Introduction to Control Systems for AS5040

em
1 What is a System?
A system is a collection of interconnected components that work together to achieve

st
a specific function. Systems exist in various domains, including mechanical, electrical,
biological, economic, and social structures.

Sy
In a physical sense, a system can be defined as any process or collection of elements
that transform inputs into outputs. Examples include:

• A car’s engine converting fuel into mechanical power.

ol
• An electrical circuit converting voltage into a desired current output.
tr
• A human body metabolizing food for energy can be considered as a complex system.
on
2 Open-loop and Closed-loop Systems
rC

2.1 Open-loop System

An open-loop system operates without feedback, meaning it does not adjust its actions
Fo

based on the actual output. Examples:

• Washing machines operating on a fixed cycle.

• Sprinkler systems working on a timer.

t
ou

• Traditional traffic lights with fixed intervals.

nd

2.2 Closed-loop System

A closed-loop system continuously monitors its output and makes adjustments using
Ha

feedback. Examples:

• Thermostats adjusting heating based on temperature readings.

• Cruise control in cars adjusting throttle based on speed measurements.

• Autonomous drones using real-time sensor feedback for navigation.

1
 Figure 1: Open Loop System

em
st
Sy
Figure 2: Closed-loop System

ol
3 Derivation of Transfer Function for a Sample Sec-
tr
ond Order System (Mass-Spring-Damper System)
on
Consider a classical mass-spring-damper system subjected to an external force. The
parameters are defined as follows:
rC

• M : Mass (kg)

• B: Damping coefficient (N·s/m)

Fo

• K: Spring stiffness (N/m)

• x(t): Displacement of mass (m) — system output

t

• F (t): External force applied (N) — system input

ou

The equation of motion for the system, derived using Newton’s Second Law, is given
by:
nd

M ẍ(t) + B ẋ(t) + Kx(t) = F (t)

Taking the Laplace Transform of both sides of the equation (assuming zero initial
Ha

conditions), we get:
M s2 X(s) + BsX(s) + KX(s) = F (s)
The transfer function is defined as the ratio of output X(s) to input F (s):

X(s)
G(s) =
F (s)

Rearranging the transformed equation:

M s2 + Bs + K X(s) = F (s)

2
 em
st
Sy
Figure 3: Second Order step response

ol
X(s) 1
G(s) = =tr 2
F (s) M s + Bs + K
To express this transfer function in a standard second-order form, divide both numer-
on
ator and denominator by M :

1/M
rC

G(s) = B K
s2 +M s+ M

Recall that the standard form of a second-order system transfer function is:
Fo

ωn2
G(s) =
s2 + 2ζωn s + ωn2

By comparing, we identify the natural frequency ωn and damping ratio ζ as:

t
ou

r
2 K K
ωn = ⇒ ωn =
M M
nd

B B
2ζωn = ⇒ ζ= √
M 2 KM
Ha

Thus, the final transfer function of the mass-spring-damper system in standard form
is:
ωn2
G(s) =
s2 + 2ζωn s + ωn2
The system parameters can be physically interpreted as follows.

• ωn : The natural frequency at which the system would oscillate if undamped.

• ζ: The damping ratio that indicates how oscillations decay over time.

Classification of System Behavior Based on ζ:

3
 Damping Ratio (ζ) System Response
ζ=0 Pure oscillation (undamped)
0<ζ<1 Underdamped (oscillations present)
ζ=1 Critically damped (fastest settling without overshoot)
ζ>1 Overdamped (slow, no oscillations)

3.1 Time Domain Specifications

3.1.1 Rise Time (Tr )

em
Definition: The time taken for the response to go from 10% to 90% of its final value.
Formula (for underdamped systems, 0 < ζ < 1):

st
1.8
Tr ≈ (1)
ωn

Sy
Example: For ωn = 5 rad/sec:
1.8
Tr ≈ = 0.36 sec (2)
5

ol
3.1.2 Settling Time (Ts ) tr
Definition: The time required for the response to stay within 2% of its final value.
on
Formula:
4
Ts ≈ (3)
ζωn
rC

Example: For ζ = 0.5, ωn = 5:

4
Ts ≈ = 1.6 sec (4)
0.5 × 5
Fo

3.1.3 Overshoot (Mp )

Definition: The maximum peak value of the response exceeding the final value, ex-
t

pressed as a percentage.
ou

Formula:  
− √ ζπ
1−ζ 2
Mp = e × 100% (5)
nd

Example: For ζ = 0.5:

 
− √0.5×π 2
Mp = e 1−0.5
× 100% (6)
Ha

Mp = e−1.81 × 100% ≈ 16.3% (7)

4 Closed-Loop System: Block Diagram and Transfer

Function Derivation
Block Diagram Representation:

The diagram contains:

4
 em
st
Figure 4: Representing closed-loop systems using block diagram.

