Transforms Lecture Notes 2020 21 PDF
Transforms Lecture Notes 2020 21 PDF
6 Transfer Functions 49
6.1 Transfer functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
6.2 Poles and zeros . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
6.3 Modelling linear systems by transfer functions . . . . . . . . . . . . . . . . . . . . . . . . . . 53
8 The Z Transform 59
8.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
8.2 Definition of Z transform . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
8.3 Properties of the Z transform . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
8.4 The inverse Z transform . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
8.5 Inverse techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
1
2 CONTENTS
8.6 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
10 Z Transfer Functions 73
10.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
10.2 Residue method for inverse of Z transform . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
10.3 The impulse response . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
10.4 The Response of a single (first-Order) zero or pole . . . . . . . . . . . . . . . . . . . . . . . 77
10.5 Exercises . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
Appendices 97
In this lecture note we study an analytic technique for finding formulas for solutions of certain differential
equations using an operation called the Laplace transform. This technique is particularly effective on
linear, constant coefficient differential equations and is quite different from other methods. The Laplace
transform lets us replace the operations of integration and differentiation with algebraic computations.
As a technique for solving initial value problems, the Laplace transform is sometimes more efficient and
sometimes, however, less efficient than the methods we learned before.
The importance of the Laplace transform is that it is useful in several different types of applications.
For example, the Laplace transform lets us deal efficiently with linear, constant coefficient differential
equations that haves discontinuous forcing functions. These discontinuities include simple jumps that
model the action of a switch. Using the Laplace transforms, we can also make a meaningful mathematical
model of the impulse force provided by a hammer blow or an explosion. The Laplace transform can also
find solutions for a forced harmonic oscillator using solutions generated by a completely different forcing
term. This technique is useful when dealing with a system that is modelled by an unknown linear, constant
coefficient differential equations for which we provide the forcing function.
The strategy of the Laplace transform (or other similar integral transforms) is to transform the difficult
differential equations into simple algebra problems where solutions can be easily obtained. One then applies
the inverse Laplace transform to retrieve the solutions of the original problems. This can be illustrated by
the diagram below as follows:
1.1 Introduction
The Laplace transform is one of many different types of integral transforms. In general, integral transforms
address the question: How much is given function f (t) like a particular standard function? For example,
if f (t) represents a radio signal, we might want to compare it to the function sin ωt, a sine wave with
frequency ω/(2π). By adjusting the parameter ω, we could test how well f (t) fits sine waves with different
frequencies. Ideally, for each value of ω, we would like a number that indicates how much f (t) is like sin ωt.
One way to accomplish this comparison is to compute the integral
Z N
f (t) sin(ωt) dt,
−N
for large N . If f (t) is oscillating with frequency ω/(2π) and is positive when sin ωt is positive, then this
integral is very large. If f (t) has some other frequency, then the signs of f (t) and sin ωt at some time t, so
3
4 CHAPTER 1. THE LAPLACE TRANSFORM
there is cancellation in the integral and its value is smaller. To use this idea in differential equations, it is
natural to compare f (t) to the function that comes up most often, the exponential function. This leads to
the definition of the Laplace transform.
It is an integral transform of f (t) in which, by the defining integral, the variable t is integrated out so that
only s remains. The input to this integral is a function f (t), and the output is a function F (s), called the
Laplace transform of f (t). We adopt the convention that if the input function is f (t), then its Laplace
transform, L[f (t)], will be denoted by F (s). The variable t is typically time, and we speak of f (t) as being
defined in the time domain. Similarly, we speak of F (s) as being defined in the transform domain.
The two different notations, F (s) and L[f (t)], may appear confusing at first. The F (s) notation is
the standard function notation, where the only variable that remains after the right-hand side of (1.2.1) is
integrated will be s. The L[f (t)] is an operator notation, which is used when we want to draw attention to
the function f (t) in the integral. If we change f (t), then the integral will change, which is what the L[f (t)]
notation is pointing out. Sometimes L[f (t)] is written in L[f ] without the variable t.
R∞
• L[0] = 0, since 0
0e−st dt = 0
Actually, the principle use of the Laplace transform is the conversion of differential equations into algebraic
equations, which are then easier to solve. Before giving an example of this process, we first calculate the
following elementary transforms.
Example 1.2.1 The Laplace transform of the function f (t) = 1 is an improper integral that convergences
if s > 0. Thus,
Z ∞ Z ∞
L[f (t)] = L[1] = F (s) = f (t)e−st dt = 1e−st dt
0 0
T −sT
1 e 1
Z
= lim e−st dt = − lim = (1.2.3)
T →∞ 0 s T →∞ s s
since
lim e−sT = 0, when s>0
T →∞
a
Example 1.2.2 The Laplace transform of f (t) = sin at is the function F (s) = s2 +a 2 , again given by an
1
1 s s>0
1
t s2 s>0
2
t2 s3 s>0
n!
tn sn+1 n = 0, 1, 2, · · ·
Γ(a)
ta sa+1 a>0
a
sin at s2 +a2 s>0
s
cos at s2 +a2 s>0
1
e−at s+a a<s
n!
tn e−at (s+a)n+1 s>a
2as
t sin at (s2 +a2 )2
s2 −a2
t cos at (s2 +a2 )2
b
e−at sin(bt) (s+a)2 +b2 a<s
s+a
e−at cos(bt) (s+a)2 +b2 a<s
a
sinh at s2 −a2 a<s
s
cosh at s2 −a2 a<s
δ(t) 1
1
u(t) unit step s s>0
e−sa
u(t − a) s s>0
Example 1.3.1 From Table 1.1 it is easy to find the following results:
6
f (t) = t3 , =⇒ F (s) =
s4
4
f (t) = sin 4t, =⇒ F (s) =
s2 + 16
t s 4s
f (t) = cos , =⇒ F (s) = 2 = 2
2 s + (1/4) 4s + 1
1
f (t) = e−2t , =⇒ F (s) =
s+2
6 CHAPTER 1. THE LAPLACE TRANSFORM
Theorem 1.4.1 (Linearity) If f (t) and g(t) are functions having Laplace transforms and if α and β are
any constants then
As a consequence of this property, we say that the Laplace transform operator is a linear operator.
Solution:
From Table 1.1 we have
1 1
L[1] = , L[sin t] =
s s2 +1
so, by the linearity property,
4 5
L[4 + 5 sin t] = 4L[1] + 5L[sin t] = +
s s2 + 1
The next theorem enables us to start with known transform pairs and derive others.
Theorem 1.4.2 (First shift theorem) If f (t) is a function having Laplace transform F (s), then the
function e−at f (t) also has a Laplace transform given by
This theorem says that the Laplace transform of e−at times a function f (t) is equal to the Laplace transform
of f (t) itself, with s replaced by s + a. It is called as the First shift theorem.
Proof: A proof of the theorem follows directly from the definition of the Laplace transform, since
Z ∞ Z ∞
L[e−at f (t)] = e−at f (t)e−st dt = f (t)e−(s+a)t dt
0 0
then, Z ∞
L[f (t)] = F (s) = f (t)e−st dt
0
we see that the last integral above is in structure exactly the Laplace transform of f (t) itself, except that
s + a takes the place of s, so that
L[e−at f (t)] = F (s + a)
Solution:
From the result in Example 1.2.2 or from Table 1.1, we have
1
L[sin t] =
s2 + 1
so, by the first shift theorem,
L[e−3t sin t] = F (s + 3)
this is,
1 1
L[e−3t sin t] = = 2
(s + 3)2 + 1 s + 6s + 10
In general, we have
a
L[e−kt sin at] = , Re(s) > −k (1.4.6)
(s + k)2 + a2
s+k
L[e−kt cos at] = , Re(s) > −k (1.4.7)
(s + k)2 + a2
where in both cases k and a are real constants.
1.4. PROPERTIES OF THE LAPLACE TRANSFORM 7
Example 1.4.3 Use Theorem 1.4.2 and the known Laplace transforms of 1, t, cos ωt, and sin ωt to find
L[eat ], L[teat ], L[eλt sin ωt], L[eλt cos ωt].
Solution:
In the following table the known transform pairs are listed on the left and the required transform pairs
listed on the right are obtained by applying Theorem 1.4.2.
f (t) ↔ F (s) e−at f (t) ↔ F (s + a)
1 1
1↔ , s>0 e−at ↔ , s>a
s (s + a)
1 1
t↔ , s>0 te−at ↔ , s>a
s2 (s + a)2
ω ω
sin ωt ↔ , s>0 e−λt sin ωt ↔ ,s>λ
s2 + ω 2 (s + λ)2 + ω 2
s s+λ
cos ωt ↔ , s>0 e−λt sin ωt ↔ ,s>λ
s + ω2
2 (s + λ)2 + ω 2
Theorem 1.4.3 (The scale changing theorem) If the Laplace transform of f (t) exists, this is, L[f (t)] =
F (s), then
1 s
L[f (at)] = F , if a > 0 (1.4.8)
a a
The primary use of this theorem is to supplement our list of the Laplace transforms. For example, because
we know that L[cos t] = s/(s2 + 1), we can use this theorem to conclude that
1 (s/a) s
L[cos at] = 2
= 2
a (s/a) + 1 s + a2
Theorem 1.4.4 (The special product theorem) If f (t) is a function having Laplace transform F (s) =
L[f (t)], then the function tn f (t) (n = 1, 2, · · · ) also have Laplace transforms, given by
dn F (s)
L[tn f (t)] = (−1)n
dsn
This theorem says that differentiating the transform of a function with respect to s is equivalent to multi-
plying the function itself by tn .
Example 1.4.4 Determine L[t sin 3t].
Solution:
Using the result in Example 1.2.2,
3
L[sin 3t] = F (s) =
s2 +9
so, by the derivative theorem,
dF (s) 6s
L[t sin 3t] = − = 2 .
ds (s + 9)2
Example 1.4.5 Determine L[tn ], where n is a positive integer.
Solution:
Using the result in Example 1.2.1,
1
L[1] = F (s) =
s
so, by the derivative theorem,
dn 1
n!
L[tn ] = (−1)n n = n+1 .
ds s s
Finally, we summarize the main properties of the Laplace transform we learned so far:
this is
df
L = sF (s) − f (0) (1.5.9)
dt
Solution:
2
Since f (0) = 0, f ′ (0) = 0, f ′′ (t) = 2, and L[2] = 2L[1] = , we obtain from (1.5.10)
s
2 2
L[f ′′ ] = s2 F (s) − sf (0) − f ′ (0) = s2 F (s), since L[2] = = s2 L[f ] ⇒ L[t2 ] =
s s3
in agreement with Table 1.1. This example illustrates that in general there are several ways of obtaining
the transform of given functions.
Example 1.5.2 Let f (t) = t sin at. Find L[f ] by using the Laplace transform of derivatives if given
s
L[cos at] = 2 .
s + a2
Solution:
We have f (0) = 0 and
so that by (1.5.10),
L[f ′′ ] = 2aL[cos at] − a2 L[f ] = s2 L[f ]
Using the formula for the Laplace transform of cos at, we thus obtain
2as
(s2 + a2 )L[f ] = 2aL[cos at] =
(s2 + a2 )
We first begin by taking the Laplace transform of both sides of the differential equation
1
s2 Y (s) − sy(0) − y ′ (0) + Y (s) = (1.6.12)
s
Using the initial conditions, Eq.1.6.12 becomes
1
(s2 + 1)Y (s) − s − 1 = (1.6.13)
s
The unknown in this equation is Y (s) = L[y(t)], the Laplace transform of the unknown function y(t). So,
instead of having to solve a differential equation to determine y(t), we need only to solve an algebraic
equation for Y (s), getting
s2 + s + 1 1 1
Y (s) = = + 2 (1.6.14)
s(s2 + 1) s s +1
where the function on the right are obtained by a partial fraction decomposition. By inspection and
memory, we notice that the first being the transform of 1 and the second the transform of sin t. Hence,
(1.6.14) determines y(t) = 1 + sin t.
That is the basic idea behind using the Laplace transform for solving initial value problems. Clearly,
the more transforms we know, and the more we know the behaviour of the Laplace transform, in general,
the greater the range of initial value problems we can solve with this technique.
Before considering specific examples, there are a few remarks worth noting at this stage.
• As we already noted in section 1.6, a distinct advantage of using the Laplace transform is that it
enable us to replace the operation of differentiation by an algebraic operation. Consequently, by
taking the Laplace transform of each term in a differential equation, it is converted into an algebraic
equation in the variable s. The desired time response is then obtained by taking the inverse Laplace
transform.
• The Laplace transform method yields the complete solution to the linear differential equation, with
the initial conditions automatically included. No determination of a general solution of homogeneous
equation.
• The Laplace transform method is ideally suited for solving initial value problems, that is, linear
differential equations in which all the initial conditions y(0), y ′ (0), and so on, at time t = 0 are
specified. The method is less attractive for boundary value problems.
1.7 Exercises
1. Use the definition (1.2.1) to obtain the Laplace transform of
(a) f (t) = t (b) f (t) = cos t
2. Use the definition (1.2.1) to find the Laplace transform for the piecewise continuous function
0, if t ≤ 2,
f (t) =
1, if t > 2,
3. Determine the Laplace transforms of the following functions by using the table
2t
(a) sin 6t (b) cos 4t (c) sin (d) e−3t
3
1 cos 7t
(e) e3t (f ) (g) t cos 3t (h)
e4t e5t
2 2 s 1
(a) L[2] = (b) L[2t] = (c) L[cos at] = (d) L[eat ] =
s s2 s 2 + a2 s−a
1.7. EXERCISES 11
7. Find he Laplace transform for the following functions (where a and ω are constant)
(a) sin 4t cos 4t (b) cos2 t (c) cos at sinh at (d) cosh2 t
(e) 5e−at sin ωt (f ) e−kt (a cos t + b sin t) (g) e−3t cos πt (h) sin2 t
8. The Laplace transform of y(t) is Y (s). y(0) = 3, y ′ (0) = 1. Find the Laplace transforms of the following
expressions
tet ,
t, 0 ≤ t < 1, 0 ≤ t < 1,
(c) f (t) = (d) f (t) =
1, t ≥ 1. et , t ≥ 1.
10. Suppose f (t) is continuous on [0, T ] and f (t + T ) = f (t) for all t ≥ 0. (We say in this case that f (t) is periodic
with period T ). Show that
Z T
1
F (s) = e−st f (t) dt, s > 0.
