System Modelling and
Closed Loop Control
Transfer Functions
A transfer function describes how a system responds to an input signal in the Laplace
domain. It gives the output 𝑌𝑌 𝑠𝑠 for a given input U 𝑠𝑠 in the Laplace domain. 𝐺𝐺 𝑠𝑠 is
the transfer function.
𝑌𝑌 𝑠𝑠
𝐺𝐺 𝑠𝑠 =
𝑈𝑈 𝑠𝑠
Model Laplace Model Temporal
Input ℒ −1
(t) (s) (s) Response
Example-1
Transfer Function: A relation between input and output of a linear system
Example: Input: force f(t), output: displacement x(t)
x(t)
𝑑𝑑 2 𝑥𝑥 𝑑𝑑𝑑𝑑
𝑓𝑓 𝑡𝑡 = 𝑚𝑚𝑥𝑥̈ + 𝑏𝑏𝑥𝑥̇ + 𝑘𝑘𝑥𝑥 𝑓𝑓 𝑡𝑡 = 𝑚𝑚 + 𝑏𝑏 + 𝑘𝑘𝑥𝑥
𝑑𝑑𝑑𝑑 𝑑𝑑𝑑𝑑
Laplace Transform: 𝐹𝐹(𝑠𝑠) = 𝑚𝑚𝑠𝑠 2 𝑋𝑋(𝑠𝑠) + 𝑏𝑏𝑏𝑏𝑏𝑏(𝑠𝑠) + 𝑘𝑘𝑘𝑘(𝑠𝑠)
Example-1
𝑋𝑋 𝑠𝑠
Transfer Function: 𝐺𝐺 𝑠𝑠 =
𝐹𝐹 𝑠𝑠
𝐹𝐹(𝑠𝑠) = 𝑚𝑚𝑠𝑠 2 𝑋𝑋(𝑠𝑠) + 𝑏𝑏𝑏𝑏𝑏𝑏(𝑠𝑠) + 𝐾𝐾𝑋𝑋(𝑠𝑠)
F(s) = 𝑚𝑚𝑠𝑠 2 + 𝑏𝑏𝑏𝑏 + 𝐾𝐾 X(s)
1
𝐺𝐺 𝑠𝑠 =
𝑚𝑚𝑠𝑠 2 + 𝑏𝑏𝑏𝑏 + 𝐾𝐾
Example-2
Example: Input: force f(t), output: velocity v(t)
v(t)
𝑑𝑑 2 𝑥𝑥 𝑑𝑑𝑑𝑑
𝑓𝑓 𝑡𝑡 = 𝑚𝑚 + 𝑏𝑏 + 𝑘𝑘𝑥𝑥
𝑑𝑑𝑑𝑑 𝑑𝑑𝑑𝑑
𝑑𝑑𝑣𝑣
𝑓𝑓 𝑡𝑡 = 𝑚𝑚 + 𝑏𝑏𝑣𝑣 + 𝑘𝑘 � 𝑣𝑣𝑣𝑣𝑣𝑣 Laplace Transform: 𝐹𝐹 𝑠𝑠 = 𝑚𝑚𝑚𝑚𝑚𝑚 𝑠𝑠 + 𝑏𝑏𝑏𝑏 𝑠𝑠 + 𝑘𝑘
1
𝑉𝑉(𝑠𝑠)
𝑑𝑑𝑑𝑑 𝑆𝑆
Example-2
𝑉𝑉 𝑠𝑠
Transfer Function: 𝐺𝐺 𝑠𝑠 =
𝐹𝐹 𝑠𝑠
1
𝐹𝐹(𝑠𝑠) = 𝑚𝑚𝑠𝑠𝑠𝑠 𝑠𝑠 + 𝑏𝑏𝑏𝑏 𝑠𝑠 + 𝑘𝑘 𝑉𝑉(𝑠𝑠)
𝑆𝑆
𝑠𝑠
𝐺𝐺 𝑠𝑠 =
𝑚𝑚𝑠𝑠 2 + 𝑏𝑏𝑏𝑏 + 𝑘𝑘
Example-3
Example: Input: Voltage v(t), output: voltage 𝑣𝑣𝐿𝐿 (t)
𝑑𝑑𝑖𝑖 𝑡𝑡 𝑑𝑑𝑖𝑖 𝑡𝑡
𝑣𝑣 𝑡𝑡 = 𝑅𝑅𝑅𝑅 𝑡𝑡 + 𝐿𝐿 𝑣𝑣𝐿𝐿 𝑡𝑡 = 𝐿𝐿
𝑑𝑑𝑑𝑑 𝑑𝑑𝑑𝑑
Laplace Transform: 𝑉𝑉 𝑠𝑠 = 𝑅𝑅𝑅𝑅 𝑠𝑠 + 𝐿𝐿𝐿𝐿𝐿𝐿 𝑠𝑠 𝑉𝑉𝐿𝐿 𝑠𝑠 = 𝐿𝐿𝐿𝐿𝐿𝐿 𝑠𝑠 => 𝐼𝐼 𝑠𝑠 =
𝑉𝑉𝐿𝐿 (𝑠𝑠)
𝑉𝑉 𝑠𝑠 = 𝐼𝐼 𝑠𝑠 [𝑅𝑅 + 𝐿𝐿𝐿𝐿] 𝐿𝐿𝐿𝐿
Example-3
𝑉𝑉𝐿𝐿 𝑠𝑠
Transfer Function: 𝐺𝐺 𝑠𝑠 =
𝑉𝑉 𝑠𝑠
𝑉𝑉𝐿𝐿 (𝑠𝑠)
𝑉𝑉 𝑠𝑠 = 𝐼𝐼 𝑠𝑠 𝑅𝑅 + 𝐿𝐿𝐿𝐿 ; 𝐼𝐼 𝑠𝑠 =
𝐿𝐿𝐿𝐿
𝑉𝑉𝐿𝐿 (𝑠𝑠) 𝑉𝑉𝐿𝐿 𝑠𝑠 𝐿𝐿𝐿𝐿
𝑉𝑉 𝑠𝑠 = 𝑅𝑅 + 𝐿𝐿𝐿𝐿 ⇒ =
𝐿𝐿𝐿𝐿 𝑉𝑉 𝑠𝑠 𝑅𝑅+𝐿𝐿𝐿𝐿
Transfer Function Poles and Zeros
