Chapter 2 Digital Models of Control Systems

In the analysis and design of the control system, the mathematical model of the system
must be established first. The mathematical model of a control system is a numerical
expression that describes the relationship between physical quantities (or variables) within the
system. Under static conditions (that is, the derivatives of variables are zero), the algebraic
equations that describe the relationship between variables are called static mathematical
models, and the differential equations that describe the relationship between derivatives of
variables are called dynamic mathematical models. If the initial conditions of the input and
variables are known, and the differential equation is solved, the expression of the output of
the system can be obtained, and the performance of the system can be analyzed. Therefore,
establishing the mathematical model of the control system is the primary task of analyzing
and designing the control system.
There are two methods to establish the mathematical model of the control system, the
analytical method, and the experimental method. The analysis method is to analyze the
motion mechanism of each part of the system and write the corresponding motion equations
according to the physical laws or chemical laws they are based on. For example, there are
Kirchhoff's laws in electricity, Newton's laws in mechanics, thermodynamics, etc. The
experimental method is to artificially apply a certain test signal to the system, record its
output response, and approximate it with an appropriate mathematical model. This method is
called system identification. In recent years, system identification has developed into an
independent sub-discipline. This chapter studies the establishment of mathematical models of
systems using analytical methods.
In automatic control theory, mathematical models take many forms. Commonly used
mathematical models in the time domain include differential equations, difference equations
and state equations; transfer functions and structural diagrams in the complex number domain;
frequency characteristics in the frequency domain, etc. This chapter only studies the
establishment and application of mathematical models such as differential equations, transfer
functions and structural diagrams, and other mathematical models will be described in detail
in later chapters.

Mathematical basic-complex number

Basic concept
A complex number z in rectangular (Cartesian) form is expressed as
𝑧 = 𝑥 + 𝑗𝑦 {2 − 1}

𝑗 = √−1, 𝑖𝑚𝑎𝑔𝑖𝑛𝑎𝑟𝑦 𝑛𝑢𝑚𝑏𝑒𝑟

𝑥, 𝑦 𝑎𝑟𝑒 𝑟𝑒𝑎𝑙 𝑛𝑢𝑚𝑏𝑒𝑟
x and y are denoted as 𝑥 = 𝑅𝑒{𝑧}, 𝑦 = 𝐼𝑚{𝑧}
The magnitude (or modulus) of a complex number 𝑧 = 𝑥 + 𝑗𝑦 is denoted by z and is
defined as

|𝑧| = √𝑥 2 + 𝑦 2 {2 − 2}
 Figure 2.1 complex number addition
Addition of complex number magnitude
If 𝑧1 is a real number
If 𝑧2 is a complex number
Like the vector addition, the addition of the complex number follows the triangle
inequality rule:
|𝑧1 + 𝑧2 | ≤ |𝑧1 | + |𝑧2 | {2 − 3}

The complex conjugate of 𝑧 = 𝑥 + 𝑗𝑦 is denoted by z and defined as 𝑧̅ = 𝑥 − 𝑗𝑦.

The product of a complex number and its conjugate is a positive real number, equal to the
square of the magnitude of z,
𝑧𝑧̅ = (𝑥 + 𝑗𝑦)(𝑥 − 𝑗𝑦) = 𝑥 2 + 𝑦 2 = |𝑧|2 {2 − 4}

The rectangular form of a complex number 𝑧 = 𝑥 + 𝑗𝑦.

Figure 2.2 rectangular form of complex number

𝑥 = 𝑟𝑐𝑜𝑠𝜃 {2 − 5}

𝑦 = 𝑟𝑠𝑖𝑛𝜃 {2 − 6}

𝑒 𝑗𝜃 = 𝑐𝑜𝑠𝜃 + 𝑗𝑠𝑖𝑛𝜃

𝑧 = 𝑥 + 𝑗𝑦 = 𝑟(𝑐𝑜𝑠𝜃 + 𝑠𝑖𝑛𝜃) = 𝑟𝑒 𝑗𝜃 {2 − 7}

𝑟 = |𝑧| = √𝑥 2 + 𝑦 2 {2 − 8}

𝐼𝑚{𝑧} 𝑦
𝜃 = 𝑡𝑎𝑛−1 ( ) = 𝑡𝑎𝑛−1 ( ) {2 − 9}
𝑅𝑒{𝑧} 𝑥

𝑧=𝑟<𝜃
 For example, using Euler’s formula to express 𝑧 = −1 − 2𝑗 in polar form
Solve:
−2
𝜃 = 𝑡𝑎𝑛−1 ( ) ≅ 1.107𝑟𝑎𝑑 ≅ 63.4349° {2 − 10}
−1
z is in the third quadrant, the phase in the positive is 𝜋 − 1.1071 ≅ 2.0344𝑟𝑎𝑑

𝑧 = −1 − 2𝑗 = √5𝑒 −2.0344𝑗 {2 − 11}

Figure 2.3 z in polar form

Complex algebra in polar form

𝑧1 = 𝑥1 + 𝑗𝑦1 = 𝑟1 𝑒 𝑗𝜃1 {2 − 12}

𝑧2 = 𝑥2 + 𝑗𝑦2 = 𝑟2 𝑒 𝑗𝜃2 {2 − 13}

𝑧1 𝑧2 = 𝑟1 𝑟2 𝑒 𝑗𝜃1 𝜃2 {2 − 14}
|𝑧1 𝑧2 | = 𝑟1 𝑟2 = |𝑧1 ||𝑧2 | {2 − 15}

arg(𝑧1 𝑧2 ) = 𝜃1 + 𝜃2 = arg(𝑧1 ) + arg(𝑧2 ) {2 − 16}

Example 1
−1−2𝑗
Using Euler’s formula to express 𝑧 = in polar form
−1+𝑗

−1 − 2𝑗 = √5𝑒 −2.0344𝑗 {2 − 17}

3𝜋
−1 + 𝑗 = √2𝑒 4 𝑗 {2 − 18}

−1 − 2𝑗 √5𝑒 −2.0344𝑗 5
𝑧= = 3𝜋 = √ 𝑒 −4.3906𝑗 {2 − 19}
−1 + 𝑗 𝑗 2
√2𝑒 4
Z is in third quadrant, so the phase in the positive is 4.3906j

2.1 differential equation of physical systems：

RLC passive network
 The device in the figure is a passive network composed of a resistor R, an inductor L,
and a capacitor C.
Try to write the differential equation of the network with 𝑢𝑖 (𝑡) as input and 𝑢𝑜 (𝑡) as
output:

Assuming the loop current 𝑖(𝑡)，the loop equation can be written from Kirchhoff's law
as:
𝑑𝑖(𝑡) 1
𝐿 + 𝑅𝑖(𝑡) + ∫ 𝑖(𝑡)𝑑𝑡 = 𝑢𝑖 (𝑡) {2 − 20}
𝑑𝑡 𝐶
1
𝑢𝑜 (𝑡) = ∫ 𝑖(𝑡)𝑑𝑡 {2 − 21}
𝐶
By eliminating the intermediate variable 𝑖(𝑡), the differential equation describing the
relationship between the input and output of the network is obtained as:

𝑑2 𝑢𝑜 (𝑡) 𝑑𝑢𝑜 (𝑡)

𝐿𝐶 2
+ 𝑅𝐶 + 𝑈𝑜 (𝑡) = 𝑢𝑖 (𝑡) {2 − 22}
𝑑𝑡 𝑑𝑡

Figure 2.4 RLC passive network

Example 2：

Spring-mass-damper mechanical displacement system:

Mechanical system：𝑋𝑟 is input signal，𝑋𝑐 is output. Utilizing Newton’s law yields
• • •
𝐾1 (𝑋𝑟 -X𝑐 ) + 𝐵1 (𝑋𝑟 − 𝑋𝑐 ) = 𝐾2 𝑋𝑐 + 𝐵2 𝑋𝑐 {2 − 23}

• •
(𝐵1 + 𝐵2 )𝑋𝑐 + (𝐾1 + 𝐾2 )𝑋𝑐 = 𝐵1 𝑋𝑟 + 𝐾1 𝑋𝑟 {2 − 24}

Figure 2.5 Spring-mass-damper mechanical displacement system

 Electrical system: Utilizing Kirchhoff’s law yields
1 1
𝑅2 𝑖 + ∫ 𝑖𝑑𝑡 + 𝑅1 𝑖 + ∫ 𝑖𝑑𝑡 = 𝑈𝑟 {2 − 25}
𝐶2 𝐶1

