Chapter 9: Frequency Response Analysis

9.1 To find the sinusoidal transfer function G(j), replace s with j

4
G ( j ) 
0.2 j  1

4
or G ( j ) 
1  j 0.2

Copyright © 2016 John Wiley & Sons

 Chapter 9

9.2 To find the sinusoidal transfer function G(j), replace s with j

3 j  1
G ( j ) 
2( j ) 2  6 j  40

Substituting j2 = –1 we obtain

1  j 3
G ( j ) 
40  2 2  j 6

2
Copyright © 2016 John Wiley & Sons
 Chapter 9

9.3 First, we determine the sinusoidal transfer function of the given I/O equation:

Y (s) 3 3
G ( s)    the sinusoidal transfer function is G ( j ) 
U ( s) 2s  10 10  j 2

We know from Eq. (9-17) that the frequency response has the form

yss (t )  G( j) U 0 sin( t   ) where  = 4 rad/s, U0 = 18

We use the sine function because the input is a sine function. Finally, we need the magnitude and
phase angle of the sinusoidal transfer function G(j) at input frequency  = 4 rad/s

32  02 3
Magnitude: G( j 4)    0.2343
10  8
2 2
164

0 8
Phase angle:   arg[ 3]  arg[10  j8]  tan 1    tan 1    –38.66 deg or –0.6747 rad
3  10 

Because the input magnitude is U0 = 18, the frequency response is

yss (t )  4.2167 sin( 4t  0.6747)

3
Copyright © 2016 John Wiley & Sons
 Chapter 9

9.4 The sinusoidal transfer function of the given I/O equation:

0.8s  2 2  j 0.8
G ( s)   the sinusoidal transfer function is G ( j ) 
4 s  s  30
2
30  4 2  j

We know from Eq. (9-18) that the frequency response has the form

yss (t )  G( j) U 0 cos(t   ) where  = 5 rad/s, U0 = 0.3

We use the cosine function because the input is a cosine function. We need the magnitude and
phase angle of the sinusoidal transfer function G(j) at input frequency  = 5 rad/s

22  42 20
Magnitude: G ( j 5)    0.0637
(70)  5
2 2
4925

arg[ 2  j 4] 4  5 
Phase angle:    tan 1    tan 1    63.43 – 175.91 = –112.48 deg
arg[ 70  j 5] 2   70 

Because the input magnitude is U0 = 0.3, the frequency response is

yss (t )  0.0191 cos(5t  1.9631)

4
Copyright © 2016 John Wiley & Sons
 Chapter 9

9.5 The sinusoidal transfer function G(j) is obtained by replacing s with j

2s  3 2 j  3 3  j 2
G(s)   G ( j )  
s 2  8s ( j )  8 j   2  j8
2

32  42 25
Magnitude: G ( j 2)   = 0.3032
(4)  16
2 2
272

arg[ 3  j 4] 4  16 
Phase angle:    tan 1    tan 1    53.13 – 104.04 = –50.91 deg
arg[ 4  j16] 3 4

5
Copyright © 2016 John Wiley & Sons
 Chapter 9

9.6 The sinusoidal transfer function is

0.5 j  0.4 0.4  j 0.5

G ( j )  or G ( j ) 
( j ) 2  6 j  8 8   2  j 6

We know from Eq. (9-17) that the frequency response has the form

yss (t )  G( j) U 0 sin( t   ) where  = 5 rad/s, U0 = 2.7

We use the sine function because the input is a sine function. We need the magnitude and phase
angle of the sinusoidal transfer function G(j) at input frequency  = 5 rad/s

0.42  2.52 6.41

Magnitude: G ( j 5)    0.0734
(17) 2  302 1189

arg[ 0.4  j 2.5]  2 .5  1  30 

Phase angle:    tan 1    tan    80.91 – 119.54 = –38.63 deg
arg[ 17  j 30]  0.4    17 

Because the input magnitude is U0 = 2.7, the frequency response is

yss (t )  0.1982 sin( 5t  0.6742)

6
Copyright © 2016 John Wiley & Sons
 Chapter 9

9.7 The sinusoidal transfer function is

2.4 2.4
G ( j )  or G ( j ) 
0.6( j )  8 j  36
2
36  0.6 2  j8

We know from Eq. (9-18) that the frequency response has the form

yss (t )  G( j) U 0 cos(t   ) where  = 20 rad/s, U0 = 30.2

We use the cosine function because the input is a cosine function. We need the magnitude and
phase angle of the sinusoidal transfer function G(j) at input frequency  = 20 rad/s

2.42  02 2.4
Magnitude: G( j 20)    0.00926
(204) 2  1602 67,216

arg[ 2.4]  0  1  160 

Phase angle:    tan 1    tan    –141.89 deg (–2.477 rad)
arg[ 204  j160]  2 .4    204 

Because the input magnitude is U0 = 30.2, the frequency response is

yss (t )  0.2796 cos( 20t  2.4765)

7
Copyright © 2016 John Wiley & Sons
 Chapter 9

9.8 The sinusoidal transfer function is

2 j  6 6  j 2
G ( j )  or G ( j ) 
( j )3  2 j  26 26  j (2   3 )

We know from Eq. (9-17) that the frequency response has the form

yss (t )  G( j) U 0 sin( t   ) where  = 3 rad/s, U0 = 0.6

We use sine because the input is a sine function. We need the magnitude and phase angle:

62  62 72
Magnitude: G( j3)    0.2539
26  (21)
2 2
1117

arg[ 6  j 6] 6   21 
Phase angle:    tan 1    tan 1    45 – – 38.93 = 83.93 deg
arg[ 26  j 21] 6  26 

Because the input magnitude is U0 = 0.6, the frequency response is

yss (t )  0.1523 sin( 3t  1.4648)

8
Copyright © 2016 John Wiley & Sons
 Chapter 9

9.9 a) We begin with the sinusoidal transfer function

0.6 0.6
G ( j )  or G ( j ) 
0.25 j  1 1  j 0.25

The magnitude in decibels (dB) is computed by taking 20log10 of G(j)

G( j ) dB  20 log 10 G( j )  20 log 10 (0.6)  20 log 10 12  (0.25 )2 (A)

For very low frequencies we have   0 so Eq. (A) becomes

G( j) dB  20 log 10 (0.6)  20 log 10 (1) = 20log10(0.6) dB

b) For very high frequencies we have    and Eq. (A) becomes

G( j) dB  20 log 10 (0.6)  20 log 10 (0.25)

or, G( j) dB  20 log 10 (0.6)  20 log 10 (0.25)  20 log 10 ()

Re-write the previous equation as G( j ) dB  20 log 10 (2.4)  20 log 10 ( ) (B)

Equation (B) is a straight line plotted on a log-log scale where the x-axis is log10(). Hence
the slope of the asymptote for    is –20 dB/decade.

c) The flat (low-frequency) asymptote is 20log10(0.6). The sloping high-frequency asymptote is

20 log 10 (2.4)  20 log 10 ( ) . Set these two asymptotes equal to determine the intersection.

20 log 10 (0.6)  20 log 10 (2.4)  20 log 10 ( ) or 20 log 10 ( )  20 log 10 (2.4)  20 log 10 (0.6)

Re-write as 20 log 10 ( )  20 log 10 (2.4 / 0.6)

Therefore the two asymptotes intersect at corner frequency c = 2.4/0.6 = 4 rad/s

9
Copyright © 2016 John Wiley & Sons
 Chapter 9

9.10 The transfer function for the RL system is

1 1
G(s)   (using L = 0.02 H and R = 1.5 )
Ls  R 0.02 s  1.5

By definition the bandwidth is the frequency range 0    B where the magnitude remains
within 3 dB from its DC gain. To compute magnitude we need the sinusoidal transfer function

1 1
G ( j )  or G ( j )  (A)
0.02 j  1.5 1.5  j 0.02

The magnitude of the DC gain (in dB) is 20log10(1/1.5) = –3.5218 dB. Hence we want to find
the cutoff frequency B where the magnitude is –6.5218 dB. Therefore the absolute value of the
sinusoidal transfer function at the cutoff frequency is 106.5218/ 20 = 0.47196.

The magnitude of G(j) is determined by using Eq. (A)

12  02 1
G( j )   (B)
1.52  (0.02 )2 2.25  0.0004 2

Setting Eq. (B) to 0.47196 and solving for  yields the cutoff frequency B = 74.822 rad/s. The
cutoff frequency in Hz is determined by dividing rad/s by 2.

Bandwidth: 0 <  < 11.91 Hz

10
Copyright © 2016 John Wiley & Sons
 Chapter 9

9.11 The mathematical model of the mechanical system is mx  bx  kx  kxin (t )

X ( s) k
 the transfer function is G ( s )  
X in ( s ) ms  bs  k
2

Using m = 2 kg, b = 20 N-s/m and k = 500 N/m the sinusoidal transfer function becomes
500
G ( j ) 
500  2 2  j 20

We know from Eq. (9-17) that the frequency response has the form

xss (t )  G( j) U 0 sin( t   ) where  = 50 rad/s, U0 = 0.04

We use sine because the input is a sine function. We need the magnitude and phase angle:

5002  02 500
Magnitude: G ( j 50)    0.1085
(4500)  1000
2 2
2.125(107 )

arg[ 500  j 0]  0  1  1000 

Phase angle:    tan 1    tan    –167.47 deg
arg[ 4500  j1000]  500    4500 

Because the input magnitude is U0 = 0.04, the frequency response is

xss (t )  0.0043sin( 50t  2.9229) m

11
Copyright © 2016 John Wiley & Sons
 Chapter 9

9.12 The DC gain of the transfer function in Problem 9.11 is k/k = 1, or 0 dB. At the cutoff
frequency (bandwidth) the magnitude is –3 dB or 10-3/20 = 0.708. The magnitude of the
sinusoidal transfer function in Problem 9.11 is set equal to 0.708

5002  02 500
G( jB )    0.708
(500  2 )  (20B )
2 2
B
2
4  1600B2  250,000
4
B

Solving for bandwidth we obtain B = 22.80 rad/s

The resonant frequency for a second-order system is  r   n 1  2 2

For the given system we have n  k / m  15.8114 rad/s and   b / 2 km  0.3162

Hence, the resonant frequency is r = 14.1421 rad/s

The maximum transmissibility occurs at the resonant frequency; therefore we evaluate the
magnitude of the sinusoidal transfer function at  = r = 14.1421 rad/s