1 − e−sT 0
Hint: You may write
X∞ Z (n+1)T
F (s) = e−st f (t) dt.
n=0 nT
Note that
et − e−t et + e−t
sinh t = , cosh t =
2 2
Solution:
3. (a) 6 , (b) 2 s , (c) 6 , (d) 1 , (e) 1 , (f) 1 , (g) s2 −9 , (h) s+5
s2 +36 s +16 9s2 +4 s+3 s−3 s+4 (s2 +9)2 (s+5)2 +49
6 − 4 , (b) 8 −3s
4. (a) + 11 − 12 , (c) 3 + 4 , (d) 1 + 2 − 8 , (e) 2 (a cos b + s sin b) (f) e2
s3 s s2 +16 s s s−2 s2 +1 s s2 s5 s2 +a2 s +1
4(s−3) 3(s−1) 8s−44
5 (a) 48
5 , (b) 2 , (c) 2 , (d) 2 (e) 12
5 (f) 1 − 2 s−1
(s−3) s −6s+34 s −2s−15 s −6s+13 (s+3) s−1 s −2s+5
2 3 s2 −2 , (e) a(s+k)+b
7 (a) 1 2 4 , (b) use cos2 t = 1+cos 2t , (c) as4 −2a4 , (d) 5ω , (f) .
2 s +16 2 s +4a s(s2 −4) (s+a)2 +ω 2 (s+k)2 +1
8. (a) sY − 3, (b) s2 Y − 3s − 1, (c) (s2 + 2s + 3)Y − 3s − 7, (d) (3s2 − s + 2)Y − 9s
12 CHAPTER 1. THE LAPLACE TRANSFORM
Chapter 2
This correspondence between the functions F (s) and f (t) is called the inverse Laplace transformation, f (t)
being the inverse transform of F (s), and L−1 being referred to as the inverse Laplace transform operator.
L−1 [αF (s) + βG(s)] = αf (t) + βg(t) = αL−1 [F (s)] + βL−1 [G(s)]
so that the inverse Laplace transform operator L−1 is also a linear operator.
• The shifting theorem from the previous section can also be useful when looking for inverse Laplace
transforms. We restate it using the current terminology:
13
14 CHAPTER 2. THE INVERSE OF THE LAPLACE TRANSFORM
• The scale changing theorem from the previous section can be useful when we are looking for inverse
Laplace transforms. We restate it using the current terminology:
h s i
if L−1 [F (s)] = f (t), then L−1 F = af (at)
a
3s2 + 6s + 2 1 1 1
= + +
s3 + 3s2 + 2s s s+1 s+2
The inverse Laplace transform of the partial fractions is easily found.
2
−1 3s + 6s + 2 −1 1 −1 1 −1 1
L =L +L +L
s3 + 3s2 + 2s s s+1 s+2
= 1 + e−t + e−2t
5e−3s
Example 2.3.2 Determine the inverse Laplace transform
(s + 2)2
Solution:
Removing the e−3s term from the expression leaves
−1 5
L = 5te−2t
(s + 2)2
Thus the inverse Laplace transform of the original expression is
−1 5e−3s
L = 5(t − 3)e−2(t−3) u(t − 3)
(s + 2)2
Example 2.3.3 Find the inverse Laplace transform of
3
F (s) =
s2 − 2s − 8
16 CHAPTER 2. THE INVERSE OF THE LAPLACE TRANSFORM
Solution:
Since
3 3
F (s) = =
s2 − 2s − 8 (s − 1)2 − 9
Recalling that
3
L[sinh 3t] =
s2 −9
the inverse using the first shift theorem we obtain
3
L−1 = et sinh 3t, where a = −1
(s − 1)2 − 9
1 − s(5 + 3s)
F (s) = . (2.3.2)
s [(s + 1)2 + 1]
Solution:
One form for the partial fraction expansion of F is
A Bs + C
F (s) = + . (2.3.3)
s (s + 1)2 + 1
However, we see from the table of Laplace transforms that the inverse transform of the second fraction on
the right of (2.3.3) will be a linear combination of the inverse transforms
of
s+1 1
and
(s + 1)2 + 1 (s + 1)2 + 1
respectively. Therefore, instead of (2.3.3) we write
A B(s + 1) + C
F (s) = + . (2.3.4)
s (s + 1)2 + 1
2A = 1
A−C = 3
5A + 2B + C = −7
Therefore
1 −1 1 7 s+1 5 −1 1
L−1 (F ) = L − L−1 − L
2 s 2 (s + 1)2 + 1 2 (s + 1)2 + 1
1 7 −t 5
= − e cos t − e−t sin t
2 2 2
2.4. THE CONVOLUTION THEOREM 17
This is an odd looking definition but it turns out to have considerable use both in Laplace transform theory
and in the modelling of linear engineering systems. It should be noted that the variable of integration is
x. As far as the integration process is concerned the t-variable is (temporarily) regarded as a constant.
Example 2.4.1 Find the convolution of 2t and t3 .
Solution:
Theorem 2.4.1 Convolution theorem: Let f (t) and g(t) be piecewise continuous, with L[f (t)] = F (s)
and L[g(t)] = G(s). The convolution theorem allows us to find the inverse Laplace transform of a product
of transforms, F (s)G(s), this is
3
Example 2.4.2 Use the convolution theorem to find the inverse Laplace transform of .
s(s2 + 4)
Solution:
3 1 1
Let F (s) = and G(s) = 2 . Then f (t) = 3 and g(t) = sin 2t. So,
s s +4 2
3
L−1 = L−1 [F (s)G(s)] = (f ∗ g)(t)
s(s2 + 4)
Z t
sin 2v 3 t
Z
= 3 dv = sin 2v dv
0 2 2 0
t
3 − cos 2v
3
= = (1 − cos 2t)
2 2 0 4
1
A useful special case of this theorem is obtained by choosing g(t) = 1 so that G(s) = . We then have the
s
following statements:
Z t
−1 1
if F (s) = L[f (t)], then L F (s) = f (z)dz, (2.4.7)
s 0
or, equivalently, Z t
1
if L[f (t)] = F (s), then L f (z)dz = F (s)
0 s
1
We will now use this theorem to recalculate L−1 .
(s2 + 1)2
1
Example 2.4.3 Find the inverse Laplace transform of .
(s2 + 1)2
Solution:
We apply the theorem to f (t) = g(t) = sin t and use the fact that L−1 [1/(s2 + 1)] = sin t to obtain
Z t
1
L−1 = sin z sin(t − z)dz
(s2 + 1)2 0
1 t
1
Z
L−1 = [cos(2z − t) − cos t]dz
(s2 + 1)2 2 0
1 t
Z t
1
Z
= cos(2z − t)dz − cos t dz (2.4.8)
2 0 2 0
Remember we are integrating with respect to z, so here t plays the role of a constant in this process. The
first integral in the right hand side can be evaluated with the change of variable u = 2z − t, so that
Z t
1 t
Z
cos(2z − t)dz = cos udu = sin t
0 2 −t
The second integral can be evaluated immediately, so we have
−1 1 1
L = (sin t − t cos t)
(s2 + 1)2 2
as anticipated.
1
Whenever we have the term in a Laplace transform, we should consider using the special case of the
s
convolution theorem, namely, (2.4.7). We now give an example of its use.
2.5. THE LAPLACE TRANSFORMS OF INTEGRALS 19
1
Example 2.4.4 Find the inverse Laplace transform of .
s(s − 1)2
Solution:
We seek f (t) for which
1
L[f (t)] =
s(s − 1)2
If we write this as
−1 1 −1 1 1
f (t) = L =L (2.4.9)
s(s − 1)2 s (s − 1)2
1
we recognize that this is in the form of (2.4.7). Because we know that L[tet ] = , we have
(s − 1)2
Z t
1 1
f (t) = L−1 = zez dz
s (s − 1)2 0
Example 2.4.5 Use the convolution theorem to find the inverse transform of
s
H(s) =
(s − 1)(s2 + 1)
Solution:
Begin by choosing two functions of s, that is, F (s) and G(s)
1 s
F (s) = and G(s) =
s−1 s2 + 1
Since, by inspection, we can write down their inverse Laplace transforms
then we have
dg
= f (t), g(0) = 0
dt
20 CHAPTER 2. THE INVERSE OF THE LAPLACE TRANSFORM
Solution:
In this case, f (t) = t3 + sin 2t, giving
6 2
F (s) = L[f (t)] = L[t3 ] + L[sin 2t] = 4
+ 2
s s +4
so, by (2.5.10), Z t
3 1 6 2
L (τ + sin 2τ )dτ = F (s) = 5 + 2
0 s s s(s + 4)
1
Example 2.5.2 Let L[f (t)] = . Find f (t).
s(s2 + a2 )
Solution:
From Table 1.1, we have
−1 1 1
L 2 2
= sin at
s +a a
From this and equation (2.5.11) we obtain the answer
1 t
1 1
Z
−1
L = sin aτ dτ = 2 (1 − cos at)
s(s2 + a2 ) a 0 a
(a) Since Z ∞
a sin at a π s
L[sin at] = 2 ⇒ L = du = − tan−1
s + a2 t 0
2
u +a 2 2 a
2.7. EXERCISES 21
(b) Let
s+a
G(s) = ln
s+b
b−a 1 1
G′ (s) = = +
(s + a)(s + b) s+a s+b
from which we see that
−1 −1 ′ e−bt − e−at
−1 −1 s+a
L [G(s)] = L ln = L [G (s)] =
s+b t t
2.7 Exercises
1. Find the inverse Laplace transforms of the following functions:
8 2 1 5
(a) (b) (c) (d)
s s3 s+2 s−3
2 2 4 1 3
(e) (f ) (g) − 3 (h)
(s + 1)2 s2+4 s s 2s + 1
5s 3 7 5
(i) (j) (k) (l)
2s2 + 18 (s − 3)5 s2 − 16 7s2 + 6
3 2(s + 1) 5 4s − 3
(a) (b) (c) (d)
s2 − 4s + 13 s2 + 2s + 10 s2 + 2s − 3 s2 − 4s − 5
s+1 3 2(s − 3) 3s + 2
(e) (f ) (g) (h)
s2 + 2s + 10 s2 + 6s + 13 s2 − 6s + 13 s2 − 8s + 25
3. Express the following as partial fractions and hence find the inverse Laplace transforms
5s + 2 3s + 4 3s + 3 2s + 5
(a) (b) (c) (d)
(s + 1)(s + 2) (s + 2)(s + 3) (s − 1)(s + 2) s+2
4s − 5 3s3 + s2 + 12s + 2 5s2 − 2s − 19 5s2 + 8s − 1
(e) (f ) (g) (h)
s2 − s − 2 (s − 3)(s + 1)3 (s + 3)(s − 1)2 (s + 3)(s2 + 1)
5. Use the convolution theorem to compute the inverse Laplace transform of the function, even if another
method would work
1 s2 2
(a) F (s) = (b) F (s) = (c) F (s) =
(s + 4)(s2 − 4)
2 (s − 3)(s2 + 5) s3 (s2
+ 5)
1 1 1
(d) F (s) = e−4s (e) F (s) = (f ) F (s) = 2
s(s + 1) (s − 2)2 (s + 3)2 s (s + 4)
Solution:
2 −1t
1. (a) 8, (b) t2 , (c) e−2t , (d) 5e3t , (e) 2te−t , (f) sin 2t, (g) 4 − t , (h) 3 e 2 , (i) 5 cos 3t, (j) 1 e3t t4 , (k) 7 sinh 4t.
2 2 2 8 4
2. (a) e2t sin 3t, (b) 2e−t cos 3t, (c) 5 e−t sinh 2t, (d) 4e2t cosh 3t + 5 e2t sinh 3t, (e) e−t cos 3t, (f) 3 e−3t sin 2t, (g) 2e3t cos 2t, (h) 3e4t cos 3t +
2 3 2
14 e4t sin 3t.
3
3. (a) 8e−2t − 3e−t , (b) 5e−3t − 2e−2t , (c) 2et + e−2t , (d) 2δ(t) + e−2t , (e) e2t + 3e−t , (f) 2e3t + e−t − 4te−t + 3 t2 e−t , (g) 2e−3t + 3et − 4tet ,
2
(h) 2e−3t + 3 cos t − sin t.
√ √ √ √
4 (a) e−4t cos 2t+ 1 e−4t sin 2t, (b) 4/ 7e3t sinh( 7t), (c) e−2t (cos 4t− 7 sin 4t), (d) 2 e−t + 1 te−t − 2 cos 2t− 3 sin 2t (e) 3.5t2 et , (f) 2t2 e−t 2
2 4 25
√ 5 25 50
−5πt 1 2t −2t 1 3t √ 3 5 √ 2 √ t 2 2 4−t
(g) e sinh πt. (5) (a) (e −e − 2 sin(2t)), (b) (e + 5 cos( 5t)) + sin( 5t), (c) cos( 5t) + − , (d) −(e − 1)u(t − 4),
32 14 14 25 5 25
−4t
(e) 2 (e−3t − e2t ) + t (e2t + e−3t ), (f) t + e − 1
125 25 4 16 16
Chapter 3
3.1 Introduction
So far we have seen how to find the Laplace transform of a function of time and how to find the inverse
Laplace transform. We now apply this to finding the particular solution of differential equations. As we
mentioned before, the initial conditions are automatically satisfied when solving an equation using the
Laplace transform. They are contained in the transform of the derivative terms.
In this section we employ the Laplace transform to solve constant coefficient ordinary differential equa-
tions. In particular we shall consider initial value problems. We shall find that the initial conditions are au-
tomatically included as part of the solution process. The idea is very simple here. We first take the Laplace
transform of each term in the differential equation, then, if the unknown function is y(t), on taking the
transform, we transform the differential equation into an algebraic equations involving Y (s) = L[y(t)] only.
This equation is solved for Y (s) which is then inverted to produce the required solution y(t) = L−1 {Y (s)}.