𝐶1 𝑈c1 = 𝐶2 𝑈c2 {2 − 26}

𝑈c = 𝑅1 𝑖 + 𝑈c1 {2 − 27}

(𝑅1 + 𝑅2 )𝑖 + 𝑈c1 + 𝑈c2 = 𝑈𝑟 {2 − 28}

𝐶1
𝑈𝑟 − (1 + ) 𝑈𝑐
𝑖= 𝐶2 {2 − 29}
𝐶1
𝑅1 + 𝑅2 − (1 + ) 𝑅1
𝐶2
• 1 1 • 1
(𝑅1 + 𝑅2 )𝑈𝑐 + ( + ) 𝑈𝑐 = 𝑅1 𝑈𝑟 + 𝑈𝑟 {2 − 30}
𝐶1 𝐶2 𝐶1

2.2 Linear approximations of physical systems:

◆ A great majority of physical systems are linear within some range of the variables
◆ Systems ultimately become nonlinear as the variables are increased without limit
◆ A linear system satisfies the properties of superposition and homogeneity

Figure 2.6 Linear approximations of physical systems

Principle of superposition

𝑦 = 𝑓(𝑥) {2 − 30}

𝑖𝑓 𝑦1 = 𝑓(𝑥1 ) 𝑦2 = 𝑓(𝑥2 ) {2 − 31}

𝑡ℎ𝑒𝑛 𝑦1 + 𝑦2 = 𝑓(𝑥1 + 𝑥2 ) {2 − 32}

Principle of homogeneity

𝑦 = 𝑓(𝑥) {2 − 33}

𝑖𝑓 𝑥2 = 𝛼𝑥1 {2 − 34}

𝑡ℎ𝑒𝑛 𝑦2 = 𝛼𝑦1 {2 − 35}

Linearization of nonlinear differential equations:

Strictly speaking, actual physical components or systems are nonlinear. For example, the
stiffness of a spring is related to its deformation, so the spring coefficient K is actually a
function of its displacement x, rather than a constant; parameter values such as resistance,
capacitance, inductance, etc. are related to the surrounding environment (temperature,
humidity, pressure, etc.) and flow through their currents are related and not constant; the
friction, dead zone and other nonlinear factors of the motor itself will complicate its motion
equation and become a nonlinear equation. Of course, under certain conditions, to simplify
the mathematical model, their influence can be ignored, and these components are regarded as
linear components, which is a commonly used linearization method. In addition, there is a
linearization method, called the tangent method or the small deviation method. This
linearization method is especially used for the nonlinear characteristic function with
continuous change. Its essence is in a very small range, a straight line replaces the nonlinear
line.

Nonlinear function with only one variable 𝑦 = 𝑓(𝑥)

The normal operating point is designated by 𝑥0. When the function is continuous and
differential over the range of interest, the Taylor series expansion about the operating point
may be utilized. Then we have

1 ′′
𝑦 = 𝑓(𝑥) = 𝑓(𝑥0 ) + 𝑓 ′ (𝑥 − 𝑥0 ) + 𝑓 (𝑥0 )(𝑥 − 𝑥0 )2 + ⋯ ⋯ {2 − 36}
2!
 When the deviation from the operating point is smaller, and the high-order terms are
neglected,

𝑦 − 𝑦0 = 𝑓(𝑥) − 𝑓(𝑥0 ) = 𝑓 ′ (𝑥0 )(𝑥 − 𝑥0 ) {2 − 37}

∆𝑦 = 𝑦 − 𝑦0 = 𝑓(𝑥) − 𝑓(𝑥0 ) ∆𝑥 = 𝑥 − 𝑥0 𝑘 = 𝑓 ′ (𝑥0 ) {2 − 38}

∆𝑦 = 𝑘∆𝑥 {2 − 39}

If 𝑧 = 𝑓(𝑥, 𝑦), the Taylor series expansion about operating point is useful for a linear
approximation to the nonlinear function

𝑧 = 𝑓(𝑥, 𝑦)

𝜕𝑓 𝜕𝑓
𝑧 = 𝑓(𝑥0 , 𝑦0 ) + [ (𝑥 − 𝑥0 ) + (𝑦 − 𝑦0 )]
𝜕𝑥 𝜕𝑦

1 𝜕2𝑓 𝜕𝑓 𝜕𝑓 𝜕2𝑓
+ [ 2 (𝑥 − 𝑥0 )2 + 2 (𝑥 − 𝑥0 )(𝑦 − 𝑦0 ) + 2 (𝑥 − 𝑥0 )(𝑦 − 𝑦0 ) + 2 (𝑦 − 𝑦0 )2 ] + ⋯ ⋯ {2 − 40}
2! 𝜕𝑥 𝜕𝑥𝜕𝑦 𝜕𝑥𝜕𝑦 𝜕𝑦

When the deviation from the operating point is smaller, and the high-order terms are
neglected, then

∆𝑧 = 𝑧 − 𝑧0 = 𝑓(𝑥, 𝑦) − 𝑓(𝑥0 , 𝑦0 ) = 𝐾1 (𝑥 − 𝑥0 ) + 𝐾2 (𝑦 − 𝑦0 ) {2 − 41}

∆𝑥 = 𝑥 − 𝑥0 ∆𝑦 = 𝑦 − 𝑦0 {2 − 42}

𝜕𝑓
𝐾2 = | {2 − 43}
𝜕𝑦 𝑥0 ,𝑦0

∆𝑧 = 𝐾1 ∆𝑥 + 𝐾2 ∆𝑦 {2 − 44}

Example 3:
Pendulum oscillator

Figure 2.7 Pendulum oscillator

Linear approximation：

𝜕𝑠𝑖𝑛𝜃
𝑇 − 𝑇0 ≅ 𝑀𝑔𝐿 | (𝜃 − 𝜃0 ) {2 − 45}
𝜕𝜃 𝜃=𝜃0

𝑊ℎ𝑒𝑛 𝑇0 = 0 𝜃0 = 0

𝑇 ≅ 𝑀𝑔𝐿𝑐𝑜𝑠0°𝜃 = 𝑀𝑔𝐿𝜃 {2 − 46}

−𝜋 𝜋
≤𝜃≤ {2 − 47}
4 4

2.3 Laplace transforms and inverse Laplace transforms

Complex Domain Mathematical Model of Control System

The differential equation of the control system is a mathematical model that describes
the performance of the system in the time domain. Under the given external action and initial
conditions, the output response of the system can be obtained by solving the differential
equation. This method is more intuitive, especially with the help of a computer, the result can
be obtained quickly and accurately. However, if the structure of the system changes or a
certain parameter change, it is necessary to rewrite and solve the differential equation, which
is not convenient to design and analyze the system.

When solving the differential equation of the linear system by the Laplace transform
method, the mathematical model of the control system in the complex number domain—the
transfer function can be obtained. The transfer function can not only characterize the dynamic
performance of the system, but also can be used to study the influence of the system structure
or parameter changes on the system performance. The frequency method and the root locus
method, which are widely used in classical control theory, are established based on transfer
function, which is the most basic and important concept in classical control theory.

One of the purposes of establishing the mathematical model of the control system is to
quantitatively study the working characteristics of the control system with mathematical
methods. After the system differential equations are written out, as long as the input and
initial conditions are given, the differential equations can be solved, and the characteristics of
the system output changing with time can be understood. There are two methods for solving
linear ODEs: the classical method and the Laplace transform method. This section only
studies the method of solving differential equations by Laplace transform method, and
analyzes the composition of differential equation solutions, which lays the foundation for the
concept of transfer function in the future.

What is Laplace transform?

➢ A math method
➢ An integral transformation between functions.
➢ Transform differential equations in the time domain into algebraic equations in the
complex domain
➢ Simply the solution of differential equation.