Transfer function is a rational function in the complex variable s = 𝜎𝜎 + 𝑗𝑗𝜔𝜔
𝑋𝑋(𝑠𝑠) 𝑏𝑏𝑚𝑚 𝑠𝑠 𝑚𝑚 + 𝑏𝑏𝑚𝑚−1 𝑠𝑠 𝑚𝑚−1 + ⋯ + 𝑏𝑏0
𝐺𝐺 𝑠𝑠 = =
𝑌𝑌(𝑠𝑠) 𝑎𝑎𝑛𝑛 𝑠𝑠 𝑛𝑛 + 𝑎𝑎𝑛𝑛−1 𝑠𝑠 𝑛𝑛−1 + ⋯ + 𝑎𝑎0
The zeros 𝑧𝑧𝑖𝑖 are the roots of 𝑋𝑋 𝑠𝑠 = 0
Thus: lim 𝑋𝑋 𝑠𝑠 = 0
𝑠𝑠→𝑧𝑧𝑖𝑖
The poles 𝑝𝑝𝑖𝑖 are the roots of Y 𝑠𝑠 = 0
Thus: lim 𝑋𝑋 𝑠𝑠 = ∞
𝑠𝑠→𝑝𝑝𝑖𝑖
Poles and Zeros
Consider the following equation:
𝑠𝑠 𝑠𝑠 + 5
𝐹𝐹 𝑠𝑠 = 2
𝑠𝑠 + 4𝑠𝑠 + 8
Zeros: -5, 0
Poles: -2+2j, -2-2j
Order of the Transfer Function
First order transfer function:
Order is obtained from the highest derivative in the denominator (power of s)
Standard form:
𝐺𝐺
𝐺𝐺 𝑠𝑠 =
𝜏𝜏𝑠𝑠 + 1
Example:
1
𝐼𝐼 𝑠𝑠 1 𝑅𝑅 𝐿𝐿
𝐺𝐺 𝑠𝑠 = = = ⇒ 𝜏𝜏 =
𝑉𝑉 𝑠𝑠 𝐿𝐿𝑠𝑠 + 𝑅𝑅 𝐿𝐿𝐿𝐿 𝑅𝑅
+1
𝑟𝑟
Order of the Transfer Function
Second order transfer function
Standard form:
𝑤𝑤𝑛𝑛2
𝐺𝐺 𝑠𝑠 = 2
𝑠𝑠 + 2𝜉𝜉𝑤𝑤𝑛𝑛 𝑠𝑠 + 𝑤𝑤𝑛𝑛2
where 𝒘𝒘𝒏𝒏 is the natural frequency, and 𝝃𝝃 is the damping ratio.
Example:
1 𝑘𝑘
1 𝑚𝑚 1 𝑚𝑚
𝐺𝐺 𝑠𝑠 = = =
𝑚𝑚𝑠𝑠 2 + 𝑏𝑏𝑏𝑏 + 𝑘𝑘 𝑠𝑠 2 + 𝑏𝑏 𝑠𝑠 + 𝑘𝑘 𝑘𝑘 𝑠𝑠 2 + 𝑏𝑏 𝑠𝑠 + 𝑘𝑘
𝑚𝑚 𝑚𝑚 𝑚𝑚 𝑚𝑚
𝑘𝑘 𝑘𝑘
𝑤𝑤𝑛𝑛2 = ⇒ 𝑤𝑤𝑛𝑛 = ∶ 𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁𝑁 𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 (𝑟𝑟𝑟𝑟𝑟𝑟/𝑠𝑠)
𝑚𝑚 𝑚𝑚
𝑏𝑏 𝑘𝑘 𝑏𝑏 𝑏𝑏
2𝜉𝜉𝑤𝑤𝑛𝑛 = ⇒ 2𝜉𝜉 = ⇒ 𝜉𝜉 = ∶ 𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷𝐷 𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟𝑟
𝑚𝑚 𝑚𝑚 𝑚𝑚 2 𝑚𝑚𝑚𝑚
Temporal Response
Step-1: Define the input signal in the Laplace domain
Temporal Response
Step-1: Define the input signal in the Laplace domain
Temporal Response
Step-2: In order to find the temporal response, apply a force signal to the transfer
function.
