YTU FACULTY OF ELECTRICAL & ELECTRONICS ENGINEERING

DEPARTMENT OF CONTROL & AUTOMATION ENGINEERING

KOM3712 CONTROL SYSTEM DESIGN, HOMEWORK-1&2
Name, Surname: Marking for HW-1: Marking for HW-2:
Student number: Problem 1: 30 Problem 4: 30
Signature: Date: 14 May, 2022
th
Problem 2: 30 Problem 5: 30
Due Date: 28th May, 2022 Problem 3: 40 Problem 6: 40

Important Notes:
- Upload your solutions as one pdf file with the name “studentnumber-familyname-hw1n2.pdf”to the University’s online
system. Do your solutions on this file (you may paste your hand written solutions and MATLAB plots). You may add extra
pages for each problem, correct the page numbers accordingly.
- You may work in groups. However, each student must turn in his/her own individual homework. Otherwise you may face
the undesirable consequences!
- If cheating or copying is suspected, all students involved will receive zero for the homework grade.
- Please note that your report should be well-organized, neat and legible. It represents you as a person.
Definitely No Copying, No Cheating, No Plagiarism… they are all treated as unethical and misconduct. They are all actual,
intended or attempted deception and dishonest actions.

Problem-1. Design a Lag Compensator so that the negative

R(s) C(s)
feedback control system given, where 𝐺𝑝 (𝑠)
+– 𝐺𝑐 (𝑠)
𝐾(𝑠 + 10)(𝑠 + 12)
𝐺𝑐 (𝑠)𝐺𝑝 (𝑠) = 𝐺(𝑠) =
𝑠(𝑠 + 3)(𝑠 + 6)(𝑠 + 9)
(a) Use frequency response methods to design a lag
compensator to yield 𝑒𝑠𝑠 = 0.001 (𝐾𝑣 = 1000) and
15% overshoot for the step response. Make any required second-order approximations.

You may take the following steps:

1. Draw Bode plots in MatLAB for 𝐾 that satisfies the steady-state error requirement.
2. Determine the frequency where the PM is 10° greater than the PM that yields the desired transient response.
3. Select a lag compensator whose magnitude response yields a total Bode magnitude plot that goes through 0 dB
at the frequency found in step-2, so that it becomes the gain crossover frequency, 𝜔𝑃𝑀 .
Hint: Draw the Bode magnitude plot of the compensator starting from high frequency back to one tenth (0.1 times)
of the frequency found in step-2, referred to as 𝜔𝑃𝑀 , at the level of negative of the magnitude found at the frequency
where PM is 60°. Then draw a line with −20 dB/decade slope towards low frequency up to 0 dB level is reached.
Set the gain of the compensator 𝐾𝑐 to 0 dB at low frequencies so that the steady-state error characteristics are not
affected.
(b) Use MatLAB or any other computer program to plot step response of the lag-compensated system and test
your second-order approximation (i.e. you will see that the response can be acceptable although the
requirement of 15% overshoot is not well-satisfied).
Solution 1.

page 1 of 6
𝜋 4 −𝜁𝜋⁄√1−𝜁 2 −ln(%𝑂𝑆/100)
𝑇𝑝 = , 𝑇𝑠 ≅ 𝜁𝜔 , %OS= 100. 𝑒 ,𝜁 = , wish you all the success, S.N. Engin ©
𝜔𝑛 √1−𝜁 2 𝑛 √𝜋2+ln2(%𝑂𝑆/100)
Problem-2. Consider the control system given where, C(s)
R(s)
+– 𝐺𝑐 (𝑠) 𝐺𝑝 (𝑠)
𝐾
𝐺𝑐 (𝑠)𝐺𝑝 (𝑠) = 𝐺(𝑠) =
𝑠(𝑠 + 5)(𝑠 + 20)

The uncompensated system has over 50% overshoot and a

peak time of 0.5 second (plot it in MatLAB to see it) when 𝐾𝑣 = 10. Do the following:
(a) Use frequency response methods to design a lead compensator to reduce the percent overshoot to 10%,
while keeping the peak time and steady-state error about the same or less. Make any required second-
order approximations.
(b) Use MatLAB or any other computer program to plot step response of the lead-compensated system and
test your second-order approximation (you will see that the response is very good in terms of the speed
requirement and acceptable for the percent overshoot requirement).
Solution 2.

page 2 of 6
𝜋 4 −𝜁𝜋⁄√1−𝜁 2 −ln(%𝑂𝑆/100)
𝑇𝑝 = , 𝑇𝑠 ≅ 𝜁𝜔 , %OS= 100. 𝑒 ,𝜁 = , wish you all the success, S.N. Engin ©
𝜔𝑛 √1−𝜁 2 𝑛 √𝜋2+ln2(%𝑂𝑆/100)
Problem-3. Given the following open-loop plant,