500
G ( j r )   1.6667  max TR = 1.6667
4  1600r2  250,000
4
r

12
Copyright © 2016 John Wiley & Sons
 Chapter 9

9.13 The mathematical model of the mechanical system is mx  (b1  b2 ) x  kx  f a (t )

The transfer function and sinusoidal transfer function are

X (s) 1 1
G (s)   and G ( j ) 
Fa ( s ) 0.6 s  3.4 s  80
2
80  0.6 2  j 3.4

a) We know from Eq. (9-17) that the frequency response has the form

xss (t )  G( j) U 0 sin( t   ) where  = 8 rad/s, U0 = 2

We use sine because the input is a sine function. We need the magnitude and phase angle:

12  02 1
Magnitude: G( j8)    0.0201
41.62  27.22 2470.4

arg[1  j 0] 0  27.2 

Phase angle:    tan 1    tan 1    –33.1785 deg
arg[ 41.6  j 27.2] 1  41.6 

Because the input magnitude is U0 = 2, the frequency response is

xss (t )  0.0402 sin( 8t  0.5791) m

b) The largest output amplitude will occur when  = r (resonant frequency). The resonant
frequency is  r   n 1  2 2 .

For the system we have n  k / m  11.5470 rad/s and   (b1  b2 ) / 2 km  0.2454

The resonant frequency is r = 10.8295 rad/s.

13
Copyright © 2016 John Wiley & Sons
 Chapter 9

9.14 The transfer function and sinusoidal transfer function from Problem 9.2 are

3s  1 1  j 3
G(s)  and G ( j ) 
2 s  6 s  40
2
40  2 2  j 6

The MATLAB commands to compute the magnitude and phase angle with  = 2 rad/s are

>> w = 2; % Input frequency, rad/s

>> numG = 1 + j*3*w; % Numerator of G ( j ) (complex)
>> denG = 40 – 2*w^2 + j*6*w; % Denominator of G ( j ) (complex)
>> magG = abs(numG)/abs(denG); % Compute the magnitude (absolute value)
>> phiG = angle(numG)–angle(denG); % Compute the phase angle (in radians)

Executing the above commands yields magG = 0.1780 and phiG = 1.0469 rad

We can verify this solution using MATLAB’s bode command as follows:

>> sysG = tf([3 1],[2 6 40]); % Define transfer function G(s)

>> w = 2; % Input frequency, rad/s
>> [mag,phase] = bode(sysG,w); % Magnitude and phase at  = 2 rad/s

Executing the above commands yields mag = 0.1780 and phase = 59.982 deg ( = 1.0469 rad)

14
Copyright © 2016 John Wiley & Sons
 Chapter 9

9.15 The following MATLAB commands create the Bode diagram for the transfer function
G(s) for Problem 9.11

>> sysG = tf(500,[2 20 500]); % define transfer function G(s)

>> bode(sysG) % create and plot Bode diagram

Because the system input is xin (t )  0.04 sin 50t (m) we estimate the magnitude and phase angle
using the Bode diagram for frequency  = 50 rad/s: Magnitude = –20 dB, Phase = –165 deg (or,
–2.88 rad). The absolute-value magnitude is 10-20/20 = 0.1. The frequency response is

xss (t )  0.004 sin( 50t  2.88) m

Bode Diagram
20

0
 = 50 rad/s
Magnitude (dB)

-20

-40

-60

-80
0

-45
Phase (deg)

-90

-135

-180
0 1 2 3
10 10 10 10
Frequency (rad/sec)

A more accurate calculation of the frequency response can be obtained by using MATLAB’s
bode command with input  = 50 rad/s:

>> sysG = tf(500,[2 20 500]); % define transfer function G(s)

>> w = 50; % define input frequency  = 50 rad/s
>> [mag,phase] = bode(sysG,w) % compute magnitude and phase

The result is mag = 0.1085 and phase = –167.47 deg (–2.9229 rad). Hence a more accurate
frequency response is
xss (t )  0.0043sin( 50t  2.9229) m

15
Copyright © 2016 John Wiley & Sons
 Chapter 9

9.16 The bode diagram is computed using the MATLAB commands and system G(s)

>> sysG = tf(500,[2 20 500]); % define transfer function G(s)

>> bode(sysG) % create and plot Bode diagram

We estimate the cutoff frequency where we observe the 3 dB drop from the DC gain (0 dB) to be
approximately B = 23 rad/s (bandwidth). The peak magnitude occurs at magnitude of roughly
4.5 dB at a frequency of 15 rad/s. Hence the resonant frequency is about r = 15 rad/s and the
peak transmissibility is about 104.5/20 = 1.68.

Bode Diagram
20

0 3 dB drop
Magnitude (dB)

-20

-40

-60

-80
0

-45
Phase (deg)

-90

-135

-180
0 1 2 3
10 10 10 10
Frequency (rad/sec)

16
Copyright © 2016 John Wiley & Sons
 Chapter 9

9.17 The MATLAB commands to simulate the frequency response are:

>> sysG = tf(500,[2 20 500]); % define transfer function G(s)

>> t = 0:0.001:1; % time vector
>> w = 50; % input frequency
>> x_in = 0.04*sin(w.*t); % input displacement xin(t)
>> [x,t] = lsim(sysG,x_in,t); % Simulate frequency response x(t)
>> plot(t,x_in,t,x) % Plot input and output