Sy
• A summing junction to compute the error e(s).
• Controller C(s) acting on error e(s) to produce control input u(s).

ol
• Process G(s) generating output y(s) from u(s).
tr
• Sensor/Measurement H(s) providing measured output ym (s).
on
Step-by-Step Transfer Function Derivation:

1. From the summing junction:

rC

e(s) = yref (s) − ym (s)

2. Controller output:
Fo

u(s) = C(s) · e(s)

3. Process output:
y(s) = G(s) · u(s) = G(s) · C(s) · e(s)
t
ou

4. Measured output:

ym (s) = H(s) · y(s) = H(s) · G(s) · C(s) · e(s)

nd

5. Substituting ym (s) back into the error equation:

Ha

e(s) = yref (s) − H(s)G(s)C(s)e(s)

6. Solving for e(s):

e(s) [1 + H(s)G(s)C(s)] = yref (s)
yref (s)
e(s) =
1 + H(s)G(s)C(s)
7. System output:
G(s)C(s)
y(s) = G(s)C(s)e(s) = yref (s)
1 + H(s)G(s)C(s)

5
 Closed-Loop Transfer Function:

y(s) G(s)C(s)
=
yref (s) 1 + H(s)G(s)C(s)
Note: If H(s) = 1 (ideal sensor), this simplifies to:

G(s)C(s)
1 + G(s)C(s)

em
This transfer function relates the output y(s) to the reference input yref (s). By choos-
ing appropriate C(s) (controller design), we can shape system response to meet desired
specifications like stability, settling time, and peak overshoot.

st
5 Closed-Loop Transfer Function of a Second-Order

Sy
System with Proportional Controller

ol
Consider the open-loop second-order system obtained from a mass-spring-damper system,
given in standard form as: tr
ωn2
G(s) = 2
s + 2ζωn s + ωn2
on
where:
r
K
rC

• ωn = is the natural frequency of the open-loop system,

M
B
• ζ= √ is the damping ratio of the open-loop system.
2 KM
Fo

Now, we introduce a Proportional Controller, which is defined as C(s) = Kp , and

assume a unity feedback i.e., H(s) = 1 configuration.
Then, the closed-loop transfer function is given by
t
ou

ωn 2
y(s) Kp s2 +2ζωn s+ωn
2
Gcl (s) = = ω 2
yref (s) 1 + Kp s2 +2ζωnn s+ω2
nd

On simplification we get
Ha

Kp ωn2
Gcl (s) = 2 .
s + 2ζωn s + ωn2 + Kp ωn2

Consider the standard form of a second-order system as follows.

2
ωn,cl
2
.
s2 + 2ζcl ωn,cl s + ωn,cl

By comparison, we identify:
2
p
ωn,cl = ωn2 (1 + Kp ) ⇒ ωn,cl = ωn 1 + Kp

6
 2ζcl ωn,cl = 2ζωn
Solving for ζcl :
ζω ζ
ζcl = p n =p
ωn 1 + Kp 1 + Kp
Observing the expressions for the closed-loop system we notice the following.
• Increasing Kp increases the natural frequency ωn,cl , making the system faster.

• Increasing Kp decreases the damping ratio ζcl , potentially making the system more

em
oscillatory.

5.1 Achieving Desired Time Domain Specifications via Propor-

st
tional Controller

Sy
In the previous section, we observed that introducing a Proportional Controller with gain
Kp to the open-loop second-order system under unity feedback modifies the system’s
dynamics. Specifically, it alters the closed-loop system’s natural frequency and damping
ratio. These changes directly influence the Time Domain Specifications, such as rise

ol
time, peak time, overshoot, and settling time. These specifications are critical in practical
control design to ensure that the system responds within desired performance limits.
tr
Recall: Time Domain Specifications for Standard Second-Order Systems (Un-
on
derdamped, 0 < ζ < 1):

• Rise Time (tr ): Approximate for 0.1 ≤ ζ ≤ 0.6:

rC

1.8
tr ≈
ωn
Fo

• Peak Time (tp ):

π
tp = p
ωn 1 − ζ 2
t

• Maximum Peak Overshoot (Mp ):

ou

√−πζ
Mp = e 1−ζ 2 × 100%
nd

• Settling Time (5% Criterion) (t5%

s ):

3
Ha

t5%
s ≈
ζωn

• Settling Time (2% Criterion) (t2%

s ):

4
t2%
s ≈
ζωn

Impact of Controller Gain Kp on Time Domain Specifications:

By replacing ωn and ζ with the modified ωn,cl and ζcl :

7
 • Rise Time:
1.8 1.8
tr ≈ = p
ωn,cl ωn 1 + Kp
• Peak Time:
π π
tp = p = q
ωn,cl 1 − ζcl2 ζ2
p
ωn 1 + Kp 1 − 1+Kp