𝑋𝑋 𝑠𝑠 1
𝐺𝐺 𝑠𝑠 = =
𝐹𝐹 𝑆𝑆 𝑚𝑚𝑠𝑠 2 + 𝑏𝑏𝑏𝑏 + 𝑘𝑘
For an impulse input f t = 𝛿𝛿 𝑡𝑡 , 𝐹𝐹 𝑠𝑠 = 1 and the temporal response is
1
𝑋𝑋 𝑠𝑠 =
𝑚𝑚𝑠𝑠 2 + 𝑏𝑏𝑏𝑏 + 𝑘𝑘
For a step-type input f t = 1𝑁𝑁, 𝐹𝐹 𝑠𝑠 = 1/𝑠𝑠 and the temporal response is
1 1
𝑋𝑋 𝑠𝑠 = 2
( )
𝑚𝑚𝑠𝑠 + 𝑏𝑏𝑏𝑏 + 𝑘𝑘 𝑠𝑠
Temporal Response
For a sinusoidal input f t = 5 sin 𝑡𝑡 𝑁𝑁 and the temporal response is
1
Laplace transform of sin 𝑡𝑡 =
𝑠𝑠 2 +1
1 5
𝑋𝑋 𝑠𝑠 = 2
( 2 )
𝑚𝑚𝑠𝑠 + 𝑏𝑏𝑏𝑏 + 𝑘𝑘 𝑠𝑠 + 1
Temporal Response
Step-3: Calculate the inverse transform of the resulting function
For a step-type input f t = 1/𝑠𝑠, 𝐹𝐹 𝑠𝑠 = 1 and the temporal response is
−1 1 1
𝑥𝑥 𝑡𝑡 = ℒ { 2 }
𝑚𝑚𝑠𝑠 + 𝑏𝑏𝑏𝑏 + 𝑘𝑘 𝑠𝑠
Exercise
𝑉𝑉 𝑠𝑠
Find the transfer function G 𝑠𝑠 = between the force r(t) and the velocity of mass
𝑅𝑅 𝑠𝑠
m1.
Find the differential equation the velocity of each mass
Calculate the Laplace transform
Calculate the transfer function
Pole Locations (First Order System)
First Order Systems
Consider the given RL circuit
1
𝐼𝐼 𝑠𝑠 1 𝑅𝑅 𝐿𝐿
𝐺𝐺 𝑠𝑠 = = = ⇒ 𝜏𝜏 = 𝑉𝑉(𝑠𝑠) = (𝑅𝑅 + 𝐿𝐿𝐿𝐿)𝐼𝐼(𝑠𝑠)
𝑉𝑉 𝑠𝑠 𝐿𝐿𝑠𝑠 + 𝑅𝑅 𝐿𝐿𝐿𝐿 𝑅𝑅
+1
𝑟𝑟
Pole Locations (First Order System)
Impulse response v t = 𝛿𝛿 𝑡𝑡 , 𝑉𝑉 𝑠𝑠 = 1
1 1 1 1
𝐼𝐼 𝑠𝑠 = = ( )
𝑅𝑅 𝜏𝜏𝑠𝑠 + 1 𝜏𝜏𝑅𝑅 𝑠𝑠 + 1
𝜏𝜏
The pole is s=−1/𝜏𝜏
1 −𝑡𝑡
The time response: 𝑖𝑖 𝑡𝑡 = 𝑒𝑒 𝜏𝜏
𝜏𝜏𝑅𝑅
If 𝜏𝜏 > 0, the pole is negative, on the left-half of s-plane
Pole Locations (First Order System)
The pole is s=−1/𝜏𝜏
1 −𝑡𝑡
The time response: 𝑖𝑖 𝑡𝑡 = 𝑒𝑒 𝜏𝜏
𝜏𝜏𝜏𝜏
If 𝜏𝜏 < 0, the pole is positive, on the left-half of s-plane
>0
1 −𝑡𝑡
𝑖𝑖 𝑡𝑡 = 𝑒𝑒 𝜏𝜏
𝜏𝜏𝜏𝜏
Pole Locations (Second Order System)
Second Order Systems
𝑤𝑤𝑛𝑛2
𝐺𝐺 𝑠𝑠 = 2
𝑠𝑠 + 2𝜁𝜁𝑤𝑤𝑛𝑛 𝑠𝑠 + 𝑤𝑤𝑛𝑛2
1
𝑠𝑠
x(t)
𝑘𝑘
1 𝑚𝑚
𝐺𝐺 𝑠𝑠 =
𝑘𝑘 𝑠𝑠 2 + 𝑏𝑏 𝑠𝑠+ 𝑘𝑘
𝑚𝑚 𝑚𝑚
1 1 𝑤𝑤𝑛𝑛2
𝑋𝑋 𝑠𝑠 = 𝐺𝐺 𝑠𝑠 𝐹𝐹 𝑠𝑠 𝑋𝑋 𝑠𝑠 = ( ) ( 2 2 )
𝑠𝑠 𝑘𝑘 𝑠𝑠 + 2𝜁𝜁 𝑤𝑤𝑛𝑛 𝑠𝑠 + 𝑤𝑤𝑛𝑛
Pole Locations (Second Order System)
The poles of the transfer function are:
−2𝜁𝜁 ± 2𝜁𝜁𝑤𝑤𝑛𝑛 2 − 4𝑤𝑤𝑛𝑛2
𝑠𝑠 = = −𝜁𝜁𝑤𝑤𝑛𝑛 ± 𝑤𝑤𝑛𝑛 𝜁𝜁 2 − 1
2
𝑠𝑠1 = 𝑤𝑤𝑛𝑛 (−𝜁𝜁 + 𝜁𝜁 2 − 1) 𝑠𝑠2 = 𝑤𝑤𝑛𝑛 (−𝜁𝜁 − 𝜁𝜁 2 − 1)
𝜁𝜁>0 𝑠𝑠1 and 𝑠𝑠2 are real number
𝜁𝜁 <0 𝑠𝑠1 and 𝑠𝑠2 have imaginary part