The plot (below) shows the frequency response. The solid line is the input xin (t )  0.04 sin 50t
(m) and the dashed line is the position of the mass x(t) (in m). We see that at steady state the
response x(t) has an amplitude of about 0.004 m and a period equal to the input (i.e., input
frequency  = 50 rad/s). The “peaks” of the response x(t) are nearly aligned with the “valleys” of
the input xin(t) and hence the output is nearly (but not quite) 180 deg out-of-phase with the input
(the phase is about –165 deg). Therefore an estimate of the frequency-response equation is

xss (t )  0.004 sin( 50t  2.9) m

0.06 Input x (t)

in
Output x(t)
Input xin(t) and output position x(t), m

0.04

0.02

-0.02

-0.04

-0.06
0 0.2 0.4 0.6 0.8 1
Time, s

17
Copyright © 2016 John Wiley & Sons
 Chapter 9

9.18 The transfer function of the mechanical system in Problem 9.13 is

X ( s) 1 1
G ( s)   
Fa ( s ) ms  (b1  b2 ) s  k 0.6 s  3.4 s  80
2 2

We can verify the frequency response results in Problem 9.13 using MATLAB’s bode command:

a) The input force is f a (t )  2 sin 8t N. The magnitude and phase of the sinusoidal transfer
function at  = 8 rad/s can be computed using the MATLAB commands below:

>> sysG = tf(1,[0.6 3.4 80]); % define transfer function G(s)

>> w = 8; % define input frequency  = 8 rad/s
>> [mag,phase] = bode(sysG,w) % compute magnitude and phase

The result is mag = 0.0201 and phase = –33.1786 deg (or –0.5791 rad). The amplitude of the
output is (2 N)(0.0201) = 0.0402. The frequency response is

xss (t )  0.0402 sin( 8t  0.5791) m (matches Problem 9.13a)

b) We can create the Bode diagram (below) using the MATLAB command bode(sysG)

Bode Diagram
-30

-40
Magnitude (dB)

-50

-60

-70

-80
0

-45
Phase (deg)

-90

-135

-180
0 1 2
10 10 10
Frequency (rad/sec)

The peak magnitude occurs at a frequency in the range between 10-12 rad/s. We can use
MATLAB’s bode command as in part (a) to compute magnitudes at frequencies between 10-12
rad/s to determine the peak magnitude. The resonant frequency is approximately r = 10.8 rad/s
which verifies the computation in Problem 9.13b.

18
Copyright © 2016 John Wiley & Sons
 Chapter 9

9.19 The transfer function (from Problem 9.13) relating displacement x to the input force fa is

X (s) 1 1
G1 ( s )   
Fa ( s ) ms  (b1  b2 ) s  k 0.6 s  3.4 s  80
2 2

The following MATLAB command creates the first Bode diagram (below):

>> sysG1 = tf(1,[0.6 3.4 80]); % define transfer function G1(s)

>> bode(sysG1) % create the Bode diagram

Bode Diagram
-30

-40
Magnitude (dB)

-50

-60

-70

-80
0

-45
Phase (deg)

-90

-135

-180
0 1 2
10 10 10
Frequency (rad/sec)

Bode diagram for transfer function G1(s) = X(s)/Fa(s)

The second desired transfer function is G2 ( s)  FT ( s) / Fa ( s) where fT is the force transmitted to

the base. The force transmitted to the fixed base depends on the displacement and velocity of the
mass as well as damper b2 and spring k (see Fig. P9.13):

fT  b2 x  kx
FT ( s )
The transfer function relating transmitted force fT to displacement x is  b2 s  k
X (s)

Finally, the desired transfer function G2 ( s)  FT ( s) / Fa ( s) can be obtained by multiplying

FT (s) / X (s) and G1 ( s)  X ( s) / Fa ( s) as shown below

FT ( s ) FT ( s ) X ( s ) b2 s  k 0.4 s  80
G2 ( s )    
Fa ( s ) X ( s ) Fa ( s ) ms  (b1  b2 ) s  k 0.6 s 2  3.4 s  80
2

19
Copyright © 2016 John Wiley & Sons
 Chapter 9

The following MATLAB command creates the second Bode diagram (below):

>> sysG2 = tf([0.4 80],[0.6 3.4 80]); % define transfer function G2(s)
>> bode(sysG2) % create the Bode diagram

Bode Diagram
20

0
Magnitude (dB)

-20

-40

-60

-80

-100
0

-45
Phase (deg)

-90

-135

-180
0 1 2 3 4
10 10 10 10 10
Frequency (rad/sec)

Bode diagram for transfer function G2(s) = FT(s)/Fa(s)

From the two Bode (magnitude) diagrams we see that the resonant frequency is about 10.8 rad/s
for both systems G1(s) and G2(s). The MATLAB command can compute the bandwidth

>> wB1 = bandwidth(sysG1) % compute bandwidth B for G1(s)

>> wB2 = bandwidth(sysG2) % compute bandwidth B for G2(s)

Both bandwidths are 17.2 rad/s.

For very low-frequency inputs, the Bode diagram for system G1(s) shows a magnitude of about
–38 dB (or, absolute value of 0.0126). Hence the steady-state displacement for a (low-frequency)
2-N sinusoidal force is (2 N)(0.0126) = 0.025 m, which is the static deflection for a 2-N step
input; i.e., fa / k = (2 N)/(80 N/m) = 0.025 m.

