School of Engineering

Module assignment ER3502

BEng (Hons) Mechanical & Manufacturing Engineering

Control Engineering

Author: Dean Woolley Supervisor: Martin Craig

Introduction to Control Systems: systems can be described by partial differential
equations. A model can be a continuous-time (CT)
A control system can be defined as a physical or a discrete-time (DT) model. A continuous-time
relationship of devices that determines the model gives the relationship between its
behaviour of other devices or systems to obtain a continuous input and output in the form of a
differential equation. [ CITATION Adv \l 2057 ]
specific result. The simplest system can be made
up of an input, a process and an output, with both Linear Systems:
variables to the system being signals. The result at
A linear system can be defined if it has two
the end is achieved via control loops, with the
mathematical properties, homogeneity and
process variable at any given point being
additivity. If both properties can be proven within
maintained. An electronic system can be
a system, then that system is linear. Within a
demonstrated in a block diagram, where each
system the static characteristics plot the values of
block is represented with input / output.
an output signal verses an input signal. If the static
Input Output characteristics are straight lines, the system is
linear.
Control System
Open Loop Control System:

The open loop control system is identified as a

Figure 1: Simple block diagram non-feed back control system, where the system
considers the input, but doesn’t respond on the
The block diagram is also recognised as the
feed back to pick up the output, therefore the
‘mathematical’ model. The parts of the system on
output has no effect on the control action. The
show are the properties of the signal transfer. The
system operates on a fixed situation, where there
diagram allows both dynamic and static behaviour
are no disturbances or variations within the
to be investigated and provides a base for the
system. As the open loop feed back system doesn’t
design of the control system. [ CITATION htt8 \l
use feedback to determine if the desired output
2057 ]
has been achieved, there is an assumption the
Types of Models: inputs objective was successful, due to the fact any
errors made cannot be corrected. Therefore any
A model is static if its output depends only on the compensation for external disturbances cannot be
actual value of its input signal. For example, a accounted for. [ CITATION htt9 \l 2057 ]
resistance where the input signal is the voltage
and the output signal is the current is a static
system. A model is dynamic if its output depends
on previous signal values as well. An electrical θⅈ Forward Path θ0
circuit consisting of serially connected resistor and
capacitor is a dynamical system, since the voltage 1
drop on the capacitance depends on the charge, s ( s+1 )
and thus on the previous values of the current. A
model can be linear or non-linear. A model can be H
deterministic or stochastic. The signals of a
deterministic model can be described by analytical Feedback Path
relationships. In a stochastic model, the signals can
be given by probabilistic variables and contain Figure 2 : A continuous closed loop system
uncertainties. Spatially, a model can have either
lumped or distributed parameters. Lumped Closed Loop Control System:
parameter systems can be described by ordinary
differential equations, while distributed parameter
Figure 2 is an example of a closed loop system. The the Laplace domain, the ODE becomes a
system works by using an open loop for the polynomial of s. The differential equation becomes
forward path, with one or more feedback paths a simple polynomial multiplication and division
from the output and its input. The system can also within the time domain and Laplace domain.
be termed a ‘feedback’ system. This improves the However as the input and output signals are also in
stability and reduces errors. The concept of the the Laplace domain, for the systems response to
closed loop system is to reach and sustain the be a valid time-dependent signal, an inverse
required output condition by comparing it with the Laplace transform must first take place. The
actual condition. This is achieved by an error signal Laplace transform of a function of time f ( t ) is
being generated. This error signal being the given by the integral: [ CITATION htt12 \l 2057 ]
difference of the output and the referenced input.
∞
A closed loop system can be defined by its control
action and it being dependent on the output in L(f ( t ) )=∫ f ( t ) ⅇ−st ⅆt=F ( s )
0
some format. [ CITATION htt10 \l 2057 ]
Poles & Zeros:
Main Characteristics:
Poles and zeros of a transfer function can often be
1. The stability of an unstable system is
complex. The dynamics of the system can be
improved.
represented by plotting values on the s-plane.
2. Increases or reduces the sensitivity of the
Poles and zeros help characterize the system, by
system.
reconstructing the input / output differential
3. To execute a reliable and repetitive
equation. Zero locations are plotted using a O
system.
(circle), whilst pole locations are plotted using a X
4. To strengthen the process from external
(cross). The values of the poles are from the
disturbances.
Laplace variable S, causing the transfer function to
5. The systems input can be adjusted
automatically, reducing errors. be infinite and zero (roots of the denominator &
nominator). The plotted locations can determine
Closed-Loop Summing Point: the systems response characteristics. The poles
and zeros are properties of the transfer function,
