Chapter 7: Analytical Solution of Linear Dynamic Systems (2nd Edition)

7.1 Re-write the first-order ODE in standard form (divide all terms by 10):

0.2 y + y = 0.3u

The single root is r = –1/0.2 = –5 and therefore the response is a stable exponential function. The
time constant is  = 0.2 s and hence the system reaches steady state in tS = 4 = 0.8 s. The steady-
state value for a constant input u = 6 is ySS = (0.3)(6) = 1.8. A hand-drawn sketch of the response
would match the response plot (below) and include labels for the initial condition y(0) = –2,
settling time tS = 0.8 s, and steady-state response ySS = 1.8.

1.5

0.5
Output y

-0.5

-1

-1.5

-2

-2.5
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Time, s
 Chapter 7

7.2 SS a) The given I/O equation is 0.4 y + 0.2 y = 0.8u

Re-write the first-order ODE in standard form (divide all terms by 0.2):

2 y + y = 4u or  y + y = 4u

The characteristic equation is 2r + 1 = 0 , so the single root is r = –0.5. The total response is
stable because the single (real) root is negative. The time constant is  = 2 s, and hence the
system reaches steady state in tS = 4 = 8 s. The steady-state value for a constant (step) input u =
–1.5 is yss = (4)(–1.5) = –6.

To make a hand-drawn sketch, we would draw an exponential decay from the initial condition,
y(0) = 1, to the steady-state value, yss = –6. The exponential response would reach steady-state at
about tS = 8 s. A hand-drawn sketch would match the response plot shown below (created using
MATLAB).

2
 Chapter 7

7.3 SS a) First, we obtain the roots of the characteristic equation (below):

2r 2 + 12r + 68 = 0

The two roots are complex: r1, 2 = −3  j 5 . Hence, the homogeneous response does exhibit
oscillations. Another way to show this is to re-write the I/O equation in the standard form for a
second-order system:

y + 6 y + 34 y = 0 or y + 2 n y + n2 y = 0 Hence  = 0.514 < 1 (underdamped)

b) For an underdamped second-order system the settling time is approximately

4 4
tS = = = 1.3333 s
 n 0.514 34

c) Because the second-order system is underdamped, the homogeneous response will

exhibit decaying oscillations at frequency d = 5 rad/s (or, period = 1.257 s). The
homogeneous response will begin at the initial condition y(0) = 3 with “zero slope” and
exhibit a damped sinusoidal response that decays to zero in about 1.3333 s (a bit longer
than one period). A sketch would match the plot:

2.5

2
Output y

1.5

0.5

-0.5
0 0.5 1 1.5 2
Time, s

3
 Chapter 7

7.4 a) First, we obtain the roots of the characteristic equation (below):

4r 2 + 22r + 18 = 4(r + 1)(r + 4.5) = 0

The two roots are real and negative: r1 = –1 and r2 = –4.5 and hence the homogeneous response is
comprised of two decaying exponential functions (no oscillations).

b) The homogeneous response has the form yH (t ) = c1e−t + c2e−4.5t and the slowest
exponential mode “dies out” at time tS = 4 s. Hence the settling time for the system is 4 s.

c) The homogeneous response will begin at the initial condition y(0) = 3 with “zero slope”
and then exponentially decay to zero in about 4 s. A sketch would match the plot below.

2.5

2
Output y

1.5

0.5

0
0 1 2 3 4 5 6
Time, s

4
 Chapter 7

7.5 In all cases the denominator polynomial (characteristic equation) will have the form

( s + r1 )(s + r2 ) = 0

a) Roots r1 = −2.5 and r2 = −0.2 : (s + 2.5)(s + 0.2) = s 2 + 2.7s + 0.5 (denominator)

c
Hence the transfer function has the form G ( s) =
s + 2.7 s + 0.5
2

Because the DC gain of this transfer function is G(s = 0) = c / 0.5 = 0.5 we obtain c = 0.25 and
the transfer function is

0.25
G( s) =
s + 2.7 s + 0.5
2

b) Roots r1 = −3 and r2 = −10 : (s + 3)(s + 10) = s 2 + 13s + 30 (denominator)

Because the DC gain of this transfer function is 6 we obtain

180
G(s) = Note: G(s = 0) = 180/30 = 6
s + 13s + 30
2

c) Roots r1, 2 = −2  j 4 : (s + 2 + j 4)(s + 2 − j 4) = s 2 + 4s + 20 (denominator)

Because the DC gain of this transfer function is 125 we obtain

2500
G ( s) = Note: G(s = 0) = 2500/20 = 125
s + 4s + 20
2

d) Roots r1, 2 = −0.4  j1.6 : (s + 0.4 + j1.6)(s + 0.4 − j1.6) = s 2 + 0.8s + 2.72

Because the DC gain of this transfer function is 0.02 we obtain

0.0544 0.0544
G( s) = Note: G(s=0) = = 0.02
s + 0.8s + 2.72
2
2.72

5
 Chapter 7

7.6 a) We can obtain the steady-state response to a constant input by using the DC gain:

84 84
G( s) = → G ( s = 0) = = 0.2593
(3s + 21s + 36)(s + 2s + 9)
2 2
(36)(9)

Therefore the steady-state output is ySS = (4)(0.2593) = 1.0370

b) The four characteristic roots are determined by setting the denominator to zero

(3s 2 + 21s + 36)(s 2 + 2s + 9) = 0 or 3(s + 3)(s + 4)(s 2 + 2s + 9) = 0

Hence the four roots (poles) are r1 = −3 , r2 = −4 , and r3, 4 = −1  j 2.8284

The three exponential decay modes for the transient response are e−3t , e−4t (for the two real
roots) and e − t (for the two complex roots). Because the slowest exponential mode is e − t , the
overall settling time is tS = 4 s because e−4 is “small.”

