎣θ ⎦ ⎢⎣ mlgθ ⎥⎦
2 ⎥ ⎢ &&⎥

⎣
Pre‐multiply by inverse of the coefficient matrix
inv([(M+m),m*l;‐m*l,m*l^2])
ans =
[ 1/(M+2*m), ‐1/l/(M+2*m)]
[ 1/l/(M+2*m), (M+m)/m/l^2/(M+2*m)]
For values of M=2, m=0.5, l=1, g=9.8
ans =
0.3333 ‐0.3333
0.3333 1.6667
Hence
x ⎤ ⎡1/ 3 −1/ 3⎤ ⎡ f ⎤
⎡ &&
⎢θ&&⎥ = ⎢1/ 3 5 / 3 ⎥ ⎢ 49 /10θ ⎥
⎣ ⎦ ⎣ ⎦⎣ ⎦
x ⎤ ⎡1/3*f-49/30θ ⎤
⎡ &&
⎢θ&&⎥ = ⎢ 1/3*f+49/6θ ⎥
⎣ ⎦ ⎣ ⎦
The state space model is:
⎡ x&4 ⎤ ⎡1/3*f-49/30x1 ⎤
⎢ x& ⎥ = ⎢ 1/3*f+49/6x ⎥
⎣ 2⎦ ⎣ 1⎦
Or:
10‐97


⎡ x&2 ⎤ ⎡ 1/3*f+49/6x1 ⎤
⎢ x& ⎥ = ⎢1/3*f-49/30x ⎥
⎣ 4⎦ ⎣ 1⎦
⎡ x&1 ⎤ ⎡ 0 1 0 0 ⎤ ⎡ x1 ⎤ ⎡ 0 ⎤
⎢ x& ⎥ ⎢ 49/6 0 0 0 ⎥⎥ ⎢⎢ x2 ⎥⎥ ⎢⎢1/ 3⎥⎥
⎢ 2⎥ = ⎢ + f
⎢ x&3 ⎥ ⎢ 0 0 0 1 ⎥ ⎢ x3 ⎥ ⎢ 0 ⎥
⎢ ⎥ ⎢ ⎥⎢ ⎥ ⎢ ⎥
⎣ x&4 ⎦ ⎣-49/30 0 0 0 ⎦ ⎣ x4 ⎦ ⎣1/ 3⎦
⎡ 0 1 0 0⎤
⎢ 49/6 0 0 0 ⎥⎥
A=⎢
⎢ 0 0 0 1⎥
⎢ ⎥
⎣-49/30 0 0 0⎦
⎡ 0 ⎤
⎢1/ 3⎥
B=⎢ ⎥
⎢ 0 ⎥
⎢ ⎥
⎣1/ 3⎦
C = [1 0 1 0]
D=0
10‐98


Use ACSYS State tool and follow the design process stated in Example 10-17-1:
10‐99


The A matrix is:

Amat =

0 1.0000 0 0
8.1667 0 0 0
0 0 0 1.0000
‐1.6333 0 0 0


ans =

s^4‐49/6*s^2


Abar =

0 0 0 0
0 0 0 0
0 0 2.8577 0
0 0 0 ‐2.8577

Eigen Vectors are

T =

0 0 0.3239 ‐0.3239
0 0 0.9256 0.9256
1.0000 ‐1.0000 ‐0.0648 0.0648
0 0.0000 ‐0.1851 ‐0.1851


a =
x1 x2 x3 x4
x1 0 1 0 0
x2 8.167 0 0 0
x3 0 0 0 1
x4 ‐1.633 0 0 0

b =
u1
x1 0
x2 0.3333
10‐100


x3 0
x4 0.3333

c =
x1 x2 x3 x4
y1 1 0 1 0

d =
u1
y1 0


ans =

s^4‐49/6*s^2


Transfer function:
4.441e‐016 s^3 + 0.6667 s^2 ‐ 2.22e‐016 s ‐ 3.267
‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐
s^4 ‐ 8.167 s^2

Pole, Zero Form:

Zero/pole/gain:
4.4409e‐016 (s+1.501e015) (s+2.214) (s‐2.214)
‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐‐
s^2 (s‐2.858) (s+2.858)