20(𝑠 + 2)
𝐺(𝑠) =
𝑠(𝑠 + 5)(𝑠 + 7)
design a controller to yield a 10% overshoot and a settling time of 2 seconds. Place the third pole 10 times as far
from the imaginary axis as the dominant pole pair.
(a) Use the Controllable Canonical Form for state feedback and find the feedback gain vector K 𝑐 = [𝑘1𝑐 𝑘2𝑐 𝑘3𝑐 ]
by means of coefficient matching method.
(b) Find the same gain vector by means of Ackerman’s formula.
(c) Verify the results in MatLAB with the following function: K=acker(Ac, Bc, D) or K=place(Ac, Bc, D), where
D is the vector consisting of the three desired poles, i.e. D = [s1 conj(s1) 10*real(s1)]
(d) Use the Observable Canonical Form for state feedback and find the feedback gain vector K 𝑜 = [𝑘1𝑜 𝑘2𝑜 𝑘3𝑜 ]
by means of K=acker(Ao, Bo, D) or K=place(Ao, Bo, D) in MatLAB.
(e) Find the similarity transformation (from CCF to OCF) matrix P by means of controllability matrices (CM)
of the two forms as P = CMo CMc−1 and
(f) Find K 𝑜 = K 𝑐 P−1 and check if you found the same result of (d).
(g) Do the computer simulation of system before and after the full state feedback.
Solution 3.

page 3 of 6
𝜋 4 −𝜁𝜋⁄√1−𝜁 2 −ln(%𝑂𝑆/100)
𝑇𝑝 = , 𝑇𝑠 ≅ 𝜁𝜔 , %OS= 100. 𝑒 ,𝜁 = , wish you all the success, S.N. Engin ©
𝜔𝑛 √1−𝜁 2 𝑛 √𝜋2+ln2(%𝑂𝑆/100)
Problem-4. The translational system given has the following mass, friction
and spring values: 𝑀 = 2kg, 𝑓𝑣 = 4N-s/m and 𝐾 = 8N/m,
(a) Write the equation of motion (as a differential equation).
(b) Represent the system in state-space (where the states are 𝑥1 = 𝑥
(displacement in m) and 𝑥2 = 𝑣 (velocity in m/s), the input 𝑢 is force
𝑓 and the output 𝑦 is the displacement (𝑥)
(c) Write the transfer function of 𝐺1 (𝑠) = 𝑋(𝑠)/𝐹(𝑠).
(d) Write the transfer function of 𝐺2 (𝑠) = 𝑉(𝑠)/𝐹(𝑠).
(e) Plot the step response of displacement vs. time.
(f) What would be the output 𝑥(𝑡) in meters in steady-state when the applied input is 𝑓(𝑡) = 2 N.
(g) What would be the peak value of output 𝑥(𝑡) in meters when the applied input is 𝑓(𝑡) = 2 N.
(h) Plot the step response of velocity vs. time.
(i) What would be the output 𝑣(𝑡) in meters per sec in steady-state when the applied input is 𝑓(𝑡) = 2 N.
(j) What would be the peak value of output 𝑣(𝑡) in meters per sec when the applied input is 𝑓(𝑡) = 2 N.
Solution 4.