6
 Chapter 7

7.7 a) The mathematical model of the mechanical rotor is J + b = Tin (t )

Because angular position  is not in the ODE, we can write the model as a first-order system

J  + b = Tin (t ) or, with numerical values 0.06 + 0.008 = Tin (t )

Dividing all terms by 0.008, we obtain 7.5 +  = 125Tin (t )

The total response of this first-order system is

 (t ) = c1e−t /7.5 + ss (A)

where  = 7.5 s is the time constant and ss is the steady-state angular velocity. The steady-state
value can be obtained from the ODE by setting  = 0 and Tin(t) = 0.4 N-m.

Therefore ss = 50 rad/s

b) Given the initial condition (0) = 20 rad/s, we can use Eq (A) to obtain constant c1 = –30
rad/s. Therefore, the total response is

(t ) = 50 − 30e−t /7.5 rad/s

We want (t) = 1.2(0) = 24 rad/s = 50 − 30e−t /7.5 rad/s, or e−t /7.5 = 0.8667 .

Solving for time, we find t = 1.073 s

7
 Chapter 7

7.8 The mathematical model of the RC circuit is RCeO + eO = ein (t )

Therefore the transfer function is

EO ( s) 1 1
G (s) = = = (using RC = 0.01 s)
Ein ( s ) RCs + 1 0.01s + 1

Clearly the DC gain is 1 and the time constant is  = RC = 0.01 s. Because the input is a constant
0.4 V, the steady-state output voltage is also 0.4 V. The output voltage will show an exponential
rise from its initial value (zero) to 0.4 V and reach steady state in settling time tS = 4 = 0.04 s. A
sketch should show eO(0) = 0, eO () = 0.4 V, and settling time (i.e., match the plot below).

0.5

0.4
Output voltage, V

0.3

0.2

0.1

0
0 0.01 0.02 0.03 0.04 0.05
Time, s

8
 Chapter 7

7.9 SS The mathematical model of the RC circuit is RCeO + eO = ein (t ) . Because the circuit
stores no initial energy, we can use the transfer function

EO ( s) 1 1
G (s) = = = (using RC = 0.01 s)
Ein ( s ) RCs + 1 0.01s + 1

The voltage input is a 12-V pulse that lasts 2(10-4) s. The time constant is for the RC circuit is  =
RC = 0.01 s, so the settling time for a step input is 4 = 0.04 s. Clearly the 12-V pulse duration is
much shorter than the settling time. In fact, the ratio of pulse time to settling time is 0.0002/0.04
= 0.005, so we can consider the 12-V pulse input as an impulse input with area (or weight) = (12
Ab
V)(0.0002 s) = 0.0024 V-s. For a first-order system, the impulse response is eO (t ) = e−t /

where A = 0.0024 V-s is the area or weight of the pulse, b = 1 is the DC gain of the transfer
function, and the time constant is  = 0.01 s. Hence the output voltage at t = 0 is eO(0) = (0.0024
V-s)/0.01 s = 0.24 V. The impulse response will begin at 0.24 V and then exhibit an exponential
decay to zero. The settling time is tS = 4 = 0.04 s. A sketch would match the response plot
shown below (created using MATLAB).

9
 Chapter 7

7.10 The two mathematical models of the electrical system are

LIL + R1I L = ein (t ) (Switch 1) and LIL + ( R1 + R2 ) I L = 0 (Switch 2)

First, consider the response for time 0  t  1 (Switch 1) when ein(t) = 4 V. The time constant of
the “Switch 1” model is  1 = L / R1 = 0.2/1.6 = 0.125 s. Hence the settling time is 41 = 0.5 s
which is less than the switching time (1 s), so the current reaches a steady-state value. The
steady-state current for the “Switch 1” model is I LSS = ein / R1 = 4/1.6 = 2.5 A.

When time t > 1 s, we have the “Switch 2” model with time constant  2 = L /( R1 + R2 ) = 0.0357
s. Hence the settling time is 42 = 0.1429 s (relative to the switch time t = 1), and the current IL
decays to zero and reaches zero at approximately time t = 1.1429 s. A sketch will match the plot
below.

2.5

2
Current, A

1.5

0.5

0
0 0.5 1 1.5
Time, s

10
 Chapter 7

7.11 The characteristic equation is the denominator polynomial of G(s) set to zero:

2s 2 + 3s + 24 = 0 or s 2 + 1.5s + 12 = 0

The roots (poles) are s1, 2 = −0.75  j3.3819 . Because the roots (or poles) are complex, the
transient response does exhibit oscillations.

Method 2: the standard second-order form is s 2 + 2 n s + n2 = 0 . Hence n = 12 = 3.464

rad/s and  = 0.217 < 1 (underdamped) and the transient response shows oscillations.

11
 Chapter 7

4 Y ( s)
7.12 SS The system transfer function is G( s) = =
s + 2s + 36 U ( s)
2

The input is a step function with a magnitude of 18.

The maximum output (peak response) will occur at maximum overshoot (if the 2nd-order system
is underdamped). The poles of the transfer function are determined from

s 2 + 2s + 36 = 0 → the poles are s = −1  j5.9161

Because the poles are complex, the 2nd-order system is underdamped. The undamped natural
frequency is n = 36 = 6 rad/s, and the damping ratio is  = 2 / (2n ) = 0.1667 . The
maximum overshoot solely depends on the damping ratio:

M OS = e− / 1− 2
= 0.5880 [ or, 58.8% overshoot of steady-state value ]

The steady-state value is easily computed from the transfer function’s DC gain (= 4/36) and the
step input:

 4 
yss =   (18) = 2
 36 

Finally, the maximum (peak) response is ymax = yss (1 + M OS ) = 3.176

12
 Chapter 7

7.13 To start, we will determine the DC gains, undamped natural frequencies, and damping
ratios for each candidate transfer function:

0.9
Option (a): G ( s ) =
s + 2.5s + 36
2

→ DC gain = 0.9/36 = 0.025, n = 36 = 6 rad/s,  = 2.5/(2n) = 0.2083

1
Option (b): G ( s) =
s + 4.8s + 40
2

→ DC gain = 1/40 = 0.025, n = 40 = 6.3246 rad/s,  = 4.8/(2n) = 0.3795

1.5
Option (c): G ( s) =
s + 1.2s + 60
2

→ DC gain = 1.5/60 = 0.025, n = 60 = 7.7460 rad/s,  = 1.2/(2n) = 0.0775

Therefore all three transfer functions have the correct DC gain (note that the steady-state response
to a unit-step input is 0.025).