Pole Locations (Second Order System)
1 1 𝑤𝑤𝑛𝑛2
𝑋𝑋 𝑠𝑠 = ( )( 2 )
𝑘𝑘 𝑠𝑠 𝑠𝑠 + 2𝜁𝜁𝑤𝑤𝑛𝑛 𝑠𝑠 + 𝑤𝑤𝑛𝑛2
Case: 𝜉𝜉 ≥ 1
𝑠𝑠1 = 𝑤𝑤𝑛𝑛 (−𝜁𝜁 + 𝜁𝜁 2 − 1) 𝑠𝑠2 = 𝑤𝑤𝑛𝑛 −𝜁𝜁 − 𝜁𝜁 2 − 1
Roots are negative real numbers:
1 𝑘𝑘1 𝑘𝑘2 𝑘𝑘3
𝑋𝑋 𝑠𝑠 = ( + + )
𝑚𝑚 𝑠𝑠 𝑠𝑠 + 𝑎𝑎1 𝑠𝑠 + 𝑎𝑎2
Pole Locations (Second Order System)
Example: m = 1 kg, k = 1 N/m
b = 3 Ns/m, ζ = 1.5; b = 2.1 Ns/m, ζ = 1.05; b = 2 Ns/m, ζ = 1
Overdamped System Critically Damped System
Pole Locations (Second Order System)
Case: 0 < 𝜁𝜁 < 1
𝑠𝑠1 = 𝑤𝑤𝑛𝑛 (−𝜁𝜁 + 𝜁𝜁 2 − 1) 𝑠𝑠2 = 𝑤𝑤𝑛𝑛 −𝜁𝜁 − 𝜁𝜁 2 − 1
Roots are complex conjugate numbers with a negative real part. Thus:
1 1
𝑥𝑥 𝑡𝑡 = [1 − 𝑒𝑒 −𝜁𝜁𝑤𝑤𝑛𝑛 𝑡𝑡 sin( 𝑤𝑤𝑛𝑛 1 − 𝜁𝜁 2 𝑡𝑡 + cos −1 (𝜁𝜁))]
𝑘𝑘 1 − 𝜁𝜁 2
Pole Locations (Second Order System)
Underdamped System
b = 1.5, ζ = 0.75 b = 1, ζ = 0.5; b = 0.5, ζ = 0.25, b = 0.1, ζ = 0.05
Block Diagram Models
Represents the relationship of a system variables graphically
Example: The relation between the input voltage and the position of a DC motor
Basic Building Elements
Block diagram elements illustration based on MATLAB Simulink
Transfer function XG(s)=Y
Gain y 𝛼𝛼X=y
Sum Z Z
Z=X+Y - Z=X-Y
Block Diagram of a DC motor
Electric circuit characteristics
𝑉𝑉 𝑠𝑠 = 𝐼𝐼 𝑠𝑠 𝑅𝑅 + 𝐿𝐿𝐿𝐿 + 𝑉𝑉𝑏𝑏 (𝑠𝑠)
𝑉𝑉𝑏𝑏
𝑉𝑉 𝑠𝑠 − 𝑉𝑉𝑏𝑏 𝑠𝑠
𝐼𝐼 𝑠𝑠 =
𝐿𝐿𝐿𝐿 + 𝑅𝑅
Back electromagnetic force (EMF) voltage
𝑉𝑉𝑏𝑏 = 𝐾𝐾𝑣𝑣 𝑤𝑤(𝑠𝑠)
Block Diagram of a DC motor
Mechanical characteristics 𝑇𝑇(𝑠𝑠)
� 𝑇𝑇 𝑠𝑠 = 𝐽𝐽𝐽𝐽 𝑠𝑠 𝑠𝑠
𝑇𝑇 𝑠𝑠 − 𝑏𝑏𝑏𝑏 𝑠𝑠 − 𝑇𝑇𝑑𝑑 𝑠𝑠 = 𝐽𝐽𝐽𝐽 𝑠𝑠 𝑠𝑠
𝑇𝑇 𝑠𝑠 − 𝑇𝑇𝑑𝑑 𝑠𝑠 = 𝑤𝑤 𝑠𝑠 [𝐽𝐽𝐽𝐽 + 𝑏𝑏]
𝑇𝑇 𝑠𝑠 −𝑇𝑇𝑑𝑑 𝑠𝑠
w 𝑠𝑠 =
𝐽𝐽𝐽𝐽+𝑏𝑏
𝑇𝑇𝑑𝑑 (𝑠𝑠)
Torque Constant
𝑇𝑇 𝑠𝑠 = 𝐾𝐾𝑡𝑡 𝐼𝐼(𝑠𝑠)
Torque constant
Block Diagram of a DC motor
Four expressions are generated for DC motor:
𝑉𝑉 𝑠𝑠 −𝑉𝑉𝑏𝑏 𝑠𝑠
𝐼𝐼 𝑠𝑠 = 𝑉𝑉𝑏𝑏 = 𝐾𝐾𝑣𝑣 𝑤𝑤 𝑠𝑠
𝐿𝐿𝐿𝐿+𝑅𝑅
𝑇𝑇 𝑠𝑠 −𝑇𝑇𝑑𝑑 𝑠𝑠
w 𝑠𝑠 = 𝑇𝑇 𝑠𝑠 = 𝐾𝐾𝑡𝑡 𝐼𝐼(𝑠𝑠)
𝐽𝐽𝐽𝐽+𝑏𝑏
𝑇𝑇𝑑𝑑 𝑠𝑠
𝑉𝑉 𝑠𝑠 𝐼𝐼 𝑠𝑠 𝑇𝑇 𝑠𝑠 1 w 𝑠𝑠
1 -
- K t 𝑠𝑠
𝐿𝐿𝐿𝐿 + 𝑅𝑅 𝐽𝐽𝐽𝐽 + 𝑅𝑅
𝑉𝑉𝑏𝑏 𝑠𝑠
K v 𝑠𝑠
Open loop
Open-loop control
In the absence of a controller and without feedback, the disturbance T(s) directly
influences the output Y (s).