Figure 2.8 Laplace transforms

Definition:
+∞
𝑳[𝒇(𝒕)] = 𝑭(𝒔) = ∫𝟎 𝒇(𝒕)𝒆−𝒔𝒕 𝒅𝒕 where s is a complex number

The Laplace transform method substitutes relatively easily solved algebraic equations for
more difficult differential equations
𝑝(𝑠) 𝑏0 𝑠 𝑚 + 𝑏0 𝑠 𝑚−1 + ⋯ + 𝑏𝑚−1 𝑠 + 𝑏𝑚 𝑘(𝑠 + 𝑧1 )(𝑠 + 𝑧2 ) ⋯ (𝑠 + 𝑧𝑚 )
𝐹(𝑠) = = = {2 − 48}
𝑞(𝑠) 𝑠 𝑛 + 𝑎1 𝑠 𝑛−1 + ⋯ + 𝑎𝑛−1 𝑠 + 𝑏𝑛 (𝑠 + 𝑝1 )(𝑠 + 𝑝2 ) ⋯ (𝑠 + 𝑝𝑛 )
➢ Characteristic equation: The denominator polynomial q(s) set equal to zero
➢ Poles: The roots of the characteristic equation
➢ Zeroes: The roots of the numerator polynomial p(s)

• Laplace Transform:
 ∞
𝐹(𝑠) = ∫ 𝑓(𝑡)𝑒 −𝑠𝑡 𝑑𝑡 {2 − 49}
−∞

• When t is large enough, the equation

|𝑓(𝑡)| ≤ 𝑀𝑒𝑐𝑡 {2 − 50}

• In the complex plane, 𝑅𝑒(𝑠) > 𝑐 exists, converge in the right plane
• Linear properties:
𝐿[𝑎𝑓1 (𝑡) + 𝑏𝑓2 (𝑡)] = 𝑎𝐹1 (𝑠) + 𝑏𝐹2 (𝑠) {2 − 51}

• Shifting theorem:

𝐿[𝑒 𝑎𝑡 𝑓(𝑡)] = 𝐹(𝑠 − 𝑎) {2 − 52}

𝐿[𝑓(𝑡 − 𝜏0 )] = 𝑒 −𝜏0 𝑠 𝐹(𝑠) {2 − 53}

• Integrate method
𝑑𝑓(𝑡)
𝐿[ ] = 𝑠𝐹(𝑠) − 𝑓(0) {2 − 54}
𝑑𝑡

𝑑2 𝑓(𝑡)
𝐿[ ] = 𝑠 2 𝐹(𝑠) − 𝑠𝑓(0) − 𝑓 ′ (0) {2 − 55}
𝑑𝑡 2

𝐿[𝑓 (𝑛) (𝑡)] = 𝑠 𝑛 𝐹(𝑠) − 𝑠 𝑛−1 𝑓(0) − ⋯ − 𝑓 (𝑛−1) (0) {2 − 56}

• In the equation 𝑓(0), 𝑓 ′ (0), … , 𝑓 𝑛−1 (𝑜) is the derivative value when 𝑡 = 0
• Final-value theorem:
𝑙𝑖𝑚 𝑓(𝑡) = 𝑙𝑖𝑚𝑠𝐹(𝑠) {2 − 57}
𝑛→∞ 𝑠→0

• If 𝐹(𝑠) = 𝐿 [𝑓(𝑡)]，when 𝑡→, comes to final-value theorem.

• When calculating the steady-state error

𝑒(∞) = lim 𝑒(𝑡) = lim 𝑠𝐸(𝑠) {2 − 58}
𝑡→∞ 𝑠→∞

Typical System Laplace Transform

1. Step input
 f(t)

O t

Figure 2.9 Step input

Unit step function
1 𝑡≥0
𝑓(𝑡) = 1(𝑡) = { {2 − 59}
0 𝑡<0
Laplace Transform
∞ ∞
1 ∞
𝐹(𝑠) = 𝐿[𝑓(𝑡)] = ∫ 𝑓(𝑡)𝑒 −𝑠𝑡 𝑑𝑡 = ∫ 1 ∗ 𝑒 −𝑠𝑡 𝑑𝑡 = − 𝑒 −𝑠𝑡 |
0 0 𝑠 0
1 1
= − [0 − 1] = {2 − 60}
𝑠 𝑠

2. Ramp input

Figure 2.10 Ramp input

𝑡 𝑡≥0 {2 − 61}
𝑓(𝑡) = 𝑡 ∙ 1(𝑡) = {
0 𝑡<0
∞ ∞
1 ∞
𝐹(𝑠) = 𝐿[𝑓(𝑡)] = ∫ 𝑓(𝑡)𝑒 −𝑠𝑡 𝑑𝑡 = ∫ 𝑡 ∙ 𝑒 −𝑠𝑡 𝑑𝑡 = − (𝑠𝑡 + 1)𝑒 −𝑠𝑡 | {2 − 62}
0 0 𝑠2 0

3. Parabola input

f(t)

O t

Figure 2.11 Parabola input

 𝑡2
𝑓(𝑡) = { 2 𝑡 ≥ 0 {2 − 63}
0 𝑡<0

Laplace Transform
+∞ +∞
1 2 −𝑠𝑡
𝐹(𝑠) = 𝐿[𝑓(𝑡)] = ∫ 𝑓(𝑡) 𝑒 −𝑠𝑡 𝑑𝑡 = ∫ 𝑡 𝑒 𝑑𝑡
0 0 2
1 1 2 −𝑠𝑡 +∞ +∞ 1 1 1
=− 2 [2 𝑡 𝑒 | 0 − ∫0 𝑡 ∙ 𝑒 −𝑠𝑡 𝑑𝑡] = − [0 − 0 − 2 ] = 3 {2 − 64}
𝑠 𝑠 𝑠 𝑠

4. Exponential input
−𝑎𝑡
𝑓(𝑡) = {𝑒 𝑡 ≥ 0 (𝑎 𝑖𝑠 𝑟𝑒𝑎𝑙 𝑛𝑢𝑚𝑏𝑒𝑟) {2 − 65}
0 𝑡<0
Laplace Transform
+∞ +∞
−𝑎𝑡 ] −𝑎𝑡 −𝑠𝑡
1 ∞
𝐹(𝑠) = 𝐿[𝑒 =∫ 𝑒 𝑒 𝑑𝑡 = ∫ 𝑒 −(𝑎+𝑠)𝑡 𝑑𝑡 = 𝑒 −(𝑎+𝑠)𝑡 |
0 0 −(𝑎 + 𝑠) 0
1
= {2 − 66}
𝑎+𝑠

5. Sine input:
𝑠𝑖𝑛𝜔𝑡 𝑡 ≥ 0 {2 − 67}
𝑠𝑖𝑛𝜔𝑡 = {
0 𝑡<0
place transform:
∞ ∞
−𝑠𝑡
1 𝑗𝜔𝑡
𝐹(𝑠) = 𝐿[𝑠𝑖𝑛𝜔𝑡] = ∫ 𝑠𝑖𝑛𝜔𝑡𝑒 𝑑𝑡 = ∫ (𝑒 − 𝑒 −𝑗𝜔𝑡 )𝑒 −𝑠𝑡 𝑑𝑡
0 0 2𝑗
1 1 1 𝜔
= [ − ]= 2 {2 − 68}
2𝑗 𝑠 − 𝑗𝜔 𝑠 + 𝑗𝜔 𝑠 + 𝜔2

6. Unit impulse function (𝜹 𝒇𝒖𝒏𝒄𝒕𝒊𝒐𝒏)

 (t)

O t
Figure 2.12 Unit impulse function
 +∞
∞ 𝑡=0
𝛿(𝑡) = { 𝑎𝑛𝑑 ∫ 𝛿(𝑡)𝑑𝑡 = 1 {2 − 69}
0 𝑡≠0 −∞
+∞
𝐹(𝑠) = 𝐿[𝛿(𝑡)] = ∫ 𝛿(𝑡)𝑒 −𝑠𝑡 𝑑𝑡 = 1 {2 − 70}
0

2.3.2 Inverse Laplace Transform

Definition

1 𝜎+𝑗𝜔
𝐿−1 [𝐹(𝑠)] = 𝑓(𝑡) = ∫ 𝐹(𝑠)𝑒 −𝑠𝑡 𝑑𝑡 {2 − 71}
2𝜋𝑗 𝜎−𝑗𝜔

Normally get 𝐹(𝑠) from 𝑓(𝑡) by Rational fraction functions division

But sometimes 𝐹(𝑠) is hard to be calculated, in this case we check the table.
Normally 𝐹(𝑠) is
𝑏0 𝑠 𝑚 + 𝑏1 𝑠 𝑚−1 + ⋯ + 𝑏𝑚−1 𝑠 + 𝑏𝑚
𝐹(𝑠) = {2 − 72}
𝑠 𝑛 + 𝑎1 𝑠 𝑛−1 + ⋯ + 𝑎𝑛−1 𝑠 + 𝑎𝑛
In the equation, 𝑎1, 𝑎2, … , 𝑎𝑛 and 𝑏1, 𝑏2, … , 𝑏𝑚 are all real numbers, 𝑚, 𝑛 are all
positive numbers, and 𝑚 < 𝑛.
Normally 𝐹(𝑠) can be separated to
𝐹(𝑠) = 𝐹1 (𝑠) + 𝐹2 (𝑠) + ⋯ + 𝐹𝑛 (𝑠) {2 − 73}

Due to the linear properties

𝐿−1 [𝐹(𝑠)] = 𝐿−1 [𝐹1 (𝑠)] + 𝐿−1 [𝐹2 (𝑠)] + ⋯ + 𝐿−1 [𝐹𝑛 (𝑠)] = 𝑓1 (𝑡) + 𝑓2 (𝑡) + ⋯ + 𝑓𝑛 (𝑡){2 − 74}

