Chapter Four

Transient and Steady-State Response Analysis

Ogata

Introduction

4-1
Ogata

Dorf

The Concept of Stability

Dorf

Absolute Stability and Relative Stability

4-2
Definitions of Absolute Stability

Definition 1

Dorf

Dorf

Example

Golnaraghi-Ogata

A liquid-level system is shown in Figure 4-45. If ℎ, 𝑞𝑖 , 𝑞𝑜 , 𝑅, and 𝐶 are the fluid height in the
tank, the volume flow rate at the valve inlet, the volume flow rate at the valve outlet, the
valve resistance, and capacitance of the tank. If 𝑞𝑖 is considered the input and ℎ the output,
the transfer function of the system is
𝐻(𝑠) 𝑅
= (*)
𝑄𝑖 (𝑠) 𝑅𝐶𝑠+1

If, however, 𝑞𝑜 is taken as the output, the input being the same, then the transfer function is
𝑄𝑜 (𝑠) 1
= (**)
𝑄𝑖 (𝑠) 𝑅𝐶𝑠+1

4-3
This system is stable, since for any bounded input 𝑞𝑖 , the height ℎ is bounded. This can be
demonstrated by assuming that initially the input 𝑞𝑖 and the outputs 𝑞𝑜 and ℎ have constant
values (note that 𝑞𝑜 must equal 𝑞𝑖 for the system to be in a stationary state). If the value of
the input 𝑞𝑖 increases by a step change, the output 𝑞𝑜 will increase gradually (equation (**))
until it reaches its steady-state value, 𝑞𝑜 = 𝑞𝑖 ; during this period, and before 𝑞𝑜 reaches 𝑞𝑖 ,
the output ℎ increases (equation (**)). The output ℎ reaches its steady-state value when 𝑞𝑜 =
𝑞𝑖 .

This system can become unstable, if the outlet valve is closed (𝑞𝑜 = 0). In this case, for any
nonzero bounded input 𝑞𝑖 , the liquid level ℎ will increase without bound until the tank
overflow. Thus, although the input is bounded, the output is unbounded, and the system is
unstable.

Golnaraghi-Ogata

Definition 2

Ogata

Ogata

Hendrick

Example

4-4
Hendrick

Ogata

Steady-State Error

Ogata

As we will see later, for a linear time-invariant system, we can determine that the system is
stable (in the absolute sense) by determining that all transfer function poles lie in the left-half
s-plane, or equivalently, that all the eigenvalues of the system matrix 𝑨 lie in the left-half s-
plane. Given that all the poles (or eigenvalues) are in the left-half s-plane, we investigate
relative-stability by examining the relative locations of the poles (or eigenvalues).

4-5
Ogata

First-Order Systems

𝐶(𝑠)
The pole of is determined by setting its denominator equal to 0, i.e. 𝑇𝑠 + 1 = 0,
𝑅(𝑠)
1
which gives 𝑠=− .
𝑇

𝑡
𝑐 (𝑡 ) = 1 − 𝑒 − 𝑇 , for 𝑡 ≥ 0 (5 − 3)
1
(−𝑇)𝑡
=1− 𝑒 = 1 − 𝑒 pole × 𝑡

4-6
4-7
4-8
Ogata

Note: The DC gain of a system is defined as the gain of the system that relates the constant
steady-state output to the constant steady-state input. Thus, for a system that has transfer
function 𝐺(0), the DC gain can be determined as follows:

𝐶 (𝑠) = 𝐺 (𝑠)𝑅(𝑠)
𝐴
For a constant input 𝑟(𝑡) = 𝐴, we have 𝑅 (𝑠) = , and
𝑠

4-9
 𝐺(𝑠)
𝐶 (𝑠) = 𝐴
𝑠

From the final value theorem

𝐺 (𝑠)
𝑐𝑠𝑠 = lim 𝑠𝐶(𝑠) = lim 𝐴𝑠 = lim 𝐴𝐺 (𝑠) = 𝐴𝐺 (0)
𝑠→0 𝑠→0 𝑠 𝑠→0

𝐴𝐺(0)
DC gain = = 𝐺 (0)
𝐴

∴ DC gain = 𝐺 (0)

For the first-order system given by equation (5-1), rewritten here for convenience
𝐶(𝑠) 1
=
𝑅(𝑠) 𝑇𝑠+1

1
the DC gain is = 1.
𝑇(0)+1

Note: The general form of the transfer function of a first-order system is

𝐶(𝑠) 𝐾
=
𝑅(𝑠) 𝑇𝑠+1

𝐾 1
The DC gain in this case is = 𝐾. The time constant is 𝑇, and the pole is 𝑠 = − .
𝑇(0)+1 𝑇
The response of this system to a unit-step input is
𝑡
𝑐 (𝑡) = 𝐾 (1 − 𝑒 −𝑇 )

Second-Order Systems

In this section, we shall obtain the response of a typical second-order control system to a step
input, ramp input, and impulse input. The closed-loop transfer function of the system shown
in Figure 5-6 a is

𝐶(𝑠) 2
𝜔𝑛
= 2 (5.10)
𝑅(𝑠) 𝑠 2 +2𝜁𝜔𝑛 𝑠+𝜔𝑛

which is shown in Figure 5-6 b, where 𝜁 and 𝜔𝑛 are called the damping ratio and undamped
natural frequency, respectively. The form (5-10) is called the standard form of the second-
order system.
4-10
 Figure 5.6. Second-order system

𝐶(𝑠)
The poles of are
𝑅(𝑠)
𝑠1,2 = −𝜁𝜔𝑛 ± 𝑗𝜔𝑛 √1 − 𝜁 2 = −𝜎 ± 𝑗𝜔𝑑 (*)

where 𝜎 and 𝜔𝑑 are the attenuation and the damped natural frequency, respectively. These
𝜎 𝜁𝜔𝑛
poles are located at a radius 𝜔𝑛 in the s-plane and at angle 𝛽 = cos −1 = cos −1 =
𝜔𝑛 𝜔𝑛

𝜔𝑑 𝜔𝑛 √1−𝜁 2 √1−𝜁 2
cos −1𝜁 = tan−1 = tan−1 = tan−1 , as shown in Figure (6-21).
𝜎 𝜁𝜔𝑛 𝜁

Figure 6-21 Complex poles

Note: If the values of 𝜁 and 𝜔𝑛 are known, the corresponding quadratic factor can be formed
as

4-11
 𝑠 2 + 2𝜁𝜔𝑛 𝑠 + 𝜔𝑛2 (**)

and the complex-conjugate poles can be deduced using formula (*). On the other hand, if the
complex-conjugate poles −𝜎 ± 𝑗𝜔𝑑 are known, then the corresponding quadratic factor can
be formed as

(𝑠 − (−𝜎 + 𝑗𝜔𝑑 ))(𝑠 − (−𝜎 − 𝑗𝜔𝑑 )) (***)

and the values of 𝜁 and 𝜔𝑛 can be deduced using the formulas (Figure 6-21)

𝜔𝑛2 = 𝜎 2 + 𝜔𝑑2 ⟹ 𝜔𝑛 = √𝜎 2 + 𝜔𝑑2

𝜎 𝜎
𝜎 = 𝜁𝜔𝑛 ⟹ 𝜁= =
𝜔𝑛 √𝜎 2 + 𝜔2
𝑑

Mandal Ogata

Locus of Roots for the Second-Order Standard System and their effect on the Transient
Response of the system to a step input

1- Constant 𝝎𝒏 and Variable 𝜻: Locus and Transient Response

From equation (*), it is apparent that, with 𝜔𝑛 held constant, if 𝜁 is varied from 0 to +∞, the
two roots of the characteristic equation will originate from ±𝑗𝜔𝑛 corresponding to 𝜁 = 0,
move along the arc of a semicircle of radius 𝜔𝑛 and meet at −𝜔𝑛 on the real axis
corresponding to 𝜁 = 1. When 𝜁 is further increased, the two branches move away from each
other in opposite directions along the real axis; one branch terminates at 𝑠 = 0 and the other
terminates at 𝑠 = −∞ as 𝜁 approaches +∞ (see figure 4.5(a)).