The summing point on a closed loop feedback and therefore of the differential equation
system ascertains the error from the actual output describing the input-output system dynamics. To
and the desired output. By comparing the systems achieve a complete system, they combine with
set point to the actual value, a positive or negative gain constant, K where they characterize the
error signal is generated, allowing the controller to differential equation.
respond.
The S Plane is plotted when the poles and zeros
Closed-Loop Transfer Function: have been determined. It consists of an imaginary
and real axis, which references the variable Z.
The systems behaviour is the mathematical
[ CITATION htt13 \l 2057 ]
relationship between the input and output. This is
known as the transfer function. [ CITATION htt11 \l Stability in the S Plane:
2057 ]
The boundary between the areas of stability and
Laplace Transform: instability is identified in the s plane by the
imaginary axis. This can be defined as:
Created and named after Pierre-Simon Laplace,
the Laplace transform is a method of transforming s= jω
an ordinary differential equation (ODE) to an
algebraic equation. Signals and systems are Stability is recognised in the s plane by the
expressed as functions of a complex variable s, as placement of the poles. If the poles of the closed
oppose to a function of time. On transformation to loop system are in the left hand side of the s plane
(negative and real), the system is stable. If the the tolerance around the final value. The tolerance
poles are in the right hand side (positive and real), is generally 2% and 5%. Settling time is
the system is unstable. If the poles appear on the represented by ts, and together with time
imaginary axis and none appear in the right hand constant are independent of the system gain. The
side, the system is marginally stable. effect this has on the system means, that when the
gain changes, both settling time and constant time
The Concept of Stability: will never change.
A system can be defined as stable if it has the
ability to return to its equilibrium state after Therefore in this instance:
initially moving away from its equilibrium state.
The stability of the non-linear system depends on 4 4
ts = = =8
the input signal and the point of operation. With ζ ωn ( 0.5× 1 )
this example, the stability is typically a state within
the system, but not of the whole system. The Peak Time (tp): Is the time needed for the
stability of the linear system, is characteristic for
response to achieve the peak value at the first
the system. Stability is based on the parameters
and structure of the system, it does not depend on stage
the input.
π
Performance Specification: tp = = 3.6
ωd
Having established an underdamped second order
Describing Function:
system, the following can be identified from the
mathematical model and it’s step response As linear control system works on the principle of
function. From the graph produced in Appendix A superposition, where the two inputs are applied at
using MATLAB, we are able to determine four the same time, giving an output from the sum of
primary functions: two outputs. Non-linear systems due to the
complexity cannot use this theory, therefore an
Rise Time (tr): For an underdamped system, the indefinite extended version of the frequency
rise time is the time required for the response to response called ‘describing function method’ can
be adopted. Initially Invented by Nikolay
rise from 0% - 100% during the first cycle of
Mitrofanovich Kryloy and Nikolay Bogoliubov, and
oscillation. Can be expressed as: further developed by Ralph Kochenburger, the
Describing Function can be identified as a process
1−ⅇ−ζ ωnt
used for investigating problems within non-linear
c (t)= sin ( ( ωn √ 1−ζ 2 ) t+θ )
√ 1−ζ 2 systems in control engineering. It is termed an
approximation method for analysis, with the main
π uses being to determine the limit cycle oscillation
π− of the non-linear system. Other applications such
3
Where tr = = 2.41 seconds as non-linear system response to sinusoidal input,
3
√ 2
jump resonance and sub-harmonic content can all
be determined by the describing function method.
[ CITATION htt14 \l 2057 ]
Peak Overshoot: Also known as maximum Jump Resonance:
overshoot, can be defined as the variance of peak
There is a interrupted change in the steady state
time to the final value.
response of the system when a parameter is
altered.
a1
% Overshot = ×100 , Limit Cycle Oscillation:
B
Limit cycle oscillation is distinguished by a
T sustained amplitude and frequency identified from
x 100, a 1=16.28 %
−ζωn ( )
where a =Be 2
the non-linear properties of the control system.
1
The limit cycles are recognizable from linear
Settling Time (ts): Is the time required for the oscillations, in the sense that the amplitude is
response to reach the steady state, and stay within
unconstrained to the initial conditions. [ CITATION operation, a switch and storage mechanism (such
htt15 \l 2057 ] as a transistor and a capacitor, or shutter and a
Sub Harmonic Content: film strip) form a sequence of samples of the
If the input of a system contains two frequencies, continuous input waveform. The output of the
the non-linearity may cause cross modulation sampling process is called pulse amplitude
effecting components in the output. For example, modulation (PAM) because the successive output
where m ω1 ±n ω 2, and m , n are undesirable intervals can be described as a sequence of pulses
components, but can exist at frequencies below with amplitudes derived from the input waveform
those of the applied signals. samples. The analog waveform can be
Non-Linearities are spilt into two identifiable approximately retrieved from a PAM waveform by