Example 4:
𝑝(𝑠) 𝑠+3
𝐹(𝑠) = = {2 − 75}
𝑞(𝑠) (𝑠 + 1)(𝑠 + 2)
Expanding equation (1) in a partial fraction expansion
𝑘1 𝑘2
𝐹(𝑠) = + {2 − 76}
𝑠+1 𝑠+2
➢ Where 𝑘1 and 𝑘2 are the coefficients
➢ The coefficients 𝑘𝑖 are called residues
The residues are evaluated by multiplying through by the denominator factor of equation
(1) corresponding to 𝑘𝑖 and setting 𝑠 equal to the root
(𝑠 − 𝑠1 )𝑝(𝑠) (𝑠 + 1)(𝑠 + 3)
𝑘1 = |𝑠=𝑠1 = | =2 {2 − 77}
𝑞(𝑠) (𝑠 + 1)(𝑠 + 2) 𝑠=−1
(𝑠 − 𝑠2 )𝑝(𝑠) (𝑠 + 2)(𝑠 + 3)
𝑘1 = |𝑠=𝑠2 = | = −1 {2 − 78}
𝑞(𝑠) (𝑠 + 1)(𝑠 + 2) 𝑠=−2
 2 −1
𝐹(𝑠) = + {2 − 79}
𝑠+1 𝑠+2
𝐹(𝑡) = 2𝑒 −𝑡 − 𝑒 −2𝑡 {2 − 80}
Steady-state or final value

𝐹(𝑡) = 2𝑒 −𝑡 − 𝑒 −2𝑡
Steps of Solving Differential Equation using Laplace Transform
(1) Laplace transform the equation, transform the differential equation to s-domain
algebra equation. the initial condition is the value when 𝑡 = 0
(2) Solve the algebra equation and get the s-domain solution.
(3) Do inverse Laplace transform and get the solution of differential equation.

Example 5:
Solve the differential equation:
𝑥̈ (𝑡) + 2𝑥̇ (𝑡) + 𝑥(𝑡) = 0, 𝑥̇ (0) = 0, 𝑥(0) = 𝑥0 {2 − 81}

Laplace transform:

𝑠 2 𝑋(𝑠) − 𝑠𝑥(0) − 𝑥̇ (0) + 2𝑠𝑋(𝑠) − 2𝑥(0) + 𝑋(𝑠) = 0 {2 − 82}

Take in the initial condition:
𝑠+2 𝑥0 𝑥0
𝑋(𝑠) = 𝑥0 = + {2 − 83}
𝑠2 + 2𝑠 + 1 (𝑠 + 1)2 𝑠+1
Do inverse Laplace transform:
𝑥(𝑡) = 𝑥0 (𝑡 + 1)𝑒 −𝑡 (𝑡 ≥ 0) {2 − 84}

2.4 The transfer function of linear systems

Definition：

The transfer function of a linear, stationary (constant parameter) system is defined as the
ratio of the Laplace transform of the output variable to the Laplace transform of the input
variable, with all initial conditions assumed to be zero.
𝐿𝑎𝑝𝑙𝑎𝑐𝑒 𝑡𝑟𝑎𝑛𝑠𝑓𝑜𝑟𝑚𝑠 𝑜𝑓 𝑜𝑢𝑡𝑝𝑢𝑡 𝐶(𝑠)
𝑇𝑟𝑎𝑛𝑠𝑓𝑒𝑟 𝐹𝑢𝑛𝑐𝑡𝑖𝑜𝑛 = | = {2 − 85}
𝐿𝑎𝑝𝑙𝑎𝑐𝑒 𝑡𝑟𝑎𝑛𝑠𝑓𝑜𝑟𝑚𝑠 𝑜𝑓 𝑖𝑚𝑝𝑢𝑡 𝑖𝑛𝑖𝑡𝑖𝑎𝑙 𝑐𝑜𝑛𝑑𝑖𝑡𝑖𝑜𝑛𝑠 𝑅(𝑠)

Clarification:
(1) Transfer function is an expression in s domain, which relates the outputs and inputs of
a linear time-invariant system through system structural parameters, regardless of the
input form. It only applies to linear time-invariant systems.
(2) The degree of the denominator is always greater than or equal to the degree of the
numerator, 𝑛 ≥ 𝑚. The highest order of the denominator polynomial is 𝑛, so the
 system is called n-order system.
(3) The communication function only describes the relationship between the input and
output of the system but does not reflect any information about the internal structure
of the system. Therefore, it is completely possible for different physical systems to
have the same form of transfer function, which creates conditions for mathematical
simulation.
(4) For same system, different transfer function, they have same denominator, the only
difference is the numerator. Denominator is characteristics equation, noted as 𝐷(𝑠).
𝐷(𝑠) = 𝑎0 𝑠 𝑛 + 𝑎1 𝑠 𝑛−1 + ⋯ + 𝑎𝑛 {2 − 86}

(5) The description of transfer function has some limitations: it can only study single-
input and single-output systems, and transfer matrix is used to represent multi-input
and multi-output systems. It can only represent the relationship between input and
output, and other variables inside the system cannot be known (the deficiency of
classical control theory), only the system characteristics of zero initial state can be
studied, and the system motion characteristics of non-zero initial state cannot be
reflected.

Consider the dynamic system represented by the differential equation：

𝑑𝑛 𝑑𝑛−1 𝑑
𝑎0 𝑛
𝑐(𝑡) + 𝑎1 𝑛−1
𝑐(𝑡) + ⋯ + 𝑎𝑛−1 𝑐(𝑡) + 𝑎𝑛 𝑐(𝑡)
𝑑𝑡 𝑑𝑡 𝑑𝑡
𝑑𝑚 𝑑𝑚−1 𝑑
= 𝑏0 𝑚
𝑟(𝑡) + 𝑏1 𝑚−1
𝑟(𝑡) + ⋯ + 𝑏𝑚−1 𝑟(𝑡) + 𝑏𝑚 𝑟(𝑡) {2 − 87}
𝑑𝑡 𝑑𝑡 𝑑𝑡
Where 𝑐(𝑡) is the response，𝑟(𝑡) is the input. All coefficient is constant.

If the initial conditions are all zero, then the transfer function is
𝐶(𝑠) 𝑏0 𝑠 𝑚 + 𝑏1 𝑠 𝑚−1 + ⋯ + 𝑏𝑚−1 𝑠 + 𝑏𝑚 𝑀(𝑠)
𝐺(𝑠) = = = {2 − 88}
𝑅(𝑠) 𝑎0 𝑠 𝑛 + 𝑎1 𝑠 𝑛−1 + ⋯ + 𝑎𝑛−1 𝑠 + 𝑎𝑛 𝑁(𝑠)
𝑀(𝑠) = 𝑏0 𝑠 𝑚 + 𝑏1 𝑠 𝑚−1 + ⋯ + 𝑏𝑚−1 𝑠 + 𝑏𝑚 {2 − 89}

𝑁(𝑠) = 𝑎0 𝑠 𝑛 + 𝑎1 𝑠 𝑛−1 + ⋯ + 𝑎𝑛−1 𝑠 + 𝑎𝑛 {2 − 90}

𝐶(𝑠) 𝑏0 𝑠 𝑚 + 𝑏1 𝑠 𝑚−1 + ⋯ + 𝑏𝑚−1 𝑠 + 𝑏𝑚 𝑀(𝑠)

𝐺(𝑠) = = =
𝑅(𝑠) 𝑎0 𝑠 𝑛 + 𝑎1 𝑠 𝑛−1 + ⋯ + 𝑎𝑛−1 𝑠 + 𝑎𝑛 𝑁(𝑠)
𝑘(𝑠 + 𝑧1 )(𝑠 + 𝑧2 ) ⋯ (𝑠 + 𝑧𝑚 )
= {2 − 91}
(𝑠 + 𝑝1 )(𝑠 + 𝑝2 ) ⋯ (𝑠 + 𝑝𝑛 )
➢ Characteristic equation: The denominator polynomial q(s) set equal to zero
➢ Poles: The roots of the characteristic equation
➢ Zeroes: The roots of the numerator polynomial p(s)
 Figure 2.13 The transfer function of RL circuits

Transfer functions of typical components：

1. Constant Gain

Figure 2.14 Constant gain

differential equation:
𝑐(𝑡) = 𝐾𝑟(𝑡) {2 − 92}
K is constant and is called gain or proportional coefficient.
Transfer function
𝐺(𝑠) = 𝐾 {2 − 93}