4-12
 Figure 4.5(a)

There are four different cases when 𝜔𝑛 is constant and 𝜁 is variable; for each case, we shall
solve for the response of the system given by equation (5.10 Ogata) to a unit step input.

1- Underdamped case (0 < 𝜁 < 1): The closed-loop poles are given by

𝑠1,2 = −𝜁𝜔𝑛 ± 𝑗𝜔𝑛 √1 − 𝜁 2

which are complex conjugates and lie on the semicircle in the left-half s-plane as shown in
1
Figure 4.5 (a). The system is referred to as underdamped. For a unit-step input, 𝑅 (𝑠) = and
𝑠
𝐶(𝑠) can be written as
2
𝜔𝑛
𝐶(𝑠) = 2 )𝑠 (5.11)
(𝑠 2 +2𝜁𝜔𝑛 𝑠+𝜔𝑛

The inverse Laplace transform of equation (5.11) can be obtained easily if 𝐶(𝑠) is written in
the follwing form (using partial fractions)
1 𝑠+𝜁𝜔𝑛 𝜁𝜔𝑛 𝜔𝑑 1 𝑠+𝜁𝜔𝑛 𝜁 𝜔𝑑
𝐶 (𝑠) = − (𝑠+𝜁𝜔 2 2 − 2 = − (𝑠+𝜁𝜔 2 − 2
𝑠 𝑛 ) +𝜔𝑑 𝜔𝑑 (𝑠+𝜁𝜔𝑛 )2 +𝜔𝑑 𝑠 2
𝑛 ) +𝜔𝑑 √1−𝜁 2 (𝑠+𝜁𝜔𝑛 )2 +𝜔𝑑

Referring to the Laplace transform table in appendix A, it can be shown that

𝜁
𝑐 (𝑡) = ℒ −1{𝐶 (𝑠)} = 1 − 𝑒 −𝜁𝜔𝑛𝑡 cos 𝜔𝑑 𝑡 − 𝑒 −𝜁𝜔𝑛𝑡 sin 𝜔𝑑 𝑡
√1−𝜁 2

𝜁
= 1 − 𝑒 −𝜁𝜔𝑛 𝑡 (cos 𝜔𝑑 𝑡 + sin 𝜔𝑑 𝑡)
√1−𝜁 2

𝜁
= 1 − 𝑒 Re(pole)×𝑡 (cos(Img (pole) × 𝑡) + sin(Img (pole)) × 𝑡) , for 𝑡 ≥ 0
√1−𝜁 2
(5.12)

From Equation (5-12), it can be seen that the frequency of transient oscillation is the damped
natural frequency 𝜔𝑑 and thus varies with the damping ratio 𝜁. The error signal for this
system is the difference between the input and output and is

𝜁
𝑒 (𝑡) = 𝑟(𝑡) − 𝑐 (𝑡) = 𝑒 −𝜁𝜔𝑛 𝑡 (cos 𝜔𝑑 𝑡 + sin 𝜔𝑑 𝑡) , for 𝑡 ≥ 0
√1 − 𝜁2

This error signal exhibits a damped sinusoidal oscillation. At steady state, or at 𝑡 = ∞, no

error exists between the input and output.

4-13
2- Critically damped case (𝜁 = 1): The closed-loop poles are given by

𝑠1,2 = −𝜁𝜔𝑛 = −𝜔𝑛

The system is said to be critically damped one. In this case, both the roots are equal, as
1
indicated in Figure 4.5(a). For a unit-step input, 𝑅 (𝑠) = and 𝐶(𝑠) can be written as
𝑠

𝜔2
𝐶(𝑠) = (𝑠+𝜔𝑛 )2 (5.14)
𝑛 𝑠

The inverse Laplace transform of equation (5.14) may be found as

𝑐 (𝑡) = 1 − 𝑒 −𝜔𝑛𝑡 (1 + 𝜔𝑛 𝑡) = 1 − 𝑒 pole×𝑡 (1 + 𝜔𝑛 𝑡), for 𝑡 ≥ 0 (5.15)

This result can also be obtained by letting 𝜁 approach unity in Equation (5-12) and by using
the following limit:

sin 𝜔𝑑 𝑡 sin 𝜔𝑛 √1−𝜁 2 𝑡

lim = lim = 𝜔𝑛 𝑡
𝜁→1 √1−𝜁 2 𝜁→1 √1−𝜁 2

The step response is monotonic without any overshoot and undershoot.

3- Overdamped case (𝜁 > 1): The closed-loop poles are given by

𝑠1,2 = −𝜁𝜔𝑛 ∓ 𝜔𝑛 √𝜁 2 − 1

Both the roots are lying on the real axis. The location of the poles in the s-plane are indicated
1
in Figure 4.5(a). For a unit-step input, 𝑅(𝑠) = and 𝐶(𝑠) can be written as
𝑠
2
𝜔𝑛
𝐶(𝑠) = (5.16)
(𝑠+𝜁𝜔𝑛 +𝜔𝑛 √𝜁 2 −1)(𝑠+𝜁𝜔𝑛 −𝜔𝑛 √𝜁 2 −1)𝑠

The inverse Laplace transform of equation (5.16) can be obtained easily if 𝐶(𝑠) is written in
the follwing form (using partial fractions)
1 1

2√𝜁2 −1(𝜁+√𝜁2 −1) 2√𝜁2 −1(𝜁−√𝜁2 −1)

1
𝐶 (𝑠) = + −
𝑠 𝑠+𝜁𝜔𝑛 +𝜔𝑛 √𝜁 2 −1 𝑠+𝜁𝜔𝑛 −𝜔𝑛 √𝜁 2 −1

Referring to the Laplace transform table in appendix A, it can be shown that

1 2 −1)𝑡 1 2 −1)𝑡
𝑐 (𝑡) = 1 + 𝑒 −(𝜁𝜔𝑛 +𝜔𝑛 √𝜁 − 𝑒 −(𝜁𝜔𝑛 −𝜔𝑛 √𝜁
2√𝜁 2 −1(𝜁+√𝜁 2 −1) 2√𝜁 2 −1(𝜁−√𝜁 2 −1)

𝜔𝑛 𝑒 pole1 ×𝑡 𝑒 pole2 ×𝑡 𝜔𝑛 𝑒 pole1 ×𝑡 𝑒 pole2 ×𝑡

=1+ ( − )=1+ ( + ) , , for 𝑡 ≥ 0 (5-17)
2√𝜁 2 −1 −pole1 −pole2 2√𝜁 2 −1 −pole1 pole2

4-14
 where pole1 = −𝜁𝜔𝑛 + 𝜔𝑛 √𝜁 2 − 1 and pole2 = −𝜁𝜔𝑛 − 𝜔𝑛 √𝜁 2 − 1. Thus, the
response includes two decaying exponential terms.