groups, continuous and discontinuous. simple low-pass filtering, provided we choose the
Discontinuous non-linearities are more commonly sampling rate properly. The ideal form of sampling
known as ‘hard’ non-linearities as they cannot be is called instantaneous sampling.
approximated by linear function.
Discrete Time System: https://www.sciencedirect.com/topics/engineerin
g/sampling-theorem
If the input and output are sampled, then it is
known as a discrete time system, where the Zero Order Hold:
correlation between them can be identified using a
difference equation. A discrete time system is also The zero-order hold (ZOH) is a mathematical
called a sampled data system. Engineers working model of the practical signal reconstruction done
with control and its theory are able to achieve by a conventional digital-to-analog
results based on digital computers, which are converter (DAC). That is, it describes the effect of
equipped with real-time facilities.
converting a continuous-time signal to a discrete-
time signal by holding each sample value for one
Non-Linearity of Saturation:
sample interval.
Sampling:
The Z Plane:
The sampling theorem specifies the minimum-
sampling rate at which a continuous-time The boundary line between the area of stability
signal needs to be uniformly sampled so that the and instability is known as the imaginary axis in the
original signal can be completely recovered or s-plane. This boundary line can be defined as:
reconstructed by these samples alone. This is
usually referred to as Shannon's sampling s= jω, where z=ⅇ st , which by definition this
theorem in the literature. equation defines the boundary between the stable
and unstable regions in the z-plane. Equation
Sampling Process: z=ⅇ st ,can be represented by plotting an Argand
diagram.
Analog information must be transformed into a
digital format. The process starts
with sampling the waveform to produce a discrete
pulse-amplitude-modulated waveform (see Figure
2.3). The sampling process is usually described in a
time domain. This is an operation that is basic to
digital signal processing and digital
communication. Using the sampling process, we
convert the analog signal in a corresponding
sequence of samples that are usually spaced
uniformly in time. The sampling process can be
implemented in several ways, the most popular Figure 3 : Example of stability within the Z plane.
being the sample-and-hold operation. In this
Figure 3 represents the equation z = ⅇ j ωt , focussed Rise time: 1.6390, Settling time: 8.0759,
on its centre origin, with a radius of 1. The values Overshoot: 16.2929, Peak time: 3.5920.
of magnitude and angle in figure 3 define a unit
When comparing these with calculations
circle within the z-plane. This statement is true as
performed by hand it can be identified that the
the line Az, from the z-value origin is always 1, and
same results are obtained, thus suggesting both
forms an angle of ωT with the positive real axis. As
are correct, and that the system is underdamped.
the signal frequency increases, and ω gets bigger,
These results can be located in Appendix A – Task
a circle with a radius of 1 is marked. This informs A, titled: Additional Performance Specification
us that the imaginary axis in the s-plane Calculations.
transforms to the unit circle in the z-plane. In
other words, this circle forms the boundary Nyquist Stability Criterion:
between stability and instability within the z-plane.
By analysing the graphs plotted from MATLAB for
the Nyquist it can be determined that the open
loop is stable as there are no poles on the right
hand side, therefore the simple Nyquist stability
criterion is sufficient. From the graph the stability
-1 +0j lies to the left of the diagram, and the open
loop does not encircle the point -1 +0j, therefore
the closed loop is stable. Generally the majority of
Figure 4  : Summary of ‘s-z’ mapping systems are stable with small loop gains, where
stability is reached at a critical gain. By increasing
The simplest way to determine if a system is stable the gain, the system will indicate unstable
or unstable would be to ‘map’ an unstable pole patterns. However the graph from MATLAB shows
from the s-plane to the z-plane, and see where it the system remains stable at any value of the loop,
sits. If the positive s value is compatible to a point this is due to its structure. This system is known as
on the right hand side of the s-plane (unstable), a ‘structurally stable system’ . As mentioned
then s= 0, where a z value of 1 ( ⅇ 0=1 ¿, any earlier the Nyquist diagram doesn’t encircle the
positive value therefore conforms to a z value point -1 + 0j, even when the loop gain is increased.
higher than 1, which means the point would sit The increased loop gain is irrelevant to the system,
outside of the unit circle. By following this ruling as it will not become unstable, however the
system, for the z-plane, the area outside the unit stability margins will decrease.
circle is the area of instability, and the area within
Using the difference equation method to calculate
the unit circle represents stability. [ CITATION
the discrete time response (Appendix C, Task C)
htt16 \l 2057 ] means a comparison can be made against the
results obtained from the continuous control
Evaluation: system and its transfer function (Appendix A, Task
A). When the transfer function:
The calculations performed in task A (reference
appendix A) and graphs completed on MATLAB 1
C(S) = 2 was uploaded to SIMULINK.
(reference appendix D), show there is an s + s+ 1
underdamped second order system, where the The graph below was created:
value of ζ =0.5 .