An open-loop system operates without feedback and directly generates the output in
response to an input signal.
Closed-loop Control
Closed-loop control
A closed-loop system compares the output Y (s) with a desired value R(s). The
error signal E(s) is used by the controller to adjust the actuator.
E(s)=R(s)-Y(s)
Open-loop vs Closed-loop Controller
While an open-loop system requires precise knowledge of the process being
controlled, a closed-loop system can automatically compensate for changes
in the process during operation — any resulting error is detected and corrected
by the controller.
Improve noise attenuation.
Closed-loop control allows to reduce steady-state error and control of the
transient response.
Transient and steady-state response
Steady-state response:
𝑒𝑒𝑠𝑠𝑠𝑠
𝑒𝑒𝑠𝑠𝑠𝑠 = lim 𝑠𝑠𝑠𝑠(𝑠𝑠)
𝑠𝑠→0
Steady-state response: Open loop vs Closed loop
Open-loop: The error is 𝐸𝐸 𝑠𝑠 = 𝑅𝑅 𝑠𝑠 – 𝑌𝑌 𝑠𝑠
𝑌𝑌 𝑠𝑠 𝐾𝐾𝐾𝐾𝑎𝑎
=
𝑅𝑅(𝑠𝑠) 𝜏𝜏𝜏𝜏+1 𝜏𝜏1 𝑠𝑠+1
𝐾𝐾𝐾𝐾𝑎𝑎
𝐸𝐸 𝑠𝑠 = 𝑅𝑅 𝑠𝑠 − 𝑅𝑅(𝑠𝑠)
𝜏𝜏𝜏𝜏 + 1 𝜏𝜏1 𝑠𝑠 + 1
𝐾𝐾𝐾𝐾𝑎𝑎 1
𝐸𝐸 𝑠𝑠 = 𝑅𝑅 𝑠𝑠 (1 − ) 𝑅𝑅 𝑠𝑠 =
𝜏𝜏𝜏𝜏+1 𝜏𝜏1 𝑠𝑠+1 𝑠𝑠
1 𝐾𝐾𝐾𝐾𝑎𝑎 1
𝑒𝑒𝑠𝑠𝑠𝑠 = lim 𝑠𝑠𝑠𝑠(𝑠𝑠) = lim 𝑠𝑠 1− = 1 − 𝐾𝐾𝐾𝐾𝑎𝑎 𝐾𝐾 =
𝑠𝑠→0 𝑠𝑠→0 𝑠𝑠 𝜏𝜏𝜏𝜏+1 𝜏𝜏1 𝑠𝑠+1 𝐾𝐾𝑎𝑎
Steady-state response: Open loop vs Closed loop
Open-loop: The error is 𝐸𝐸 𝑠𝑠 = 𝑅𝑅 𝑠𝑠 – 𝑌𝑌 𝑠𝑠
𝑌𝑌 𝑠𝑠 𝐾𝐾𝐾𝐾𝑎𝑎
=
𝑅𝑅(𝑠𝑠) 𝜏𝜏𝜏𝜏+1 𝜏𝜏1 𝑠𝑠+1 +𝐾𝐾𝐾𝐾𝑎𝑎
𝐾𝐾𝐾𝐾𝑎𝑎 1
𝐸𝐸 𝑠𝑠 = 𝑅𝑅 𝑠𝑠 − 𝑅𝑅 𝑠𝑠 𝑅𝑅 𝑠𝑠 =
𝜏𝜏𝜏𝜏+1 𝜏𝜏1 𝑠𝑠+1 +𝐾𝐾𝐾𝐾𝑎𝑎 𝑠𝑠
𝐾𝐾𝐾𝐾𝑎𝑎
𝐸𝐸 𝑠𝑠 = 𝑅𝑅 𝑠𝑠 (1 − )
𝜏𝜏𝜏𝜏 + 1 𝜏𝜏1 𝑠𝑠 + 1 + 𝐾𝐾𝐾𝐾𝑎𝑎
1 𝐾𝐾𝐾𝐾𝑎𝑎 𝐾𝐾𝐾𝐾𝑎𝑎 1
𝑒𝑒𝑠𝑠𝑠𝑠 = lim 𝑠𝑠𝑠𝑠(𝑠𝑠) = lim 𝑠𝑠 1− =1− =
𝑠𝑠→0 𝑠𝑠→0 𝑠𝑠 𝜏𝜏𝜏𝜏+1 𝜏𝜏1 𝑠𝑠+1 +𝐾𝐾𝐾𝐾𝑎𝑎 1+𝐾𝐾𝐾𝐾𝑎𝑎 1+𝐾𝐾𝐾𝐾𝑎𝑎
Transient Response
The transient response of a system is the temporary part of its output that occurs
immediately after an input or disturbance is applied, before the system reaches
steady state.