When 𝜁 is appreciably greater than unity, one of the two decaying exponentials decreases
much faster than the other, so the faster-decaying exponential term (which corresponds to
a smaller time constant) may be neglected. That is, if pole1 is located very much closer
to the 𝑗𝜔 axis than pole2 (which means that |pole1 | ≪ |pole2 | ), then for an
approximate solution we may neglect pole2 . This is permissible because the effect of
pole2 on the response is much smaller than that of pole1 , since the term involving pole2
in equation (5-17) decays much faster than the term involving pole1 . Once the faster-
decaying exponential term has disappeared, the response is similar to that of a first-order
𝐶(𝑠)
system, and may be approximated by
𝑅(𝑠)

𝐶(𝑠) −pole1
=
𝑅(𝑠) 𝑠−pole1

−pole1
𝐶(𝑠) =
(𝑠−pole1 )𝑠

The time response 𝑐(𝑡) is then

𝑐 (𝑡) = 1 − 𝑒 pole1 ×𝑡 , for 𝑡 ≥ 0

4- Undamped case (𝜁 =0): The closed-loop poles are given by

𝑠1,2 = ±𝑗𝜔𝑛

The two poles are located on the imaginary axis as indicated in Figure 4.5(a). For a unit-step
1
input, 𝑅 (𝑠) = and 𝐶(𝑠) can be written
𝑠
2
𝜔𝑛
𝐶(𝑠) = (𝑠2 2 )𝑠 (**)
+𝜔𝑛

4-15
The inverse Laplace transform of equation (**) can be obtained easily if 𝐶(𝑠) is written in
the follwing form (using partial fractions)
1 2
𝜔𝑛
𝐶 (𝑠) = − 2
𝑠 𝑠 2 +𝜔𝑛

Referring to the Laplace transform table in appendix A, it can be shown that

𝑐 (𝑡) = 1 − cos 𝜔𝑛 𝑡 = 1 − cos(Img (pole) × 𝑡) , for 𝑡 ≥ 0 (5.13)

This result can also be obtained by substituting 𝜁 = 0 in Equation (5-12). In this case, the
response becomes undamped, and oscillations continue indefinitely.

Figure 5-7 Unit-step response curves of the system shown in Figure 5-6

4-16
2- Constant 𝜻 and Variable 𝝎𝒏 : Locus and Transient Response

The complex conjugate roots 𝑠1 and 𝑠2 are shown in Figure 4.5(b) for the case 0 < 𝜁 < 1. If
𝜁 is held constant and 𝜔𝑛 is varied, the roots will move on straight lines making an angle
𝛽 = cos −1 𝜁 as shown in Figure 4.5(b).

Figure 4.5(b)

4-17
Figure (*) shows the response of a second-order system to a step input for different values of
𝜔𝑛 and constant value of 𝜁 (𝜁 = 0.5). Note that the percentage overshoot is constant. This
constant overshoot can be demonstrated by investigating equation 5-21 Ogata, given later.

Figure (*) Step response of a second-order system with constant 𝜁 and variable 𝜔𝑛

3- Variable 𝜻 and 𝝎𝒏 but Constant 𝜻𝝎𝒏 : Locus and Transient Response

A glance at Equation (*) reveals that the real part of the roots can be kept constant when 𝜁𝜔𝑛
is maintained to a constant value with various combinations of 𝜁 and 𝜔𝑛 . The locus of roots
will appear as shown in Figure 4.5(c).

Figure 4.5(c)

4-18
Figure (**) shows the response of a second-order system to a step input for different values
of 𝜔𝑛 and 𝜁 and constant value of 𝜁𝜔𝑛 (𝜁𝜔𝑛 = 1.2). A constant value of 𝜁𝜔𝑛 will be found
to produce a constant settling time in the transient response to a step input. This constant
settling time can be demonstrated by investigating equation 5-221 Ogata, given later.

Figure (**) Step response of a second-order system with variable 𝜁 and 𝜔𝑛 keeping 𝜁𝜔𝑛
constant

Mandal Ogata

Note: For the second-order system given by equation (5-10), rewritten here for convenience

𝐶(𝑠) 2
𝜔𝑛
= 2 ,
𝑅(𝑠) 𝑠 2 +2𝜁𝜔𝑛 𝑠+𝜔𝑛

2
𝜔𝑛
the DC gain is 2 = 1.
02 +2𝜁𝜔𝑛 (0)+𝜔𝑛

Houpis

4-19
4-20
Houpis

Ogata

4-21
A specification, called decay ratio 𝑑, is given in [PID Controllers: Theory, Design, and
Tuning. Astrom], which is defined as the ratio between two consecutive maxima of the

4-22
 1
error for a step change in setpoint. The value 𝑑 = , which is called quarter amplitude
4
damping, has been used traditionally. This specification is shown in figure 4.3.

4-23
 √1 − 𝜁 2
tan 𝜔𝑑 𝑡𝑟 = −
𝜁

Consider Figure 6-21. Note that

√1 − 𝜁 2 𝜔𝑑
=
𝜁 𝜎

and
𝜔𝑑
tan(𝜋 − 𝛽) = −
𝜎
therefore

tan 𝜔𝑑 𝑡𝑟 = tan(𝜋 − 𝛽)

and

𝜔𝑑 𝑡r = 𝜋 − 𝛽

Thus, the rise time is

𝜋 − 𝛽 𝜋 − cos −1𝜁
𝑡r = =
𝜔𝑑 𝜔𝑑

Clearly, for a small value of 𝑡𝑟 , 𝜔𝑑 must be large.

Ogata

4-24
Golnaraghi

It is more difficult to determine the exact analytical expression of the rise time 𝑡r (10%-90%),
even for just the simple prototype second-order system. However, the exact value can be
determined directly from the responses of Figure 5-7 (Ogata). The plot of 𝜔𝑑 𝑡r versus 𝜁 is
shown in Figure 5-22. In this case, the relation can be approximated by a straight line over
the range of 0 < 𝜁 < 1:
0.8 + 2.5𝜁
𝑡r = 0<𝜁<1
𝜔𝑛

A better approximation can be obtained by using a second-order equation:

1 − 0.4167𝜁 + 2.917𝜁 2
𝑡r = 0<𝜁<1
𝜔𝑛

Delay time

It is also difficult to determine the exact analytical expression of the delay time 𝑡d , since we
would have to set 𝑐 (𝑡) =0.5 in equation 5-12 (Ogata) and solve for 𝑡. A easer way would be
to plot 𝜔𝑑 𝑡d versus 𝜁, as shown in Figure 5-21, and then approximate the curve by a straight
line or a curve over the range of 0 < 𝜁 < 1. From Figure 5-21, the delay time for the
prototype second-order system is approximated as
1 + 0.7𝜁
𝑡d ≈ 0<𝜁<1
𝜔𝑛

We can obtain a better approximation by using a second-order equation

4-25
 1.1 + 0.125𝜁 + 0.469𝜁 2
𝑡d = 0<𝜁<1
𝜔𝑛

The following conclusions can be made on the rise time and delay time of the prototype
second-order system:

• 𝑡r and 𝑡d are proportional to 𝜁 and inversely proportional to 𝜔𝑛 .

• Increasing (decreasing) the natural undamped frequency 𝜔𝑛 will reduce (increase) 𝑡r
and 𝑡d .

Golnaraghi

4-26
Ogata

𝜋 𝜁 𝜎
−𝜁𝜔𝑛 −( )𝜋
𝑀𝑝 = 𝑐(𝑡𝑝 ) − 1 = −𝑒 𝜔𝑑 (cos 𝜋 + sin 𝜋) = 𝑒 𝜔𝑑
√1 − 𝜁 2
𝜁
− 𝜋
=𝑒 √1−𝜁 2

𝜁
− 𝜋
√1−𝜁2
The maximum percent overshoot is 𝑒 × 100%

4-27
 𝜁
𝑐 (𝑡) = 1 − 𝑒 −𝜁𝜔𝑛𝑡 (cos 𝜔𝑑 𝑡 + sin 𝜔𝑑 𝑡) , for 𝑡 ≥ 0
√1−𝜁 2

4-28
4-29
 Figure 5-12 𝑀𝑝 versus 𝜉 curve.