Step Response Analysis:

From the MATLAB graph (reference Appendix D,

Task A), which was generated from the continuous
time function, the following have been
determined:
 Calculations where completed using the infinte
power series for verification, and each individual
time reponse was the same, which would indicate
that both methods are satisfactory. The results
from the two graphs suggests that the stability is
When compared against the results obtained for the same.
the difference equation plot for the unit step, it
can be identified that the 2 graphs are almost
identical:

REFERENCES

1. https://www.tutorialspoint.com/control_systems/control_systems_introduction.htm. [Online]

2. Advanced Textbooks in Control & Signal Processing – Control Engineering, Laszlo Keviczky, Ruth Bars, Jeno
Hetthessy.

3. http://engineering.electrical-equipment.org/panel-building/difference-between-open-loop-closed-loop-
systems. [Online]

4. https://www.electronics-tutorials.ws/systems/closed-loop-system.html. [Online]

5. https://www.tutorialspoint.com/control_systems/control_systems_time_domain_specifications.htm.
[Online]

6. https://www.sciencedirect.com/topics/engineering/laplace-transforms. [Online]

7. http://web.mit.edu/2.14/www/Handouts/PoleZero.pdf. [Online]

8. https://www.electrical4u.com/different-types-non-linearities-in-control-system. [Online]

APPENDICES
Appendix A - Task A – Linear System

Solution:
Continuous Time Position Control System:

1
A system where G ( s )= Assuming H = 1
s ( s+1 )

Complete the Closed Loop:

Where C = Output, R = Input, G = Transfer function along the forward path, H = Transfer function along the
feedback path

C(s) = G(s) {R(s) – H(s)C(s)}

C(s) = G(s)R(s) – G(s)H(s)C(s)

C(s){1 + G(s)H(s)} = G(s)R(s)

C G (s )
( s )=
R 1+G ( s ) H ( s )

The closed-loop transfer function is the forward-path transfer function divided by one plus the open-loop
transfer function.

The Transfer Function is: Equating Co-Efficient:

C G (s )
( s )= Sub in 0 for S, where C(s) = 1
R 1+G ( s ) H ( s )

1
C s (s +1)
( s )= 1 = A( s2 + s+1 ¿ Bs2 +Cs
R 1
1+ +1
s ( s +1 )
2
1 = A(0 +0+ 1¿+ B ¿

1
C(s) = 2 1 = A(1), Therefore A = 1
s + s+ 1

Completing the Square: s2 + s+1=C (s) Determine B & C from the following:

1 2 3
( )
s+
2
+ =C ( s )
4
( 1+ B=0 ) Re - arrange to get B & C

( s+0.5 )2 +0.75=C( s) ( 1+C=0 )

Using Partial Fractions expand the equation: B = -1 and C = -1

A Bs C
C ( s) = s + 2 + 2
( s + s+1 ) ( s +s +1 )
A Bs+C
C ( s) = + 2
s ( s +s +1 )
 1 √3

C(s) =
1
−1
−s−
2 +
√3 2 ( )( )
s 3
2
( s−−0.5 )2+ √
2
( s−−0.5 )2 ( √ 0.75 )
2 ( )
1 s +0.5 0.5
C(s) =
s
−¿ ( 2
( s +0.5 ) +0.75
+¿
)
( 5+0.5 )2 +0.75
This is the Unit Step response