Next, we compute the settling time and peak overshoot for each transfer function:

4 4
Option (a): tS = = = 3.2 s (good match with Figure P7.13)
n (0.2083)(6)

ymax = 0.025(1 + MOS) = 0.025 1 + e ( − / 1− 2

) = 0.038 (good match with Figure P7.13)

4 4
Option (b): tS = = = 1.667 s (does not match Figure P7.13)
n (0.3795)(6.3246)

ymax = 0.025(1 + MOS) = 0.025 1 + e ( − / 1− 2

) = 0.032 (does not match Figure P7.13)

4 4
Option (c): tS = = = 6.663 s (does not match Figure P7.13)
n (0.0775)(7.746)

ymax = 0.025(1 + MOS) = 0.025 1 + e ( − / 1− 2

) = 0.045 (does not match Figure P7.13)

Clearly Option (a) is the correct transfer function because the computed settling time and peak
response best match the values observed from Figure P7.13.

13
 Chapter 7

7.14 SS The given unit-step response (Figure P7.14) is repeated below:

10
System (a) is G ( s ) = . This transfer function is second-order (it is underdamped),
s + 0.6s + 40
2

and hence the step response will show a peak at the first one-half cycle. The given step response
does not show a peak at the first one-half cycle, so System (a) does not produce the given step
response.

8
System (b) is G ( s) = , which is third order. The three poles are
0.2s + 2.8s + 15.4s + 32
3 2

0.2s3 + 2.8s 2 + 15.4s + 32 = 0 → s = −5, −4.5  j3.4278

Therefore, the total response is

y(t ) = c1e−5t + c2e−4.5t sin(3.4278t +  ) + yss

The steady-state response, yss, is the transfer function’s DC gain = 8/32 = 0.25 (the input is a unit
step). Therefore, the steady-state response matches the value in the plot. The “slowest” pole
−4.5t
(root) corresponds to the damped sinusoidal term and therefore the exponential function e
determines the settling time. The settling time for this mode is tS = 4/4.5 = 0.8889 s. The given
response takes more than 10 s to reach steady-state, so System (b) does not produce the step
response.

8
System (c) is G ( s) = , which is third order. The three poles are
s + 1.4s + 40.48s + 32
3 2

14
 Chapter 7

s3 + 1.4s 2 + 40.48s + 32 = 0 → s = −0.8, −0.3  j 6.3174

Therefore, the total response is

y(t ) = c1e−0.8t + c2e−0.3t sin(6.3174t +  ) + yss

The given response plot appears to be the sum of an exponential function and a damped
sinusoidal function. The steady-state response, yss, is the transfer function’s DC gain = 8/32 =
0.25, which matches the value in the plot. The slowest pole corresponds to the damped sinusoidal
−0.3t
function. Its exponential function, e , has a settling time of tS = 4/0.3 = 13.333 s. This
settling time matches settling time in the given response plot. Finally, the period of the damped
sinusoidal response is 2/6.3174 = 0.995 s, and this matches the period of oscillation in the given
step response (i.e., we observe about 10 cycles in 10 s).

Hence, System (c) best matches the given unit-step response.

15
 Chapter 7

7.15 The damping ratio  is the magnitude of the real part of the root divided by the distance
from the root to the origin (see Fig. 7.17). If the complex root is s = −a  jb then we have

a
 =
a 2 + b2

a) Roots are s = −3  j 2 . Hence  = 3 / 32 + 22 = 0.8321

b) Roots are s = −2  j3 . Hence  = 2 / 22 + 32 = 0.5547

c) Roots are s = −3  j3 . Hence  = 3 / 32 + 32 = 0.7071

→ root-pair (a) has the greatest damping ratio.

16
 Chapter 7

7.16 The mathematical model of the mass-spring-damper system is

my + by + ky = f a (t )
Its transfer function is

1 Y ( s)
G (s) = =
ms + bs + k Fa ( s )
2

The DC gain is 1/k, so the steady-state response to an 18-N step input is yss = (1/k)(18 N) = 0.036
m. Solving for the spring constant we obtain k = 500 N/m.

The damping ratio  solely determines the maximum percent overshoot MOS , which is used to
compute the peak response:

ymax = yss(1 + MOS) = 0.036 1 + e ( − / 1− 2

) = 0.05 m
Therefore,

e − / 1− 2
= 0.3889

or, taking the natural logarithm,

−
= ln(0.3889) = −0.9444
1−  2

Solving for damping ratio,

0.94442
 = = 0.2879
 2 + 0.94442

Using the standard form for an underdamped system, the first-order term is

2n = b / m where n = k / m = 500 / 0.1 = 70.711 rad/s

Finally we can obtain the friction coefficient b = 4.0716 N-s/m

17
 Chapter 7

7.17 We will use the log-decrement method. The peak values and peak times of the impulse
response are estimated from the impulse-response plot:

Peak 1: x1 = 0.026 m, t1 = 0.1 s Peak 2: x2 = 0.012 m, t1 = 0.5 s

Hence the period is approximately Tperiod = 0.5 – 0.1 = 0.4 s. The log decrement is

x1 0.026
 = ln = ln = 0.7732
x2 0.012
The damping ratio is approximately


 = = 0.1221
4 2 +  2

The damped frequency is d = 2/Tperiod = 15.708 rad/s. Because damped frequency is

d = n 1 −  , the undamped natural frequency can be determined from d and  to be
2

n = d / 1 −  2 = 15.8264 rad/s

18
 Chapter 7

7.18 a) The general solution form for the homogeneous response of an underdamped system
with two complex roots at r1, 2 =   j is

yH (t ) = cet cos(t +  ) (A)