23
24CHAPTER 3. THE SOLUTION OF DIFFERENTIAL EQUATIONS USING LAPLACE TRANSFORMS
Example 3.2.2 Use the Laplace transform to solve the initial value problem
y ′′ − 6y ′ + 5y = 3e2t , y(0) = 2, y ′ (0) = 3. (3.2.1)
Solution:
Taking Laplace transforms of both sides of the differential equation in (3.2.1) yields
3
L(y ′′ − 6y ′ + 5y) = L 3e2t =
,
s−2
which we rewrite as
3
L(y ′′ ) − 6L(y ′ ) + 5L(y) = . (3.2.2)
s−2
Now denote L(y) = Y (s). Using the initial conditions in (3.2.1) we have
L(y ′ ) = sY (s) − y(0) = sY (s) − 2
and
L(y ′′ ) = s2 Y (s) − y ′ (0) − sy(0) = s2 Y (s) − 3 − 2s
Substituting from the last two equations into (3.2.2) yields
3
s2 Y (s) − 3 − 2s − 6 (sY (s) − 2) + 5Y (s) =
s−2
Therefore
3
(s2 − 6s + 5)Y (s) = + (3 + 2s) + 6(−2) (3.2.3)
s−2
so
3 + (s − 2)(2s − 9)
(s − 5)(s − 1)Y (s) = ,
s−2
and
3 + (s − 2)(2s − 9)
Y (s) = .
(s − 2)(s − 5)(s − 1)
Using partial fraction method, it yields
1 1 1 5 1
Y (s) = − + + ,
s−2 2s−5 2s−1
and taking the inverse transform of this yields
1 5
y = −e2t + e5t + et
2 2
as the solution of (3.2.1).
It is not necessary to write all the steps that we used to obtain (3.2.3). To see how to avoid this, let us
apply the method of Example 3.2.2 to the general initial value problem
ay ′′ + by ′ + cy = f (t), y(0) = k0 , y ′ (0) = k1 (3.2.4)
Taking Laplace transforms of both sides of the differential equation in (3.2.4) yields
aL(y ′′ ) + bL(y ′ ) + cL(y) = F (s). (3.2.5)
Now let Y (s) = L(y). Using the initial conditions in (3.2.4), we obtain that
L(y ′ ) = sY (s) − k0 and L(y ′′ ) = s2 Y (s) − k1 − k0 s.
Substituting these into (3.2.5) yields
a s2 Y (s) − k1 − k0 s + b (sY (s) − k0 ) + cY (s) = F (s).
(3.2.6)
The coefficient of Y (s) on the left is the characteristic polynomial
p(s) = as2 + bs + c
of the complementary equation for (3.2.4). Using this and moving the terms involving k0 and k1 to the
right side of (3.2.6) yields
p(s)Y (s) = F (s) + a(k1 + k0 s) + bk0 . (3.2.7)
This equation corresponds to (3.2.3) of Example 3.2.2. Having established the form of this equation in the
general case, it is preferable to go directly from the initial value problem to this equation. You may find it
easier to remember (3.2.7) rewritten as
p(s)Y (s) = F (s) + a (y ′ (0) + sy(0)) + by(0). (3.2.8)
3.2. SOLVING ODES BY THE LAPLACE TRANSFORMS 25
Example 3.2.3 Use the Laplace transform to solve the initial value problem
Solution:
The characteristic polynomial is
and
8
L(8e−2t ) = ,
s+2
so (3.2.8) becomes
8
(2s + 1)(s + 1)Y (s) = + 2(2 − 4s) + 3(−4).
s+2
Solving for Y (s) yields
4 (1 − (s + 2)(s + 1))
Y (s) = .
(s + 1/2)(s + 1)(s + 2)
Using partial fraction expansion, it yields
4 1 8 8 1
Y (s) = − + ,
3 s + 1/2 s + 1 3 s + 2
t
1 2 3 4 5 6 7
−1
−2
−3
−4
Solution:
The characteristic polynomial is
p(s) = s2 + 2s + 2 = (s + 1)2 + 1
and
1
L(1) = ,
s
so (3.2.8) becomes
1
(s + 1)2 + 1 Y (s) = + 1 · (1 − 3s) + 2(−3).
s
26CHAPTER 3. THE SOLUTION OF DIFFERENTIAL EQUATIONS USING LAPLACE TRANSFORMS
.5
t
1 2 3 4 5 6 7
−1
−2
−3
−4
1
Figure 3.2: A plot of y = 2 − 27 e−t cos t − 52 e−t sin t
Example 3.2.5 (RL circuit with ramp input) Use the Laplace transform to solve
di
iR + H = t, i(0) = 0, t≥0
dt
Solution:
Let L[i(t)] = I(s) and take the Laplace transform of both sides of the equation
1
I(s)R + H(sI(s) − i(0)) =
s2
Note that i(0) = 0 and so
1
I(s)(R + Hs) =
s2
1 A B C
I(s) = 2 = + 2+
s (R + Hs) s s R + Hs
Evaluating the constants A, B and C by partial fractions gives
H 1 H2 −H 1 H
I(s) = − 2
+ 2
+ 2 = 2 + + 2 R
R s Rs2
R s Rs R (R + Hs) R H +s
x′ = x − 2y,
y ′ = 5x − y,
Solution:
If we apply the Laplace transform to both sides of these equations, we obtain
L[x′ ] = L[x − 2y]
L[y ′ ] = L[5x − y]
Using the initial condition, we transform this system into
sX(s) + 1 = X(s) − 2Y (s)
(3.3.11)
sY (s) − 2 = 5X(s) − Y (s)
where X(s) = L[x(t)] and Y (s) = L[y(t)]. We may solve for Y (s) from the top equation, Y (s) = (1/2)[(1 −
s)X(s) − 1], and substitute the result into the bottom equation to find
s+5
X(s) = − 2 (3.3.12)
s +9
If we rewrite this expression as
s 5 3
X(s) = − −
s2 + 9 3 s2 + 9
we see that x(t) is
5
x(t) = − cos 3t − sin 3t (3.3.13)
3
We have two choices for determining y(t). One is to substitute the expression for X(s) form (3.3.12) into
the equation in (3.3.11) to obtain an expression for Y (s), and then find its inverse Laplace transform.
However, here it is easier simply to use the explicit solution for x(t) from (3.3.13) in the first of our original
differential equations in (3.3.11) to obtain
1 ′ 1 5
y(t) = (x − x ) = − cos 3t − sin 3t − 3 sin 3t + 5 cos 3t
2 2 3
or
7
y(t) = 2 cos 3t −
sin 3t
3
Coupled electrical circuits and mechanical vibrating systems involving several masses in springs offer ex-
amples of engineering systems modelled by systems of differential equations.
Example 3.3.2 Electrical circuit
Consider the RL (resistance/inductance) circuit with a voltage v(t) applied as shown in Fig.3.3. If i1 and
i2 denote the currents in each loop we obtain, using Kirchhoff ’s voltage law:
1. in the right loop
di1
L1 + R2 (i1 − i2 ) + R1 i1 = v(t)
dt
2. in the left loop
di2
L2 + R2 (i2 − i1 ) = 0
dt
Suppose, in the circuit, that L1 = 0.8 henry, L2 = 1 henry, R1 = 1.4Ω, R2 = 1Ω . Assume zero initial
conditions: i1 (0) = i2 (0) = 0. Suppose that the applied voltage is constant: v(t) = 100 volts (t ≥ 0). Solve
the problem by Laplace transforms.
Solution:
Begin by obtaining V (s), the Laplace transform of v(t). We have, from the definition of the Laplace
transform: Z ∞ −st ∞
−st e 100
V (s) = 100e dt = 100 =
0 −s 0 s
This is simply the Laplace transform of the step function of height 100.
Now insert the parameter values into the differential equations and obtain the Laplace transform of each
equation. Denote by I1 (s), I2 (s) the Laplace transforms of the unknown currents. (These are equivalent
to X(s) and Y (s) which we used before.
di1
0.8 + i1 − i2 + 1.4i1 = v(t)
dt
di2
+ i2 − i1 = 0
dt
28CHAPTER 3. THE SOLUTION OF DIFFERENTIAL EQUATIONS USING LAPLACE TRANSFORMS
Now solve these equations for I1 (s) and I2 (s). Put each expression into partial fractions and finally take
the inverse Laplace transform to obtain i1 (t) and i2 (t). We use partial fractions to get
3.4 Exercises
1. Use Laplace transforms to solve the following simple equations
(a) x′ + x = 3, x(0) = 1
(b) x′′ + x = 2t, x(0) = 0, x′ (0) = 5
(c) x′′ + x′ − 2x = 1 − 2t, x(0) = 6, x′ (0) = −11
(d) 2x′′ + x′ − x = 27 cos 2t + 6 sin 2t, x(0) = −1, x′ (0) = −2
3. Solve the equations (you may use the convolution theorem to solve these equations)
Solution:
1. (a) e−3t (3 cos 2t + 8 sin 2t, (b) y(t) = e3t − 3t − 1, (c) y(t) = (4t − 1)et + 1 e4t , (d) y(t) = 3et (cos t + sin t) − et cos 2t.
3
2. (a) x(t) = 3 − 2e−t , (b) 3 sin t + 2t, (c) t + 6e−2t , (d) −3 cos 2t + 2e−t .
4. (a) y1 = et + 1 e−t − 1 cos t − 1 sin t, y2 = −et + 1 e−t + 1 cos t + 1 sin t, (b) y1 = 1 e−t sin 2t, y2 = e−t (cos 2t − 1 sin 2t), (c) x(t) =
2 2 2 2 2 2 2 2
1 − 2et + cos t − 2√ sin t, y(t) = 2et − sin t √ − 2 cos t, (d) x = cos t, y = − cos t − sin t, (e) x(t) = et − t − 1, y(t) = 2t − 3 + 3et , (f)
x(t) = 3 cos t + cos( 3t), y(t) = 3 cos t − cos( 3t).
30CHAPTER 3. THE SOLUTION OF DIFFERENTIAL EQUATIONS USING LAPLACE TRANSFORMS
Chapter 4
15
10
2 4 6 8 10
31
32 CHAPTER 4. THE SYSTEM WITH THE STEP FUNCTION U (T )
1.2
1.0
0.8
0.6
0.4
0.2
-10 -5 5 10
Solution:
The Laplace transform of u(t) is
∞ N
1
Z
−st
L[u(t)] = e dt = lim − e−st
0 N →∞ s 0
• If we are given a function f (t), then the function f (t)u(t) is 0 for t < 0 and the original function f (t)
for t ≥ 0. The graph of tu(t) is shown in Fig.4.3.
• Because u(t) has its jump at t = 0, the function u(t − a) has its jump at t = a. (Why?) The graph
of u(t − 3) is shown in Fig.4.4.
• If we are given a function f (t), then the function f (t)u(t − a) is 0 for t < a and the original function
f (t) for t ≥ a. Thus, the function E(t) in (4.1.1) can be written in terms of the unit step function
by E(t) = 12u(t − 2).
• If we are given a function f (t), then the function f (t − a)u(t − a) is the original function f (t) shifted
to the right by a units, if a > 0. If a < 0, the function is shifted to the left by a unit. The graph of
(t − 1)u(t − 1) is shown in Fig.4.5.
Solution:
To sketch the graph of u(t − 1) − u(t − 3), we first recognize that there are two possible jumps, at t = 1
and t = 3. Thus, we should analyze u(t − 1) − u(t − 3) in three regions t < 1, 1 < t < 3 and 3 < t.
4.2. UNIT STEP FUNCTION U (T ) 33
Figure 4.4: The graph of the unit step function u(t − 3).
10
-5 5 10 15
Figure 4.5: The graph of the unit step function (t − 5)u(t − 5).
If t < 1, then both u(t − 1) and u(t − 3) contribute 0, so the function is 0 here. At t = 1 the contribution
from u(t − 1) is 1. If 1 < t < 3, then u(t − 1) = 1, but u(t − 3) = 0, so their difference is 1. If 3 ≤ t, both
u(t − 1) and u(t − 3) contribute 1, their difference is 0. Combining this information, we find
0 if t < 1
u(t − 1) − u(t − 3) = 1 if 1 ≤ t < 3 (4.2.2)
0 if t ≥ 3
The graph of such a function is shown in Fig.4.6. Physically, this might correspond to a pulse of magnitude
1 lasting for 2 time units, starting at t = 1 and ending at t = 3.
1.2
1.0
0.8
0.6
0.4
0.2
-4 -2 2 4
Figure 4.6: The graph of the unit step function u(t − 1) − u(t − 3).
Theorem 4.2.1 (The second shift theorem) If L[f (t)] = F (s) then for a positive constant a
Proof: By definition
Z ∞
L[f (t − a)u(t − a)] = f (t − a)u(t − a)e−st dt
Z0 ∞
= f (t − a)e−st dt
a
34 CHAPTER 4. THE SYSTEM WITH THE STEP FUNCTION U (T )
It should be remembered that in all of problems involving the second shift theorem the function MUST
be in the form
u(t − a)f (t − a)
before taking the transforms. If it is not in that form we must have to put it into the right form! In order
to illustrate the use of this form, we present the following examples.
Example 4.2.3 Find the Laplace transform of each of the following functions
1. g(t) = 2(t − 6)3 u(t − 6) − (8 − e12−3t )u(t − 4)
2. g(t) = −t2 u(t − 3) + cos(t)u(t − 5)
Solution:
1. There are two terms in this function. The first is simply a step function so we can use (4.2.3) on this
term
f (t) = 2t3 ⇒ f (t − 6) = 2(t − 6)3
and this has been shifted by the correct amount.The second term uses the following function
2. There are two terms and neither has been shifted by the proper amount. The first term needs to be
shifted by 3 and the second needs to be shifted by 5. So, we need to add in the shifts, and then take
them back to the original forms. Here they are
(a + b)2 = a2 + 2ab + b2
The second term can be dealt with in one of two ways. The first would be to use the formula
to break it up into cosines and sines with arguments of t − 5 which can be shifted as expected.
However, there is an easier way to do this one. From the table of Laplace transforms we have
s cos θ − ω sin θ
L[cos(ωt + θ)] =
s2 + ω 2
and using that we can see that if
It seems the two functions which have been shifted here are
f (t) = −(t2 + 6t + 9)
f (t) = cos(t + 5)
s cos 5 − sin 5
2 6 9 −3s
G(s) = − 3
+ 2+ e + e−5s
s s s s2 + 1
Note that sin 5 and cos 5 are just two constants here.
Solution:
The first thing that we need to do is to write the function in terms of the step functions.