Using Inverse Laplace Transforms:

1 b s−a
=1 = ⅇ at sin ( bt ) =ⅇ at cos ( bt )
s ( s−a )2+ b2 2
( s−a ) + b 2

Therefore the second order response to a unit step can be written as:

1 −0.5 t √3 t
C (t) = 1−ⅇ−0.5 t cos (( √ 0⋅ 75 2) t ) +
√3
ⅇ sin
2( )
Additional Performance Specification Calculations:

Rise Time:

1−ⅇ−ζ ωnt

c (t)= sin ( ( ωn √ 1−ζ 2 ) t+θ ) and

√ 1−ζ 2
−θ
Rise time (tr) = π
wn √ 1−ζ 2

We know ωd is damped frequency, where ωd = ( ωn √1−ζ 2 ), where ωn = 1 and ζ = 0.5, therefore:

ωd = 0.866 or √ 3 and θ = cos−(1¿ζ ) ¿ = π
2 3

π
π−
3
Where tr = = 2.41 seconds
√3
2
Peak Overshoot:

a1
% overshoot = ×100 , which equates to:
B
T
x 100, where ζ = 0.5 and ω n = 1, which is obtained when comparing the following:
−ζωn ( )
2
a 1=Be

K ω2 n 1
2 2 , with the transfer function
G ( s )= 2
S + 2ζ ωn+ ω n s + s +1
Therefore transient oscillation is ωd , then:

2π 2π
T= = = 7.26
ωd ωn √ 1−ζ 2
1

1ⅇ−0.5 x ( 3.63) ×100 a

Therefore: , 1 = 16.28, or 16.3 (1DP)
1

This can also be expressed as:

2

e−ζπ ∕ √ 1−ζ × 100

2

e−0.5 π ∕ √ 1−0.5 × 100

= 16.30

Settling Time:

2%

0.02 B = B ⅇ−ζωn ts can be written as Natural logs ln 50=ζ ωnts

1
ts= ( )
ζ ωn
ln 50 = 7.82

5%

0.05 B = B ⅇ−ζωn ts can be written as Natural logs ln 20=ζ ωnts

1
ts= ( )
ζ ωn
ln 20 = 5.99

Appendix B - Task B – Nonlinear System

Solution:

Determine the frequency response of the linear system:

1
s ( s+1 )

k
G ( s ) H ( s )=
s ( s +1 )

Where k = 1

k
G ( s ) H ( s )= 2
( s + s)
 k k
Sub in jω for s, G ( jω ) H ( jω) = =
( jω )2+ jω 2
−ω + jω

Rationalizing: Multiply top & bottom by (−ω2 + jω )

k (−ω2− jω )
G ( jω ) H ( jω) =
( −ω2 + jω) (−ω2− jω )

k ( −ω2 − jω )
G ( jω ) H ( jω) =
ω4 + ω2

Therefore: |G−( jω ) H ¿

K
Therefore: |G−( jω ) H ¿ =
ω + ω2
4

Determine Angle:

∠ G ( jω ) H ( jω)=tan −1 ( ba )
K− (−ω )

∠ G ( jω ) H ( jω)=tan −1 ( ) ω 4 +ω2
K (−ω 2 )
ω 4 +ω2

k (−ω )
¿ tan
−1
( ( ))
k −ω 2

¿ tan−1
( ωω ) 2

For Nyquist plotting, Where K=1 

Nyquist Plot Table

ωω 0.2 0.5 1 1.5 2 5 7 8 20 50 100

(rad/s)
|G(jω)H(jω)|Gj𝜔Hj𝜔 4.902 1.788 0.707 0.369 0.223 0.0392 0.0202 0.0155 0.00249 0.000399 9.999E-
05

 (K=1) 
∠G(jω)H(jω)∠Gj𝜔Hj𝜔 -101.309 -116.565 -135 -146.309 -153.434 -168.690 -171.869 -172.874 -177.135 -178.854 -179.427

 (deg) 

For plotting on angle, it is (ans-180). 