Example 6:
operational amplifier

Figure 2.15 Operational amplifier

𝑈2 𝑅𝑓
= =𝐾 {2 − 94}
𝑈1 𝑅1
Potentiometer
A potentiometer is a device that converts linear displacement or angular displacement
into voltage. In the control system, a single potentiometer is used as a signal conversion
device, and a pair of potentiometers can form an error detection device
At no-load, the relationship between the brush angular displacement 𝜃(𝑡) of a single
potentiometer and the output voltage u(t) is as follows. It can be seen from the figure that the
𝐸
output voltage is 𝑢(𝑡) = 𝐾1 𝜃(𝑡), where𝐾1 = is the output voltage corresponding to the
𝜃𝑚𝑎𝑥
unit angular displacement of the brush, which is called the potentiometer transfer coefficient,
where E is the power supply voltage of the potentiometer, and 𝜃𝑚𝑎𝑥 is the maximum working
angle of the potentiometer. Then the transfer function of the potentiometer can be obtained

Figure 2.16 Potentiometer

𝑈(𝑠) 𝐸𝑚
= =𝐾=𝐺 {2 − 95}
𝜃(𝑆) 𝜃𝑚
2. Poles at origin

Figure 2.17 Poles at origin

Differential equation:

𝑐(𝑡) = ∫ 𝑟(𝑡)𝑑𝑡 {2 − 96}

Transfer function:
1
𝐺(𝑠) = {2 − 97}
𝑠
Example 7:
integrator

Figure 2.18 Integrator

Transfer function:
1
𝑈2 (𝑠) 𝐶𝑓 𝑠 1 1
= = = {2 − 98}
𝑈1 (𝑠) 𝑅1 𝑅1 𝐶𝑓 𝑠 𝑇𝑖 𝑠

Example Light oil cylinder

 Figure 2.19 Light oil cylinder
Flow rate:

𝑑𝑥
𝑄𝑓 (𝑡) = 𝐴 {2 − 99}
𝑑𝑡

Transfer function

𝑋(𝑠) 1⁄ 𝐾
= 𝐴= {2 − 100}
𝑄𝑓 (𝑠) 𝑠 𝑠

Example small inertia motor:

Figure 2.20 small inertia motor

Transfer function:

𝜃𝑚 (𝑠) 𝐾𝑚
= {2 − 101}
𝑈𝑎 (𝑠) 𝑠

3. Zeros at Origin

Figure 2.21 Zeros at origin

Differential equation:
𝑑𝑟(𝑡)
𝑐(𝑡) = {2 − 102}
𝑑𝑡
transfer function:
 𝐺(𝑠) = 𝑠 {2 − 103}

Example 8:
tachometer generator:

Figure 2.22 tachometer generator

transfer function:

𝑢(𝑡) = 𝐾𝑡 𝜃̇ (𝑡) {2 − 104}

𝑈(𝑠)
= 𝐾𝑡 𝑠 {2 − 105}
𝜃(𝑠)
4. Poles at Real Axis

Figure 2.23 Poles at real axis

differential equation:
𝑑𝑐(𝑡)
𝑇 + 𝑐(𝑡) = 𝑟(𝑡) {2 − 106}
𝑑𝑡
transfer function:
1
𝐺(𝑠) = {2 − 107}
𝑇𝑠 + 1
Example 9:
operational amplifier

Figure 2.24 Operational amplifier

transfer function:
 1
𝑅𝑓 𝐶 𝑠
𝑓
⁄ 1
(𝑅𝑓 + 𝐶 𝑠) 𝑅𝑓
𝑈2 (𝑠) ⁄𝑅 𝐾
𝑓 1
= = = {2 − 108}
𝑈1 (𝑠) 𝑅1 𝑅𝑓 𝐶𝑓 𝑠 + 1 𝑇𝑠 + 1

5. Zeros at Real Axis

Figure 2.25 Zeros at real axis

differential equation:
𝑑𝑟(𝑡)
𝑐(𝑡) = 𝜏 + 𝑟(𝑡) {2 − 109}
𝑑𝑡
transfer function:
𝐺(𝑠) = 𝜏𝑠 + 1 {2 − 110}
RC network feedback is applied to the amplifier when the gain K is sufficiently large
Example 10:

Figure 2.26 example

𝑈2 (𝑠) 𝐾 𝐾(𝑅𝐶𝑠 + 1) 𝑅𝐶𝑠 + 1
= = = ≈ 𝑅𝐶𝑠 + 1 = 𝜏𝑠 + 1 {2 − 111}
𝑈1 (𝑠) 1 + 1 𝑅𝐶 1
𝐾 𝑅𝐶𝑠 + 1 + 𝐾 𝑠+ +1
𝑅𝐶𝑠 + 1 𝐾 𝐾
6.Complex Conjugate Poles

Figure 2.27 Complex conjugate poles

differential equation:
𝑑2 𝑐(𝑡) 𝑑
2
+ 2𝜉𝜔𝑛 𝑐(𝑡) + 𝜔𝑛2 𝑐(𝑡) = 𝜔𝑛2 𝑟(𝑡) {2 − 112}
𝑑𝑡 𝑑𝑡
transfer function:
𝜔𝑛2
𝐺(𝑠) = {2 − 113}
𝑠 2 + 2𝜉𝜔𝑛 𝑠 + 𝜔𝑛2
𝜉 : damping ratio
𝜔𝑛 : natural frequency
Example:

Figure 2.28 Complex conjugate poles

𝑈2 (𝑠) 1 1⁄ 𝜔𝑛2
= = 𝐿𝑠 = 2 {2 − 114}
𝑈1 (𝑠) 𝐿𝐶𝑠 2 + 𝑅𝐶𝑠 + 1 𝑠 2 + 𝑅 𝑠 + 1 𝜔𝑛 + 𝑠 2 + 2𝜔𝑛 𝑠
𝐿 𝐿𝐶
Example 11:
Spring Mass Damper System

Figure 2.29 Spring mass damper system

Newton’s second law:

∑ 𝐹 = 𝑚𝑎 {2 − 115}

𝑑 𝑑2
𝐹(𝑡) − 𝑘𝑦(𝑡) − 𝑓 𝑦(𝑡) = 𝑚 2 𝑦(𝑡) {2 − 116}
𝑑𝑡 𝑑𝑡
Do LaPlace transform:
(𝑚𝑠 2 + 𝑓𝑠 + 𝑘)𝑌(𝑠) = 𝐹(𝑠) {2 − 117}
1𝑘
𝑌(𝑠) 1 𝑘 𝑚 𝐾𝜔𝑛2
𝐺(𝑠) = = = = {2 − 118}
𝐹(𝑠) 𝑚𝑠 2 + 𝑓𝑠 + 𝑘 𝑠 2 + 𝑓 𝑠 + 𝑘 𝑠 2 + 2𝜉𝜔𝑛 𝑠 + 𝜔𝑛2
𝑚 𝑚
7. Complex Conjugate Zeros
 Figure 2.30 Complex conjugate zeros
differential equation:
𝑑2 𝑑
𝑐(𝑡) = 𝜏 2 2
𝑟(𝑡) + 2𝜉𝜏 𝑟(𝑡) + 𝑟(𝑡) {2 − 119}
𝑑𝑡 𝑑𝑡
transfer function:
𝐺(𝑠) = 𝜏 2 𝑠 2 + 2𝜉𝜏𝑠 + 1 {2 − 120}

𝜉 : damping ratio

Example 12:
If the transfer function
4𝑠 + 2
𝐺(𝑠) = {2 − 121}
(𝑠 + 1)(𝑠 + 2)
Find the poles and zeros.

p2 p1 z1
-2 -1 O

-1

Figure 2.31 example

𝐾(𝑠 + 𝑧1 )(𝑠 + 𝑧2 ) ⋯ (𝑠 + 𝑧𝑚 )
{2 − 122}
(𝑠 + 𝑝1 )(𝑠 + 𝑝2 ) ⋯ (𝑠 + 𝑝𝑛 )

Poles: 𝑝1 = −1， 𝑝2 = −2

Zeros: 𝑧1 = −0.5
Example: RLC circuit:

Ur(s) Uc(s)
G(s)
 Figure 2.32 RLC circuit
𝑑𝑈𝑐 𝑑𝑈𝑐
𝑖=𝐶 → 𝑅𝐶 + 𝑈𝑐 = 𝑈𝑟 (𝑠) {2 − 123}
𝑑𝑡 𝑑𝑡
𝑅𝐶[𝑠𝑈𝑐 (𝑠) − 𝑈𝑐 (0)] + 𝑈𝑐 (𝑠) = 𝑈𝑟 (𝑠) {2 − 124}