For very low-frequency inputs, the Bode diagram for system G2(s) shows a magnitude of 0 dB
(or, absolute value of 1). Hence the steady-state transmitted force for a (low-frequency) 2-N
sinusoidal force is also 2 N, which makes sense because at very low frequencies the mass is
nearly in static equilibrium (almost no motion) and the spring force kx balances the input force fa.

20
Copyright © 2016 John Wiley & Sons
 Chapter 9

9.20 The mathematical model of the mechanical system is mz  bz  (k1  k2 ) z  f a (t )

a) Hence, the transfer function (substituting the numerical values for m, b, k1, and k2) is

Z (s) 1 3.3333
G (s)    2
Fa ( s ) 0.3s  1.5s  140 s  5s  466.6667
2

The following MATLAB command creates the Bode diagram using the transfer function

>> sysG = tf(1,[0.3 1.5 140]); % define transfer function G(s)

>> bode(sysG) % create the Bode diagram

Bode Diagram
-20

-40
Magnitude (dB)

-60

-80

-100

-120
0

-45
Phase (deg)

-90

-135

-180
0 1 2 3
10 10 10 10
Frequency (rad/sec)

We can create the same Bode diagram using the state-space representation (SSR) for the system,
where the state variables are x1 = z and x2  z ; the input is u = fa(t), and the output is y = x1 = z.
The SSR is
 0 1  0 
State equation: x    x u Output equation: y  1 0x
 466.67  5 3.3333

The following MATLAB command creates the Bode diagram using the SSR

>> A = [ 0 1 ; -466.67 -5]; % define state matrix A

>> B = [ 0 ; 3.3333 ]; % define input matrix B
>> C = [ 1 0 ]; % define output matrix C
>> D = 0; % define direct-link term D
>> sys = ss(A,B,C,D); % define sys as SSR
>> bode(sys) % create the Bode diagram from SSR

The Bode diagram is identical to the diagram above produced using the transfer function.

21
Copyright © 2016 John Wiley & Sons
 Chapter 9

b) The MATLAB command bandwidth can be used as shown below:

>> wB = bandwidth(sysG) % compute bandwidth B for G(s)

The bandwidth is 33.24 rad/s.

The magnitude Bode diagram shows that a peak occurs at a resonant frequency of about 21 rad/s.
Repeated trials of the MATLAB command [mag,phase] = bode(sysG,w) shows that the
resonant frequency is about r = 21.3 rad/s.

22
Copyright © 2016 John Wiley & Sons
 Chapter 9

1
9.21 a) The transfer function is G ( s )  ; therefore the sinusoidal transfer function is
2 s  32
2

1 1
G ( j )  
2( j )  32 32  2 2
2

For the input frequency  = 10 rad/s, the magnitude and phase are

12  02 1
Magnitude: G ( j10)   = 0.00595
(32  200) 2  02 168

arg[1  j 0] 0  0 
Phase angle:    tan 1    tan 1    180 deg
arg[ 168  j 0] 1   168 

b) The Simulink model (below) simulates the response to the sinusoidal input u (t )  3 sin 10t .
The output y(t) is plotted below.

0.08

0.06

0.04

0.02
Output, y

-0.02

-0.04

-0.06

-0.08
0 2 4 6 8 10
Time, s

c) The simulation response plot above does not match the frequency response Eq. (9-17)
because this system is undamped and therefore the transient response does not die out. Note that
the system’s undamped natural frequency is n  32 / 2  4 rad/s and hence the response y(t)
contains two sinusoidal functions with two frequencies: 4 rad/s (undamped natural frequency) and
10 rad/s (input frequency).

23
Copyright © 2016 John Wiley & Sons
 Chapter 9

9.22 The system transfer function is

576
G (s) 
16 s  38.4 s  576
2

The two magnitude Bode plots show the same peak value, high-frequency slope, and shape; only
the resonant frequency is different. We can compute the resonant frequency from knowledge of
the undamped natural frequency n and damping ratio  . First, divide G(s) by 16:

36
G (s) 
s  2.4 s  36
2

Therefore n  36  6 rad/s, and damping ratio  = 2.4/(2n) = 0.2.

The resonant frequency is  r   n 1  2 2 = 5.755 rad/s which matches System 1.

24
Copyright © 2016 John Wiley & Sons
 Chapter 9

9.23 a) We can represent the complete system by using a state-space representation (SSR),
where we can select the states as x1 = z (position), x2  z (velocity), and x3 = I (current).
The input is u = ein(t) and the desired output is position y = x1 = z. The complete SSR is

 0 1 0   0 
x   51,667  91.667 76.667 x   0 u
  y  1 0 0x
 0  230  400  100

The following MATLAB commands compute the frequency response for  = 150 rad/s

>> A = [0 1 0; -51667 -91.667 76.667; 0 -230 -400]; % define state matrix A

>> B = [ 0 ; 0 ; 100]; % define input matrix B
>> C = [ 1 0 0 ]; % define output matrix C
>> D = 0; % define direct-link term D
>> sys = ss(A,B,C,D); % define sys as SSR
>> w = 150; % define input frequency 
>> [mag,phase] = bode(sys,w) % compute magnitude and phase