Stable system 

Determine the frequency response of the Non - linear system with describing function:

Let: |Gn ( jω )|=Gn ( ω ) , ∠ Gn ( jω )=∅ n ( ω )

Therefore: Gn ( ω ) = Fundamental amplitude of output

 Amplitude of input

And ∅ n ( ω ) = Phase shift between the fundamental component of the output & input

Describing Function – Definition:

Non Linear Element – Input

ⅇ (t )=E sin ωt

Non Linear Element – Output is m ( t ) , where m ( t ) , can be represented by its Fourier Series:

Mo
m (t) = + M 1 sin ( ωt + ∅1 ) + M 2 sin ( 2 ωt + ∅2 ) + M 3 sin ( 3 ωt+ ∅2 ) + ……………
2

The Describing Function can be defined as:

M1
Gn ( ω ) =
E

∅ n ( ω ) =∅ 1

2π
Fourier expansion of a periodic function m ( t ) of period T = is:
ω

A0 n=∞ n=∞
m ( t )= + ∑ An cos n ωt + ∑ Bnsin nωt
2 n=1 n=1

Where

T
2
An= ∫ m(t )cos nωtⅆt
T 0

And

T
2
Bn= ∫ m(t )sin nωtⅆt
T 0

From this

mn= √ A n2 + B n2 and ∅ = tan

−1
( AnBn )
Considering the Non Linearity of Saturation:

Let ⅇ (t )=E sin ωt

Then m ( t )=KE sin ωt For 0<t <t 1

¿ KE sin ω t 1 For t 1< t<T /2−t 1

¿ KE sin ωt For T /2−t 1<t <T / 2+ t 1

¿ - KE sin ω t 1 For T /2+ t 1 <t< T −t 1

¿ KE sin ωt For T −t 1 ¿ t< T

2π
Where T =
ω

Given S is the value of ⅇ (t ), at which the output saturates, then, S= E sin ω t 1

Since m ( t ) , is an odd function [ f ( t )=f (−t ) ] , then m ( t ) , contains no cosine or DC terms, or even harmonics:
n=∞
m ( t )= ∑ Bnsin nωt
n=1

T
2 T
Bn= ∫ m(t )sin nωtⅆt , as the function m ( t ) is symmetrical about , we can write:
T 0 4

T/4
2
Bn=4 x ∫ m(t)sin nωtⅆt
T 0

Therefore

t1 t/4
8 8
Bn= ∫ KE sin ωt .sin nωt dt+ ∫ KE sin ω t1 . sin nωt
T 0 T t1

8 KE
¿ ¿
T

From Trig Identities:

cos ( A−B )−cos ( A+ B )=2 sin A sin B, hence sin ωt ⋅ sin nωt=¿ ¿

Note: sin ω t 1 is a constant

t1 t /4
8 KE 1
Bn = [ ∫ cos ( n−1 ) ωt −cos ( n+1 ) ωt dt+ sinω t 1 ∫ sinnωt dt]
T 20 t1

8 KE sin ( n+1 ) ω t 1 sin ω t 1 cos ω t 1

= ¿- + ]
T 2 ( n+1 ) ω nω

2π
Since T = , and by expanding sin ω t 1 . cos nωt by use of sin ( A+ B )−sin ( A−B )
ω

= sin ( n+1 ) ω t 1−sin ( n−1 ) ω t 1, where A=nw t 1 and B=ω t 1

Where A = nω t 1 and B = ω t 1
Therefore:

2 KE sin ( n+1 ) ω t 1 sin ( n+1 ) ω t 1 sin ( n−1 ) ω t 1

Bn= ¿- + - ]
π n+1 n n
Thus giving:

sin ( n+1 ) ω t 1 sin ( n+1 ) ω t 1 n−( n−1 ) sin ( n−1 ) ω t 1

n
-
n
= sin ( n+1 ) ω t 1 [
n ( n−1 )
=
]
n(n+1)

And:

sin ( n−1 ) ω t 1 sin ( n−1 ) ω t 1 n−( n−1 ) sin ( n−1 ) ω t 1

n−1
-
n
= sin ( n−1 ) ω t 1 [
n ( n−1 )
=
n (n−1) ]
Therefore BN can be written as:

2 KE sin ( n+1 ) ω t 1
Bn= ¿+ ]
nπ n+1

To determine Gn, only B1 is required, as this is the fundamental frequency (1st harmonic), therefore, it can be
simplified to give n=1

2 KE sin 2 ω t 1
B 1=
π [
ω t 1+
2 ]
M1
The Describing Function is defined as G n ( ω ) = , and having completed the Fourier series can be
E
expressed as:

sin 2 ω t 1
Gn ( ω ) = { (
2k E
π
ω t 1+
2
,
)}
E

2k E sin 2 ω t 1
Gn ( ω ) =
π (
ω t 1+
2 )
Calculating values for ωt , where:

m ( t )=K E sin ( ωt ) , rearranging to give ωt

Therefore:

ωt=sin−1 ( KEm ) , equivalent to Theta

Table for the Non Linearity Plot on the Nyquist:

E (x) 0.5 0.6 0.7 0.725 0.8 0.9 1.0 1.5 2

 ωt1 , 1.571 0.985 0.796 0.761 0.675 0.589 0.523 0.3398 0.2526

( m=2 )

Gn ( ω ) 4 3.681 3.298 3.209 2.961 2.676 2.435 1.665 1.259

( k =4 )

The systems stability with the Non Linearity remains stable, as −Gndoes not exceed -1, therefore doesn’t
cross the real axis on the Nyquist diagram as shown in the table below:

−Gn ( ω ) -0.25 -0.27162 -0.30313 -0.31156 -0.33765 -0.37366 -0.41051 -0.60036 -0.79375

( k =4 )

Appendix C - Task C - Discrete Time System

Solution:

Open Loop Pulse Transfer Function:

1−ⅇ−Ts
G ( s )=k ( s )( s ( s1+1) )
Given K = 1

1
G ( s )=( 1−ⅇ−Ts )
( 2
s ( s+ 1 ) )
Partial Fraction Expansion:

1 A B C
2
s ( s+1 )
= (
+ 2+
s s ( s+1 ) )
1 s 2 C +s ( s+ 1 ) A+ ( s+ 1 ) B
=
s 2 ( s+1 ) s2 ( s+ 1 )

1=s 2 C+ S ( S+1 ) A + ( S+1 ) B

Equating Co-Efficient:

Sub in 0 for S

1=¿
B=1

Sub in -1 for S
2
1=(−1 ) C+ 0 (−1+1 ) A+ (−1+1 ) B
C=1

Sub in 1 for S, 1 for B & 1 for C

1=( 1 )2 1+ 1 ( 1+1 ) A + ( 1+1 ) 1

1 = 2A + 2 + 1

A = -1

Sub in Values for Equation:

1 A B C
= + 2+
s ( s+1 ) s s ( s+1 )
2

−1 1 1
G ( s )=( 1−ⅇ−T ) s

( + 2+
s s ( s+1 ) )
Taking Z Transforms:

F(s) F(z)

1 z
s z−1

1 Tz
s2 ( z−1 )2

1 z
( s+ a ) z−ⅇ−at

ⅇ−kTs z−k

1Tz z
) −z
G(z)=( 1−z
−1
(( 2
z−1 ) ( z−1 ) ( z−ⅇ−1 t ) )
Given T = 1

z−1 −1 1 x 1 z
G(z)= ( )(z
z 2
( z−1 ) ( z−1 ) ( z−0.368 ) )
−1 ( z−1 )( z−0.368 )+ ( z−0.368 ) + ( z−1 )2
G(z)=( z−1 )
( ( z−1 )2 ( z −0.368) )
 −z 2+ 1.368 z−0.368+ z−0.368+ z 2−2 z+ 1
G(z)=( z−1 )
( ( z−1 ) (z−0.368) )
Simplifies to give the open loop transfer function:

G(z)= ( z 0.368 z +0.264

−1.368 z +0.368 )
2

Determine the closed loop function from equation:

C G ( z)
( z )=
R 1+G ( z ) H ( z )

0.368 z +0.264
c
( z )=
( 2
z −1.368 z +0.368 )
R 0.368 z +0.264
1+ 2
(
z −1.368 z +0.368 )
Simplifies to give the output as:

C 0.368 z +0.264
( z )= 2 or
R z −z +0.632

C ( ) 0.368 z−1 +0.264 z−2

z=
R 1−z−1 +0.632 z −2

Determine the discrete time response using the Infinite power series method:

C 0.368 z +0.264
Where: ( z )= 2
R z −z +0.632

z 0.368+0.264
R ( z )=
z −1
and C ( z )=R ( z ) 2 (
z −z +0.632 )
Therefore:

0.368 z 2 +0.2642
c ( z) =
( 2−1 ) ( z 2−z +0.632 )

0.368 z 2+ 0.2642
c ( z) =
( z 3 −2 z 2 +1.632 z−0.638 )
Long Division ( Polynomial ) Multiply the denominator by the numerator:

0.368 z−1 + z−2+ 1.4 z−3+ 1.4 z−4 +1.147 z−5 +0.894 z−6
+0.801 z 7+ 0.868 z −8

z 3−2 z 2+1.632 z−0.632 0.368 z 2 + 0.264z

 0.368 z 2−0.736 z +0.6−0.233 z −1

0 z −¿ 0.6 +¿ 0.233 z−1 +¿ 0

z −¿ 2+¿ 1.632 z −1 +¿ 0.632 z−2

0 1.4 −¿ 1.399 z−1 −0.632 z−2 +0

1.4 −¿ 2.8 z−1 +¿ 2.285 z−2 −¿ 0.885 z−3

0 1.4 z−1 −¿ 1.648 z−2 +¿ 0.885 z−3 +¿ 0

1.4 z−1 −¿ 2.8 z−2 +¿ 2.285 z−3 −¿

0.885 z−4

0 1.147 z−2 −¿ 1.4 z−3 +¿ 0.885 z−4

1.147 z−2−¿ 2.294 z−3 +¿

−4 −5
1.872 z +0.725 z

Therefore C ( z ) = 0.368 z−1 + z−2+ 1.4 z−3+ 1.4 z−4 +1.147 z−5 +0.894 z−6 +0.801 z 7+ 0.868 z −8

Using the below equation we can express it as a difference equation for the discrete time
response:

C ( ) 0.368 z−1 +0.264 z−2

z=
R 1−z−1 +0.632 z −2

C ( z ) ( 1−z−1 +0.632 z −2) = R ( z ) ( 0.368 z−1 +0.264 z−2 )

−z−1 C ( z )+ 0.632 z−2 C ( z ) = 0.368 z−1 R ( z ) +0.26 4 z −2 R ( z )

C ( z )=z −1 C ( z )−0.632 z−2 C ( z ) + 0.368 z−1 +0.26 4 z−2 R ( z )

C ( kT ) =1c ( k −1 ) T −0.632 c ( k−2 ) T + 0.368r ( k −1 ) T +0.264 r ( k −2 ) T

Using the above and assuming zero conditions, the unit step response is determined:

Where k = 0, c (0) = 0, T = 0 and the sample rate, T = 1

k = 1, c (1T) = 1(0) – 0.632 ( 0 ) +0.368 ( 1 ) +0.264 ( 0 )

Unit Step & Zero input as sampling has not started, i.e. the system is off

Therefore - c ( 1T )=0.368

If k =2, then c ( 2T )=1 ( 0.368 )−0.632 ( 0 )+ 0.368 ( 1 ) +0.264 (1)

c ( 2T )=1

If k =3, then c ( 3T )=1 ( 1 )−0.632 ( 0.368 ) +0.368 ( 1 ) +0.264 (1)

c ( 3T )=1 .399

If k =4 , then c ( 4 T )=1 ( 1.399 )−0.632 ( 1 ) +0.368 ( 1 )+ 0.264(1)

c ( 4 T )=1.399

If k =5, then c ( 5T )=1 ( 1.399 )−0.632 ( 1.399 ) +0.368 ( 1 ) +0.264( 1)

c ( 5T )=1.1468

If k =6, then c ( 6 T ) =1 ( 1.1468 ) −0.632 ( 1.399 ) +0.368 ( 1 ) +0.264 (1)

c ( 6 T ) =0.8946

If k =7, then c ( 7 T )=1 ( 0.8946 )−0.632 (1.1468 )+ 0.368 (1 ) +0.264 (1)

c ( 7 T )=0.8018

If k =8, then c ( 8 T ) =1 ( 0.8018 ) −0.632 ( 0.8946 ) +0.368 ( 1 )+ 0.264(1)

c ( 8 T ) =0.8684

Difference Equation Plot for Unit Step Response:

C G ( z)
( z )=
R 1+G ( z ) H ( z )

Determine steady state value of using Final Value Theorem:

C ( ∞ ) =lim
z −∞
( z−1z ) CR ( z) R ( z )
C ( ∞ ) =lim
z−1
( z−1z )( 0.368
2
z +0.264 z−1
z −z+ 0.632 ) ( z )

0.368 z+ 0.264
C (∞)=
( z 2−z +0.632
=1
)
Hence there is no steady state error

Appendix D – MATLAB, Simulink, Desmos & Wolfram Alpha Verification

Task A
Task B
Task C