We see that the real part of the root will be the exponent of the exponential function while the
imaginary part will be the frequency of the sinusoidal function. Because we have two pairs of
complex roots we will have two damped sinusoidal functions in the form of Eq. (A). Using the
two root pairs given in the problem, the homogeneous response has the form

yH (t ) = c1e−1.2t cos(6.3t + 1 ) + c2e−0.4t cos(4.2t + 2 )

The unknown constants are c1, c2, 1, and 2. We could have used sine instead of cosine.

b) The two exponential functions e−1.2t and e−0.4t determine the decay rate for the two damped
sinusoidal functions. The “slowest” decaying exponential function is e−0.4t which requires about
10 s to reach zero because e−4 = 0.0183 is “small.” Hence the settling time is 10 s.

19
 Chapter 7

7.19 Because the free response is x(t) = 6cos50t , we note that the undamped natural frequency is
n = 50 rad/s.

The mechanical model of an unforced mass-spring system is mx + kx = 0 . The standard I/O
equation form of a second-order system with no damping ( = 0) is x + n2 x = 0 . Hence the
undamped natural frequency of the mass-spring system is

k
n = = 50 rad/s
m

Because mass is m = 0.2 kg, the spring constant is k = 500 N/m

20
 Chapter 7

7.20 The I/O equation is 2 y + 8 y + 6 y = 3u

a) The characteristic equation can be written from inspection:

2r 2 + 8r + 6 = 0 or 2(r + 1)(r + 3) = 0

The two characteristic roots are r1 = −1 and r2 = −3 .

b) The following transfer function can be derived from the given I/O equation:

Y (s) 3
G(s) = = 2
U ( s ) 2 s + 8s + 6

The poles of G(s) are determined by setting the denominator polynomial to zero:

2s 2 + 8s + 6 = 0

Therefore the two poles of G(s) are s1 = −1 and s2 = −3 , which are identical to the roots in (a).

c) We can obtain a SSR for the states x1 = y and x2 = y . The resulting state equation is

0 1 0
x =   x +  u
− 3 − 4 1.5

The eigenvalues are computed from the determinant I − A = 0

 −1
I − A = = 2 + 4 + 3 = 0
3 +4

The eigenvalues are 1 = −1 , 2 = −3 which are identical to the roots and poles in (a) and (b).

d) Because the two roots are real and negative the zero-input response consists of two decaying
exponential functions
y(t ) = c1e−t + c2e−3t

The response starts at y(0) = 5 with “zero slope” due to y (0) = 0 and eventually decays to zero.
The slowest exponential decay mode is e − t which takes about 4 s to reach zero.

21
 Chapter 7

7.21 SS We can use the log-decrement method. Using the first two peaks, the log decrement is

x1 0.1091
 = ln = ln = 1.1748
x2 0.0337

x2 0.0337
We can re-compute  using peaks 2 and 3:  = ln = ln = 1.1757
x3 0.0104

x3 0.0104
We may also use peaks 3 and 4:  = ln = ln = 1.1787
x4 0.0032

Averaging these three values, we obtain  = 1.1764.

The damping ratio is approximately


 = = 0.1840
4 2 +  2

The time between peaks determines the damped frequency:

2 2
d = = = 174.5329 rad/s
t2 − t1 0.044 − 0.008

Because damped frequency is d = n 1 −  2 , the undamped natural frequency is

d
n = = 177.5656 rad/s
1−  2

Using the standard form for a second-order system, we obtain the transfer function

b b X (s)
G (s) = = =
s + 2n s + 
2 2
n s + 65.354s + 31,529.5 F ( s )
2

Finally, we know that the steady-state step response is 0.012 m for a 15-N constant force input.
Using the DC gain, the steady-state response is xss = (b/31,529.5)(15) = 0.012 m. Therefore, the
numerator must be b = 25.2236. The system transfer function is

25.2236 X ( s)
G( s) = =
s + 65.354s + 31,529.5 F ( s)
2

22
 Chapter 7

7.22 First, we determine the poles of the third-order transfer function G1(s)

s3 + 20.2s 2 + 124s + 24 = 0

The three poles are s1 = −0.2 and s2 = −10  j 4.4721. The two underdamped poles show very
high damping ( = 0.913) and a relatively short settling time tS = 0.4 s. The single real pole
( s1 = −0.2 ) exhibits an exponential decay e−0.2t which takes 20 s to die out to zero. Hence the
damped sinusoidal (second-order) component of the transient response of G1(s) is barely
noticeable and the first-order component e−0.2t dominates the transient response. It is easy to see
that the single pole of the first-order transfer function G2(s) is also s = −0.2 and hence its
transient response is also e−0.2t , which matches the dominant pole of G1(s).

Finally, note that the DC gain of the third-order system is G1(s = 0) = 72/24 = 3 which is also the
DC gain of the first-order system, i.e., G2(s = 0) = 0.6/0.2 = 3. So both transfer functions show
the same steady-state response to a constant input. For these reasons the first-order system G2(s)
is an excellent low-order approximation of the third-order system G1(s).

We can obtain the third-order and first-order step responses using the following MATLAB
commands:

>> sysG1 = tf(72,[1 20.2 124 24]); % define transfer function G1(s)
>> sysG2 = tf(0.6,[1 0.2]); % define transfer function G2(s)
>> t = 0:0.01:25; % define time vector
>> [y1,t] = step(sysG1,t) ; % unit-step response of G1(s)
>> [y2,t] = step(sysG2,t) ; % unit-step response of G2(s)
>> plot(t,y1,t,y2) % plot both unit-step responses

The plot is below. The two unit-step responses are almost identical, and hence G2(s) is an
excellent low-order approximate model for system G1(s).