1. if 𝑈𝑐 (0) = 0，namely initial conditions assumed to be zero

𝑅𝐶𝑠𝑈𝑐 (𝑠) + 𝑈𝑐 (𝑠) = 𝑈𝑟 (𝑠) {2 − 125}
[𝑅𝐶𝑠 + 1]𝑈𝑐 (𝑠) = 𝑈𝑟 (𝑠) {2 − 126}
𝑈𝑐 (𝑠) 1
= {2 − 127}
𝑈𝑟 (𝑠) 𝑅𝐶𝑠 + 1
2. if 𝑈𝑐 (0) ≠ 0 ，namely initial conditions assumed to be nonzero
𝑅𝐶𝑠𝑈𝑐 (𝑠) + 𝑈𝑐 (𝑠) − 𝑅𝐶𝑈𝑐 (0) = 𝑈𝑟 (𝑠) {2 − 128}
1 𝑅𝐶
𝑈𝑐 (𝑠) = 𝑈 (𝑠) + 𝑈 (0) {2 − 129}
𝑅𝐶𝑠 + 1 𝑟 𝑅𝐶𝑠 + 1 𝑐

The transfer function of linear systems

Characteristic:
1.The transfer function of linear system depends on the structure and parameters of the system
itself, having nothing to do with the input signal.
2. The transfer function is a fractional function with complex variable-s, and has all the
properties of complex variable function；𝑚 ≤ 𝑛；all coefficients of transfer function are real

3. The transfer function descripts the cause-effect relationship between the input and output,
but it does not provide any details of the physical system structure
2.5 Block diagram models
The structure diagram and the signal flow diagram of the control system are
mathematical diagrams that describe the signal transmission relationship between the various
components of the system. They represent the causal relationship between the variables in the
system and the operations performed on each variable. A convenient way to describe complex
systems. Compared with the structure diagram, the symbol of the signal flow diagram is
simple, which is easier to draw and apply, especially in the computer simulation research of
the system and the analysis and design of the state space, the signal flow diagram can directly
give the computer simulation program and the state equation description of the system, which
shows its superiority. However, signal flow diagrams are only suitable for linear systems,
while structure diagrams can also be used for nonlinear systems.
➢ The dynamic systems are represented mathematically by a set of differential
equations
➢ The Laplace transformation reduces the problem to the solution of a set of linear
algebraic equation
➢ The block diagram representation of system variables relationships is an important
means and prevalent in control system engineering
2.5.1 The composition of the block diagram
The structure diagram of the control system is composed of many blocks that perform
single operation on the signal and some signal flow lines. It includes the following four basic
elements:

Operational block－the transfer function of the variables Write the transfer function of
the component or system in the box

R(s) C(s)
G(s)

Figure 2.33 Operational block

Signal line－the lines with arrow, indicate the direction of the signal (unidirectional)

U(s)

Figure 2.34 Signal line

Pickoff point (measurement point) －the measuring location of the signal, the magnitude
and characteristic of measured signal remain unchanged

U(s)

U(s)

Figure 2.35 Pickoff point

Summing point-the signal must be of the same nature

U(s) U(s)  B(s)

 B(s)

Figure 2.36 Summing point

When drawing the system structure diagram, first consider the load effect and write
down the differential equations or transfer functions of each component of the system, and
represent them in boxes; then, according to the signal flow direction of each component, use
signal lines to define each box. The connection will get the structure diagram of the system.
Therefore, the system structure diagram is essentially a combination of the system schematic
diagram and mathematical equations, which not only supplements the quantitative description
lacking in the schematic diagram, but also avoids the abstract operation of pure mathematics.
Its role in the system; more importantly, the transfer function of the system can be easily
obtained from the system structure diagram. Therefore, the system structure diagram is also a
mathematical model of the control system.
Although the system structure diagram is obtained from the mathematical model of the
system components, the blocks in the structure diagram are not in a one-to-one
correspondence with the actual system components. An actual component can be represented
by several boxes; and a box can also represent several components or a subsystem, or a large
complex system.
Example 13:
RC filter network
Modeling system differential equation:
𝑢𝑟 (𝑡) = 𝑅𝑖(𝑡) + 𝑢𝑐 (𝑡)
{ 𝑑𝑢𝑐 (𝑡) {2 − 130}
𝑖(𝑡) = 𝐶
𝑑(𝑡)
Do Laplace transform
1
𝑈𝑟 (𝑠) − 𝑈𝑐 (𝑠) = 𝑅𝐼(𝑠) → 𝐼(𝑠) = [𝑈𝑟 (𝑠) − 𝑈𝑐 (𝑠)]
{ 𝑅 {2 − 131}
1
𝐼(𝑠) = 𝐶𝑠𝑈𝑐 (𝑠) → 𝑈𝑐 (𝑠) = 𝐼(𝑠)
𝐶𝑠
Draw block diagram:

Ur(s) Ur(s)-Uc(s) Ur(s)-Uc(s) 1 I(s) I(s) Uc(s)

1
- R Cs
Uc(s)

Figure 2.37 separate block diagram

Connect the block diagram to get the dynamic structure diagram

Ur(s) Ur(s)-Uc(s) 1 I(s) Uc(s)

1
- R Cs

Figure 2.38 Connecting the block diagram

2.5.2 Combining blocks
In a complex system structure diagram, the connection of its blocks must be intricate, but
the basic connection methods between the blocks are only series, parallel and feedback.
Therefore, the general method of structure diagram simplification is to move the lead point or
comparison point, exchange the comparison point, and perform block operation to combine
the blocks connected in series, parallel and feedback. In the simplification process, the
principle of keeping the variable relationship before and after the transformation should be
kept equivalent. Specifically, the product of the transfer function in the forward path before
and after the transformation should remain unchanged, and the product of the transfer
function in the loop should remain unchanged.
The rule of combining blocks:
(1) Structure equivalent
(2) Unique transfer function, diverse structure block diagram
 (3) One direction transfer signal
(4) Superposition of multiple input systems

Simplification clarification
1. Combining blocks in cascade
The transfer functions are respectively several blocks in the first row, and the equivalent
blocks of several blocks connected in series are equal to the product of the transfer functions
of each block.

R(s) C(s)
G1(s) G2(s) ...... Gn(s)

R(s) C(s)
G1(s)G2(s)...Gn(s)

Figure 2.39 Combining blocks in cascade

𝑈1 (𝑠) = 𝐺1 (𝑠)𝑅(𝑠) {2 − 132}

𝑈2 (𝑠) = 𝐺2 (𝑠)𝑈1 (𝑠) = 𝐺2 (𝑠)𝐺1 (𝑠)𝑅(𝑠) {2 − 133}

𝐶(𝑠) = 𝐺3 (𝑠)𝑈2 (𝑠) = 𝐺3 (𝑠)𝐺2 (𝑠)𝐺1 (𝑠)𝑅(𝑠) {2 − 134}

𝐶(𝑠)
= 𝐺3 (𝑠)𝐺2 (𝑠)𝐺1 (𝑠) = 𝐺(𝑠) {2 − 135}
𝑅(𝑠)

𝐺(𝑠) = ∑ 𝐺𝑖 (𝑠) {2 − 136}

𝑖=1

2. Combining blocks in parallel

The transfer functions are several boxes in the first picture. If they have the same input
and the output is equal to the algebraic sum of the outputs of the two boxes, then these boxes
are called parallel connections.
The equivalent equation connected in parallel is equal to the algebraic sum of the transfer
functions of the individual blocks.
 G1(s)

R(s) C(s)
G2(s)

...
Gn(s)

R(s) C(s)
G1(s)+G2(s)+    +Gn(s)

Figure 2.40 Combining blocks in parallel

𝐶(𝑠) = 𝐶1 (𝑠) + 𝐶2 (𝑠) + 𝐶3 (𝑠) = 𝐺1 (𝑠)𝑅(𝑠) + 𝐺2 (𝑠)𝑅(𝑠) + 𝐺3 (𝑠)𝑅(𝑠)

= [𝐺1 (𝑠) + 𝐺2 (𝑠) + 𝐺3 (𝑠)]𝑅(𝑠) {2 − 137}

𝐶(𝑠)
= 𝐺1 (𝑠) + 𝐺2 (𝑠) + 𝐺3 (𝑠) = 𝐺(𝑠) {2 − 138}
𝑅(𝑠)
𝑛

𝐺(𝑠) = ∑ 𝐺𝑖 (𝑠) {2 − 139}

𝑖=1

3. Eliminating feedback loop

If the transfer function is two boxes of 𝐺(𝑠) and 𝐻(𝑠) respectively, it is called a feedback
connection. "+" means positive feedback, which means that the input signal and feedback
signal are added; "-" means subtraction, which is negative feedback.