Note: a 1% criterion is also used. Its formula is deduced as follows (Franklin):

𝑒 −𝜉𝜔𝑛𝑡𝑠 = 0.01
4.6
𝑡𝑠 =
𝜉𝜔𝑛

Decay Ratio

The decay ratio is given by [PID Controllers: Theory, Design, and Tuning. Astrom]
−2𝜋𝜁
√1−𝜁 2
𝑑= 𝑒

4-30
Rise time 𝑡r (0%-100%): The rise time is
𝜋−cos−1 𝜁 𝜋−cos−1 0.6
𝑡r = = = 0.55 s
𝜔𝑑 4

Figure 5.6. Second-order system

4-31
 Solution

The value of 𝜁 that corresponds to a maximum overshoot of 25.4% can be deduced

as follows:
𝜁
− 𝜋
𝑀𝑝 = 𝑒 √1−𝜁 2 = 0.254
𝜁
− 𝜋
√1−𝜁2 𝜁
ln 𝑒 = ln 0.254 ⟹ − 𝜋 = −1.3704 ⟹ 𝜁 = 0.4362√1 − 𝜁 2
√1−𝜁 2
⟹ 𝜁 2 = 0.1903(1 − 𝜁 2) ⟹ 2
𝜁 = 0.1599 ⟹ 𝜁 = 0.3999 ≅ 0.4.

From the response curve we have

It follows that 𝜔𝑛 = 1.14. The desired characteristic polynomial has the quadratic factor

𝑠 2 + 2𝜁𝜔𝑛 𝑠 + 𝜔𝑛2 = 𝑠 2 + 2(0.4)(1.14)𝑠 + 1.142 = 𝑠 2 + 0.91𝑠 + 1.3

Since the system is second order, the factor 𝑠 2 + 0.91𝑠 + 1.3 is the only factor that appears
in the characteristic equation. Hence, the desired characteristic equation is

4-32
 𝑠 2 + 0.91𝑠 + 1.3 = 0 (1)

From the block diagram we have

𝐶(𝑠) 𝐾
= (2)
𝑅(𝑠) 𝑇𝑠2 +𝑠+𝐾

The characteristic equation is

𝑇𝑠 2 + 𝑠 + 𝐾 = 0 (3)

To find the values of 𝐾 and 𝑇, we equate the coefficients of the like powers of 𝑠 in
equations (1) and (3), but before that, we must transform the polynomial in equation (3) to a
nominal polynomial.
1 𝐾
𝑠2 + 𝑠 + =0 (4)
𝑇 𝑇

Equating the coefficients of the like powers of 𝑠 in equations (1) and (4) yields
1
0.91 = ⟹ 𝑇 = 1.09
𝑇
𝐾
1.3 = ⟹ 𝐾 = 1.3𝑇 = 1.42
𝑇
Note: This system is a standard second-order system, therefore, the transfer function (2) can
be compared (after transforming it to the standard form) with the standard second-order
𝜔2𝑛
transfer function . Thus
𝑠2 +2𝜁𝜔𝑛 𝑠+𝜔2𝑛

𝐾
𝐶(𝑠) 𝐾 𝑇 1 1 1
= = 1 𝐾 ⟹ = 2𝜁𝜔𝑛 ⟹ 𝑇= = 2×0.4×1.14 = 1.09
𝑅(𝑠) 𝑇𝑠2 +𝑠+𝐾 𝑠2 + 𝑠+ 𝑇 2𝜁𝜔𝑛
𝑇 𝑇

𝐾
= 𝜔𝑛2 ⟹ 𝐾 = 𝑇𝜔𝑛2 = 1.09 × 1.142 = 1.42
𝑇

Note: The previous example can be considered as a design problem, where the design
parameters are 𝐾 and 𝑇 and the desired design specification is given in figure 5-52 (b).

4-33
Solution

𝜁 = 0.4

Figure Closed-loop system

The peak time 𝑡p is specified as 2 s. And so

𝜋 𝜋 𝜋
𝑡p = = = =2
𝜔𝑑 𝜔𝑛 √1−𝜁 2 𝜔𝑛 √1−0.4042

It follows that 𝜔𝑛 = 1.72. The desired characteristic polynomial has the quadratic factor

𝑠 2 + 2𝜁𝜔𝑛 𝑠 + 𝜔𝑛2 = 𝑠 2 + 2(0.4)(1.72)𝑠 + 1.722 = 𝑠 2 + 1.38𝑠 + 2.96

Since the system is second order, the factor 𝑠 2 + 0.91𝑠 + 1.3 is the only factor that appears
in the characteristic equation. Hence, the desired characteristic equation

𝑠 2 + 1.38𝑠 + 2.96 (*)

From Figure 5-53, the closed-loop transfer function is

4-34
 (**)

The characteristic equation is

𝑠 2 + 𝐾𝑘𝑠 + 𝐾 = 0 (***)

To find the values of 𝐾 and 𝑘, we equate the coefficients of the like powers of 𝑠 in
1.38
equations (*) and (***). This yields 𝐾 = 2.96 and 1.38 = 𝐾𝑘 ⟹ 𝑘 = = 0.47
𝐾

Note: This system is a standard second-order system, therefore, the transfer function (**) can
be compared (after transforming it to the standard form) with the standard second-order
𝜔2𝑛
transfer function . Thus
𝑠2 +2𝜁𝜔𝑛 𝑠+𝜔2𝑛

𝐶(𝑠) 𝐾
= ⟹ 𝐾 = 𝜔2𝑛 = 1.722 = 2.96 and 𝐾𝑘 = 2𝜁𝜔𝑛 ⟹
𝑅(𝑠) 𝑠2 +𝐾𝑘𝑠+𝐾

2×0.4×1.72
2.96𝑘 = 2 × 0.4 × 1.72 ⟹ 𝑘= = 0.47
2.96

Note: The previous example can be considered as a design problem, where the design
parameters are 𝐾 and 𝑘 and the desired design specifications are certain values of 𝑀p and
𝑡p .

Example: Consider the system

𝑠+𝑎 1
𝑅(𝑠) 𝐾 𝑌(𝑠)
𝑠+𝑏 𝑠(𝑠 + 3)

Design the controller parameters so that the closed-loop system has a 10% overshoot for a
unit step input and a settling time 1.7s (2% criterion).

4-35
Solution:
𝜁
− 𝜋
𝑀𝑝 = 𝑒 √1−𝜁 2 = 0.1 ⟹ 𝜁 = 0.59

4 4
𝑡𝑠 (2%) = = 1.7 ⟹ 𝜔𝑛 = = 3.99
𝜁𝜔𝑛 1.7𝜁

The complex-conjugate poles that corresponds to 𝜁 = 0.6 and 𝜔𝑛 = 4 are

𝑠1,2 = −𝜁𝜔𝑛 ± 𝑗𝜔𝑛 √1 − 𝜁 2 = −0.59 × 3.99 ± 𝑗3.99√1 − 0.592 = −2.35 ± 𝑗3.22

Thus, the desired characteristic polynomial has the quadratic factor

𝑠 2 + 2𝜁𝜔𝑛 𝑠 + 𝜔𝑛2 = 𝑠 2 + 2 × 0.59 × 3.99𝑠 + 3.992 = 𝑠 2 + 4.71𝑠 + 15.92

Since the system is third order, there is a third closed-loop pole (which is real). To make the
complex-conjugate poles dominant, we take the value of the third pole to be at least five
times the real part of the complex-conjugate poles. Thus,