3.5

2.5

2
Output y

Step response of G1(s)

Step response of G2(s)
1.5

0.5

0
0 5 10 15 20 25
Time, s

23
 Chapter 7

7.23 The mathematical model of the mechanical rotor in Problem 7.7 is J  + b = Tin (t )

A Simulink model of this system is shown below:

Note that we use an integrator block diagram because the rotor has initial angular velocity (thus
we cannot use a transfer function). The Step block has a magnitude of 0.4 N-m (input torque),
and the integrator is initialized at (0) = 20 rad/s. The simulation result is plotted below:

The simulation result for angular velocity matches the total response determined in Problem 7.7,
i.e., (t ) = 50 − 30e−t /7.5 rad/s. We can zoom in on the response plot at t = 1.073 s and see that
the angular velocity is 24 rad/s, which is 20% greater than the initial angular velocity.

24
 Chapter 7

7.24 SS a) The eigenvalues are computed from the determinant I − A = 0

 + 0.2 0.6
I − A = = 2 + 4.2 + 2 = 0
−2 +4

The eigenvalues are 1 = −0.5476 and 2 = −3.6524 .

b) The eigenvalues can be computed using the simple MATLAB commands; matches part (a)

>> A = [-0.2 -0.6 ; 2 -4]; % define the state matrix A

>> eig(A)

c) Because the two roots are real and negative, the free (zero-input) response will consist of two
decaying exponential functions, or y(t ) = c1e−0.5476t + c2e−3.6524t . The “slow” exponential
mode is e −0.5476t which takes about 7.3 s to decay to zero.

d) MATLAB commands are probably the easiest method: use the command initial

>> A = [-0.2 -0.6 ; 2 -4]; % define the state matrix A

>> B = [0 ; 1.5 ]; % define the input matrix B
>> C = [1 0]; % define the input matrix C
>> D = 0; % define the direct-link matrix D
>> sys = ss(A,B,C,D); % define sys as the SSR
>> x0 = [-2 ; -1]; % define the initial state vector
>> t = 0:0.01:12; % define the time vector
>> [y,t]=initial(sys,x0,t); % obtain the free response to the initial conditions
>> plot(t,y) % plot the free response

The plot (below) verifies the free response described in part (c).
0.5

-0.5
Output y

-1

-1.5

-2

-2.5
0 2 4 6 8 10 12
Time, s

25
 Chapter 7

7.25 a) The MATLAB commands define state matrix A and determine its eigenvalues:

>> A = [0 1 0 ; 0 0 1 ; -12 -20 -9]; % define the state matrix A

>> eig(A)

The three eigenvalues are –1, –2, and –6.

b) Because the three eigenvalues (roots) are real and negative, the free (zero-input) response will
consist of three decaying exponential functions, or y(t ) = c1e−t + c2e −2t + c3e −6t . The “slow”
exponential mode is e − t which takes about 4 s to decay to zero.

c) MATLAB commands are probably the easiest method: use the command initial

>> A = [0 1 0 ; 0 0 1 ; -12 -20 -9]; % define the state matrix A

>> B = [0 ; 0 ; 0.4 ]; % define the input matrix B
>> C = [1 0 0]; % define the input matrix C
>> D = 0; % define the direct-link matrix D
>> sys = ss(A,B,C,D); % define sys as the SSR
>> x0 = [2 ; -0.5 ; 0]; % define the initial state vector
>> t = 0:0.01:5; % define the time vector
>> [y,t]=initial(sys,x0,t); % obtain the free response
>> plot(t,y) % plot the free response

The plot below verifies the free response described in part (b).

1.8

1.6

1.4

1.2
Output, y

0.8

0.6

0.4

0.2

0
0 1 2 3 4 5
Time, s

26
 Chapter 7

7.26 a) The MATLAB commands define state matrix A and determine its eigenvalues:

>> A = [0 1 0 ; 0 0 1 ; -340 -88 -9]; % define the state matrix A

>> eig(A)

The three eigenvalues are –5, –2 + j8, and –2 – j8.

b) Because the three eigenvalues (roots) have negative real parts the free response will eventually
decay to zero. The free response will be the sum of one decaying exponential function and a
damped sinusoidal function, or y(t ) = c1e−5t + c2e−2t (sin 8t +  ) . The “slow” exponential mode is
e−2t which takes about 2 s to decay to zero. The free response will oscillate at frequency 8 rad/s.

c) MATLAB commands are probably the easiest method. The plot verifies part (b).

>> A = [0 1 0 ; 0 0 1 ; -340 -88 -9]; % define the state matrix A

>> B = [0 ; 0 ; 2]; % define the input matrix B
>> C = [1 0 0]; % define the input matrix C
>> D = 0; % define the direct-link matrix D
>> sys = ss(A,B,C,D); % define sys as the SSR
>> x0 = [-4 ; 0 ; 0]; % define the initial state vector
>> t = 0:0.01:5; % define the time vector
>> [y,t]=initial(sys,x0,t); % obtain the free response
>> plot(t,y) % plot the free response

0.5

-0.5

-1
Output, y

-1.5

-2

-2.5

-3

-3.5

-4
0 1 2 3 4 5
Time, s

27
 Chapter 7

7.27 a) Use the MATLAB commands to define state matrix A and determine its eigenvalues:
>> A = [0 1 0 0 ; -12 -2.6 12 2.6 ; 0 0 0 1 ; 12 2.6 -32 -2.6 ];
>> eig(A)

The four eigenvalues are 1,2 = −2.3136  j5.3864 and 3,4 = −0.2864  j 2.6271 . Because at
least two of the eigenvalues are a complex conjugate pair (here all four eigenvalues are complex,
i.e., two complex conjugate pairs), the output will exhibit oscillations during the transient
response.

b) The real part of the eigenvalues determines the decay rate to steady-state. The settling time is
dictated by the slowest eigenvalues, which in this case are 3,4 = −0.2864  j 2.6271 . The
associated decay rate for the transient response is e−0.2864t , which decays to zero ( e−4  0 ) at
time t = 13.97 s.