The MATLAB result is mag = 0.000486 and phase = –52.508 deg (–0.9164 rad). The
amplitude of the output is (8 V)(0.000486) = 0.00389 m (3.89 mm). Hence the frequency
response is

z ss (t )  0.00389 sin( 150t  0.9164) m

b) The MATLAB command (below) determines the bandwidth B = 318.33 rad/s

>> wB = bandwidth(sys) % compute bandwidth B in rad/s

25
Copyright © 2016 John Wiley & Sons
 Chapter 9

9.24 The notch filter transfer function is

LCs 2  1 E ( s)
G ( s)   O
LCs  RCs  1 Ein ( s )
2

a) The following MATLAB commands create the Bode diagram (using values for L, C, and R)

>> L = 0.005; % define inductance L

>> C = 0.02; % define capacitance C
>> R = 1; % define resistance R
>> sysG = tf([L*C 0 1],[L*C R*C 1]) % define transfer function G(s)
>> bode(sysG) % create the Bode diagram

Bode Diagram
0

-50
Magnitude (dB)

-100

-150

-200
450

405
Phase (deg)

360

315

270
0 1 2 3 4
10 10 10 10 10
Frequency (rad/sec)

Bode diagram for L = 0.005 H, C = 0.02 F, and R = 1 

Note that the magnitude plot has a DC gain of 0 dB at all frequencies except near  = 100 rad/s.
The large negative-dB “notch” at  = 100 rad/s means that there is essentially zero output when
the input frequency is near the “notch frequency”  N  1 /( LC )  10 = 100 rad/s.
4

b) When the output amplitude is reduced by one-half the corresponding change in decibels is

20 log 10 (0.5)  6 dB

Hence, we can use the magnitude Bode diagram to estimate the frequency where the magnitude
drops from 0 dB (low frequencies) to –6 dB. A “zoomed-in” view around the notch in the
magnitude plot is shown on the next page.

26
Copyright © 2016 John Wiley & Sons
 Chapter 9

Bode Diagram

-2

Magnitude (dB)
-4

-6 Stop-band
-8
frequency
-10

450

405
Phase (deg)

360

315

270
0 1 2 3
10 10 10 10
Frequency (rad/sec)

Bode diagram for L = 0.005 H, C = 0.02 F, and R = 1 (zoomed-in view)

Hence the “stop band” frequency range where the magnitude is less than –6 dB is roughly
between 60 and 140 rad/s (i.e., symmetric about the notch frequency N = 100 rad/s).

c) Executing the previous MATLAB commands with R = 0.2  creates the Bode diagram shown
below. Note that the “notch” is sharper and has a narrower stop-band frequency range. The stop-
band frequency range (i.e., magnitude less than –6 dB) is between 90 and 110 rad/s.

Bode Diagram

0
Magnitude (dB)

-2

-4

-6

-8

450

405
Phase (deg)

360

315

270
1 2 3
10 10 10
Frequency (rad/sec)

Bode diagram for L = 0.005 H, C = 0.02 F, and R = 0.2 (zoomed-in view)

27
Copyright © 2016 John Wiley & Sons
 Chapter 9

9.25 a) Since the input force is f in (t )  2 sin 50t N, we simply read the Bode diagram at input
frequency  = 50 rad/s: Magnitude of G(j50) = –16 dB and phase of G(j50) = –135 deg (or –
2.3562 rad). The absolute-value magnitude is 10–16/20 = 0.1585. Hence, the amplitude of the
frequency response is (2 N)(0.1585) = 0.317 N (amplitude of the transmitted force). The
complete frequency response of the transmitted force is

fT (t )  0.317 sin( 50t  2.3562) N

b) Transmissibility of 1.75 is equal to an amplitude ratio (magnitude) in decibels of

20 log 10 (1.75)  4.86 dB

Hence, the magnitude Bode plot cannot exceed 4.86 dB so that TR < 1.75. We see from the Bode
diagram that the magnitude exceeds roughly 5 dB for the frequency range 10 <  < 20 rad/s and
this is the frequency range that must be avoided in order to keep transmissibility less than 1.75.

28
Copyright © 2016 John Wiley & Sons
 Chapter 9

9.26 The linearized transfer function for the pneumatic system is

7.003(106 ) X ( s)
G( s)  
s( s  28.3s  15890) U ( s)
2

a) The Simulink model has a sinusoidal valve input u (t )  0.001 sin 50t m.

A plot of piston position x(t) and input u(t) is below:

0.025 Valve input u(t)

Piston output x(t)
Valve input (u) and piston position (x), m

0.02

0.015

0.01

0.005

-0.005
0 0.2 0.4 0.6 0.8 1
Time, s
8-Hz input u(t) and system response x(t)

The response plot x(t) does not oscillate about zero and is therefore different from the “standard”
frequency-response equation (9-17):

xss (t )  G( j) U 0 sin( t   )

The difference can be explained by observing the poles of the system transfer function. The
standard frequency-response equation (9-17) was derived by assuming that all poles of G(s) have
negative real parts; that is, the corresponding transient response eventually decays to zero. The
pneumatic servo transfer function G(s) has a pole at the origin (s = 0) and hence part of the
natural response is a constant that does not decay to zero. This constant is an “offset” that is
added to the sinusoidal frequency response, and hence the frequency response does not oscillate
about zero.