We had to add in a −8 in the second term since that appears in the second part and we also had to subtract
a t in the second term since the t in the first portion is no longer there. This subtraction of the t adds a
problem because the second function is no longer correctly shifted. However, this is easy to fix. Here is
the correct function
f (t) = t + (−8 − (t − 6 + 6) + (t − 6)2 )u(t − 6)
= t + (−8 − (t − 6) − 6 + (t − 6)2 )u(t − 6)
= t + (−14 − (t − 6) + (t − 6)2 )u(t − 6)
g(t) = t2 − t − 14
On the other hand, f (t − a)u(t − a) represents a translation of the function f (t) by a unit to the right
(since a > 0), so that
0 t<a
f (t − a)u(t − a) =
f (t − a) t≥a
In fact, f (t − a)u(t − a) may be interpreted as representing the function f (t) delayed in time by a units.
Thus, when considering its Laplace transform e−as F (s), where F (s) denotes the Laplace transform f (t),
the component e−as may be interpreted as a delay operator on the transform F (s), indicating that the
response of the system characterized by F (s) will be delayed in time a units. Since many practically
important systems have some form of delay inherent in their behaviour, it is clear that the result of this
theorem is very useful.
In order to illustrate the use of these results, we consider an electrical circuit that has a 12-volt battery
connected in series with a resistor and a capacitor. At t = 0, the battery is bypassed, allowing the capacitor
to begin discharging. Two seconds later the battery is inserted back into the circuit as shown in Fig.4.7.
We want to discover how the charge, q, on the capacitor behaves under this situation.
Solution:
In this example we will take R = 10, C = 1/10 and q(0) = 12C, so we want to solve
dq 0 0≤t<2
+q =
dt 1.2 t≥2
The graph of (4.2.7) is shown in Fig.4.8. (Does its behaviour agree with what you anticipated?) We shall
consider the following electrical network question before we finish this section.
4.2. UNIT STEP FUNCTION U (T ) 37
1.5
1.0
0.5
2 4 6 8 10 12 14
Example 4.2.6 Find the currents i1 (t) and i2 (t) in the network in Fig. 4.9 with L and R measured in
terms of the usual units. v(t) = 100 volts if 0 ≤ t ≤ 0.5 sec and 0 thereafter, and i(0) = 0, i′ (0) = 0.
Solution:
The model of the network is obtained from Kirchhoff’s voltage law.
1
0.8i′1 + 1(i1 − i2 ) + 1.4i1 = 100[1 − u(t − )]
2
1 · i′2 + 1(i2 − i1 ) = 0.
Division by 0.8 and ordering gives
1
i′1 + 3i1 − 1.25i2 = 125[1 − u(t − )]
2
i′2 − i1 + i2 = 0.
With i1 (0) = i2 (0) = 0 we obtain from the Laplace transform (the second shifting theorem)
1 e−s/2
(s + 3)I1 − 1.25I2 = 125 −
s s
−I1 + (s + 1)I2 = 0
Solving algebraically for I1 and I2 gives
125(s + 1)
I1 = (1 − e−s/2 )
s(s + 12 )(s + 27 )
125
I2 = (1 − e−s/2 )
s(s + 12 )(s + 27 )
The right sides without the factor (1 − e−s/2 ) have the partial fraction expansions
500 125 625
− −
7s 3(s + 21 ) 21(s + 27 )
38 CHAPTER 4. THE SYSTEM WITH THE STEP FUNCTION U (T )
125 625
i1 (t) = − (1 − e1/4 )e−t/2 − (1 − e7/4 )e−7t/2
3 21
250 250
i2 (t) = − (1 − e1/4 )e−t/2 + (1 − e7/4 )e−7t/2
3 21
Can you explain physically why both current eventually go to zero, and why i1 (t) has a sharp cusp whereas
i2 (t) has a continuous tangent direction at t = 21 ?
4.3. EXERCISES 39
4.3 Exercises
1. Find the Laplace transform of each function
3 3π 3π
(a) f (t) = u(t − 3)(t − 3) (b) f (t) = u(t − 1) sin(t − 1) (c) f (t) = u t − sin 2 t −
3 2
3π
(d) f (t) = 2t2 u(t − 5) (e) f (t) = u(t − 4)(7 − e12−3t ) (f ) f (t) = u t − sin 2t
3
3π t
(g) f (t) = et u(t − 3) (h) f (t) = u(t − 4)(7 − e5t ) (i) f (t) = u t − cos
3 2
t2
0≤t<2 cos t 0 ≤ t < 2π
(c) f (t) = (d) f (t) =
1 − t − 3t2 t≥2 2 − sin t t ≥ 2π
d2 x dx
+2 + 10x = u(t) − u(t − 5)
dt2 dt
where the right-hand side may represent the impulse being switched on at t = 0 and switched off at
t = 5.
8. (Network) Set up the model of the network in Fig.4.10 and find the currents, assuming that the
currents and the charge on the capacitor are zero when the switch is closed at t = 0.
40 CHAPTER 4. THE SYSTEM WITH THE STEP FUNCTION U (T )
(a)
y ′′ + y = u(t − 2π), y(0) = 10, y ′ (0) = 0
(b)
y ′′ + 3y ′ + 2y = e−t + 5u(t − 2), y(0) = 0, y ′ (0) = 1
(c)
y ′′ + 5y ′ + 6y = (1 − u(t − 10))et − e10 u(t − 10), y(0) = 0, y ′ (0) = 1
(d)
y ′′ + y = −2 sin t + 10u(t − π), y(0) = 0, y ′ (0) = 1
(e)
y ′′ + 3y ′ − 4y = 2et − 8e2 u(t − 2), y(0) = 2, y ′ (0) = 0
(f)
y ′′ + 6y ′ + 8y = 2u(t − 1) + 2u(t − 2), y(0) = 1, y ′ (0) = 0
(g)
0 0≤t<4
y ′′ − 2y ′ − 3y = f (t), y(0) = 1, y ′ (0) = 0, where f (t) =
12 t≥4
(h) 2
t 0≤t<3
y ′′ + 5y ′ + 6y = g(t), y(0) = 0, y ′ (0) = −4, where g(t) =
0 t≥3
(i)
0 0≤t<5
y ′′ + 6y ′ − 7y = f (t), y(0) = −2, y ′ (0) = 0, where f (t) =
2 t≥5
10. The response y(t) of a system to a forcing function yi (t) is determined by the second order differential
equation
y ′′ + 6y ′ + 10y = yi (t), (t ≥ 0)
Suppose that yi (t) is a constant stimulus applied for a limited period and characterized by
3 (0 ≤ t < a)
yi (t) =
0 (t ≥ a)
determine the response of the system at time t given that the system was initially in a quiescent
state. Show that the transient response at time T (> a) is
3 −3T
y(T ) = − e {cos T + 3 sin T − e3a [cos(T − a) + 3 sin(T − a)]}
10
Solution:
5a. y = 3(1 − cos t) − 3u(t − π)(1 + cos t) 5b. y = 3 − 2 cos t + 2u(t − 4) (t − 4 − sin(t − 4))
u(t−1) 2(t−1)
5c. y = − 15 + 3 e2t − 2t + (e − 2t + 1)
2 2 2
5d. y = 1 et + 13 e−t + 1 e2t + u(t − 2) −1 + 1 et−2 + 1 e−(t−2) + 1 et+2 − 1 e−(t−6) − 1 e2t
2 6 3 2 2 2 6 3
6. X(s) = 1 (1 − e −5s ), x(t) = 1 − 1 e −t cos 3t − 1 e −t sin 3t − 1 u(t − 5) + 1 e −(t−5) cos 3(t − 5)u(t − 5) + 1 e−(t−5) sin 3(t − 5)u(t − 5)
s(s2 +2s+10) 10 10 30 10 10 30
7. y(t) = u(t − 1)(t − 1)e 2t−2 + u(t − 2)(2 − t)e 2t−4 8. i1 + 10(i1 − i2 ) = 100t , 30i2 + 10(i2 − i1 ) + 100i2 = 0. i1 = ( 4 + 4t)e−5t + 10t2 − 4 , i2 =
′ 2 ′ ′ ′
5 5
( 4 + 2t)e−5t + 2t − 4
5 5
Chapter 5
L[δ(t)] = F (s) = 1
It was introduced by theoretical physicist Paul Dirac. In the context of signal processing it is often referred
to as the unit impulse signal (or function). Its discrete analog is the Kronecker delta function which is
usually defined on a finite domain and takes values 0 and 1 only.
The δ(t) is not a function in the traditional sense as no function defined on the real numbers has these
properties. The Dirac delta function can be rigorously defined either as a distribution or as a measure,
whose complete description would take considerably more time than we can afford here. Distribution is not
a point valued function but rather a functional. In mathematics, and particularly in functional analysis,
a functional is a map from a vector space into its underlying scalar field, or a set of functions to the real
numbers.
The delta function is sometimes thought of as an infinitely high, infinitely thin spike at the origin,
with total area one under the spike, and physically represents an idealized point mass or point charge.
41
42 CHAPTER 5. THE SYSTEM WITH THE δ(T ) FUNCTION
Therefore, we can use this property of the δ-function to model a sudden sharp force acting on a physical
system. Impulse forcing is the term used to describe a very quick push or pull on a system, such as a blow
with a hammer or the effect of an explosion. For example, suppose we have an unforced oscillator that
satisfies equation
d2 y dy
2
+2 + 26y = 0
dt dt
This equation models a unit mass attached to a spring with spring constant 26 sliding on a table with
damping coefficient 2. Now suppose we strike the mass with a hammer once at time t = 4 and shown in
Fig.5.1. We can write the forced equation as
d2 y dy
2
+2 + 26y = g(t)
dt dt
where
very big force when t = 4
g(t) =
0 6 4
when t =
To make this precise enough to have any hope of obtaining a solution, we need some sort of formula
for g(t). As a first attempt to derive such a formula, we might assume that the hammer delivers a large
constant force over a short time interval. More precisely, we assume that the hammer is in contact with
the mass during the time interval from 4 − ∆t to 4 + ∆t for some time ∆t, and that force on the mass
during this interval is the constant k. Then the forcing function becomes
k, 4 − ∆t ≤ t ≤ 4 + ∆t
g∆t (t) =
0, otherwise
Here we think of both ∆t and k as parameters. The parameters k and ∆t are closely related. The external
force is only nonzero for a time interval of length 2∆t. Hence, the smaller we take ∆t, the larger we should
take k in order to deliver the same total push.
We usually choose k = 1/(2∆t) so that k becomes large as ∆t → 0. Therefore
1
2∆t , 4 − ∆t ≤ t ≤ 4 + ∆t
g∆t (t) =
0 otherwise
With this choice of k, the area under the graph of g∆t (t) is the same the area of a rectangle with base 2∆t
and height 1/(2∆t). That is, the area is 1 no matter what we choose for ∆t as shown in Fig.5.2. Since we
0
-1 0 1 2 3 4 5 6
are modelling a hammer strike that is delivered very quickly, we would like to make the time interval ∆t
as small as possible. So we consider what happens as ∆t → 0. Informally, as ∆t → 0, we are compressing
the same amount of force into a shorter and shorter time interval.
An instantaneous force would be represented by the limit as ∆t → 0. But lim∆t→0 g∆t (t) is a confusing
object. For any t 6= 4, the limit of g∆t (t) as ∆t → 0 is 0 because for small ∆t, the time t is outside the
interval on which g∆t (t) is positive. On the other hand, lim∆t→0 g∆t (4) does not exist because g∆t (4) → ∞
as ∆t → 0. The function δ4 (t) = lim∆t→0 g∆t (t) is called the Dirac delta function. This function is zero
for all values of t except t = 4, where it is infinite.
Now we turn to the Laplace transform of the function g∆t (t). For any ∆t > 0 we compute
∞ 4+∆t
1 −st
Z Z
−st
L[g∆t (t)] = g∆t (t)e e dt
dt =
0 4−∆t 2∆t
− e−s(4−∆t)
−4s −s(4+∆t)
e e
= −
s 2∆t
− e−s(4−∆t)
−4s −s(4+∆t)
e e
lim L[g∆t (t)] = lim −
∆t→0 ∆t→0 s 2∆t
2s∆t
−1
−4s
e e
= lim e−s∆t
s ∆t→0 2∆t
= e−4s
Hence, we define the Laplace transform of δ4 (t) = δ(t − 4) as L[δ4 (t)] = L[δ(t − 4)] = e−4s . We could just
as easily put the impulse force at time t = a, obtaining a Dirac delta function δa (t) = δ(t − a) that is zero
for t 6= a and ∞ for t = a. We have the following definition:
Definition 5.2.1 The Laplace transform of δa (t) = δ(t − a) is L[δa (t)] = L[δ(t − a)] = e−as . In particular,
if a = 0, then L[δ0 (t)] = L[δ(t)] = 1.
We use δa (t) to model the force delivered by a tap of a hammer at time t = a. Using the Dirac delta
function, we can rewrite the initial value problem as
d2 y dy
2
+2 + 26y = δ4 (t), y(0) = 1, y ′ (0) = −1
dt dt
Taking the Laplace transform of both sides of the equation, we obtain
2
d y dy
L 2
+ 2L + 26L[y] = L[δ4 (t)].
dt dt
Using the formulas for the Laplace transform of a derivative and second derivative and solving for L[y], we
find
(s2 Y (s) − sy(0) − y ′ (0)) + (2sY (s) − 2y(0)) + 26Y (s) = L[δ4 ]
Substituting the initial conditions and evaluating L[δ4 ], we have
0.5
2 4 6 8 10
-0.5
-1.0
Figure 5.3: The plot of y(t) = e−t cos 5t + 15 u(t − 4)e−(t−4) sin(5(t − 4)).
As expected, the solution has a jump discontinuity at t = 4, just as the impulse force is applied. This is
clearly shown in Fig.5.3.
An importance property of the delta function that is of practical significance is the so-called sifting
property, which states that if f (t) is continuous at t = a then
Z ∞
f (t)δ(t − a)dt = f (a) (5.2.4)
−∞
This is referred to as the shifting property because it provides a method of isolating, or shifting out, the
value of a function at any particular point.