𝑠3 > 5 × (−2.35) = −11.75

Let

𝑠3 = −12

Thus, the desired characteristic polynomial has the linear factor

𝑠 − 𝑠3 = 𝑠 − (−12) = 𝑠 + 12

The desired characteristic equation is

(𝑠 2 + 4.71𝑠 + 15.92)(𝑠 + 12) =0

𝑠 3 + 16.71𝑠 2 + 72.44𝑠 + 191.04 = 0 (*)

The closed-loop transfer function is

𝑠+𝑎 1
𝑌(𝑠) 𝐾 𝐾𝑠+𝐾𝑎
𝑠+𝑏 𝑠(𝑠+3)
= 𝑠+𝑎 1 = (**)
𝑅(𝑠) 1+𝐾 𝑠+𝑏 𝑠(𝑠+3) 𝑠 3 +(3+𝑏)𝑠 2 +(3𝑏+𝐾)𝑠+𝐾𝑎

The characteristic equation is

𝑠 3 + (3 + 𝑏)𝑠 2 + (3𝑏 + 𝐾)𝑠 + 𝐾𝑎 = 0 (***)

To find the values of 𝑎, 𝑏 and 𝑘, we equate the coefficients of the like powers of 𝑠 in
equations (*) and (**). This yields
4-36
 16.71 = 3 + 𝑏 ⟹ 𝑏 = 13.71

72.44 = 3𝑏 + 𝐾 ⟹ 𝐾 = 30.87

191.04 = 𝐾𝑎 ⟹ 𝑎 = 6.19

Impulse Response of Second-Order Systems

For a unit-impulse input 𝑟(𝑡) = 𝛿(𝑡), the corresponding Laplace transform is unity, or
𝑅 (𝑠) = 1. The unit-impulse response 𝐶(𝑠) of the second-order system given by equation
(5.10 Ogata) is

4-37
Figure 5-14 Unit-impulse response curves of the system shown in Figure 5-6

4-38
Ogata

As stated in Chapter 2, the zero-input response of a system is the response of the system
when the input is zero so that the response is due to the initial conditions only. This
response is also called free, unforced, or natural response.

4-39
 Closed-loop pole location and its relation to the zero-input response of a system

Golnaraghi

Effect of a Third Pole and a Zero on the Second-Order System Response

In this section, we show that the addition of poles and zeros to the forward-path and closed-
loop transfer functions has varying effects on the transient response of the closed loop
system.

Addition of a Pole to the Forward-Path Transfer Function: Unity-Feedback System

4-40
4-41
Addition of a Pole to the Closed-Loop Transfer Function

4-42
Golnaraghi

Dorf

Example

For a third-order system with a closed-loop transfer function

4-43
Dorf

Golnaraghi

Addition of a Zero to the Closed-Loop Transfer Function

4-44
4-45
Golnaraghi

Dorf

Example

The transient response of a system with one zero and two poles maybe affected by the
1
𝜉𝜔𝑛
location of the zero. The percent overshoot for a step input as a function of , when 𝜉 ≤ 1,
𝑇𝑧
is given in Figure 5.13(a) for the system transfer function
𝜔𝑛2 (𝑇𝑧 𝑠 + 1)
𝑇(𝑠) = 2
𝑠 + 2𝜉𝜔𝑛 𝑠 + 𝜔𝑛2
The actual transient response for a step input is shown in Figure 5.13(b) for selected values
1
𝜉𝜔𝑛
of . The actual response for these selected values is summarized in Table 5.4 when 𝜉 =
𝑇𝑧
0.45.
The correlation of the time-domain response of a system with the s-plane location of the
poles of the closed-loop transfer function is very useful for selecting the specification of the
system.

4-46
4-47
 𝟏
𝝃𝝎𝒏
𝑻𝒛

Dorf

Golnaraghi

Addition of a Zero to the Forward-Path Transfer Function: Unity-Feedback System

4-48
Golnaraghi

4-49
Ogata

Higher-Order Systems

4-50
Figure 5-16

4-51
Ogata

In general, the response of system 5-32 (Ogata) to any input 𝑅(𝑠)

𝐾(𝑠 + 𝑧1)(𝑠 + 𝑧2) … (𝑠 + 𝑧𝑚 )

𝐶 (𝑠) = 𝑅 (𝑠)
(𝑠 + 𝑝1 )(𝑠 + 𝑝2 ) … (𝑠 + 𝑧𝑛 )

2
𝑎 𝑏𝑘 (𝑠+𝜁𝑘𝜔𝑘 )+𝑐𝑘 𝜔𝑘 √1−𝜁𝑘
= 𝐶𝑅 (𝑠) + ∑𝑞𝑗=1 𝑗 + ∑𝑟𝑘=1 (*)
𝑠+𝑝𝑗 𝑠2 +2𝜁𝑘 𝜔𝑘𝑠+𝜔2𝑘

where 𝐶𝑅 (𝑠) contains the terms of 𝐶 (𝑠) which originate in the poles of 𝑅 (𝑠). The response
𝑐 (𝑡), the inverse Laplace transform of 𝐶 (𝑠) is

𝑞 2
𝑐 (𝑡) = ℒ −1{𝐶𝑅 (𝑠)} + ∑𝑗=1 𝑎𝑗 𝑒 −𝑝𝑗 𝑡 + ∑𝑟𝑘=1 𝑏𝑘 𝑒 −𝜁𝑘 𝜔𝑘 𝑡 cos 𝜔𝑘√1 − 𝜁𝑘 𝑡

2
+ ∑𝑟𝑘=1 𝑐𝑘 𝑒 −𝜁𝑘𝜔𝑘𝑡 sin 𝜔𝑘√1 − 𝜁𝑘 𝑡 (**)

The two summation terms of (**) are the natural-response terms of 𝑐 (𝑡). If these terms tend
to zero as time increases, the system is stable, and these terms are called the transient
response. As stated previously, the system is stable if all closed-loop poles lie in the left-half
s-plane, i.e., 𝑝𝑗 > 0 for all 1 < 𝑗 < 𝑞 and 𝜁𝑘 𝜔𝑘 > 0 for all 1 < 𝑘 < 𝑟. Thus, the poles
𝐶(𝑠)
of the transfer function enter into the transient response of the system, while the poles of
𝑅(𝑠)
the input 𝑅 (𝑠) yields the steady-state response terms in the solution.
4-52
Ogata

Ogata

Golnaraghi

Dominant Poles and Zeros of Transfer Functions

4-53
Golnaraghi

4-54
Dorf

The concept of dominant poles can be utilized in simplification of linear systems.

Consider a system with transfer function
𝐾
𝐺 (𝑠) =
𝑠(𝑠 + 2)(𝑠 + 30)

Dorf

Ogata

Transient Response Analysis with MATLAB

4-55
4-56
4-57
Example 5-3

4-58
Figure 5-18 Unit-step response curves

4-59
4-60
4-61
Example 5-5

4-62
4-63
4-64
4-65
4-66
4-67
4-68
4-69
Ogata

Response to Initial Condition

In [Ogata], more than one method is presented for obtaining the response of a system to an
initial condition. As we shall see, in state space approaches (both case 1 and case 2), the
initial conditions of the states are converted into impulse inputs in the state-space equations
(the weight of each impulse signal equals the value of the initial condition of the associated

4-70
state). Note that the new system will have zero initial condition. However, the typical
method to find the response of a system described by a state-space model is to find the
solution of the state equation (considering both the initial condition and input) and substitute
the state vector in the output equation. This method will be given in Chapter Eight.