Transient Response
Second-order System
Consider the following second order control system:
𝑌𝑌 𝑠𝑠 𝑘𝑘
Transfer function: =
𝑅𝑅(𝑠𝑠) 𝑠𝑠 2 +𝑠𝑠𝑠𝑠+𝑘𝑘
We can write the above function in the standard formulation:
𝑌𝑌 𝑠𝑠 𝑤𝑤𝑛𝑛2
= 2
𝑅𝑅(𝑠𝑠) 𝑠𝑠 + 2𝜁𝜁𝑤𝑤𝑛𝑛 𝑠𝑠 + 𝑤𝑤𝑛𝑛2
Transient Response
𝑌𝑌 𝑠𝑠 𝑘𝑘 𝑌𝑌 𝑠𝑠 2
𝑤𝑤𝑛𝑛
= = 2
𝑅𝑅(𝑠𝑠) 𝑠𝑠 2 +𝑠𝑠𝑠𝑠+𝑘𝑘 𝑅𝑅(𝑠𝑠) 𝑠𝑠 2 +2𝜁𝜁𝑤𝑤𝑛𝑛 𝑠𝑠+𝑤𝑤𝑛𝑛
𝑤𝑤𝑛𝑛2 = 𝑘𝑘 ⇒ 𝑤𝑤𝑛𝑛 = 𝑘𝑘
𝑎𝑎
2𝜁𝜁𝑤𝑤𝑛𝑛 = 𝑎𝑎 ⇒ 𝜁𝜁 =
2 𝑘𝑘
Transient Response
𝑌𝑌 𝑠𝑠 𝑤𝑤𝑛𝑛2
=
𝑅𝑅(𝑠𝑠) 𝑠𝑠 2 + 2𝜁𝜁𝑤𝑤𝑛𝑛 𝑠𝑠 + 𝑤𝑤𝑛𝑛2
The natural frequency 𝑤𝑤𝑛𝑛 depends on 𝑘𝑘
The damping ratio 𝜁𝜁 depends on k
The response for an unit step input when 0 < 𝜉𝜉 < 1
1 −𝜁𝜁𝑤𝑤𝑛𝑛 𝑡𝑡 sin( 𝑤𝑤 1 − 𝜁𝜁 2 𝑡𝑡 + cos −1 (𝜁𝜁))
y 𝑡𝑡 = 1 − 2
𝑒𝑒 𝑛𝑛
1−𝜁𝜁
Transient Response
𝑌𝑌 𝑠𝑠 𝑤𝑤𝑛𝑛2
= 2
𝑅𝑅(𝑠𝑠) 𝑠𝑠 + 2𝜁𝜁𝑤𝑤𝑛𝑛 𝑠𝑠 + 𝑤𝑤𝑛𝑛2
1 −𝜁𝜁𝑤𝑤𝑛𝑛 𝑡𝑡 sin( 𝑤𝑤 1 − 𝜁𝜁 2 𝑡𝑡 + cos −1 (𝜁𝜁))
y 𝑡𝑡 = 1 − 2
𝑒𝑒 𝑛𝑛
1−𝜁𝜁
Measures of Performance
Rise time 𝑇𝑇𝑟𝑟 , peak time 𝑇𝑇𝑝𝑝 , and peak value 𝑀𝑀𝑝𝑝𝑝𝑝
Settling time 𝑇𝑇𝑠𝑠 : 𝑦𝑦(𝑡𝑡) within 2% of its final value
Percent overshoot P.O.
𝑇𝑇𝑟𝑟 and 𝑇𝑇𝑝𝑝 characterize the swiftness of the response
P.O. and 𝑇𝑇𝑠𝑠 characterize the closeness of the response to the input
Measures of Performance
Rise time 𝑇𝑇𝑟𝑟 , peak time 𝑇𝑇𝑝𝑝 , and peak value 𝑀𝑀𝑝𝑝𝑝𝑝
Settling time 𝑇𝑇𝑠𝑠 : 𝑦𝑦(𝑡𝑡) within 2% of its final value
Percent overshoot P.O.
𝑇𝑇𝑟𝑟 and 𝑇𝑇𝑝𝑝 characterize the swiftness of the response
P.O. and 𝑇𝑇𝑠𝑠 characterize the closeness of the response to the input
Measures of Performance
Rise time 𝑇𝑇𝑟𝑟 , peak time 𝑇𝑇𝑝𝑝 , and peak value 𝑀𝑀𝑝𝑝𝑝𝑝
Settling time 𝑇𝑇𝑠𝑠 : 𝑦𝑦(𝑡𝑡) within 2% of its final value
Percent overshoot P.O.