By the definition of the Laplace transform, we have for any a > 0
Z ∞
L[δ(t − a)] = δ(t − a)e−st dt (5.2.5)
0
We can show that equations (5.2.6) and (5.2.7) hold for a = 0, this gives
Z ∞
L[δ(t)] = δ(t)e−st dt = e−s0 = 1
0
so that
L[δ(t)] = 1, (5.2.8)
or, in inverse form,
L−1 [1] = δ(t) (5.2.9)
s2
Example 5.2.1 Determine L−1 .
s2 + 4
Solution:
Since
s2 s2 + 4 − 4 4
= =1− 2
s2 +4 s2 + 4 s +4
we have
s2
−1 −1 −1 4
L = L [1] − L
s2 + 4 s2 + 4
giving
s2
−1
L = δ(t) − 2 sin 2t
s2 + 4
5.2. THE δ(T ) FUNCTION 45
The δ(t) function may be used to study the behaviour of a circuit that has been subjected to transients.
These are generated during switching, and the high input voltages associated with them can create excessive
current in the components, damaging the circuit. Transients can also be harmful because they contain a
broad spectrum of frequencies. Introducing a transient into a circuit can therefore have the effect of forcing
the circuit with a range of frequencies. If one of these in near the natural frequency of the system, resonance
may occur, resulting in oscillations large enough to damage the system. For this reason, before a circuit is
built, engineers sometimes use a delta function to model a transient and study its effect on the circuit.
Example 5.2.2 Suppose, in the circuit of Fig.5.4, the current and charge on the capacitor are zero at
time zero. We want to determine the output voltage response to a transient modeled by δ(t).
Solution:
The output voltage is q(t)/C, so we will determine q(t). By Kirchhoff’s voltage law,
1
Li′ + Ri + q = i′ + 10i + 100q = δ(t) (5.2.10)
C
Since i = q ′ ,
q ′′ + 10q ′ + 100q = δ(t). (5.2.11)
′
We assume initial conditions q(0) = q (0) = 0.
Apply the Laplace transform to the differential equation and use the initial conditions to obtain
Then
1
Q(s) =
(s + 5)2 + 75
Since
√
1 1
L−1 2
= √ sin(5 3t)
s + 75 5 3
then
√
1 1
q(t) = L−1 2
= √ e−5t sin(5 3t)
(s + 5) + 75 5 3
The output voltage is
1 20 √
q(t) = 100q(t) = √ e−5t sin(5 3t) (5.2.13)
C 3
A graph of this output is shown in Fig.5.5. The circuit output displays damped oscillations at its natural
frequency, even though it was not explicitly forced by oscillations of this frequency. If we wish, we can
obtain the current by i(t) = q ′ (t).
46 CHAPTER 5. THE SYSTEM WITH THE δ(T ) FUNCTION
6
-2
5.3 Exercises
1. Solve the following equations
3. Using the Laplace transform and showing the details of your work, solve the initial value problems
π
(a) y ′′ + 9y = δ t − , y(0) = 2, y ′ (0) = 0
2
(b) y ′′ − 16y = 4δ(t − 3π), y(0) = 2, y ′ (0) = 0
(c) y ′′ + 4y = δ(t − π) − δ(t − 2π), y(0) = 0, y ′ (0) = 1
(d) y ′′ + 3y ′ + 2y = 10(sin t + δ(t − 1)), y(0) = 1, y ′ (0) = −1
(e) y ′′ + 16y = 4δ(t − π), y(0) = −1, y ′ (0) = 0
(f ) y ′′ + 3y ′ + 2y = u(t − 1) + δ(t − 2), y(0) = 0, y ′ (0) = 0
4. Using the Laplace transform and showing the details of your work, solve the following equations
5. Using differentiation, integration, s-shifting and convolution and showing the details, find f (t) if
L[f ] = F (s) equals
s s 2s + 6
(a) F (s) = (b) F (s) = (c) F (s) =
(s2 + 16)2 (s2 − 4)2 (s2 + 6s + 10)2
s s+1 s+a
(d) F (s) = ln (e) F (s) = ln (f ) F (s) = ln
s+1 s s+b
6. Using the Laplace transform and showing the details, solve the following high order equations
Solution:
1a. y = 1 e2t − 4e−t + 11 e−2t + 2u(t − 1)(e−(t−1) − e−2(t−1) )
2 2
1b. y = 2e−2t + 5e−t + 5 u(t − 1)(e(t−1) − e−2(t−1) )
3
1c. y = 1 e 2t 2
− e −t − 1 e−2t + 5 u(t − 1) sinh 2(t − 1) 1d. y = 1 (8 cos t − 5 sin t − sin 3t) − 2u(t − π/2) cos t
6 3 2 2 8
1e. y = 1 − cos 2t + 1 sin 2t + 1 u(t − 3π) sin 2t 1f. y = 4et + 3e−t − 8 + 2u(t − 2) sinh(t − 2)
2 2
1g. y = 1 et − 7 e−t + 2 + 3u(t − 6)(1 − e−(t−6) ) 1h. y = e2t + 7 cos 2t − sin 2t − 1 u(t − π/2) sin 2t
2 2 2
1i. y = 1 (1 + e−2t ) + u(t − 1)(e−(t−1) − e−2(t−1) ) 1j. y = 1 et + 1 e−t (2t − 5) + 2u(t − 2)(t − 2)e−(t−2)
2 4 4
2. (b) y(t) = u(t − 1)(t − 1)e2t−2 + u(t − 2)(2 − t)e2t−4 .
48 CHAPTER 5. THE SYSTEM WITH THE δ(T ) FUNCTION
Chapter 6
Transfer Functions
In this chapter we introduce the concept of a transfer function and then use this to obtain a Laplace
transform model of a linear engineering system. (A linear engineering system is one modelled by a constant
coefficient ordinary differential equation.)
dy(t)
+ y(t) = x(t), y(0) = y0 (6.1.1)
dt
and assume that it models a simple engineering system. Then x(t) represents the input to the system and
y(t) represents the output, or response of the system to the input x(t). For reason that will be explained
below it is necessary to assume that the initial conditions associated wit the differential equation are zero.
In equation (6.1.1) this means we take y0 to be zero. Taking the Laplace transform of equation (6.1.1)
yields
so that
Y (s) 1
=
X(s) 1+s
Y (s)
The function, is called a Transfer function. It is the ratio of the Laplace transform of the output to
X(s)
the Laplace transform of the input. It is often denoted by G(s). Therefore, for equation (6.1.1)
1
G(s) =
1+s
The concept of a transfer function is very useful in engineering, in particular, in digital signal processing.
It provides a simple algebraic relationship between the input and the output. In other words, it allows the
analysis of dynamic system based on the differential equation to proceed in a relatively straightforward
manner.
Example 6.1.1 Find the transform function of the following equation assume that x(t) is the input and
y(t) is the output.
dy(t)
− 4y(t) = 3x(t), y(0) = 0
dt
49
50 CHAPTER 6. TRANSFER FUNCTIONS
Solution:
Taking Laplace transform of the differential equation gives
which hold
X2 (s) = G1 (s)X1 (s), Y (s) = G2 (s)X2 (s)
Eliminating X2 (s) from these equations yields
x′ + 2x = f (t) (6.1.3)
2y ′ − y = x(t) (6.1.4)
Solution:
The input in equation (6.1.3) is f (t) and the output is x(t), this forms the input to Equation (6.1.4).
The block diagrams for Equations (6.1.3) and (6.1.4) are combined into a single block diagram
F (s) 1 Y (s)
=⇒ =⇒
(s + 2)(2s − 1)
2. Eliminating a negative feedback loop Consider the following diagram which shows a negative feedback
loop. Such loops are common in a variety of engineering systems.
R(s) X2 (s)
−−−−→ (+1) −−−−→ G(s) −−−−→ Y (s)
x
X1 (s)
y
(−1) ←−−−− H(s)
The quantities X1 (s) and X2 (s) represent intermediate signal in the system. We wish to obtain an
overall transfer function for this system relating Y (s) and R(s). For the two transfer functions the
following hold:
Y (s) = G(s)X2 (s)
X1 (s) = H(s)Y (s)
For the summing point
X2 (s) = R(s) − X1 (s)
This gives
X2 (s) = R(s) − H(s)Y (s)
Combining the above equations, we have
R(s) G(s)
−−−−→ −−−−→ Y (s)
1 + G(s)H(s)
The values of s that make G(s) zero are known as the system zeros and correspond to the roots of P (s) = 0,
that is s = z1 , z2 , · · · , zM . The values of s that make G(s) infinite are known as the system poles and
correspond to the roots of Q(s) = 0, that is s = p1 , p2 , · · · , pn . As we have seen, poles may be real or
complex. Complex poles always occur in complex conjugate pairs whenever the polynomial Q(s) has real
coefficients.
Engineers find it useful to plot these poles and zeros on an s plane diagram. A complex plane plot
is used with, conventionally, a real axis label of σ and an imaginary axis label of jω. The benefit of this
approach is that it allows the character of a linear system to be determined by examining the s plane plot.
In particular, the transient response of the system can easily be visualized by the number and positions of
the system poles and zeros.
Example 6.2.1 Find the poles of
s−2
(s2 + 2s + 5)(s + 1)
Solution:
The denominator is factorized into linear factors,
where p1 = −1 + 2j, p2 = −1 − 2j, p3 = −1. The poles are −1 + 2j, −1 − 2j and −1.
1
In fact, if X(s) = s−p 1
where the pole p1 is given by p1 = a + bj, then
Hence the real part of the pole a gives rise to an exponential term and the imaginary part b gives rise to
an oscillatory term. If a < 0 the response, x(t) will decrease to zero as t → ∞.
Example 6.2.2 The response x(t) of a system to a forcing function u(t) is determined by the differential
equation
d2 x dx du
9 2 + 12 + 13x = 2 + 3u
dt dt dt
1. Determine the transfer function characterizing the system.
2. Write down the characteristic equation of the system. What is the order of the system?
3. Determine the transfer function poles and zeros, and illustrate them diagrammatically in the s plane.
Solution:
1. Assuming all the initial conditions to be zero, taking Laplace transforms throughout in the differential
equation
d2 x dx du
9 2 + 12 + 13x = 2 + 3u
dt dt dt
leads to
(9s2 + 12s + 13)X(s) = (2s + 3)U (s)
so the system transfer function is given by
X(s) 2s + 3
G(s) = = 2
U (s) 9s + 12s + 13
9s2 + 12s + 13 = 0
9s2 + 12s + 13 = 0
which are p
−12 ± (144 − 468) −2 ± 3j
s= =
18 3
6.3. MODELLING LINEAR SYSTEMS BY TRANSFER FUNCTIONS 53
2s + 3 = 0
0.8
0.6
0.4
Imaginary Part
0.2
−0.2
−0.4
−0.6
−0.8
−1
Figure 6.1: Pole (X) and zero ( ) plot for Example 6.2.2.
Solution:
If the system has an overall transfer function H(s) then V0 (s) = H(s)V1 (s). But this particular system is
the negative feedback loop described earlier and so
K
G(s) K
H(s) = = 1+asK =
1 + G(s) 1 + 1+as K + 1 + as
and so the response to a step input u(t) is given by the convolution of h(t) with u(t)
Z t
v0 (t) = u(t − x)5e−7x dx
0
Z t
= 5e−7x dx
0
t
5 5
= − e−7x = − [e−7t − 1]
7 0 7
Chapter 7
Remember that you must specify the variables and the functions first. Here are more examples from
Matlab:
syms a s t w x
laplace(t^5) % returns 120/s^6
laplace(exp(a*s)) % returns 1/(t-a)
laplace(sin(w*x),t) % returns w/(t^2+w^2)
laplace(cos(x*w),w,t) %returns t/(t^2+x^2)
laplace(x^sym(3/2),t) %returns 3/4*pi^(1/2)/t^(5/2)
laplace(diff(sym(’F(t)’))) % returns laplace(F(t),t,s)*s-F(0)
To make the expression more readable one can use the commands, simplify and pretty. Here is an
example for the function of f (t) = −1.25 + 3.5te−2t + 1.25e−2t
>> syms t s
>> f=-1.25+3.5*t*exp(-2*t)+1.25*exp(-2*t);
>> F=laplace(f,t,s)
F =
5/(4*(s + 2)) + 7/(2*(s + 2)^2) - 5/(4*s)
>> simplify(F)
ans =
(s - 5)/(s*(s + 2)^2)
>> pretty(ans)
s - 5
----------
2
s (s + 2)
This is
s−5
F (s) =
s(s + 2)2
55
56 CHAPTER 7. LAPLACE TRANSFORMS WITH MATLAB
>> ilaplace(F,t);
>> simplify(ans);
>> pretty(ans)
2
t heaviside(t - 10) % return to the previous function f(t)=t^2*u(t-10)
7.2. THE INVERSE OF THE LAPLACE TRANSFORM 57
Therefore, you can use the Malab program to check the solutions you got in the previous sections, however,
I strongly suggest that you must do the questions by hand calculations first then do the Matlab program
for checking the solutions. Let us use the following example to end this chapter.
58 CHAPTER 7. LAPLACE TRANSFORMS WITH MATLAB
Solution:
>> syms F f t s a
>> f=t^3*exp(-2*t)+heaviside(t-5)*exp(-3*t)+heaviside(t-6)*sin(t);
>> laplace(f);
>> simplify(ans);
>> pretty(ans)
1 6 cos(6) + s sin(6)
--------------------- + -------- + -----------------
exp(5 s + 15) (s + 3) 4 2
(s + 2) exp(6 s) (s + 1)
which gives
1 −(5s+15) 6 cos(6) + s sin(6) −6s
F (s) = e + 4
+ e
s+3 (s + 2) s2 + 1
2. The Matlab codes for the inverse of transform function F (s) are
>> F=s*exp(-s)/(s^2-14*s+1)+log(s/(s+1))+log((s+a)/(s-a))
F =
log(s/(s + 1)) + log(-(a + s)/(a - s)) + s/(exp(s)*(s^2 - 14*s + 1))
>> ilaplace(F);
>> simplify(ans)
ans =
(1/exp(t) + 2*sinh(a*t) - 1)/t + heaviside(t - 1)*exp(7*t - 7)*cosh(4*3^(1/2)*(t - 1))
+ (7*3^(1/2)*sinh(4*3^(1/2)*(t - 1))*heaviside(t - 1)*exp(7*t - 7))/12
>> pretty(ans)
1
------ + 2 sinh(a t) - 1
exp(t)
------------------------ + heaviside(t - 1) exp(7 t - 7)
t
1/2
cosh(4 3 (t - 1)) +
1/2 1/2
7 3 sinh(4 3 (t - 1)) heaviside(t - 1) exp(7 t - 7)
---------------------------------------------------------
12
This gives
√ √
e−t + 2 sinh(at) − 1 7t−7
√ 7 3 sinh(4 3(t − 1))u(t − 1)e7t−7
f (t) = + u(t − 1)e cosh(4 3(t − 1)) +
t 12
Chapter 8
The Z Transform
8.1 Introduction
In this chapter we focus attention on discrete processes. With the advent of fast and cheap digital com-
puters, there has been renewed emphasis on the analysis and design of digital systems, which represent a
major class of engineering systems.