Ogata

4-71
4-72
4-73
4-74
4-75
4-76
4-77
4-78
Ogata

4-79
Golnaraghi

Stability of Linear Control Systems

As stated at the beginning of this chapter, the stability is classified into absolute
stability and relative stability. Absolute stability refers to whether the system is stable or
unstable; it is a yes or no answer. Once the system is found to be stable, it is of interest
to determine how stable it is, and this degree of stability is a measure of relative
stability.
In preparation for the definition of stability, we need to restate the zero-state response
and zero-input response for linear time-invariant systems:

Golnaraghi

Franklin

Bounded-Input, Bounded-Output (BIBO) Stability-Definition 1

A system is said to have bounded input-bounded output (BIBO) stability if every bounded
input results in a bounded output (regardless of what goes on inside the system). A test for
this property is readily available when the system response is given by convolution. If the
system has input 𝑢(𝑡), output 𝑦(𝑡), and impulse response 𝑔(𝑡), then

4-80
 ∞
𝑦(𝑡) = ∫−∞ 𝑔(𝜏)𝑢(𝑡 − 𝜏)𝑑𝜏 (3.83)

If 𝑢(𝑡) is bounded, then there is a constant 𝑀 such that |𝑢| ≤ 𝑀 < ∞, and the output is
bounded by
∞ ∞ ∞

|𝑦| = | ∫ 𝑔(𝜏)𝑢(𝑡 − 𝜏)𝑑𝜏| ≤ ∫ |𝑔(𝜏)||𝑢(𝑡 − 𝜏)| 𝑑𝜏 ≤ 𝑀 ∫ |𝑔(𝜏)| 𝑑𝜏

−∞ −∞ −∞
∞
Thus, the output will be bounded if ∫−∞|𝑔(𝜏)| 𝑑𝜏 is bounded.

On the other hand, suppose the integral is not bounded and the bounded input is

+1 if ℎ(𝜏) > 0
𝑢(𝑡 − 𝜏) = {
−1 if ℎ(𝜏) < 0

In this case,
∞
𝑦(𝑡) = ∫−∞|𝑔(𝜏)|𝑑𝜏 (3.84)

and the output is not bounded. We conclude that

the system with impulse response 𝑔(𝑡) is BIBO stable if and only if the integral
∞
∫−∞|𝑔(𝜏)|𝑑𝜏 < ∞ (*)

Franklin

4-81
Golnaraghi

Relationship between Characteristic Equation Roots and Stability

4-82
Zero-Input and Asymptotic Stability--Definition 2

4-83
4-84
Golnaraghi

4-85
Ogata

4-86
Routh's Stability Criterion

4-87
 𝑎0 𝑎2
|𝑎 𝑎3 |
1
𝑏1 = −
𝑎1
𝑎0 𝑎4
|𝑎 𝑎5 |
1
𝑏2 = −
𝑎1
𝑎0 𝑎6
|𝑎 𝑎7 |
1
𝑏3 = −
𝑎1

4-88
 𝑎1 𝑎3
|𝑏 𝑏2 |
1
𝑐1 = −
𝑏1
𝑎1 𝑎5
|𝑏 𝑏3 |
1
𝑐2 = −
𝑏1
𝑎1 𝑎7
|𝑏 𝑏4 |
1
𝑐3 = −
𝑏1

Example 5-12

4-89
Ogata

Paraskevopoulos

Example

Paraskevopoulos

4-90
Ogata

Example 5-11

Ogata

Note: The previous example can be considered as a design problem, where the design
parameters are 𝑎0 , 𝑎1 , 𝑎2 , and 𝑎3 and the desired design specification is that the system
must be absolutely stable.

Special Cases

Case 1

If a first column term in any row is zero, but some other elements of the row containing
the zero in the first column are nonzero, then the zero term is replaced by a small positive
number 𝜖 and the rest of the array is evaluated.

4-91
Golnaraghi

Example

Golnaraghi

4-92
Franklin (3rd p 218)

Franklin (3rd p 218)

Mandal (p. 137)

Case 2

If all the coefficients in any derived row are zero, or when the row consists of a single
element that is zero, it indicates that there are roots that are symmetrically located about
the origin of the s-plane. Therefore, case 2 occurs when there are factors such as
(𝑠 + 𝜎), (𝑠 − 𝜎); (𝑠 + 𝑗𝜔), (𝑠 − 𝑗𝜔); or (𝑠 + 𝜎 ± 𝑗𝜔), (𝑠 − 𝜎 ∓ 𝑗𝜔). This means that
for this case, the system is either marginally stable (if the factors are (𝑠 + 𝑗𝜔), (𝑠 − 𝑗𝜔))
or unstable (if the factors are (𝑠 + 𝜎), (𝑠 − 𝜎) or (𝑠 + 𝜎 ± 𝑗𝜔), (𝑠 − 𝜎 ∓ 𝑗𝜔)).

Mandal (p. 137)

4-93
Golnaraghi (p. 81)

Example

4-94
Example

4-95
Example

4-96
Golnaraghi (p. 84)

Ogata

Example

Fig. 5-35 Control system

4-97
Ogata

Dorf

The Relative Stability of Feedback Control Systems

4-98
Dorf

4-99
Paraskevopoulos

The Hurwitz Criterion

4-100
Paraskevopoulos

Golnaraghi

Characteristic Equation from State Equations

The transfer function in terms of state equations is given by

adj(𝑠𝑰 − 𝑨) 𝑪(adj(𝑠𝑰 − 𝑨))𝑩 + |𝑠𝑰 − 𝑨|𝑫

𝑮(𝑠) = 𝑪(𝑠𝑰 − 𝑨)−1𝑩 + 𝑫 = 𝑪 𝑩+𝑫=
|𝑠𝑰 − 𝑨| |𝑠𝑰 − 𝑨|

Setting the denominator of the transfer function matrix 𝑮(𝑠) to zero, we get the charateristic
equation

|𝑠𝑰 − 𝑨| = 0

Golnaraghi

4-101
4-102
4-103
Eigenvalues of a Matrix

The roots of the characteristic equation are often referred to as the eigenvalues of the matrix
𝑨. The eigenvalues are also called the characteristic roots.

Consider for example, the following matrix 𝑨:

0 1 0
𝑨=[ 0 0 1]
−6 −11 −6
The characteristic equation is
𝑠 −1 0
|𝑠𝑰 − 𝑨| = [0 𝑠 −1 ] = 𝑠 3 + 6𝑠 2 + 11𝑠 + 6 = (𝑠 + 1)(𝑠 + 2)(𝑠 + 3) = 0
6 11 𝑠 + 6
The eigenvalues of 𝑨 are -1, -2, and -3.

Liu

4-104
4-105
 The details of this method is given in Chapter 8.