Smat =

0 0.3333 0 2.7222
0.3333 0 2.7222 0
0 0.3333 0 ‐0.5444
0.3333 0 ‐0.5444 0

The system is therefore Not Controllable, rank of S Matrix is =

rankS =

4
10‐101


Mmat =

0 ‐8.1667 0 1.0000
‐8.1667 0 1.0000 0
0 1.0000 0 0
1.0000 0 0 0


Ptran =

0 0 0.3333 0
0 0 0 0.3333
‐3.2667 0 0.3333 0
0 ‐3.2667 0 0.3333


Abar =

0 1.0000 0 0
0 0 1.0000 0
0 0 0 1.0000
0 0 8.1667 0

Bbar =

0
0
0
1

Cbar =

‐3.2667 0 0.6667 0

Dbar =

0
10‐102


Note incorporating –K in Abar:
Abar K=

0 1.0000 0 0
0 0 1.0000 0
0 0 0 1.0000
‐k1 ‐k2 8.1667‐k3 ‐k4

System Characteristic equation is:
‐k4*s^4+(8.1667‐k3 )*s^3‐k2*s‐k1=0
From desired poles we have:
>> collect((s-210)*(s-210)*(s+20)*(s-12))
ans =
-10584000+s^4-412*s^3+40500*s^2+453600*s
Hence: k1=10584000, k2=40500, k3=412+8.1667and k4=1
10-65) If 3 and 0.707, then 1.414. The 2nd order desired characteristic equation of the
system is:
2 2 0 1
On the other hand:
0 1
6 5
where the characteristic equation would be:
5 6 0 (2)
Comparing equation (1) and (2) gives:
5 2
6 2
10‐103


which means 4 and 3
10-66) Using ACSYS we can convert the system into transfer function form.

1) Activate MATLAB
3) Type in Acsys
8) Next verify Controlability and find the A matrix
9) Follow the design procedures in section 10-17 (pole placement)
10‐104


10‐105


The A matrix is:
Amat =
-1 -2 -2
0 -1 1
1 0 -1
ans =
s^3+3*s^2+5*s+5
Abar =
-0.6145 + 1.5639i 0 0
0 -0.6145 - 1.5639i 0
0 0 -1.7709
Eigen Vectors are
T=
-0.8074 -0.8074 -0.4259
0.2756 + 0.1446i 0.2756 - 0.1446i -0.7166
-0.1200 + 0.4867i -0.1200 - 0.4867i 0.5524
State-Space Model is:
a=
x1 x2 x3
x1 -1 -2 -2
x2 0 -1 1
x3 1 0 -1
10‐106


b=
u1
x1 2
x2 0
x3 1
c=
x1 x2 x3
y1 1 1 1
d=
u1
y1 0
Continuous-time model.
ans =
s^3+3*s^2+5*s+5
Transfer function:
3 s^2 + 7 s + 4
---------------------
s^3 + 3 s^2 + 5 s + 5
Pole, Zero Form:
Zero/pole/gain:
3 (s+1.333) (s+1)
---------------------------------
(s+1.771) (s^2 + 1.229s + 2.823)

10‐107


Smat =
2 -4 0
0 1 0
1 1 -5
The system is therefore Controllable, rank of S Matrix is =
rankS =
Mmat =
5 3 1
3 1 0
1 0 0
Ptran =
-2 2 2
3 1 0
3 4 1
Abar =
0 1.0000 0
0 0 1.0000
-5.0000 -5.0000 -3.0000
10‐108


Bbar =
Cbar =
4 7 3
Dbar =
Using Equation (10-324) we get:
sI − ( A − BK ) = s 3 + (3 + k3 ) s 2 + (5 + k2 ) s + (5 + k1 ) = 0
Using a 2nd order prototype system, for 5, then 1. For overshoot of 4.33%, 0.707.
nd
Then the desired 2 order system will have a characteristic equation:
s 2 + 2ζωn s + ωn 2 = s 2 + 2 s + 2 = 0
The above system poles are: s1,2 = −1 ± j
One approach is to pick K=[k1 k2 k3] values so that two poles of the system are close to the desired
second order poles and the third pole reduces the effect of the two system zeros that are at z=-1.333 and
z=-1. Let’s set the third pole at s=-1.333. Hence
(s+1.333)*(s^2+2*s+2)= s^3+3.33*s^2+4.67*s+2.67
and K=[-2.37 -0.37 0.33].
Y 3( s + 1)
= 2
R s + 2s + 2
Use ACSYS control tool to find the time response. First convert the transfer function to a unity feedback
system to make compatible to the format used in the Control toolbox.
3( s + 1)
G=
s2 − s −1
10‐109