Digital systems operator on digital signals, which are usually generated by sampling a continuous-time
signal, that is a signal defined for every instant of a possibly infinite time interval. The sampling process
generates a discrete-time signal, defined only at the instants when sampling takes place so that a digital
sequence is generated. After processing by a computer, the output digital signal may be used to constructed
a new continuous-time signal, perhaps by the use of a zero-order hold device, and this turn might be used
to control a plant or process. Digital signal processing devices have made a major impact in many areas
of engineering, as well as in the home. For example, compact disc players, which operate using digital
technology, offer such a significant improvement in reproduction quality that recent years have seen them
are taking over from cassette tape players and vinyl record decks. DVD players are taking over from video
players and digital radios are setting the standard for broadcasting. Both of these are based on digital
technology.
The Laplace transform and its discrete-time counterpart the Z-transform are essential mathematical
tools for digital system design and analysis, and for prediction and monitoring of the stability of a system.
A working knowledge of the Z-transform is essential to the study of discrete-time filters and systems. It is
partly through the use of such transforms that we can formulate a closed-form mathematical description
of a system in the frequency domain, design the system, and then analyse and predict the stability, the
transient response and the steady state characteristics of the system.
A mathematical description of the input-output relation of a system can be formulated either in the
time domain or in the frequency domain. Time-domain and frequency domain representation methods
offer alternative insights into a system, and depending on the application it may be more convenient to
use one method in preference to the other.
Time domain system analysis methods are based on differential equations which describe the system
output as a weighted combination of the differentials (i.e. the rates of change with time) of the system input
and output signals. Frequency domain methods, mainly the Laplace transform, the Fourier transform, and
the Z-transform, describe a system in terms of its response to the individual frequency constituents of the
input signal
We have seen that the Laplace transform was a valuable aid in the analysis of continuous-time systems,
and in this chapter we develop the Z transform, which will perform the same task for discrete-time systems.
We first introduce the Z transform in connection with the solution of difference equations, and later we
show how difference equations arise as discrete-time system models.
whenever the sum exists and where z is a complex variable, as yet undefined.
59
60 CHAPTER 8. THE Z TRANSFORM
The process of taking the Z transform of a sequence thus produce a function of a complex variable z,
whose form depends upon the sequence itself. The symbol Z denotes the Z-transform operator, when it
operates on a sequence f [k], it transforms the latter into the function F (z) of the complex variable z.
Z transform
a sequence f [k] a function F (z)
Inverse of Z transform
Note that the sequence f [k] is sometimes also written in the form of fk . Thus, these two notations are the
same in this lecture note.
Let first make a revision for a geometric sequence before considering the Z transform. As we know, when
a sequence has a constant ratio between successive terms it is called a geometric progression. The constant
is called the common ratio, r. For example, 1, 2, 4, 8, · · · where the common ratio is 2 and a, ar, ar2 , ar3 , · · ·
where the common ratio is r. Let a geometric progression be a, ar, ar2 , ar3 , · · · , arn−1 then the sum of n
terms
a(rn − 1)
Sn = a + ar + ar2 + · · · + arn−1 =
(r − 1)
If n → ∞, then the sum of a geometric progression is
∞
X a
S∞ = arn = , which is valid when − 1 < r < 1(|r| < 1) (8.2.2)
n=0
1−r
This is a very useful formula when we calculate a geometric series for the Z transforms.
f [k] = 2k , (k ≥ 0)
Solution:
From the definition (8.2.1)
∞ ∞ k
X 2k X 2
Z{ 2k } = =
zk z
k=0 k=0
which we recognize as a geometric series, with common ratio r = 2/z between successive terms. The series
thus converges for |z| > 2 by equation(8.2.2), when
∞ k
X 2 1 − (2/z)k 1
= lim =
z k→∞ 1 − z/2 1 − 2/z
k=0
leading to
z
Z{ 2k } = (|z| > 2) (8.2.3)
z−2
so that
z
f [k] = 2k ⇐⇒ F (z) =
z−2
is an example of a Z transform pair.
We can generalize the result (8.2.3) in an obvious way to determine the Z transform of the sequence
Z{ ak }, where a is a real or complex constant. At once
∞
X ak 1
Z{ ak } = = (|z| > a) (8.2.4)
zk 1 − a/z
k=0
Solution:
1
Taking a = − in (8.2.4), we have
2
X (−1/2)k ∞
1 z
Z{ (− )k } = =
2
k=0
z k z − ( 12 )
so that k
1 2z 1
Z{ − }= , (|z| > )
2 2z + 1 2
Further Z transform pair can be obtained from (8.2.4) by formally differentiating with respect to a.
This gives k
d k da d z
Z{ a } = Z =
da da da z − a
leading to
z
Z{ kak−1 } = (8.2.5)
(z − a)2
In the particular case a = 1 this gives
z
Z{ k } = (8.2.6)
(z − 1)2
The Z transform table
Z Transform Table
f [k] F (z) Coverngence domain
1 k≥0 z
u[k] = z−1 |z| > 1
0 k<0
z
k (z−1)2 |z| > 1
2 z(z+1)
k (z−1)3 |z| > 1
3 z(z 2 +4z+1)
k (z−1)4 |z| > 1
z
e−ak z−e−a |z| > e
z
ak z−a |z| > |a|
az
kak (z−a)2 |z| > |a|
az(z+a)
k 2 ak (z−a)3 |z| > |a|
αk α/z
k e
z sin a
sin ak z 2 −2z cos a+1 |z| > 1
z(z−cos a)
cos ak z 2 −2z cos a+1 |z| > 1
ze−a sin b
e−ak sin bk z 2 −2ze−a cos b+e−2a |z| > 1
z 2 −ze−a cos b
e−ak cos bk z 2 −2ze−a cos b+e−2a |z| > 1
δ(k) 1
δ(n − m) z −m |z| > 1
and
Z{cf [k]} = cZ{f [k]}
Solution:
Since
z
Z{e−k } =
z − e−1
z
Z{3k} = 3Z{k} = 3
(z − 1)2
62 CHAPTER 8. THE Z TRANSFORM
Hence,
z 3z
Z{e−k + 3k} = +
z − e−1 (z − 1)2
2. The first shift theorem
If f [k] is a sequence and F (z) is its Z transform, then
Solution:
By the first shift theorem
Z{f [k + 1]} = zF (z) − zf [0]
( )
k
1 z 2z
F (z) = Z = =
2 z − 21 2z − 1
f [0] = 1
Hence
2z z
Z{f [k + 1]} = zF (z) − zf [0] = z −z =
2z − 1 2z − 1
F (z)
Z{f [k − i]u[k − i]} = z −i F (z) = i ∈ N+
zi
where F (z) is the Z transform of f [k].
Solution:
1. Multiplication by e−bk
If f [k] is a sequence, then
Z{e−bk f [k]} = F (eb z)
where F (z) is the Z transform of f [k].
In general, if f [k] is a sequence, then for a constant a
2. Multiplication by k n
If f [k] is a sequence and Z{ f [k] } = F (z), then
n
n d
Z{ k f [k] } = −z F (z) (8.3.8)
dz
∞
X
f [k] F (z) = f [k]z −k
k=0
f [k + 1] zF (z) − zf [0]
f [k + 2] z 2 F (z) − z 2 f [0] − zf [1]
f [k + 3] z 3 F (z) − z 3 f [0] − z 2 f [1] − zf [2]
f [k + m] z m F (z) − z m f [0] − · · · − zf [m − 1]
F (z)
f [k − m]u[k − m] z −m F (z) =
zm
dF (z)
kf [k] −z
Z dz
1 F (z)
f [k] − dz
k z
ak f [k] F (z/a)
Example 8.5.2 If
z+3
F (z) =
z−2
find f [k].
Solution:
Since
z+3 z 3
= +
z−2 z−2 z−2
and
z
Z −1 = 2k
z−2
Also
3 3 z
=
z−2 z z−2
64 CHAPTER 8. THE Z TRANSFORM
Therefore,
z+3
Z −1 = 2k + 3(2)k−1 u[k − 1]
z−2
Example 8.5.3 Find the sequence whose Z transform is
2z 2 − z
F (z) =
(z + 4)(z − 5)
Solution:
Since
2z 2 − z z(2z − 1)
F (z) = =
(z + 4)(z − 5) (z + 4)(z − 5)
F (z) (2z − 1)
=
z (z + 4)(z − 5)
We now use the partial fractions to make
(2z − 1) A B 1 1
= + = +
(z + 4)(z − 5) z+4 z−5 z+4 z−5
This gives
F (z) 1 1
= +
z z+4 z−5
Multiplying by z gives
z z
F (z) = +
z+4 z−5
Therefore
f [k] = Z −1 [F (z)] = (−4)k + 5k
We note that the approach here parallels that used to obtain the inverse Laplace transform of F (s).
In this case, however, we work with F (z)/z instead of F (s). Why F (z)/z and not F (z)? Examining our
standard forms we found that functions of z occur in the form
z z z
, 2
,
z−1 (z − 1) (z − e−a )
and not in the normal partial fraction form
1 1 1
, ,
z−1 (z − 1)2 (z − e−a )
8.6 Exercises
1. Given
x[n + 2] + x[n + 1] − x[n] = 2, x[0] = 3, x[1] = 5
Find x[2], x[3], x[4] and x[5].
2. Determine x[2] and x[3] given
3. Calculate the first five terms of the difference equation with given initial condition
6. If f [k] = 4(3k ), find Z{f [k]}. Use the first shift theorem to deduce Z{f [k + 1]} and show that
Z{f [k + 1]} − 3Z{f [k]} = 0.
Y (z)
9. By first resolving into partial fractions, find the inverse of Y (z) when Y (z) is given by
z
10z z z2
(a) (b) (c)
(z − 1)(z − 2) (2z + 1)(z − 3) (2z + 1)(z − 1)
2z z z
(d) (e) (f ) √
2z 2 + z − 1 z2 + 1 2
z − 2 3z + 4
2z 2 − 7z z2 z
(g) (h) (i)
(z − 1)2 (z − 3) (z − 1)2 (z 2 − z + 1) (z − 1)(z + 2)
1 2 3 2 3z + z 2 + 5z 3
(a) + 5 (b) 1+ − 7 (c)
z z z2 z z5
1+z 3z 2z 3 + 3z 2 + 4z + 5 2
2z − 7z + 7
(d) + (e) (f )
z4 5z + 1 z 2 (2z + 1) (z − 1)2 (z − 2)
z−3 z(1 − e−aT ) z 3 + 2z 2 + 1
(g) (h) (i)
z 2 − 3z + 2 (z − 1)(z − e−aT ) z3
Solutions:
(1) 0,7,-5,14. (2) 5, 29 .
2
(3) 5, 7, 2, 9, −3.
z(z−cos 3) 2 −2 cos 1
(4) (a) 2 , (b) 2 z −ze (c) z , (d) z . (e) z −3
z −2z cos 3+1 z −2ze−2 cos 1+e−4 z−4 z+3
−aT
5 (a) 1 , (b) 2z+1 , (c) z , (d)
z(1−e
, (e) 2z − 4z + 3z , (f) z2 (5) 1 (7)
(z+3)
(8) (a)
(z−1)(z−2) (z+1)(z−3) z 2 +a2 (z−1)(z−e−aT z−1/4 z−1/2 z−1 (z+2)(z−1)2 (z−1)2 (z−2)
k k k (1/3)k +(−1/3)k
4(4 ), (b) −2(2 ) + 2(3 ), (c) .
2
(9) 10(2k − u[k])
(10) (a) x[k] = 3 − 2(3k ), (b) x[k] = 10(4k ) − 5(k4k ), (c) y[k] = (−1)k − (−2)k .
(11) (a) 1 {(−1/4)k − (−1/2)k }, (b) 2(3k ) sin 1 (k + 1)π, (c) 2 (0.4)k + 1 (−0.2)k , (d) 4k+1 + 2k .
2 6 3 3
66 CHAPTER 8. THE Z TRANSFORM
Chapter 9
9.1 Introduction
First we illustrate the motivation for studying difference equations by means of an example. Suppose
that a sequence of signals {xk } is being recorded. We might attempt to process (for example, smooth or
filter) this sequence of signals {xk } using the discrete-time feedback system illustrated in Figure 9.1. At
time step k the signal {xk } enters the system as an input, and after combination with the feedback signal at
the summing junction S, proceeds to the block labeled D. This block is a unit delay block, and its function
is to hold its input signal until the clock advances one step, to step k + 1. At this time the input signal is
passed without alteration to become the signal {yk+1 }, the (k + 1)th member of the output sequence {yk }.
At the same time this signal is fed back through a scaling block of amplitude α to the summing junction
S. This process is instantaneous, and at S the feedback signal is subtracted from the next input signal
xk+1 to provide the next input to the delay block D. The process then repeats at each clock step.
To analyze the system, let {rk } denote the sequence of input signals to D, then, owing to the delay
action of D, we have
yk+1 = rk
Also, owing to feedback action
rk = xk − αyk
where α is the feedback gain. Combining the two expressions gives
yk+1 = xk − αyk
or
yk+1 + αyk = xk
This is an example of a first order difference equation, and it relates adjacent members of the sequence
{yk } to each other and to the input sequence {xk }.
Example 9.1.1 Find a difference equation to represent the system shown in Figure 9.2, having input and
output sequence {xk } and {yk }, respectively, where D is the unit delay block and a and b are constant
feedback gains.