𝑇𝑇𝑟𝑟 and 𝑇𝑇𝑝𝑝 characterize the swiftness of the response
P.O. and 𝑇𝑇𝑠𝑠 characterize the closeness of the response to the input
Measures of Performance
Rise time 𝑇𝑇𝑟𝑟 , peak time 𝑇𝑇𝑝𝑝 , and peak value 𝑀𝑀𝑝𝑝𝑝𝑝
Settling time 𝑇𝑇𝑠𝑠 : 𝑦𝑦(𝑡𝑡) within 2% of its final value
Percent overshoot P.O.
𝑇𝑇𝑟𝑟 and 𝑇𝑇𝑝𝑝 characterize the swiftness of the response
P.O. and 𝑇𝑇𝑠𝑠 characterize the closeness of the response to the input
Overshoot
For an unit step input, the percent overshoot is
𝑀𝑀𝑝𝑝𝑝𝑝 −𝑓𝑓𝑓𝑓
P.O. = 𝑥𝑥𝑥𝑥𝑥
𝑓𝑓𝑓𝑓
1 −𝜁𝜁𝑤𝑤𝑛𝑛 𝑡𝑡 sin( 𝑤𝑤 1 − 𝜁𝜁 2 𝑡𝑡 + cos −1 (𝜁𝜁))
Differentiating the equation y 𝑡𝑡 = 1 − 2
𝑒𝑒 𝑛𝑛
1−𝜁𝜁
and setting it to zero yields the peak time as.
𝜋𝜋
𝑇𝑇𝑝𝑝 =
𝑤𝑤𝑛𝑛 1−𝜁𝜁 2
Replacing 𝑇𝑇𝑝𝑝 in the equation of y 𝑡𝑡 gives the peak response:
−𝜁𝜁𝜋𝜋/ 1−𝜁𝜁 2
𝑀𝑀𝑝𝑝𝑝𝑝 = 1 + 𝑒𝑒
The percentage overshoot (𝑓𝑓𝑓𝑓 = 𝑟𝑟(𝑡𝑡) = 1)
−𝜁𝜁𝜁𝜁/ 1−𝜁𝜁2
1+𝑒𝑒 −1 −𝜁𝜁𝜁𝜁/ 1−𝜁𝜁 2
P.O. = 𝑥𝑥𝑥𝑥𝑥 ⇒ 100𝑒𝑒
1
Settling time
For a unit step input and 𝟎𝟎 < 𝝃𝝃 < 𝟏𝟏 :
Exponential component
1 −𝜁𝜁𝑤𝑤𝑛𝑛 𝑡𝑡 2 𝑡𝑡 + cos −1 (𝜁𝜁))
y 𝑡𝑡 = 1 − 2
𝑒𝑒 sin( 𝑤𝑤𝑛𝑛 1 − 𝜁𝜁
1−𝜁𝜁
Final value Scaling factor Sinusoidal component
1 −𝜁𝜁𝑤𝑤𝑛𝑛 𝑡𝑡 sin( 𝑤𝑤 2 𝑡𝑡 + cos −1 (𝜁𝜁))
y 𝑡𝑡 = 1 − 2
𝑒𝑒 𝑛𝑛 1 − 𝜁𝜁
1−𝜁𝜁
=0: Steady state =0.02x1: settling time range
Settling time
For a unit step input and 𝟎𝟎 < 𝝃𝝃 < 𝟏𝟏 :
When 𝑡𝑡 = 𝑇𝑇𝑠𝑠 , the response is within 2% of its final value, thus:
ln 𝑒𝑒 −𝜁𝜁𝑤𝑤𝑛𝑛 𝑇𝑇𝑠𝑠 < ln 0.02
or
𝜁𝜁𝑤𝑤𝑛𝑛 𝑇𝑇𝑠𝑠 ≈ 4
thus
4
𝑇𝑇𝑠𝑠 = = 4𝜏𝜏
𝜁𝜁𝑤𝑤𝑛𝑛
The settling time is equal to 4 times the time constant.