Overshoot is about 2%. You can adjust K values to obtain alternative results by repeating this process.
10-67) a) According to the circuit:
If , , then
10‐110


or
2 1
Therefore:
2 1
1 1
0
0
b) Uncontrollability condition is:
According to the state-space of the system, C is calculated as:
2 1
As 0, because , then the system is controllable
c) Unobservability condition is:
According to the state-space of the system, C is calculated as:
Since det H 0, because R 0 or L ∞, then the system is observable.
10‐111


d) The same as part (a)
1 1
0
1
0
1 1
1
For controllability, we define G as:
1 1
If ,and then 0, which means the system is not controllable.
For observability, we define H as:
1
1
1
If , then 0, which means the system is not observable.
10‐112

B.C Kuo Solutions 1

Uploaded by

Copyright:

Available Formats

B.C Kuo Solutions 1

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

B.C Kuo Solutions 1

Uploaded by

Copyright:

Available Formats

Ninth Edition

Farid Golnaraghi • Benjamin C. Kuo

'Generated transfer function:'

'Generated transfer function:'

'Generated transfer function:'

'Generated transfer function:'

In all cases substitute and simplify. The use MATLAB to verify.

'Generated transfer function:'

'Generated transfer function:'

'Generated transfer function:'

'Generated transfer function:'

lim ω →0 G ( jω ) = 5; φ = tan −1 1 = −90o

zeta=0.5; %asuuming a value for zeta <1

wn=2*pi*10 %asuuming a value for wn

zeta=1.5; %asuuming a value for zeta >1

T=3.5 %assuming value for parameter T

Steps for plotting |G|:

(1) For ω < 0.1, asymptote is

Break point: ω = 0.5

Steps for plotting G

b) Let’s convert the transfer function to the following form:

Steps for plotting |G|:

(1) Asymptote: ω < 1 |G(jω)| 2.5 / ω

Steps for plotting G(jω):

c) Convert the transfer function to the following form:

(1) the breakpoint: ω = 1 and slope is zero

Steps for plotting G:

(1) phase starts from -180o due to

Steps for plotting the |G|:

(1) Asymptote for <1 is zero

(2) Breakpoint: = 1, slope = -1 or -10 dB/decade

Steps for plotting G:

at = 5. For > 5, the phase remains at -180o.

'Generated transfer function:'

'Generated transfer function:'

'Generated transfer function:'

'Generated transfer function:'

'Generated transfer function:'

'from pade approzimation:'

2-10) We know that:

Partial integration of equation (1) gives:

Differentiation of both sides of equation (1) with respect to s gives:

Comparing with equation (1), we conclude that:

2-11) Let g(t) = ∞

Using Laplace transform and differentiation property, we have X(s) = sG(s)

Therefore G(s) = , which means:

2-12) By Laplace transform definition:

Now, consider τ = t - T, then:

f(t) = g1(t) g2(t) =

By Laplace transform definition:

By using time shifting theorem, we have:

Let’s consider g(t) = g1(t) g2(t)

By inverse Laplace Transform definition, we have

2-14) a) We know that

When s Æ ∞ , it can be written as:

b) By Laplace transform differentiation property:

L2= 1/s - 8/s/(s^2+16)

'verified as L2 equals L3'

2‐16) (a) (b) (c)

%section (e) requires assignment of T and a numerical loop calculation

wn=2pi10 %asuuming a value for wn

f7=(2+2spade(exp(-1s),1)+4pade(exp(-2s),1))/(s^2+3s+2) %using Pade command

f7n=(-2s^3+6s+12)/(s^4+6s^3+13s^2+12*s+4) %generating sumbolic function for

Eq=5(pade(exp(-2s),1))/(s^2+s+1); %Pade approximation oreder 1 used