Solution:
Introducing intermediate signal sequences {rk } and {vk } as shown in Figure 9.2, at each step the outputs
of the delay blocks are
yk+1 = vk (9.1.1)
67
68 CHAPTER 9. SOLVING DIFFERENCE EQUATIONS WITH THE Z TRANSFORMS
vk+1 = rk (9.1.2)
and at the summing junction
rk = xk − avk + byk (9.1.3)
From (9.1.1),
yk+2 = vk+1
which on using (9.1.2) gives
yk+2 = rk
Substituting for rk from (9.1.3) then gives
Solution:
Taking Z transform of both sides of the equation we obtain
zZ{y[k]} − 4z − 3Z{y[k]} = 0
This becomes
(z − 3)Z{y[k]} = 4z
so that
4z
Z{y[k]} =
z−3
Inverting this, we find
4z
y[k] = Z −1
= 4(3)k
z−3
9.2. SOLVING THE DIFFERENCE EQUATIONS 69
Solution:
Taking the Z transform both sides we have
Let Y (z) = Z{y[k]} and use the first shift theorem we obtain
Hence
(z 2 − 5z + 6)Y (z) = 2z
So that
2z
Y (z) =
z 2 − 5z + 6
Dividing both sides by z gives
Y (z) 2 2
= 2 =
z z − 5z + 6 (z − 2)(z − 3)
2 2
= −
z−3 z−2
2z 2z
Y (z) = −
z−3 z−2
Inverting gives the solution as
2z 2z
y[k] = Z −1
− = 2(3k ) − 2(2k )
z−3 z−2
Example 9.2.3 Solve the difference equation
Solution:
Transforming the equation we have
z
2F (z) − 3(zF (z) − zf [0]) = (−1)−1
z+1
z
F (z)(2 − 3z) = − − 6z
z+1
z 6z
F (z) = +
(z + 1)(3z − 2) 3z − 2
z 3z 6z
=− + +
5(z + 1) 5(3z − 2) 3z − 2
in partial fractions. Thus,
z z 2z
F (z) = − + +
5(z + 1) 5(z − 2/3) z − 2/3
z 11z
=− +
5(z + 1) 5(z − 2/3)
Hence,
k
1 11 2
f [k] = − (−1)k +
5 5 3
We use the following example involving with complex numbers to end this section.
70 CHAPTER 9. SOLVING DIFFERENCE EQUATIONS WITH THE Z TRANSFORMS
y[k + 2] + 2y[k] = 0
√
given that y[0] = 1 and y[1] = 2
Solution:
Taking Z transforms , we obtain
This gives
1 z z
Y (z) = (1 + j) √ − (1 − j) √
2j z − 2ejπ/2 z − 2e−jπ/2
Taking inverse transform it gives
2k/2
y[k] = (1 + j)ejkπ/2 − (1 − j)e−jkπ/2
2j
which is
kπ kπ
y[k] = 2k/2 cos + sin
2 2
This series of numbers gives a limited description of f (t). An alternative way of representing the sampled
function is to define the continuous time sampled version of f (t) as fˆ(t) where
∞ ∞
fˆ(t) =
X X
f (t)δ(t − kT ) = f (kT )δ(t − kT ) (9.3.5)
k=0 k=0
The representation (9.3.5) may be interpreted as defining a row of impulses located at the sampling points
and weighted by the appropriate sampled values as illustrated in Figure 9.3. Taking the Laplace transform
of fˆ(t), we have
Z ∞ X ∞
!
L[fˆ(t)] = f (kT )δ(t − kT ) e−st dt
0 k=0
∞
X Z ∞
= f (kT ) δ(t − kT )e−st dt
k=0 0
9.4. EXERCISES 71
giving
∞
L[fˆ(t)] =
X
f (kT )e−kst (9.3.6)
k=0
where, F (z) denotes the Z transform of the sequence {f (kT )}. We can therefore view the Z transform of
a sequence of samples in discrete time as the Laplace transform of the continuous-time sampled function
fˆ(t) with an appropriate change of variable z = esT .
9.4 Exercises
1. Using the Z transform method solve the following difference equations
3. The dynamics of a discrete time system are determined by the difference equation
where in is the current in the (n + 1)th loop, and R1 and R2 are constant resistors.
(a) Show that this may be written as
where
−1 R1
α = cosh 1+
2R2
(b) By solving the equation in (5a), show that
6. The signal f (t) = t is sampled at intervals T to generate the sequence {f (kT )}. Show that
Tz
Z{f (kT )} =
(z − 1)2
Solution:
1. (c) y(k) = 41 (2)k − 14 (−2)k , (d) y(k) = 2(1/2)k + 3k , (e) k5 − 254 4
(−4)k + 25
k k 4 k 16 k 3 k 1
2. (a) 3 − 2(3) , (b) 5(4) − 5(4) (k − 1), (c) 2 − 1, (d) 15 (1/2) − 20 (−3) + 12 , (e) 2(1/4)k − 4(1/2)k + 3
k 1 k 1
3. 2 − 2 (3) + 2
4. (a) 52 (−1/2)k − 10 9
(1/3)k + 12 , (b) 25 (3)k − 5(2)k + 52 , (c) 15
4
(1/2)k − 23 (2)k + 25 (3)k , (d) 1 − k, (e)
1 k 1 k 12 k 2 k
1 − 2 (−2) − 2 (2) − k, (f) 5 (2) − 2k − 5 (−1/2) − 1
Chapter 10
Z Transfer Functions
10.1 Introduction
When considering continuous time linear systems modelled by differential equations, we introduced the
concept of the system (Laplace) transfer function. This is a powerful tool in the description of such
systems. In the same way, we can identify a Z transfer function for a discrete time linear time invariant
system modelled by a difference equation, and we can arrive at results analogous to those in Laplace
transform.
Let us consider the general difference equation with input uk and output yk . Both uk and yk are causal
sequences throughout. Such a difference equation model takes the form
an yk+n + an−1 yk+n−1 + · · · + a0 yk = bm uk+m bm−1 uk+m−1 + · · · + b0 uk
where k ≥ 0 and n, m (with n ≥ m) are positive integers and that ai and bj are constants. Assuming the
system to be initially in a quiescent state, we take Z transform to give
(an z n + an−1 z n−1 + · · · + a0 )Y (z) = (bm z m + bm−1 z m−1 + · · · + b0 )U (z)
Therefore, the Z transfer function G(z) is defined as
Y (z) bm z m + bm−1 z m−1 + · · · + b0
G(z) = = (10.1.1)
U (z) an z n + an−1 z n−1 + · · · + a0
On writing
P (z) = bm z m + bm−1 z m−1 + · · · + b0
Q(z) = an z n + an−1 z n−1 + · · · + a0
the discrete transfer function may be expressed as
P (z)
G(z) =
Q(z)
The equation
Q(z) = 0
is called the characteristic equation of the discrete system, its order, n, determines the order of the system,
and its roots are referred to as the poles of the transfer function. Likewise, the roots of
P (z) = 0
are referred to as the zeros of the discrete transfer function.
Example 10.1.1 Find the z transfer function to represent the system modelled by the difference equation
yk+2 + 3yk+1 − yk = uk (10.1.2)
Solution:
Taking the z transform throughout in (10.1.2), under the assumption of a quiescent initial condition, we
obtain
z 2 Y (z) + 3zY (z) − Y (z) = U (z)
The z transfer function follows
Y (z) 1
G(z) = = 2
U (z) z + 3z − 1
73
74 CHAPTER 10. Z TRANSFER FUNCTIONS
lim G(z) = ∞
z→z1
Solution:
z k+1
G(z) = z k−1 F (z) =
(z + 3)2
There is a pole of second order at z = −3:
d d k+1
R−3 = [(z + 3)2 G(z)] = [z ] = (k + 1)z k
dz dz
At z = −3,
R−3 = (k + 1)(−3)k
Hence
f [k] = (k + 1)(−3)k
Solution:
Consider
z k−1
G(z) = z k−1 F (z) =
(z + 1)(z + 2)
which has a simple pole at z = 0 when k = 0, but not at z = 0 for k ≥ 1. Thus the cases k = 0 and k ≥ 1
must be considered separately.
For k = 0,
1
G(z) =
z(z + 1)(z + 2)
1
Rz=0 =
2
Rz=−1 = − 1
1
Rz=−2 =
2
X 1 1
f [0] = R= −1+ =0
2 2
For k ≥ 1
z k−1
G(z) =
(z + 1)(z + 2)
Rz=−1 =(−1)k−1
Rz=−2 = − (−2)k−1
Hence,
f [k] = (−1)k−1 − (−2)k−1
Consider a system with transfer function G(z), so that the z transform Y (z) of the output sequence yk
corresponding to an input sequence uk with z transform U (z) is
If the input sequence yk is the impulse sequence δk and the system is initially quiescent, then the output
sequence yδk is called the impulse response of the system. Hence
That is, the z transfer function of the system is the z transform of the impulse response. Alternatively, we
can say that the impulse response of a system is the inverse z transform of the system transfer function.
Substituting (10.3.4) into (10.3.3), we obtain
Thus the z transform of the system output in response to any input sequence uk is the product odf the
transform of the input sequence with the transform of the system impulse response. The result (10.3.5)
shows the underlying relationship between the concepts of impulse response and transfer function, and
explains why the impulse response (or the transfer function) is thought of as characterizing a system. In
simple terms, if either of these is known then we have all the information about the system for any analysis
we may wish to do.
Example 10.3.1 Find the impulse response of the system with z transfer function
z
G(z) =
z2 + 3z + 2
Solution:
Using (10.3.4), we have
z z
Yδ (z) = =
z2 + 3z + 2 (z + 2)(z + 1)
Resolving Yδ (z)/z into partial fractions gives
Yδ (z) 1 1 1
= 2 = −
z z + 3z + 2 (z + 1) (z + 2)
Example 10.3.2 A system has the impulse response sequence
yδk = ak − (0.5)k
where a > 0 is a real constant. What is the nature of this response when (a) a = 0.4, and (b) a = 1.2?
Find the step response of the system in both cases.
Solution:
When a = 0.4,
yδk = 0.4k − 0.5k
and since both 0.4k → 0 and 0.5k → 0, as k → ∞, we see that the terms of the impulse response sequence
go to zero as k −→ ∞.
On the other hand, when a = 1.2, since (1.2)k → ∞ as k → ∞, we see that in this case the impulse
response sequence terms become unbounded, implying that the system blow up.
In order to calculate the step response, we first determine the system transfer function G(z) using
(10.3.4), as
G(z) = Yδ (z) = Z(ak − 0.5)
giving
z z
G(z) = −
z − a z − 0.5
10.4. THE RESPONSE OF A SINGLE (FIRST-ORDER) ZERO OR POLE 77
so that
Y (z) z z
= −
z (z − a)(z − 1) (z − 0.5)(z − 1)
a 1 1 1 1
= − + −2 +
a − 1 z − a z − 0.5 1−a z−1
giving
a z z 1 z
Y (z) = − + −2 +
a − 1 z − a z − 0.5 1−a z−1
which on taking inverse transforms gives the step response as
a 1
yk = ak − 0.5k + −2 +
a−1 1−a
Considering the above output sequence, we see that when a = 0.4, since 0.4k → 0 and 0.5k → 0 as k → ∞,
the output sequence terms tend to the constant value
1
−2 + = 0.3333
1 − 0.4
However, in the case of a = 1.2, since (1.2)k → ∞ as k → ∞, the output is unbounded, and again the
system blows up.
The effect of a first order zero is to introduce a deep or trough in the frequency spectrum of the signal
at the frequencies 0 or π radians, where π corresponds to half the sampling frequency. The effect of a first
order pole is to introduce a peak in the frequency spectrum of the signal at frequencies 0 or π. These are
illustrated in the following examples.
Example 10.4.1 Consider the first order feed forward filter of Fig 10.1 given by
Solution:
Taking the Z-transform of Eq. (10.4.6) yields
Figure 10.2: Illustration of the impulse response, the pole-zero diagram, and the frequency response of a
first order system with a single zero, for the varying values of the zero α.
Equating H(z) = 0 yields a zero at z = −α. Substituting z = e−jω in Eq. (10.4.7) yields the frequency
response
H(ejω ) = 1 + αe−jω (10.4.8)
Now at an angular frequency ω = 0, H(0) = 1 + α and at ω = π, H(π) = 1 − α. Hence when α = 1,
H(π) = 0. Fig. 10.2 shows the variation of the frequency response of the first order single-zero system with
the radius of the zero α. On the frequency axis the angular frequency ω = π corresponds to a frequency of
Fs
Hz where Fs is the sampling rate. Note that for the positive values of α the zero is on the left hand half
2
of the z-plane and the system has a low-pass frequency response, and conversely for the negative values of
α the zero is on the right half of the z-plane and the system has a high-pass frequency response.
Example 10.4.2 Consider the first order feedback system of Fig 10.3 given by
Solution:
Taking the z-transform of Eq. (10.4.9) yields
Figure 10.4: Illustration of the impulse response, the pole-zero diagram, and the frequency response of a
first order system with a single zero, for the varying values of the zero α.
Y (z) 1 z
H(z) = = =
X(z) 1 − αz −1 z−α
The transfer function H(z) has a pole at z = α and a zero at the origin. Substituting z = ejω , we get the
frequency response
1
H(ejω ) =
1 − αe−jω
1 1
Now at ω = 0, H(0) = and at ω = π, H(π) = . Fig. 10.4 shows the variation of the
1−α 1−α
frequency response of the first order single-pole system with the pole radius α. On the frequency axis the
Fs
angular frequency ω = π corresponds to a frequency of Hz where Fs is the sampling rate. Note that
2
for the positive values of α the pole is on the right hand half of the z-plane and the system has a low-pass
frequency response, and conversely for the negative values of α the pole is on the left hand half of the
z-plane and the system has a high-pass frequency response.