Liu

4-106
Ogata

Steady-State Errors in Unity-Feedback Control Systems

4-107
Step Input – Static Position Error Constant 𝑲p

The steady-state error of the system for a step input of magnitude 𝐴 (𝑟(𝑡) = 𝐴𝑢(𝑡)) is
𝑠 𝐴 𝐴 𝐴
𝑒𝑠𝑠 = lim = =
𝑠→0 1+𝐺(𝑠) 𝑠 1+lim 𝐺(𝑠)
𝑠→0
1+𝐺(0)

The static position error constant 𝐾𝑝 is defined by

𝐾𝑝 = lim 𝐺(𝑠) = 𝐺(0)

𝑠→0

Thus, the steady-state error in terms of the static position error 𝐾𝑝 is given by
𝐴
𝑒ss =
1+𝐾𝑝

For a type 0 system,

4-108
 𝐾(𝑇𝑎 𝑠 + 1)(𝑇𝑏 𝑠 + 1) …
𝐾𝑝 = lim =𝐾
𝑠→0 (𝑇1 𝑠 + 1)(𝑇2 𝑠 + 1) …

and
𝐴 𝐴
𝑒ss = =
1 + 𝐾𝑝 1 + 𝐾

For a type 1 or higher system,

𝐾 (𝑇𝑎 𝑠 + 1)(𝑇𝑏 𝑠 + 1) …
𝐾𝑝 = lim =∞
𝑠→0 𝑠 𝑁 (𝑇1 𝑠 + 1)(𝑇2 𝑠 + 1) …

and
𝐴 1
𝑒ss = = =0
1 + 𝐾𝑝 1 + ∞

Ramp Input – Static Velocity Error Constant

The steady-state error of the system for a ramp (velocity) input with a slope 𝐴 (𝑟(𝑡) = 𝐴𝑡) is
𝑠 𝐴 𝐴 𝐴
𝑒𝑠𝑠 = lim = lim =
𝑠→0 1 + 𝐺(𝑠) 𝑠 2 𝑠→0 𝑠 + lim 𝑠𝐺(𝑠) lim 𝑠𝐺(𝑠)
𝑠→0 𝑠→0

The static velocity error constant 𝐾𝑣 is defined by

𝐾𝑣 = lim 𝑠𝐺(𝑠)
𝑠→0

Thus, the steady-state error in terms of the static velocity error 𝐾𝑣 is given by
𝐴
𝑒ss =
𝐾𝑣

4-109
For a type 0 system,

𝐾 (𝑇𝑎 𝑠 + 1)(𝑇𝑏 𝑠 + 1) …
𝐾𝑣 = lim 𝑠 =0
𝑠→0 (𝑇1𝑠 + 1)(𝑇2𝑠 + 1) …

and
𝐴 𝐴
𝑒ss = = =∞
𝐾𝑣 0

For a type 1 system,

𝐾(𝑇𝑎 𝑠 + 1)(𝑇𝑏 𝑠 + 1) … 𝐾(𝑇𝑎 𝑠 + 1)(𝑇𝑏 𝑠 + 1) …
𝐾𝑣 = lim 𝑠 = lim =𝐾
𝑠→0 𝑠(𝑇1𝑠 + 1)(𝑇2𝑠 + 1) … 𝑠→0 (𝑇1𝑠 + 1)(𝑇2𝑠 + 1) …

and
𝐴 𝐴
𝑒ss = =
𝐾𝑣 𝐾

For a type 2 or higher system,

𝐾 (𝑇𝑎 𝑠 + 1)(𝑇𝑏 𝑠 + 1) … 𝐾(𝑇𝑎 𝑠 + 1)(𝑇𝑏 𝑠 + 1) …

𝐾𝑣 = lim 𝑠 = lim =∞
𝑠→0 𝑠 𝑁 (𝑇1𝑠 + 1)(𝑇2𝑠 + 1) … 𝑠→0 𝑠 𝑁−1(𝑇1𝑠 + 1)(𝑇2𝑠 + 1) …

and
𝐴 𝐴
𝑒ss = = =0
𝐾𝑣 ∞

4-110
Parabolic Input – Static Acceleration Error Constant

The steady-state error of the system for a parabolic (acceleration) input with a second-order
𝐴
derivative 𝐴 (𝑟(𝑡) = 𝑡 2) is
2

𝑠 𝐴 𝐴 𝐴
𝑒𝑠𝑠 = lim = lim =
𝑠→0 1 + 𝐺(𝑠) 𝑠 3 𝑠→0 𝑠 2 + lim 𝑠 2 𝐺(𝑠) lim 𝑠 2𝐺(𝑠)
𝑠→0 𝑠→0

The static acceleration error constant 𝐾𝑎 is defined by

𝐾𝑎 = lim 𝑠 2𝐺(𝑠)
𝑠→0

Thus, the steady-state error in terms of the static acceleration error 𝐾𝑎 is given by
𝐴
𝑒ss =
𝐾𝑎

Note that the acceleration error, the steady-state error due to a parabolic input, is an error in
position.

For a type 0 system,

𝐾(𝑇𝑎 𝑠 + 1)(𝑇𝑏 𝑠 + 1) …
𝐾𝑎 = lim 𝑠 2 =0
𝑠→0 (𝑇1𝑠 + 1)(𝑇2𝑠 + 1) …

and
𝐴 𝐴
𝑒ss = = =∞
𝐾𝑎 0

For a type 1 system,

𝐾(𝑇𝑎 𝑠 + 1)(𝑇𝑏 𝑠 + 1) … 𝐾 (𝑇𝑎 𝑠 + 1)(𝑇𝑏 𝑠 + 1) …
𝐾𝑎 = lim 𝑠 2 = lim 𝑠 =0
𝑠→0 𝑠(𝑇1𝑠 + 1)(𝑇2𝑠 + 1) … 𝑠→0 (𝑇1 𝑠 + 1)(𝑇2 𝑠 + 1) …
4-111
and
𝐴 1
𝑒ss = = =∞
𝐾𝑣 0

For a type 2 system,

𝐾 (𝑇𝑎 𝑠 + 1)(𝑇𝑏 𝑠 + 1) … 𝐾 (𝑇𝑎 𝑠 + 1)(𝑇𝑏 𝑠 + 1) …

𝐾𝑣 = lim 𝑠 2 = lim =𝐾
𝑠→0 𝑠 2(𝑇1𝑠 + 1)(𝑇2𝑠 + 1) … 𝑠→0 (𝑇1 𝑠 + 1)(𝑇2 𝑠 + 1) …

and
𝐴 𝐴
𝑒ss = =
𝐾𝑣 𝐾

For a type 3 or higher system,

𝐾 (𝑇𝑎 𝑠 + 1)(𝑇𝑏 𝑠 + 1) … 𝐾 (𝑇𝑎 𝑠 + 1)(𝑇𝑏 𝑠 + 1) …

𝐾𝑣 = lim 𝑠 2 = lim =∞
𝑠→0 𝑠 𝑁 (𝑇1𝑠 + 1)(𝑇2𝑠 + 1) … 𝑠→0 𝑠 𝑁−2(𝑇1𝑠 + 1)(𝑇2𝑠 + 1) …

and
𝐴 1
𝑒ss = = =0
𝐾𝑣 ∞

4-112
 Table 5.1 Steady-state errors in terms of 𝐾
Parabolic Input
Step Input Ramp Input
𝐴
𝑟(𝑡) = 𝐴𝑢(𝑡) 𝑟(𝑡) = 𝐴𝑡 𝑟 (𝑡 ) = 𝑡 2
2
𝐴
Type 0 system ∞ ∞
1+𝐾
𝐴
Type 1 system 0 ∞
𝐾
𝐴
Type 2 system 0 0
𝐾

Ogata

Dorf

Example

4-113
Solution

The system is a type 1 system.

The static position error is

5(𝑠 + 8)
𝐾𝑝 = lim 𝐺(𝑠) = lim =∞
𝑠→0 𝑠→0 𝑠(𝑠 + 1)(𝑠 + 4)(𝑠 + 10)

1 1
and the steady-state error to a step input = = =0
1+𝐾𝑝 1+∞

The static velocity error is

5(𝑠 + 8) 5(0 + 8)
𝐾𝑣 = lim 𝑠𝐺(𝑠) = lim 𝑠 = =1
𝑠→0 𝑠→0 𝑠(𝑠 + 1)(𝑠 + 4)(𝑠 + 10) (0 + 1)(0 + 4)(0 + 10)
1 1
and the steady-state error to a ramp input = = =1
𝐾𝑣 1

Example

Solution

The system is a type 0 system.