28
 Chapter 7

7.28 The system transfer function is

800
G(s) =
2s + 24s + 144s + 640
3 2

The characteristic roots are the poles of the transfer function, i.e., 2s3 + 24s2 + 144s + 640 = 0.
Using MATLAB, the roots are

>> denG = [2 24 144 640]; % denominator of G(s)

>> roots(denG)

The three roots are –8, –2 + j6, and –2 – j6. Because the two of the roots are complex the
transient response will exhibit oscillations with frequency 6 rad/s (period = 2/6 = 1.047 s). The
real root will contribute an exponential mode e−8t which dies out in 0.5 s. The complex roots
will contribute a damped sinusoidal mode with an exponential envelop of e−2t (dies out in 2 s).
Because the DC gain is G(s = 0) = 800/640 = 1.25 and the input is a unit step, the steady-state
response is ySS = 1.25. MATLAB’s step command can be used to obtain a numerical solution:

>> sysG = tf(800,[2 24 144 640]); % define transfer function G(s)

>> t = 0:0.002:3; % define time vector
>> [y,t] = step(sysG); % step response
>> plot(t,y) % plot y(t)

1.6

1.4

1.2

1
Output, y

0.8

0.6

0.4

0.2

0
0 0.5 1 1.5 2 2.5 3
Time, s

29
 Chapter 7

7.29 The system transfer function for the RC circuit is

EO ( s) 1 1
G (s) = = =
Ein ( s ) RCs + 1 0.01s + 1

The Simulink model simulates the step response to a 0.4-V input. The output voltage response
eO(t) matches the sketch in Problem 7.8.

0.5

0.4
Output voltage, V

0.3

0.2

0.1

0
0 0.01 0.02 0.03 0.04 0.05
Time, s

30
 Chapter 7

7.30 The 1-DOF mechanical system transfer function is

Y (s) 0.25
G( s) = = 2
U ( s) s + 2s + 9

a) Because the system is underdamped (note the poles are − 1  j 2.8284 ) we can compute the
step-response performance metrics using Table 7-4 (summarized below)

Undamped natural frequency n = 9 = 3 rad/s, damping ratio  = 2/(2n) = 0.3333.


Peak time t p = = 1.1107 s
n 1 −  2

Steady-state response = (step input in N)(DC gain) = (4 N)(0.25/9) = 0.1111 m

− / 1− 2
Maximum overshoot M OS = e = 0.3293 (or 32.9%)
→ peak value = (1.3293)(0.1111 m) = 0.1477 m

4
Settling time t S = =4s
 n

2
Period of oscillation Tperiod = = 2.2214 s
n 1 −  2
2 1−  2
Number of cycles to S-S N cycles = = 1.8 cycles


An accurate sketch would show these response values - see simulation plot in part (b).

b) The below Simulink model will produce the step response - see plot of y(t) on the next page.

31
 Chapter 7

0.16

0.14

0.12

0.1
Position, y, m

0.08

0.06

0.04

0.02

0
0 1 2 3 4 5 6
Time, s

32
 Chapter 7

7.31 The I/O equation of the 1-DOF mechanical system can be derived from the transfer
function G(s) in Problem 7.30:

y + 2 y + 9 y = 0.25u

a) The system is underdamped (note the roots are − 1  j 2.8284 ) and we can use the second-
order underdamped metrics from Table 7-4 (see Problem 7.22)

Undamped natural frequency n = 9 = 3 rad/s, damping ratio  = 2/(2n) = 0.3333.

Steady-state response is zero because input u = 0 and the damped sinusoid decays to zero.

4
Settling time t S = =4s
 n

2
Period of oscillation Tperiod = = 2.2214 s
n 1 −  2
2 1−  2
Number of cycles to S-S N cycles = = 1.8 cycles


An accurate sketch would show these response values - see simulation plot in part (b) on
the next page.

b) We cannot use transfer functions because the system has non-zero initial conditions. The
below Simulink model uses integrator blocks and will produce the free response. Note that the
input is zero and the integrators must have initial conditions y(0) = 0.04 m and y (0) = 0 .

33
 Chapter 7

0.04

0.03

Position, y, m 0.02

0.01

-0.01

-0.02
0 1 2 3 4 5 6
Time, s

The simulation result from executing the Simulink model is plotted above. The simulation
verifies the settling time, period, and number of cycles to steady state.

34
 Chapter 7

7.32 The mathematical model of the frictionless mechanical system is

J + k = Tin (t )

a) Using the numerical values for J and k the transfer function is

( s ) 1
G ( s) = =
Tin ( s ) 0.2s + 100
2

Hence the undamped natural frequency is n = 100 / 0.2 = 22.3607 rad/s. The damping ratio 
is zero because there is no friction. Hence, the homogeneous (or natural) response will be an
undamped harmonic (sinusoidal) function with the form

 H (t ) = c1 sin 22.3607t + c2 cos 22.3607t rad

The forced response or particular solution (due to the sinusoidal input torque) will be a sinusoidal
function with frequency 3 rad/s (same as the input) with unknown coefficients:

P (t ) = c3 sin 3t + c4 cos3t rad

Hence the total solution is the sum of sinusoidal function at two frequencies, n and 3 rad/s.

 (t ) = c1 sin 22.3607t + c2 cos22.3607t + c3 sin 3t + c4 cos3t rad

b) The Simulink model (below) will produce the desired system response.

The complete response (t) is shown on the next page. Note that the response is a summation of
two harmonic (sinusoidal) functions: the first harmonic function has a frequency of 3 rad/s
(period = 2.1 s) due to the torque input and the second harmonic function has frequency 22.36
rad/s (period = 0.28 s) due to k/J. This dual-frequency harmonic function matches the result in
part (a).