page 4 of 6
𝜋 4 −𝜁𝜋⁄√1−𝜁 2 −ln(%𝑂𝑆/100)
𝑇𝑝 = , 𝑇𝑠 ≅ 𝜁𝜔 , %OS= 100. 𝑒 ,𝜁 = , wish you all the success, S.N. Engin ©
𝜔𝑛 √1−𝜁 2 𝑛 √𝜋2+ln2(%𝑂𝑆/100)
Problem-5. The figure below on the left shows an armature controlled dc servomotor driving a load through a gear
train, which is commonly used in a closed-loop control system. The schematic diagram below on the right
represents the armature circuit rotating simply due to the voltage 𝑒𝑎 (𝑡) applied and the fixed magnetic field B by a
permanent magnet. The armature voltage as an electrical parameter 𝑒𝑎 (𝑡) is considered to be the input to the
system. The resistance and inductance of the armature circuit are 𝑅𝑎 and 𝐿𝑎 , respectively. 𝑣𝑏 (𝑡) is the back emf
and directly proportional to the rotational speed of the armature as 𝑣𝑏 (𝑡) = 𝐾𝑏 𝜔𝑚 (𝑡), where 𝐾𝑏 is a constant of
proportionality called the back emf constant. The torque developed by the motor is proportional to the armature
current, 𝑇𝑚 (𝑡) = 𝐾𝑡 𝐼𝑎 (𝑡), where 𝐾𝑡 is the constant of proportionality and called the motor torque constant. When
the motor drives a load, the equivalent inertia and viscous damping at the armature are 𝐽𝑚 and 𝐷𝑚 , respectively.
These entities include the corresponding armature and load parameters.
𝜃𝑚 (𝑠) 𝜃 (𝑠) 𝜔𝑚(𝑠) 𝜔𝐿 (𝑠)
(a) Obtain the transfer function of 𝐺1 (𝑠) = , 𝐺2 (𝑠) = 𝐸𝐿 (𝑠), 𝐺3 (𝑠) = and 𝐺3 (𝑠) = in terms of
𝐸𝑎 (𝑠) 𝑎 𝐸𝑎 (𝑠) 𝐸𝑎 (𝑠)
electrical and mechanical parameters 𝐾𝑡 , 𝐾𝑏 , 𝐽𝑚 , 𝐷𝑚 , 𝑅𝑎 , 𝐿𝑎 .
(b) Represent this dc servomotor with a negative feedback block diagram that produces the transfer functions of
𝐺1 (𝑠) and 𝐺3 (𝑠). Hint: The expected block diagram is given below with labelled blocks and signals for your
convenience. Hence, all you need to do is determining the functions of each block.
(c) Obtain a state-space representation of the system choosing the state variables as 𝑥1 = 𝜔𝑚 (motor shaft’s
angular speed in rad/s) and 𝑥2 = 𝑖𝑎 (armature current in Amps), the input 𝑢 is the applied voltage 𝑒𝑎 and the
output 𝑦 is the angular speed (𝜔𝑚 ) in terms of electrical and mechanical parameters 𝐾𝑡 , 𝐾𝑏 , 𝐽𝑚 , 𝐷𝑚 , 𝑅𝑎 , 𝐿𝑎 .

Electrical Torque Mechanical

parameters Constant parameters Integrator

𝐸𝑎 (𝑠) 𝐼𝑎 (𝑠) 𝑇𝑚 (𝑠) 𝜔𝑚 (𝑠) 𝜃𝑚 (𝑠)

+–
𝑉𝑏 (𝑠)

back emf
constant
Solution 5.

page 5 of 6
𝜋 4 −𝜁𝜋⁄√1−𝜁 2 −ln(%𝑂𝑆/100)
𝑇𝑝 = , 𝑇𝑠 ≅ 𝜁𝜔 , %OS= 100. 𝑒 ,𝜁 = , wish you all the success, S.N. Engin ©
𝜔𝑛 √1−𝜁 2 𝑛 √𝜋2+ln2(%𝑂𝑆/100)
Problem-6. Consider the block diagram given below, which represents a negative feedback control system with all
undesirable inputs, i.e. disturbance and noise.
(a) Obtain the output signal 𝑌(𝑠) with respect to each input components.
(b) Obtain the error signal 𝐸(𝑠) with respect to each input components.
(c) Obtain the control signal 𝑈(𝑠) with respect to each input components.

Hint: You may define the loop transfer function as 𝐿(𝑠) = 𝐾𝐺𝑐 (𝑠)𝐺𝑝 (𝑠)𝐻(𝑠) and use it in the output equation for
convenience. The the output will be in the following form,
𝑁𝑢𝑚1 (𝑠) 𝑁𝑢𝑚2 (𝑠) 𝑁𝑢𝑚3 (𝑠)
𝑌(𝑠), 𝐸(𝑠), 𝑈(𝑠) = ± 𝑅(𝑠) ± 𝑁(𝑠) ± 𝐷(𝑠)
𝐷𝑒𝑛1 (𝑠) 𝐷𝑒𝑛2 (𝑠) 𝐷𝑒𝑛3 (𝑠)
(d) Obtain the sensitivity functions of the output to each input (r, n and d). Hint: when you evaluate for example
the sensitivity of y to r, assume that all other inputs are zero, i.e. d=n=0.
(e) Comment on the sensitivity functions you found in (d) like “in order to reduce sensitivity of y to r we should
increase loop transfer function”, “for large loop transfer functions, S becomes 1, which means ideal
(instantaneous) set-point tracking would occur that is 𝑦 = 𝑟 and it means the control objective is achieved”
or “in order to reduce sensitivity of y to r increase gain of the controller but this causes the sensitivity of y to
n approach −1, making 𝑦 = −𝑛, which is not desirable, since it implies that the output is directly 100%
related to the noise signal.”
Solution 6.

page 6 of 6
𝜋 4 −𝜁𝜋⁄√1−𝜁 2 −ln(%𝑂𝑆/100)
𝑇𝑝 = , 𝑇𝑠 ≅ 𝜁𝜔 , %OS= 100. 𝑒 ,𝜁 = , wish you all the success, S.N. Engin ©
𝜔𝑛 √1−𝜁 2 𝑛 √𝜋2+ln2(%𝑂𝑆/100)