R(s) E(s) C(s)

G(s) R(s) G( s ) C(s)
 1 G( s) H ( s)
B(s)
H(s)
Figure 2.41 Eliminating feedback loop

𝐶(𝑆) = 𝐺(𝑆)𝐸(𝑆) = 𝐺(𝑆)[𝑅(𝑆) − 𝐻(𝑆)𝐶(𝑆)]

= 𝐺(𝑆)𝑅(𝑆) − 𝐺(𝑆)𝐻(𝑆)𝐶(𝑆) {2 − 140}

𝐶(𝑠) 𝐺(𝑠)
= {2 − 141}
𝑅(𝑠) 1 + 𝐺(𝑠)𝐻(𝑠)
If 𝐻(𝑠) = 1 ⟹ unity feedback
𝐶(𝑠) 𝐺(𝑠)
= {2 − 142}
𝑅(𝑠) 1 + 𝐺(𝑠)

4. Moving summing point ahead

 R(s) C(s) R(s) C(s)
G(s) G(s)
 
B(s) 1 B(s)
G(s)
Figure 2.42 Moving summing point ahead

𝐶(𝑠)
[R(s)𝐺1 (𝑠) − 𝐶(𝑠)] × 𝐺2 (𝑠) = 𝐶(𝑠) → 𝑅(𝑠)
𝐺1 (𝑠)𝐺2 (𝑠)
= {2 − 143}
1 + 𝐺2 (𝑠)

𝐶(𝑠) 𝐶(𝑠) 𝐺1 (𝑠)𝐺2 (𝑠)

[𝑅(𝑠) − ] × 𝐺1 (𝑠) × 𝐺2 (𝑠) = 𝑐(𝑠) → = {2 − 144}
𝐺1 (𝑠) 𝑅(𝑠) 1 + 𝐺2 (𝑠)

5. Moving summing point behind

Figure 2.43 Moving summing point behind

𝐶(𝑠) 𝐺1 (𝑠)𝐺2 (𝑠)

[𝑅(𝑠) − 𝐶(𝑠)] × 𝐺1 (𝑠) × 𝐺2 (𝑠) = 𝐶(𝑠) → = {2 − 145}
𝑅(𝑠) 1 + 𝐺1 (𝑠)𝐺2 (𝑠)

𝐶(𝑠) 𝐺1 (𝑠)𝐺2 (𝑠)

[𝑅(𝑠) × 𝐺1 (𝑠) − 𝐶(𝑆) × 𝐺1 (𝑠)] × 𝐺2 (𝑠) = 𝐶(𝑠) → = {2 − 146}
𝑅(𝑠) 1 + 𝐺1 (𝑠)𝐺2 (𝑠)

6. Moving pickoff point ahead

Figure 2.44 Moving pickoff point ahead

 𝐶(𝑠) 𝐺1 (𝑠)𝐺2 (𝑠)
[𝑅(𝑠) − 𝐶(𝑠)] × 𝐺1 (𝑠) × 𝐺2 (𝑠) = 𝐶(𝑠) → = {2 − 147}
𝑅(𝑠) 1 + 𝐺1 (𝑠)𝐺2 (𝑠)

𝑅(𝑠)𝐺1 (𝑠) 𝐶(𝑠) 𝐺1 (𝑠)𝐺2 (𝑠)

𝑀(𝑠) = → = {2 − 148}
1 + 𝐺1 (𝑠)𝐺2 (𝑠) 𝑅(𝑠) 1 + 𝐺1 (𝑠)𝐺2 (𝑠)

7. Moving pickoff point behind

Figure 2.45 Moving pickoff point behind

𝐶(𝑠) 𝐺1 (𝑠)𝐺2 (𝑠)
= {2 − 149}
𝑅(𝑠) 1 + 𝐺1 (𝑠)𝐻(𝑠)

Figure 2.46 Combining blocks

 Figure 2.47 Moving a summing junction

Figure 2.48 Moving a pickoff point

Notes in Block diagram simplification

(1) Series, parallel and feedback three typical structures can be directly used formula; You
can't use the formula directly if it's not a typical structure.
(2) Transform from inside to outside: transform the multi-loop structure from the inner loop
to the outer loop, reduce the inner loop one by one, until transformed into an equivalent
square.
(3) Multiple input transformation: the system has multiple inputs, so each input must be
transformed one by one to obtain the respective transfer function.

2.5.4 Transfer Function in Control System with Disturbance

Typical block diagram in control system with disturbance
 Figure 2.49 Typical block diagram in control system with disturbance

𝑅(𝑠)——input signal

𝑁(𝑠)——disturbance signal

𝐶(𝑠)——output signal

𝐵(𝑠)——feedback signal

𝐸(𝑠)——error signal

(1) Open-loop transfer function

Figure 2.50 Open-loop transfer function

Break the main-feedback loop
𝐵(𝑠)
= 𝐺1 (𝑠)𝐺2 (𝑠)𝐻(𝑠) {2 − 150}
𝐸(𝑠)
𝐺1 (𝑠)𝐺2 (𝑠)𝐻(𝑠) is called the open-loop transfer function
(2) Close-loop transfer function without 𝑁(𝑠)

Figure 2.51 Close-loop transfer function without N(s)

When 𝑁(𝑠) = 0, close-loop function from 𝐶(𝑠) to 𝑅(𝑠)
𝐶(𝑠) 𝐺1 (𝑠)𝐺2 (𝑠)
𝜑𝐶.𝑅 (𝑠) = = {2 − 151}
𝑅(𝑠) 1 + 𝐺1 (𝑠)𝐺2 (𝑠)𝐻(𝑠)
 (3) Close-loop transfer function without 𝑅(𝑠)

Figure 2.52 Close-loop transfer function without R(s)

When 𝑅(𝑠) = 0, close-loop function from 𝑁(𝑠)
𝐶(𝑠) 1 𝐺1 (𝑠)𝐺2 (𝑠) 𝐺2 (𝑠)
𝜑𝐶.𝑁 (𝑠) = = = {2 − 152}
𝑁(𝑠) 𝐺1 (𝑠) 1 + 𝐺1 (𝑠)𝐺2 (𝑠)𝐻(𝑠) 1 + 𝐺1 (𝑠)𝐺2 (𝑠)𝐻(𝑠)

(4) Output of the disturbance system

Combining above two systems:
𝐶(𝑠) = 𝜑𝐶.𝑁 (𝑠)𝑁(𝑠) + 𝜑𝐶.𝑅 (𝑠)𝑅(𝑠)
𝐺2 (𝑠) 𝐺1 (𝑠)𝐺2 (𝑠)
= 𝑅(𝑠) + 𝑁(𝑠) {2 − 153}
1 + 𝐺1 (𝑠)𝐺2 (𝑠)𝐻(𝑠) 1 + 𝐺1 (𝑠)𝐺2 (𝑠)𝐻(𝑠)

Approximation of Disturbance System, if in the system, |G1(s)G2(s)H(s)|>>1

and|G1(s)H(s)|>>1 then, the System can be approximated to

𝐺2 (𝑠) 𝐺1 (𝑠)𝐺2 (𝑠)

𝐶(𝑠) ≈ 𝑅(𝑠) + 𝑁(𝑠)
1 + 𝐺1 (𝑠)𝐺2 (𝑠)𝐻(𝑠) 1 + 𝐺1 (𝑠)𝐺2 (𝑠)𝐻(𝑠)
1 1
= 𝑅(𝑠) + 0 × 𝑁(𝑠) = 𝑅(𝑠) {2 − 154}
𝐻(𝑠) 𝐻(𝑠)

𝐸(𝑠) = 𝑅(𝑠) − 𝐵(𝑠) = 𝑅(𝑠) − 𝐻(𝑠)𝐶(𝑠) = 0 {2 − 155}

Example 14:
𝐶(𝑠) 𝐶(𝑠) 𝐶(𝑠)
Calculate the following transfer function: 𝑅(𝑠) , 𝑁 ,
1 (𝑠) 𝑁2 (𝑠)

𝐶(𝑠)
(1)
𝑅(𝑠)
(2) 𝐶(𝑠)/𝑁2 (𝑠)

Figure 2.53 simplification

N2(s) signal can only one direction transform

Figure 2.54 simplification

𝐶(𝑠)
= −1 {2 − 156}
𝑁2 (𝑠)

2.6 Signal-flow Graph Models

➢ Block diagrams are adequate for the representation of the interrelationships of

controlled and input variables
➢ For a system with complex interrelationships, the block diagram reduction procedure
is cumbersome and often quite difficult to complete
➢ An alternative method, called the signal-flow graph method, provides the relation
between system variables without requiring any reduction procedure or manipulation
of the flow graph

The basic conception

Having a linear equation
𝑛

𝑥𝑖 = ∑ 𝑎𝑖𝑗 𝑥𝑗 𝑖 = 1,2, ⋯ , 𝑛 {2 − 157}

𝑗=1

Utilizing the Laplace transform, yielding

𝑛

𝑥𝑖 (𝑠) = ∑ 𝐺𝑖𝑗 (𝑠)𝑥𝑗 (𝑠) 𝑖 = 1,2, ⋯ , 𝑛 {2 − 158}

𝑗=1

The relationship between the variables is clearly portrayed by the signal-flow graph
 ➢ Node：the input and output points or junctions

➢ Branch: unidirectional path segment, which relates the dependency of an input and an
output variable in a manner equivalent to a block of a block diagram

➢ Gain：the cause-effect of branch or path or loop

Example:

Mixed node a53

a32
input node a43
(source) a44 1
a12 5
2 3
x4 x5 x6
1
x1 x2 a23 x3 a34 4 a45
a24
a25

Figure 2.55 Signal-flow Graph

Source node: all branches leaving

Terminal node：all branches entering

Mixed node：some branches leaving, some branches entering.