The static position error is

20
𝐾𝑝 = lim 𝐺(𝑠) = lim = 0.4
𝑠→0 𝑠→0 𝑠 2 + 14𝑠 + 50

1 1
and the steady-state error to a step input = = =0.71
1+𝐾𝑝 1+0.4

The static velocity error is

20
𝐾𝑣 = lim 𝑠𝐺(𝑠) = lim 𝑠 =0
𝑠→0 𝑠→0 𝑠 2 + 14𝑠 + 50
1 1
and the steady-state error to a ramp input = = =∞
𝐾𝑣 0

4-114
 𝐾2 𝐾 𝐾𝐾1 𝑠 + 𝐾𝐾2
𝐺 (𝑠) = 𝐺𝑐 (𝑠)𝑃 (𝑠) = (𝐾1 + ) =
𝑠 𝜏𝑠 + 1 𝑠(𝜏𝑠 + 1)
𝐾𝐾1
if 𝐾2 = 0 (type 0 system)
𝜏𝑠 + 1
= 𝐾𝐾1 𝑠 + 𝐾𝐾2
if 𝐾2 > 0 (type 1 system)
𝑠(𝜏𝑠 + 1)
{
If the steering command is a step input, then the static position error is
𝐾𝐾1 if 𝐾2 = 0 (type 0 system)
𝐾𝑝 = 𝐺 (0) = { ∞ if 𝐾2 > 0 (type 1 system)

and the steady-state error is

𝐴
𝐴 if 𝐾2 = 0 (type 0 system)
𝑒ss = ={ 𝐾𝐾1
1 + 𝐾𝑝 0 if 𝐾2 > 0 (type 1 system)

If the steering command is a ramp input, the static velocity error is

𝐾𝐾1 𝑠 + 𝐾𝐾2 𝐾𝐾1 𝑠 + 𝐾𝐾2
𝐾𝑣 = lim 𝑠𝐺(𝑠) = lim 𝑠 = lim = 𝐾𝐾2
𝑠→0 𝑠→0 𝑠(𝜏𝑠 + 1) 𝑠→0 𝜏𝑠 + 1

and the steady-state error is

𝐴 𝐴
𝑒ss = =
𝐾𝑣 𝐾𝐾2

4-115
Note: In the previous example, we want to increase the gain factor 𝐾𝐾2 in order to
increase 𝐾𝑣 and reduce the steady-state error. However, an increase in 𝐾𝐾2 results in an
attendant decrease in the systems damping ratio 𝜉 and therefore a more oscillatory
response to a step input. Thus, we want a compromise that provides the largest 𝐾𝑣 based
on the smallest 𝜉 allowable.

4-116
4-117
system in Figure (*). The closed-loop transfer function is
𝑌(𝑠) 𝐺(𝑠)
=
𝑅(𝑠) 1 + 𝐻 (𝑠)𝐺(𝑠)

4-118
 Figure (*)

𝑌(𝑠) 𝑍(𝑠) 𝐺(𝑠)

= where 𝑧(𝑠) = (*)
𝑅(𝑠) 1+𝑍(𝑠) 1+𝐺(𝑠)(𝐻(𝑠)−1)

The equivalent unity feedback system is shown in Figure (**). The loop transfer

Figure (**)

function of this system is 𝑍(𝑠). It follows that the error constants for nonunity
feedback systems are given as

Note: The nonunity feedback system shown in Figure (*) can be converted to the unity
feedback system shown in Figure (***) (Ogata, Paraskevopoulos); however, this
configuration is different from the standard form given in Figure 5-46 (Ogata) because of the

4-119
 1
existence of the prefilter , which means that the error constants must be defined
𝐻(𝑠)
differently.

Figure (***)

Example

Solution

Method one: Establishing the error signal and applying the final-value theorem

(a)

The tracking error is

𝐸 (𝑠) = 𝑅 (𝑠) − 𝑌 (𝑠) = 𝑅 (𝑠) − 𝑇(𝑠)𝑅(𝑠) = (1 − 𝑇(𝑠))𝑅(𝑠)

4-120
The steady-state tracking error is

𝑒ss = lim 𝑠(1 − 𝑇(𝑠))𝑅(𝑠)

𝑠→0

1
For a unit step input (𝑅 (𝑠) = ):
𝑠

1
𝑒ss = lim 𝑠(1 − 𝑇(𝑠)) = 1 − 𝑇(0)
𝑠→0 𝑠
The closed loop transfer function is
𝐾
𝑠 𝑠 + 2)
( 𝐾(𝑠 + 0.1)
𝑇(𝑠) = 𝐺𝑝 (𝑠) = 𝐺𝑝 (𝑠)
𝐾 𝑠+3 𝑠(𝑠 + 2)(𝑠 + 0.1) + 𝐾(𝑠 + 3)
1+
𝑠(𝑠 + 2) (𝑠 + 0.1)
𝐾(0 + 0.1) 0.1𝐾
𝑇(0) = 𝐺𝑝 (0) = 𝐺 (0) = 0.033𝐺𝑝 (0)
0(0 + 2)(0 + 0.1) + 𝐾(0 + 3) 3𝐾 𝑝

𝑒ss = 1 − 𝑇(0) = 1 − 0.033𝐺𝑝 (0)

When 𝐺𝑝 (𝑠) = 1

𝑒ss = 1 − 0.033 × 1 = 0.967

(b)

𝑒ss = 1 − 0.033𝐺𝑝 (0) = 0

1
𝐺𝑝 (𝑠) = = 30
0.033

Method Two: Manipulation of Block Diagram and Applying Formula (*)

Moving the summing point ahead of 𝐺𝑐 (𝑠)

𝐺𝑝 (𝑠) 𝐾
𝑠(𝑠 + 2)

1 𝑠+3
𝐺𝑝 (𝑠) 𝑠 + 0.1

4-121
 𝐾
𝐺𝑝 (𝑠)
𝑠(𝑠 + 2)

𝑠+3
𝐺𝑝 (𝑠)(𝑠 + 0.1)

𝑧(𝑠)

𝐾
𝐺𝑝 (𝑠)
𝑠(𝑠 + 2)
𝑍(𝑠) =
𝐾 𝑠+3
1 + 𝐺𝑝 (𝑠) ( − 1)
(
𝑠 𝑠+2 ) 𝐺 𝑝 (𝑠)(𝑠 + 0.1)
𝐺𝑝 (𝑠)𝐾(𝑠 + 0.1)
=
𝑠(𝑠 + 2)(𝑠 + 0.1) + 𝐾 (𝑠 + 3) − 𝐺𝑝(𝑠)𝐾(𝑠 + 0.1)

0.1𝐺𝑝 (0)𝐾 0.1𝐺𝑝 (0)

𝐾𝑝 = 𝑍(0) = =
3𝐾 − 0.1𝐺𝑝 (0)𝐾 3 − 0.1𝐺𝑝 (0)

1 1
𝑒ss = =
1 + 𝐾𝑝 0.1𝐺𝑝 (0)
1+
3 − 0.1𝐺𝑝 (0)

When 𝐺𝑝 (𝑠) = 1
1 1
𝑒ss = = 0.1𝐺𝑝(0) = 0.967
1+𝐾𝑝 1+3−0.1𝐺
𝑝 (0)

(b)
1
𝑒ss = =0
0.1𝐺𝑝 (0)
1+
3 − 0.1𝐺𝑝 (0)

𝐺𝑝 (𝑠) = 30

4-122
Performance Indices

4-123
4-124
Dorf

4-125
 Problems

Ogata

4-126
4-127
4-128
4-129
4-130
Ogata

4-131
Dorf (chapter 5)

4-132
4-133
4-134
4-135
4-136
Dorf (chapter 5)

4-137
Dorf (chapter 6)

4-138
4-139
4-140
4-141
Dorf (chapter 6)

4-142