PID Controllers
𝑘𝑘𝑝𝑝
1
𝑘𝑘𝑖𝑖
𝑠𝑠
𝑘𝑘𝑑𝑑 𝑠𝑠
The command signal is a function of
• The magnitude of the current error: proportional gain
• The integral of the error over time: integral gain
• The time rate change of the error: derivative gain
PID Controller
𝑘𝑘𝑝𝑝
𝐸𝐸(𝑠𝑠) 1 𝑈𝑈(𝑠𝑠)
𝑘𝑘𝑖𝑖
𝑠𝑠
𝑘𝑘𝑑𝑑 𝑠𝑠
𝑡𝑡
𝑢𝑢 𝑡𝑡 = 𝑘𝑘𝑝𝑝 𝑒𝑒 𝑡𝑡 + 𝑘𝑘𝑖𝑖 � 𝑒𝑒 𝑡𝑡 𝑑𝑑𝑑𝑑 + 𝑘𝑘𝑑𝑑 𝑠𝑠𝑠𝑠 𝑡𝑡
0
𝑘𝑘𝑖𝑖 1
𝑈𝑈 𝑠𝑠 = 𝑘𝑘𝑝𝑝 𝐸𝐸 𝑠𝑠 + 𝐸𝐸 𝑠𝑠 + 𝑘𝑘𝑑𝑑 𝑠𝑠𝑠𝑠(𝑠𝑠)
𝑠𝑠
𝑘𝑘𝑖𝑖 1
𝑈𝑈 𝑠𝑠 = 𝐸𝐸 𝑠𝑠 (𝑘𝑘𝑝𝑝 + + 𝑘𝑘𝑑𝑑 𝑠𝑠)
𝑠𝑠
Effect of 𝑘𝑘𝑝𝑝
Transfer function is
𝑘𝑘𝑝𝑝
𝑇𝑇 𝑠𝑠 = 𝑌𝑌 𝑠𝑠 = 𝑇𝑇 𝑠𝑠 𝑅𝑅(𝑠𝑠)
𝑠𝑠+𝑘𝑘𝑝𝑝 +2
The time constant is 1/(𝑘𝑘𝑝𝑝 + 2)
The steady-state error for 𝑥𝑥(𝑡𝑡) = 1 is
𝑘𝑘𝑝𝑝 1 𝑘𝑘𝑝𝑝
𝑒𝑒 ∞ = 𝑢𝑢 𝑡𝑡 − lim 𝑠𝑠𝑌𝑌 𝑠𝑠 = 1 − lim 𝑠𝑠 =1−
𝑠𝑠→0 𝑠𝑠→0 𝑠𝑠 + 𝑘𝑘𝑝𝑝 + 2 𝑠𝑠 𝑘𝑘𝑝𝑝 + 2
Effect of the integrator on the steady-state error
Transfer function is
𝑘𝑘𝑖𝑖 𝑘𝑘𝑖𝑖
𝑇𝑇 𝑠𝑠 = =
𝑠𝑠 𝑠𝑠+2 +𝑘𝑘𝑖𝑖 𝑠𝑠 2 +2𝑠𝑠+𝑘𝑘𝑖𝑖
Steady-state error
𝑘𝑘𝑖𝑖 1
𝑒𝑒 ∞ = 𝑢𝑢 𝑡𝑡 − lim 𝑠𝑠𝑠𝑠 𝑠𝑠 = 1 − lim 𝑠𝑠 =0
𝑠𝑠→0 𝑠𝑠→0 𝑠𝑠 𝑠𝑠 + 2 +𝑘𝑘𝑖𝑖 𝑠𝑠
Effect of the integrator and 𝑘𝑘𝑖𝑖 on the transient response
Transfer function:
𝑘𝑘𝑖𝑖 𝑘𝑘𝑖𝑖
𝑇𝑇 𝑠𝑠 = = 2
𝑠𝑠 𝑠𝑠 + 2 + 𝑘𝑘𝑖𝑖 𝑠𝑠 + 2𝑠𝑠 + 𝑘𝑘𝑖𝑖
where 𝑤𝑤𝑛𝑛 = 𝑘𝑘𝑖𝑖 and 𝜁𝜁 = 1/ 𝑘𝑘𝑖𝑖
The damping ration increases with 𝑘𝑘𝑖𝑖
− 𝜁𝜁𝜁𝜁/ 1−𝜁𝜁 2
Overshoot = 100𝑒𝑒 = 100𝑒𝑒 −𝜋𝜋/ 𝑘𝑘𝑖𝑖 −1
The overshot increases with 𝑘𝑘𝑖𝑖
Effect of the derivative gain 𝑘𝑘𝑑𝑑
Transfer function is
𝑠𝑠𝑘𝑘𝑑𝑑 + 𝑘𝑘𝑝𝑝
𝑇𝑇 𝑠𝑠 = 2
𝑠𝑠 + 2 + 𝑘𝑘𝑑𝑑 𝑠𝑠 + 1 + 𝑘𝑘𝑝𝑝
Steady-state error, 𝑘𝑘𝑑𝑑 has no affect on steady-state error.
𝑠𝑠𝑠𝑠𝑑𝑑 + 𝑘𝑘𝑝𝑝 1 𝑘𝑘𝑝𝑝
𝑒𝑒 ∞ = 𝑢𝑢 𝑡𝑡 − lim 𝑠𝑠𝑠𝑠 𝑠𝑠 = 1 − lim 𝑠𝑠 2 =
𝑠𝑠→0 𝑠𝑠→0 𝑠𝑠 + 2 + 𝑘𝑘𝑑𝑑 𝑠𝑠 + 1 + 𝑘𝑘𝑝𝑝 𝑠𝑠 𝑘𝑘𝑝𝑝 + 1
Effect of the derivative gain 𝑘𝑘𝑑𝑑
From the characteristic equation:
𝜁𝜁 = (2 + 𝑘𝑘𝑑𝑑 )/2( 1 + 𝑘𝑘𝑑𝑑 ), 𝑤𝑤𝑛𝑛 = 1 + 𝑘𝑘𝑝𝑝
1−𝜁𝜁 2
The overshoot: 100𝑒𝑒 −𝜁𝜁𝜁𝜁/ decreases with 𝑘𝑘𝑑𝑑
The settling time : 𝑇𝑇𝑠𝑠 = 4𝑤𝑤𝑛𝑛 𝜁𝜁 decreases with 𝑘𝑘𝑑𝑑