35
 Chapter 7

-3
x 10
8

Angular position,, rad 4

-2

-4

-6

-8
0 2 4 6 8 10
Time, s

36
 Chapter 7

7.33 The I/O equation of a linear second-order mechanical system is

f a (t )
x + 2 n x + n2 x =
m

The numerical values determined in Problem 7.17 are  = 0.1221 (damping ratio) and n =
15.8264 rad/s (undamped natural frequency), and mass m = 0.2 kg. Hence the transfer function is

X (s) 1/ m 5
G (s) = = 2 = 2
Fa ( s ) s + 2 n s + n s + 3.865s + 250.47
2

The Simulink model below shows that two step functions are added to create a pulse input. One
step input is delayed by 0.01 s.

The magnitude of the pulse input was varied until the peak response of x(t) nearly matched 0.026
m (the peak response from the plot in Problem 7.17). A 10-N pulse input with pulse width of
0.01 s produced the following response (below) which nearly matches the plot in Problem 7.17.
Because the pulse magnitude is 10 N and pulse width is 0.01 s, the strength of the impulse input
is (10 N)(0.01 s) = 0.1 N-s.

0.03

0.025

0.02

0.015
Displacement, x, m

0.01

0.005

-0.005

-0.01

-0.015

-0.02
0 0.5 1 1.5 2 2.5 3
Time, s

37
 Chapter 7

7.34 SS a) The system transfer function is

780 Y ( s)
G(s) = =
s + 29.5s + 119s + 765.5s + 975 U (s)
4 3 2

The four poles (roots) are computed from s + 29.5s + 119s + 765.5s + 975 = 0
4 3 2

Poles: s = −1.5, −26, −1  j 4.899

The “fast” pole (or root) at s = –26 will have a very fast exponential decay, e−26t , that reaches
steady-state in about 4/26 = 0.154 s. Hence this mode is “fast” and can be neglected.

Consequently, the denominator of the 3rd-order model must be

(s + 1.5)(s + 1 − j 4.899)(s + 1 + j 4.899) = (s + 1.5)(s 2 + 2s + 25) = s3 + 3.5s 2 + 28s + 37.5

The DC gain of the original 4th-order transfer function is 780/975 = 0.8. The third-order transfer
function must have the same DC gain. Therefore, its numerator must be (0.8)(37.5) = 30. The
third-order transfer function is

30 Y ( s)
G3 ( s) = =
s + 3.5s + 28s + 37.5 U ( s)
3 2

b) Next, we execute the below Simulink model to determine the unit-step responses of the
original 4th-order system and reduced-order (3rd-order) system.

A plot showing both the 4th-order and 3rd-order step responses is on the next page.

38
 Chapter 7

The two responses show a good match.

c) It is impossible to create an accurate 2nd-order reduced-order system for G(s). The step-
response plot in part (b) shows that the total response is a sum of an exponential function
(i.e., mode e−1.5t ) and a damped sinusoidal function (i.e., e−t sin(4.899t +  ) ). Both the
exponential and damped sinusoidal functions have the same time scale (the exponential
function reaches steady-state in 2.667 s, while the damped sinusoid reaches steady-state in 4
s). Therefore, we cannot remove either of these modes from the total response. An
underdamped second-order response will show a peak overshoot at the first one-half cycle,
and the total response in part (b) does not exhibit this behavior.

39
 Chapter 7

7.35 The mathematical model of the thermal system is (see Example 4.7)

REQ CT + T = REQ q BH + Ta

6
1 1 1 1 1
The equivalent thermal resistance of all surfaces is = + ++ =
REQ R1 R2 R6 i =1 Ri

Or, REQ = 0.018014 deg C-s/J. Hence the time constant of the first-order system is  = REQC =
8,700.8 s. The settling time is tS = 4 = 34,803 s = 9.67 hrs. At steady-state we have T = 0 so
the first-order model becomes

T () = REQ qBH + Ta = (0.018014 deg C-s/J)(1000 J/s) + 10 deg C = 28.01 deg C.

The settling time and steady-state temperature match the numerical results from Problem 6.25. A
sketch of T(t) would be an exponential rise from 10 deg C to 28 deg C with settling time of 9.7
hrs (solution from Problem 6.26 is repeated below).

30

28

26
Room temperature, deg C

24

22

20

18

16

14

12

10
0 5 10 15 20
Time, hrs

40
 Chapter 7

7.36 The mathematical model of the RC circuit (see Problem 3.32) is RCeC + eC = 0

Using the values R = 10  and C = 185(10-6) F, the time constant is  = RC = 0.00185 s. Hence
the settling time is tS = 4 = 0.0074 s. Clearly at steady state eC = 0 and therefore eC () = 0 .
A sketch of the capacitor voltage eC(t) would be an exponential decay from eC(0) = 200 V to zero
at approximately 0.0074 s. The mathematical solution is eC (t ) = 200e−t /  = 200e−540.54t V.

The current is I (t ) = CeC which is the product of the slope (time derivative) of eC(t) and
capacitance C. The derivative is eC (t ) = −1.0811(105 )e−540.54t so I (t ) = CeC = − 20e−540.54t A.
A sketch of the current I(t) would be an exponential rise from –10 A to zero at about tS = 0.0074
s. Sketches would match the plots below.

220
200
180
160
Capacitor voltage, e , V
C

140
120
100
80
60
40
20

0
0 0.002 0.004 0.006 0.008 0.01
Time, s

-5
Current, A

-10

-15

-20

0 0.002 0.004 0.006 0.008 0.01

Time, s

41
 Chapter 7

7.37 SS a) The power amplifier is a first-order I/O system: 0.003eO + eO = 25ein (t )

Clearly the time constant is  = 0.003 s and therefore the settling time is tS = 4 = 0.012 s. The
DC gain of the amplifier transfer function is 25 and therefore the steady-state voltage for a 0.2 V
step input is 5 V. The voltage response eO(t) will exponentially reach 5 V in 0.012 s. A hand-
sketch would match the below plot.