• Maximum Peak Overshoot:

em
ζ
−π √
1+Kp
√−πζcl s
ζ2
1−ζ 2 1−
1+Kp
Mp = e cl × 100% = e × 100%

st
• Settling Time (5% Criterion):
3 3 3

Sy
t5%
s ≈ = ζ
=
√
p
ζcl ωn,cl · ωn 1 + Kp ζωn
1+Kp

• Settling Time (2% Criterion):

ol
4 4
t2%
s ≈
tr =
ζcl ωn,cl ζωn
on
Important Observations and Design Insights:
• Rise time and peak time decrease (i.e., faster response) as Kp increases because
rC

of increased ωn,cl .
• Overshoot may increase with higher Kp due to reduced ζcl , leading to more
oscillations.
Fo

• Interestingly, the settling time based on the 5% and 2% criteria remains unaffected
by Kp , since ζcl ωn,cl = ζωn .
t

By tuning the proportional gain Kp , we directly influence how fast and how smooth
ou

the system response is. Proper choice of Kp allows achieving the desired trade-off between
response speed and overshoot control.
nd

6 Some Common Controller Structures

Ha

6.1 Proportional Controller

The key idea is that the control input to the system is “proportional” to the error between
desired and output values, i.e.
u(t) ∝ e(t).
The controller transfer function is of the form:
GC (s) = KP . (8)
The controller gain KP , can be tuned to achieve desired performance.
Other than Proportional controller, other controllers also have been explored.

8
 6.2 Integral Controller
The control input is proportional to the integral of the error till the current time, i.e.
Z t
u(t) ∝ e(τ )dτ.
t0

The controller transfer function is of the form:

KI
GC (s) = . (9)

em
s
This controller forces ess to zero and keeps it there.

st
6.3 Derivative Controller

Sy
The control input in this controller scheme is proportional to the derivative of the error
at that time, i.e.
d
u(t) ∝ e(τ )|t .
dτ

ol
The controller transfer function is:
tr
GC (s) = sKD . (10)
on
The derivative controller improves the transient response and reduces the oscillations.
There may, however, exist nonzero steady state errors.
rC

6.4 PI Controller
The PI controller is an additive combination of proportional and integral controllers. The
Fo

controller transfer function Gc (s) is of the form:

 
KI
KI K P s + KP
Gc (s) = KP + = . (11)
t

s s
ou

The overall transfer function is:

 
KI
KP s + K GS (s)
nd

P
G(s) =   . (12)
KI
s + KP s + KP
G S (s)
Ha

• Negative, small and real open loop zero (OLZ) is introduced.

• Open loop pole (OLP) at origin is added.

• Steady state error is zero, since system order increases by 1.

• Oscillations due to KP may exist.

9
 6.5 PD Controller
Similar to PI controller, PD controller is an additive combination of proportional and
derivative controllers. The transfer function is given as:
 
KP
Gc (s) = KD s + KP = KD s + . (13)
KD

The overall transfer function is:

em
 
KP
KD s + KD
GS (s)
G(s) =   . (14)
KP
1 + KD s + KD
G S (s)

st
• A negative OLZ is introduced in the system.

Sy
• OLZ is selected to get the desired transient response.

• Generally, the damping increases with increasing KD .

ol
• For a second order system, ζnew = 2ζ+KD ωn
2
when KP = 1.

• Presence of a derivative term makes the system susceptible to noise and saturation
effects.
tr
on
6.6 PID Controller
rC

One of the most versatile and powerful controller, and is best of both the worlds. Con-
troller gains KP , KI and KD can be tweaked and tuned to get the desired transient and
steady state properties. The transfer function is of the form:
Fo

KI KD s2 + KP s + K I
GC (s) = KP + + KD s = . (15)
s s
The overall transfer function is:
t

(KD s2 + KP s + KI ) GS (s)
ou

G(s) = . (16)
s + (KD s2 + KP s + KI ) GS (s)

• Two negative OLZ and one OLP at origin is introduced.

nd

• One OLZ and the origin OLP improve the steady state error (similar to the PI
Ha

controller.)

• The other OLZ affects the transient properties

6.7 Rate-feedback Controller

To improve the transient response, the PD controller is a very good choice. However,
cascade controllers like PD are not the only option. Moreover, PD controllers suffer from
noise due to increased bandwidth. The transient response can be improved by placing a

10
 rate sensor, which acts as a differentiator, in the inner loop (Fig. 5). The general transfer
function for the open-loop branch of rate-feedback controller is:

KK1 G1 (s)
GS (s) = . (17)
1 + sK1 Kf G1 (s)

The overall transfer function is:

KK1 G1 (s)
G(s) = . (18)
1 + (sKf + K)K1 G1 (s)

em
Generally, K1 G1 (s) is taken as a single unit. Increasing Kf , the damping of the system
increases and the max-overshoot reduces. The bandwidth of the system reduces which

st
improves the stability and noise susceptibility.

Sy
ol
tr
on
rC

Figure 5: Rate-feedback compensation

t Fo
ou
nd
Ha

11