Summing point + Pickoff point

Path: a branch or a continuous sequence of branches that can be traversed from one node to
another node, with no node being met twice along the path
𝑥1 → 𝑥2 → 𝑥3 → 𝑥4 → 𝑥5 𝑥1 → 𝑥2 → 𝑥5

Forward path：a path that traverses from source node to terminal node

𝑥1 → 𝑥2 → 𝑥3 → 𝑥4 → 𝑥5 → 𝑥6
𝑥1 → 𝑥2 → 𝑥4 → 𝑥5 → 𝑥6
𝑥1 → 𝑥2 → 𝑥5 → 𝑥6
Loop: a closed path that originates and terminates on the same node, with no other nodes
being met twice along the path
Self-loop: a loop only with a branch
𝑥2 → 𝑥4 → 𝑥3 → 𝑥2
𝑥4 → 𝑥4
Non touching loops: loops do not have a common node
Touching loops: loops share one or more common nodes
𝑥2 → 𝑥3 → 𝑥2 𝑎𝑛𝑑 𝑥4 → 𝑥4
𝑥2 → 𝑥3 → 𝑥2 𝑎𝑛𝑑 𝑥3 → 𝑥4 → 𝑥3

Plotting signal flow graph

1. According to differential equation
For linear equations with differential or integral item, it is transformed to algebraic
equation by the Laplace transformation, and then plot the signal flow diagram
➢ Firstly, specifies a node for each variable of the system
➢ The variables are arranged from left to right in the order according to the causal
relationship
➢ Based on mathematical equations, each node is properly connected by using the
branch with branch gain

Example：a system equations is shown as follows

𝑎𝑥 − 𝑥1 + 𝑏𝑥2 = 0
{ 𝑟 {2 − 159}
𝑐𝑥𝑟 + 𝑑𝑥1 − 𝑥2 = 0

Try to plot the signal flow graph

Solve：equations are rewritten as

𝑥 = 𝑎𝑥𝑟 + 𝑏𝑥2
{ 1 {2 − 160}
𝑥2 = 𝑐𝑥𝑟 + 𝑑𝑥1

Firstly, plot nodes 𝑥1 , 𝑥2 , 𝑥3

Then, plot branches according to equations

Figure 2.56 Plotting signal flow graph

 𝑥 = 𝑎𝑥𝑟 + 𝑏𝑥2
{ 1 {2 − 161}
𝑥2 = 𝑐𝑥𝑟 + 𝑑𝑥1

2. According to system structure block

Nodes：mark signals with the small circle flag on the signal line

Branches：Mark a line segment with the transfer function block instead of the block of
configuration diagram

Draw the signal flow graph from the structure diagram should be as concise as possible. the
number of nodes and branches that can be reduced

Simplified rules of the signal flow graph

Mason’s Rule
From a complex system signal flow diagram, the transfer function of the system can be
obtained through simplification, and the equivalent transformation rules of the structure
diagram are also applicable to the simplification of the signal flow diagram, but this process is
still rather troublesome after all. The commonly used Mason's rule gain formula directly
calculates the transfer function from the source node to the sink node without simplifying the
signal flow diagram, which provides convenience for the wide application of signal flow
diagrams. Due to the corresponding relationship between the system structure diagram and
the signal flow diagram, the Mason gain formula can also be directly used in the system
structure diagram.
The source of Mason's gain formula is the result of skillfully linking the numerator
polynomial and denominator polynomial of the solution with the signal flow graph when
solving the linear equation system according to Gramer's rule.
Mason’s Rule is the most powerful method when transforming block diagram to
transform function.
 𝑛
𝐶(𝑠) 1
𝛷(𝑠) = = ∑ 𝑃𝑘 ∆𝑘 {2 − 162}
𝑅(𝑠) ∆
𝑘=1

K = number of forward paths

𝑇𝑘 = the kth forward-path gain

∆= 1 − ∑ 𝑙𝑜𝑜𝑝 𝑔𝑎𝑖𝑛𝑠 + ∑ 𝑛𝑜𝑛𝑡𝑜𝑢𝑐ℎ𝑖𝑛𝑔 − 𝑙𝑜𝑜𝑝 𝑔𝑎𝑖𝑛𝑠 𝑡𝑎𝑘𝑒𝑛 𝑡𝑤𝑜 𝑎𝑡 𝑎 𝑡𝑖𝑚𝑒

− ∑ 𝑛𝑜𝑛𝑡𝑜𝑢𝑐ℎ𝑖𝑛𝑔 − 𝑙𝑜𝑜𝑝 𝑔𝑎𝑖𝑛𝑠 𝑡𝑎𝑘𝑒𝑛 𝑡ℎ𝑟𝑒𝑒 𝑎𝑡 𝑎 𝑡𝑖𝑚𝑒

+ ∑ 𝑛𝑜𝑛𝑡𝑜𝑢𝑐ℎ𝑖𝑛𝑔 − 𝑙𝑜𝑜𝑝 𝑔𝑎𝑖𝑛𝑠 𝑡𝑎𝑘𝑒𝑛 𝑓𝑜𝑢𝑟 𝑎𝑡 𝑎 𝑡𝑖𝑚𝑒

∆𝑘 = ∆ − ∑ loop gain terms in ∆ that touch the 𝑘𝑡ℎ forward path. In other words, ∆𝑘 is
formed by eliminating from ∆ those loop gains that touch the kth forward path.

Mason’s Gain Formula Clarification

Forward path: The signal channel from the input to the output in one direction (along the
arrow) is called the forward path.
Closed-loop: A closed channel is a closed circuit in which a signal can flow in the direction of
an arrow and pass through any element no more than once in a structure diagram.
Non-touching loop: A circuit with no contact in common with each other.

Example: calculate the transfer function

Figure 2.57 Block Graph

(1) One forward path
𝑃1 = 𝐺1 𝐺2 𝐺3 𝐺4 𝐺5 𝐺6
(2) Single-close loop
𝐿1 = −𝐺1 𝐺2 𝐺3 𝐺4 𝐺5 𝐺6 𝐻1

𝐿2 = −𝐺2 𝐺3 𝐻2

𝐿3 = −𝐺4 𝐺5 𝐻3

𝐿4 = −𝐺3 𝐺4 𝐻4
 (3) Two loops are non-touching

∑ 𝐿𝑏 𝐿𝑐 = 𝐿2 𝐿3 = (−𝐺2 𝐺3 𝐻2 )(−𝐺4 𝐺5 𝐻3 ) = 𝐺2 𝐺3 𝐺4 𝐺5 𝐻2 𝐻3 {2 − 163}

(4) Characteristic equation

∆= 1 − ∑ 𝐿𝑎 + ∑ 𝐿𝑏 𝐿𝑐
= 1 + 𝐺1 𝐺2 𝐺3 𝐺4 𝐺5 𝐺6 𝐻1 + 𝐺2 𝐺3 𝐻2 + 𝐺4 𝐺5 𝐻3 + 𝐺3 𝐺4 𝐻4 + 𝐺2 𝐺3 𝐺4 𝐺5 𝐻2 𝐻3 {2 − 164}

(5) Cofactor of the path

∆1 = 1 {2 − 165}

𝐶(𝑠)
=
𝑅(𝑠)
𝐺1 𝐺2 𝐺3 𝐺4 𝐺5 𝐺6
{2 − 166}
1 + 𝐺1 𝐺2 𝐺3 𝐺4 𝐺5 𝐺6 𝐻1 + 𝐺2 𝐺3 𝐻2 + 𝐺4 𝐺5 𝐻3 + 𝐺3 𝐺4 𝐻4 + 𝐺2 𝐺3 𝐺4 𝐺5 𝐻2 𝐻3
Mason’s Gain Formula Notes:
k forward channels refer to the total number of forwarding loop from the input signal to the
output signal. The node can be passed only once and cannot be repeated.
Non-touching loops: Loops that are non-touching to each other,  and 𝑘 should be calculated
correctly.
The sign of the feedback loop should be considered.