29
Copyright © 2016 John Wiley & Sons
 Chapter 9

b) We can change the input frequency in the Simulink model to a 20-Hz input sinusoidal
function u (t )  0.001 sin 126t . The corresponding system response is below:

0.025 Valve input u(t)

Piston output x(t)

Valve input (u) and piston position (x), m

0.02

0.015

0.01

0.005

-0.005

-0.01

-0.015
0 0.1 0.2 0.3 0.4 0.5
Time, s
20-Hz input u(t) and system response x(t)

The response plot shows that input u(t) and output x(t) are nearly 180 deg out of phase (i.e., the
input “peaks” are aligned with the output “valleys.”). The phase plot in the Bode diagram
(produced by MATLAB) verifies the –180 deg phase at input frequency  = 126 rad/s. Note that
the undamped natural frequency of G(s) is n  15,890  126.06 rad/s which is equal to the
input frequency.

>> sysG = tf(7.003e6,[1 28.3 15890 0]) % define transfer function G(s)
>> bode(sysG) % create the Bode diagram

Bode Diagram
40

20
Magnitude (dB)

-20

-40

-60
-90

-135
Phase (deg)

-180

-225

-270
1 2 3
10 10 10
Frequency (rad/sec)

Bode diagram with piston position as the output

c) The transfer function with velocity as the output can be obtained by multiplying the position-
output transfer function G(s) by s

30
Copyright © 2016 John Wiley & Sons
 Chapter 9

7.003(106 )
G2 ( s)  (velocity is the output)
s 2  28.3s  15890

The Simulink diagram with the velocity-output transfer function is shown below. Note that the
pneumatic servo transfer function is 2nd-order (no pole at s = 0). The velocity response is plotted
below. Note that velocity oscillates about zero.

0.8

0.6

Piston velocity (x-dot), m/s

0.4

0.2

-0.2

-0.4

-0.6
0 0.2 0.4 0.6 0.8 1
Time, s

Velocity response to an 8-Hz valve input

The frequency response of the velocity can be computed from the corresponding Bode diagram of
transfer function G2(s), shown below:

Bode Diagram
70

60
Magnitude (dB)

50

40

30

20

10
0

-45
Phase (deg)

-90

-135

-180
1 2 3
10 10 10
Frequency (rad/sec)

Bode diagram with piston velocity as the output

At input frequency  = 50 rad/s (~8 Hz) the magnitude is 54 dB (or 520.1) and the phase is –6
deg (or –0.105 rad). Hence the velocity frequency response is

xss (t )  0.520 sin( 50t  0.105) m/s

31
Copyright © 2016 John Wiley & Sons
 Chapter 9

9.27 The mathematical model of the 1-DOF nonlinear suspension system is

4500 4500
mx   Fd ( z)  kz or, mx  x  kx  xin (t )  kxin (t )
z  v
2 2
z 2  v 2

where z  x  xin (t ) is the relative velocity and v = 0.2 m/s. The Simulink model is below. Note
that a user-defined function computes the nonlinear damping force given relative velocity
z  x  xin (t ) .

The following M-file “loops” through a range of input frequencies 0.1    100 rad/s, executes
the Simulink model, stores the steady-state amplitude ratio (x/xin), and creates the magnitude
Bode diagram (in dB).

% Mfile for Problem 9.27

r2d = 180/pi; % rad-to-deg

% input frequencies
Nw = 50;
w = logspace(0,2,Nw); % log-spaced frequencies from 0.1 to 100 rad/s

% Loop for frequency

for i=1:Nw
w_in = w(i);
f_in_Hz = w_in/(2*pi);

% re-compute step size based on period

Tperiod = 2*pi/w_in; % period, sec
t_stop = round(w_in*3)*Tperiod;
dt = Tperiod/500;

% execute Simulink model

sim P9_27

Npts = numel(x); % number of data pts

N_half = round(Npts/2);
amp_x_SS = max(x(N_half:Npts));
amp_xin = max(x_in);
magdB(i) = 20*log10(amp_x_SS/amp_xin); % magnitude, dB
end

% Bode magnitude plot

semilogx(w,magdB)
grid
xlabel('Frequency, \it\omega\rm, rad/s')
ylabel('Magnitude, dB')

32
Copyright © 2016 John Wiley & Sons
 Chapter 9

The Bode magnitude plot from running the M-file is shown below:

-5
Magnitude, dB

-10

-15

-20

-25

-30

-35 0 1 2
10 10 10
Frequency, , rad/s
Bode magnitude plot of 1-DOF system with nonlinear damper

We see that the magnitude exhibits a DC gain of 0 dB, or unity transmissibility at low
frequencies. The system has a small 1-dB resonant peak at about 5 rad/s and drops off at high
frequencies. Therefore this nonlinear suspension system has good damping and attenuates high-
frequency input.

33
Copyright © 2016 John Wiley & Sons
 Chapter 9

9.28 The state equation of the ¼-car system (see Problem 2.30 and 6.23) is

 0 1 0 0   0 
 k / m  b / m k1 / m1 b / m1   0 
x   1 1 1
x u
 0 0 0 1   0 
   
 k1 / m2 b / m2  (k1  k2 ) / m2  b / m2  k2 / m2 

The state variables are x1  z1 , x2  z1 , x3  z2 , and x4  z2 . The input is u = zin(t).

a) An impulsive input zin(t) (a road bump) will excite the natural frequencies of the system. The
natural frequencies are associated with the imaginary parts of the characteristic roots, or poles, or
eigenvalues. Because we have a SSR it is easiest to determine the eigenvalues of the 4  4 A
matrix using the MATLAB commands:

>> m1 = 250; % mass of ¼ car, kg

>> m2 = 30; % mass of wheel/axle, kg
>> k1 = 1.6e4; % suspension stiffness, N/m
>> b = 980; % suspension friction, N-s/m
>> k2 = 1.6e5; % tire stiffness, N/m
>> Arow1 = [ 0 1 0 0 ]; % A (row 1)
>> Arow2 = [ -k1/m1 -b/m1 k1/m1 b/m1 ]; % A (row 2)
>> Arow3 = [ 0 0 0 1 ]; % A (row 3)
>> Arow4 = [ k1/m2 b/m2 (-k1-k2)/m2 -b/m2 ]; % A (row 2)
>> A = [Arow1 ; Arow2 ; Arow3 ; Arow4]; % A matrix
>> lambda = eig(A) % eigenvalues of the A matrix

The four eigenvalues are complex: 1, 2  16.6503  j 74.0592 and 3, 4  1.6430  j 7.5193 .
Hence the vibration frequencies are the imaginary parts: 1 = 7.519 rad/s and 2 = 74.059 rad/s.

a) The system output is the ¼-car mass position: y  z1  x1

y  1 0 0 0x  [0]u

The following added MATLAB commands produce the Bode diagram (using the SSR)

>> B = [ 0 ; 0 ; 0 ; k2/m2 ]; % B matrix

>> C = [ 1 0 0 0 ]; % C matrix (output, y = z1)
>> D = 0; % D matrix (null)
>> sys = ss(A,B,C,D); % create SSR system
>> bode(sys); % draw Bode diagram

34
Copyright © 2016 John Wiley & Sons
 Chapter 9

Bode Diagram
20

0
Magnitude (dB)
-20

-40

-60

-80

-100
0
Phase (deg)

-90

-180

-270
0 1 2 3
10 10 10 10
Frequency (rad/sec)

The magnitude plot of the Bode diagram shows two resonant peaks at approximately  = 7.4
rad/s and  = 70 rad/s, which are very close to the natural (damped) frequencies in part (a). The
magnitudes at these resonant frequencies are about 8 dB (at  = 7.4 rad/s) and –18 dB (at  = 70
rad/s). Hence the two transmissibility values are 108/20 = 2.512 and 10-18/20 = 0.126.

35
Copyright © 2016 John Wiley & Sons
 Chapter 9

9.29 The mathematical model of the 1-DOF mechanical system is

mx  bx  kx  bxb (t )  kxb (t )

The system’s transfer function is

X ( s) bs  k
G ( s)  
X b ( s ) ms  bs  k
2

Because we are given the amplitude of the frequency response we must compute the magnitude
of the sinusoidal transfer function:

bj  k k  jb
G ( j )  
 m  bj  k k  1.4 2  jb
2

where we have substituted mass m = 1.4 kg. The magnitude of the sinusoidal transfer function
for input frequency  = 5 rad/s (first table entry) is

k 2  (5b) 2 0.5070
Magnitude: G ( j 5)   = 1.0140
(k  (1.4)( 25))  (5b)
2 2 0.5

or, squaring both sides:


k 2  25b 2  (1.0140) 2 k 2  70k  1225  25b 2  (A)

Similarly we can compute the magnitude with input frequency  = 30 rad/s (second table entry)

k 2  (30b) 2 0.7008
Magnitude: G ( j 30)   = 1.4016
(k  (1.4)(900))  (30b)2 2 0.5
or


k 2  900b2  (1.4016)2 k 2  2520k  1,587,600  900b2  (B)

While Equations (A) and (B) provide two equations with two unknowns (k and b) the equations
are nonlinear. Perhaps a simpler method is to use matrix methods with a third equation using the
third table entry,  = 60 rad/s:

k 2  (60b) 2 0.4975
Magnitude: G ( j 60)   = 0.9950
(k  (1.4)(3600)) 2  (60b) 2 0.5
or


k 2  3600b2  (0.9950)2 k 2  10,080k  25,401,600  3600b2  (C)

36
Copyright © 2016 John Wiley & Sons
 Chapter 9

Equations (A)-(C) provide three equations with three unknowns k2, k, and b2. A matrix
formulation is

  0.0282 71.9737  0.7049  k 2   1259.54 

 0.96448 4950.4961  868.0343  k   3.1188125(106 ) 
    
 0.009975 9979.4520 35.9100  b 2  2.5148219(107 )

Multiplying the inverse of the 3 3 matrix by the right-hand side vector yields the solution:

k 2  6.2436(106 )
   
 k    2500.3 
b 2   3729.408 
   

Hence the composite spring constant is k = 2500.3 N/m and the friction coefficient is b = 61.07
N-s/m. Note that there is some loss of numerical accuracy due to carrying finite digits for the
very large numbers. As a check, we can use the MATLAB command bode as follows:

>> m = 1.4; % mass, kg

>> k = 2500.3; % stiffness, N/m
>> b = 61.07; % friction, N-s/m
>> sysG = tf([b k],[m b k]); % create transfer function G(s)
>> [mag,phase] = bode(sysG,5); % magnitude and phase for  = 5 rad/s

The magnitude is mag = 1.0140 which matches the first entry in the table. We can use these
MATLAB commands to verify the other entries in the table for  = 30, 60, and 100 rad/s.

37
Copyright © 2016 John Wiley & Sons