4
Amplifier output voltage, V

0
0 0.005 0.01 0.015
Time, s

b) We can compute the steady-state spool-valve position from the individual DC gains of the
three transfer functions: power amp DC gain = 25, solenoid DC gain = 1.6, and spool-valve
DC gain = 1/1800 = 5.5556(10–4). The product of the three DC gains is 0.0222. Hence the
steady-state valve position is the product of the total DC gain (0.0222) and magnitude of the
step input voltage, ein(t) = 0.2 V, or, (0.0222)(0.2) = 0.0044 m = 4.4 mm.

c) The I/O equation for the spool valve is 0.035z + 7 z + 1800z = f (t )

Dividing by 0.035 we obtain the standard form for a second-order system

z + 200z + 51,428.6z = 28.571 f (t ) or z + 2n z + n2 z = 28.571 f (t )

The undamped natural frequency is n = 226.7787 rad/s and the damping ratio is  = 200/(2n) =
0.441 and hence the spool valve step response is a damped sinusoidal that eventually settles at the
steady-state value of 0.0044 m (4.4 mm) computed in part (b). The transient step-response
characteristics can be computed using the performance equations in Table 7-4:


Peak time: t p = = 0.0154 s
n 1 −  2

42
 Chapter 7

− / 1− 2
Maximum overshoot: M OS = e = 0.2136, or zmax = (1.2136)(0.0044) = 0.0053 m

4
Settling time: tS = = 0.04 s
 n

2
Period of oscillation: Tperiod = = 0.0309 s
n 1 −  2

2 1−  2
Number of cycles to steady state: Ncycles = ~ 1.3 cycles

A hand-sketch would show (with labels) the above calculated values and match the below plot.

-3
x 10
6
Valve position, z, m (approximate sketch)

0
0 0.01 0.02 0.03 0.04 0.05 0.06
Time, s

Finally, we can verify the approximate hand-sketch (i.e., step response of the 2nd-order spool
valve model) by a numerical simulation of the complete system using Simulink (next page):

43
 Chapter 7

The step response of the complete Simulink model is below. Note that the steady-state value
(0.0044 m) matches the approximate sketch. However, the addition of two first-order models
(power amp and solenoid) has delayed the response. Note that the amplifier and solenoid have
time constants of 0.003 and 0.002 s, respectively, which correspond to settling times around
0.012 and 0.008 s. Therefore, the second-order spool valve response is delayed (compare the
peak times between the approximate sketch and the full simulation). The peak response of the
full simulation is also diminished (damped) compared to the approximate sketch.

-3
x 10
6
Valve position, z, m (Simulink result)

0
0 0.01 0.02 0.03 0.04 0.05 0.06
Time, s

44
 Chapter 7

7.38 The measured motor characteristics are a) steady-state ss = 85 rad/s for a 2-V step voltage
input, b) settling time tS = 0.6 s, and c) the speed response from zero to 85 rad/s is an exponential
rise.

Because the response is exponential we can model the system using the “standard form” of a first-
order transfer function:

 (s) a
G (s) = =
Ein ( s ) s + 1

The time constant is  = tS/4 = 0.15 s and the DC gain is G(0) = a = (85 rad/s)/(2 V) = 42.5.
Hence an appropriate transfer function model of the DC motor is

 ( s) 42.5
G ( s) = =
Ein ( s ) 0.15s + 1

The following MATLAB commands simulate the step response of this transfer function model.

>> sysG = tf(42.5,[ 0.15 1 ]); % define transfer function G(s)

>> t = 0:0.001:1; % define time vector t
>> e_in = 2*ones(size(t)); % define 2-V step input ein(t)
>> [w,t] = lsim(sysG,e_in,t); % obtain response (t), output speed

The plot of the simulation results (below) verifies the experimental characteristics of the DC
motor.

90

80

70
DC motor speed, rad/s

60

50

40

30

20

10

0
0 0.2 0.4 0.6 0.8 1
Time, s

45
 Chapter 7

7.39 The model of the mass-spring-damper system is mx + bx + kx = f a

The standard form of the second-order system is

b k 1 1
x + x + x = f a or x + 2 n x + n2 x = fa
m m m m

The mass is m = 3 kg. Maximum overshoot for a step input only depends on damping ratio 
(see Table 7-4). Substitute each value of b, k and compute damping ratio:

b 90
Option A:  = = = 0.2444
2 mk 2 (3)(11,300)

b 88
Option B:  = = = 0.2724
2 mk 2 (3)(8,700)

Because Option B has the largest damping ratio it has the smallest percent overshoot to a step
input.

46
 Chapter 7

7.40 a) The measured output shows a classic underdamped response which we can model with
a second-order transfer function:

( s ) an2
G( s) = =
Tin ( s) s 2 + 2 n s + n2

The steady-state output is 0.012 rad for a step input torque of 0.6 N-m. Hence the DC gain of the
transfer function is a = 0.012/0.6 = 0.02.

The peak response is 0.016 rad which is a 33.333% overshoot of the steady-state value of 0.012
rad. The maximum overshoot is

M OS = e− / 1− 2
= 0.333333 → damping ratio  = 0.3301

The period of the response is 1.1 s. Therefore the damped frequency is d = 2 /1.1 = 5.712
rad/s. The undamped natural frequency is n = d / 1 −  2 = 6.0512 rad/s. The second-order
transfer function is

( s ) 0.73234
G ( s) = = 2
Tin ( s ) s + 3.995s + 36.617

b) The following MATLAB commands simulate the step response of this transfer function
model.

>> sysG = tf(0.73234,[1 3.995 36.617]); % define transfer function G(s)

>> t = 0:0.002:3; % define time vector t
>> T_in = 0.6*ones(size(t)); % define 0.6 N-m step input Tin(t)
>> [theta,t] = lsim(sysG,T_in,t); % obtain response (t)

The simulation result (plotted below) shows a good match with Fig. P7.40; therefore the second-
order transfer function G(s) is an accurate representation of the mechanical system.
0.02

0.018

0.016
Angular displacement, rad

0.014

0.012

0.01

0.008

0.006

0.004

0.002

0
0 0.5 1 1.5 2 2.5 3
Time, s

47