80 CHAPTER 10. Z TRANSFER FUNCTIONS
10.5 Exercises
1. Find the impulse response for the systems with z transfer function
z z2
(a) G(z) = 2
(b) G(z) =
8z + 6z + 1 z 2 − 3z + 3
z2 5z 2 − 12z
(c) G(z) = 2 (d) G(z) = 2
z − 0.2z − 0.08 z − 6z + 8
2. Find the transfer function of each of the following discrete time systems
(a) y[k + 2] − 3y[k + 1] + 2y[k] = u[k] (b) y[k + 2] − 3y[k + 1] + 2y[k] = u[k + 1] − u[k]
(c) 9y[k + 2] + 9y[k + 1] + 2y[k] = u[k] (d) 4y[k + 2] − 3y[k + 1] − y[k] = u[k + 1] − 2u[k]
z = −1 + j, and z = −1 − j
7 1 2
3a. [2 − ( )n ]u(n), 3b. [2 + √ sin(n + 1)π/3]u(n), 3c. [cos(nπ/2) − sin(nπ/2)]u(n)
8 2 3
Chapter 11
ans =
(z*sin(1))/(z^2 - 2*cos(1)*z + 1)
>> ztrans(f1,z)
ans =
z/(z - 1/4)
>> pretty(ans)
z
-------
z - 1/4
Compute the Z-transform of this expression calling the ztrans function with one argument. If you do
not specify the transformation index, ztrans uses the variable n, for example
>> syms f a n z
>> f=a^n;
>> ztrans(f,z)
ans =
-z/(a - z)
Compute the following Z-transforms that involve the Heaviside function and the binomial coefficient
>> syms n z
ztrans(heaviside(n - 3), n, z)
ans =
(1/(z - 1) + 1/2)/z^3
81
82 CHAPTER 11. Z TRANSFORM WITH MATLAB
ans =
1/factorial(k)
ans =
2*k + 3
ans =
Solution:
Taking Z transform we obtain
1 4z
Y (z) = 6z +
6z 2 − 5z + 1 4z + 1
>> iztrans((6*z+4*z/(4*z-1))/(6*z^2-5*z+1),k)
ans =
10*(1/2)^k - 18*(1/3)^k + 8*(1/4)^k
This gives
k k k
1 1 1
y[k] = 10 − 18 +8
2 3 4
Chapter 12
12.1 Introduction
The Fourier transform arises in a natural way by inquiring about the behaviour of a Fourier series as the
period of the function being expanded goes to infinity. In other words, what form does a Fourier series
assume if the function being expanded is not periodic? In practical, there are many waveforms are not
periodic. Examples are pulse signals and noise signals. The functions shown in Fig.12.1 are examples of
non-periodic signals. In this chapter we shall see how Fourier techniques can still be useful by introducing
the Fourier transform which is used extensively in communications engineering and signal processing. For
example, it can be used to analyse the processes of modulation, which involves superimposing an audio
signal on to a carrier signal, and demodulation, which involves removing the carrier signal to leave the
audio signal.
where
1 ∞
Z
A(ω) = f (t) cos ωtdt
π −∞
(12.2.2)
1 ∞
Z
B(ω) = f (t) sin ωtdt
π −∞
Provided
1. f (t) and f ′ (t) are piecewise continuous in every finite interval.
R∞
2. −∞ |f (t)|dt exists.
then the above Fourier integral representation of f (t) holds. At a point of discontinuity of f (t) the integral
representation converges to the average value of the right and left hand limits. As with Fourier series, an
equivalent complex representation exists which is, in fact, more commonly used.
83
84 CHAPTER 12. THE FOURIER TRANSFORM
where Z ∞
F (ω) = f (t)e−jωt dt (12.2.4)
−∞
There is no universal convention concerning the definition of these integrals and a number of variants are
1
still correct. For instance, some books use the factor in the second integral rather than the first, while
2π
1
others place a fact √ in both, giving some symmetry to the equations. There is also variation in the
2π
location of the factors e−jωt and ejωt . We shall use definition (12.2.3) throughout but it is important to
be aware of possible differences when consulting texts.
Equations (12.2.3) and (12.2.4) form what is called a Fourier transform pair. The Fourier transform
of f (t) is F (ω) which is sometimes written as F{f (t)}. Similarly f (t) in equation (12.2.3) is the inverse
Fourier transform of F (ω), usually denoted by F −1 {F (ω)}. Engineers refer to the variable ω in the
transform function as the frequency of the signal f .
The Fourier transform of f (t) is defined to be
Z ∞
F{f (t)} = F (ω) = f (t)e−jωt dt (12.2.5)
−∞
Recall the Laplace transform, there are some similarities between equation (12.2.5) and the definition
of the Laplace transform of f (t)
Z ∞
L[f (t)] = f (t)e−st dt (12.2.6)
0
We note that apart from the limits of integration, the substitution jω = s in equation (12.2.5) results in
the Laplace transform in equation (12.2.5). There is indeed an important relationship between the two
transforms which we will discuss it in due course. We note that equation (12.2.3) provides a formula for
inverse Fourier transform of F (ω), although the integral is frequently difficult to evaluate.
Solution:
Using equation (12.2.5), its Fourier transform is given by
Z ∞ Z ∞
−jωt
F (ω) = f (t)e dt = e−at e−jωt dt since f (t) = 0 when t < 0
−∞ 0
Z ∞ ∞
−1 −(a+jω)t
= e−(a+jω)t dt = e
0 a + jω 0
1
= since e−(a+jω)t → 0 as t → ∞
a + jω
This is,
1
F (ω) =
a + jω
Example 12.2.2 Let a and k be positive numbers, and let
k −a ≤ t < a
f (t) =
0 t < −a and t ≥ a
Solution:
In fact, f (t) is a pulse function, which can be written in terms of the unit step function as
Thus,
2k
F (ω) = sin(aω)
ω
12.3.1 Linearity
If f and g are functions of t and k is a constant, then
and
F{kf } = kF{f }
Both of these properties follow directly from the definition and linearity properties of integrals, and mean
that F{} is a linear operator.
Solution:
We saw in example 12.2.1 that
1
F{u(t)e−t } =
1 + jω
Furthermore,
∞ ∞ ∞
e−2(jω)t 1
Z Z
−2t −2t −jωt −(2+jω)t
F{u(t)e }= u(t)e e dt = e dt = =
−∞ 0 −(2 + jω) 0 2 + jω
Therefore,
1 1 2 + jω + 1 + jω 3 + 2jω
F{u(t)e−t + u(t)e−2t } = + = =
1 + jω 2 + jω (1 + jω)(2 + jω) 2 − ω 2 + 3jω
F{ejat f (t)} = F (ω − a)
where a is a constant.
is given by
6 sin 2ω
f (ω) =
ω
and then use the first shift theorem to find the Fourier transform e−jt f (t).
86 CHAPTER 12. THE FOURIER TRANSFORM
Solution:
2 2
e−jωt
Z
F (ω) =3 e−jωt dt = 3
−2 −jω −2
2jω
−e − e2jω
−2jω −2jω
e e 6
=3 =6 = sin 2ω
−jω −2jω ω
Since we have
6
F{f (t)} = F (ω) = sin 2ω
ω
Using the first shift theorem with a = −1 we obtain
6
F{e−jt f (t)} = F (ω + 1) = sin 2(ω + 1)
ω+1
Example 12.3.3 Find the Fourier transform of
−3t
e t≥0
f (t) =
e3t t<0
Figure 12.2: The plot for the function g(t) in Example 12.3.4.
is the Fourier transform of f (t). In equation (12.3.7), ω is a dummy variable so, for example, equation
(12.3.7) could be equivalently written as
Z ∞
1
f (t) = F (z)ejzt dz (12.3.9)
2π −∞
Then, from equation (12.3.9), replacing t by −ω, we obtain
Z ∞ Z ∞
1 1
f (−ω) = F (z)e−jωz dz = F (t)e−jωt dt
2π −∞ 2π −∞
1
which we recognize as 2π times the Fourier transform of F (t). Thus, we have the following result:
If F (ω) is the Fourier transform of f (t),
1
then f (−ω) is × (the Fourier transform of F (t))
2π
1
Example 12.3.5 Given that the Fourier transform of u(t)e−t is use the duality principle to deduce
1 + jω
1
the transform of .
1 + jt
Solution:
We know
1
f (ω) =
1 + jω
1
is the Fourier transform of f (t) = u(t)e−t . Therefore, 2πu(−ω)eω is the Fourier transform of 1+jt , this is
1
F{ } = 2πu(−ω)eω
1 + jt
88 CHAPTER 12. THE FOURIER TRANSFORM
12.4 Spectra
Plots of amplitude verse frequency and phase verse frequency are together known as the spectrum of a
waveform. Periodic functions have discrete or line spectra; this is, the spectra assume non-zero values only
at certain frequency. Only a discrete set of frequencies is required to synthesize a periodic waveform. On
the other hand when analysing non-periodic phenomena via Fourier transform techniques we find that,
in general, a continuous range of frequencies is required. Instead of discrete spectra we have continuous
spectra. The modulus of the Fourier transform, |F (ω)|, gives the spectrum amplitude while its argument
arg(F (ω)) describes the spectrum phase.
Example 12.4.1 Determine the amplitude and phase spectra of the causal signal
Solution:
From (12.2.1),
1
F{f (t)} = F (jω) =
a + jω
Thus the amplitude and argument of F (jω) are
1
|F (jω)| = √
a + α2
2
ω ω
argF (jω) = tan−1 (1) − tan = − tan−1
a a
These are the amplitude and phase spectra of f (t), and are plotted in Fig.12.3.
Figure 12.3: (a) Amplitude and (b) phase spectra of the function in Example 12.4.1.
Generally, as we have observed, the Fourier transform and thus the frequency spectrum are complex
valued quantities. In some cases, the spectrum is purely real.
was found to be
2 sin ω
F (ω) =
ω
Sketch the spectrum of f (t).
12.5. FOURIER TRANSFORMS OF SOME SPECIAL FUNCTIONS 89
Solution:
F (ω) is purely real. The spectrum of f (t) is depicted by plotting |F (ω)| against ω as shown in Fig.12.4.
Note that
sin ω
lim =1
ω→0 ω
Solution:
By definition
Z ∞
F{δ(t − a)} = δ(t − a)e−jωt dt
−∞
for any reasonably well-behaved function f (t). Using this property with f (t) = e−jωt we have
Z ∞
F{δ(t − a)} = e−jωt δ(t − a)dt = e−jωa
−∞
In particular, if a = 0, we have
F{δ(t)} = 1
This result is plotted in Fig.12.5.
Solution:
We have f (t) = δ(t − a), F (ω) = e−jωa . Applying the t − ω duality principle we find
1
f (−ω) = δ(−ω − a) = F{e−jta }
2π
Therefore,
We see that the spectrum of cos at consists of single lines at ω = ±a corresponding to a single frequency
component, see Fig.12.7.
Solution:
Subtracting the previous expression for F{ejta } and F{e−jta } and using Euler’s relations we find
that is,
ejta − e−jta π
F{ } = (δ(ω − a) − δ(ω + a))
2j j
so that
π
F{sin at} = (δ(ω − a) − δ(ω + a))
j
In the definition of the Laplace transform, the parameter s is complex and we may write s = σ + jω, so
that Z ∞
L[f (t)] = f (t)e−σt e−jωt dt
0
Thus an additional factor, e−σt , appears in the integrand of the Laplace transform. Form example,σ > 0
this represents an exponentially decaying factor, the presence of which means that the integral exists for a
wider variety of functions than the corresponding Fourier integral.
Example 12.6.1 Find the Laplace transform of (i) u(t)e−2t , (ii) u(t)e2t . Let s = jω and comment upon
the result.
Solution:
1
L[u(t)e−2t ] =
s+2
1
L[u(t)e2t ] =
s−2
92 CHAPTER 12. THE FOURIER TRANSFORM
1 1
Replacing s by jω for the above two, we have and , respectively. Now,
jω + 2 jω − 2
1
F{u(t)e−2t } =
jω + 2
so that replacing s by jω in the Laplace transform results in the Fourier transform. However, F{u(t)e2t }
1
does not exist and even though we can let s = jω in the Laplace transform and obtain , we cannot
jω − 2
interpret this as a Fourier transform.
The Fourier transform does possess certain advantages over the Laplace transform. While the Laplace
transform can only be applied to functions which are zero for t < 0, the Fourier transform is applicable to
functions with domain −∞ < t < ∞. In some applications where, for example, t represents not time but
a spatial variable, it it often necessary to work with negative values.
The inverse Fourier transform is given by
Z ∞
1
−1
F {F (ω} = F (ω)ejωt dω
2π −∞
The corresponding inverse Laplace transform requires advanced techniques in the theory of complex vari-
ables which are beyond the scope of this lecture. The existence of the above equation is not quite as
advantageous as it may seem because it is often difficult to perform the required integration analytically.
3 × 108
λ= = 75 m (12.7.10)
4 × 106
which is rather large. For audio frequencies, this value is increased by a factor of 2000, that is, 150,000m.
Consequently, to transmit audio or video signals through the atmosphere in an efficient manner, it is
essential that these signals be moved up in frequency, without losing their information content. This is the
process of modulation and requires the use of a high-frequency carrier wave, which is modified in some way
to include the information content of the original signal. At the receiving end, the signals have to brought
back to their original frequency range. this is the process of demodulation.
12.7.1 Modulation
The basic idea in modulation is to modify the carrier wave in such a way that it contains the information
in the input signal. As a result, a general expression of this modified waveform is given by
where ωc is the radian frequency of the carrier waveform, f (t) is its amplitude, φ(t) is its phase. Note that
both the amplitude and the phase of the modulated signal are assumed to be time-varying in equation
12.7. SOME APPLICATIONS OF FOURIER TRANSFORMS 93
(12.7.11). We consider two special case here. If φ(t) is kept constant but f (t) is allowed to vary according
to the modulating signal, the resulting expression is said to be an amplitude modulation (AM) signal. On
the other hand, if we keep the amplitude constant and allow φ(t) to vary in some way according to the
modulating signal, we get a phase modulated (PM) signal. A special case of phase modulation is called
frequency modulation (FM).
Amplitude modulation
Consider the case when φ(t) is set to a constant value, say 0, and the amplitude of the modulated signal
is expressed as
f (t) = A + λfm (t) (12.7.12)
where A and λ are constants and fm (t) is the modulating signal. Therefore, the resulting modulated signal
is given by the expression
m(t) = [A + λfm (t)] cos ωc t (12.7.13)
Note that the information is contained only in the modulating signal fm (t). Plots of the unmodulated and
modulated carrier are shown in Fig.12.8. It will be seen that although the unmodulated carrier is a pure
sinusoid, after modulation, its envelope takes the shape of the modulating signal.
Figure 12.8: Waveforms of the unmodulated and modulated carrier. (a) Unmodulated carrier. (b) Carrier
after amplitude modulation.
From equation (12.7.13) it is evident that the modulated signal is made up of two terms. The first is
the carrier signal A cos ωc t, while the second term is the product of the modulating signal fm (t) and the
carrier signal λ cos ωc t. The frequency spectrum of the second term can be obtained through convolution
of the modulating signal and carrier in the frequency domain. It can be shown that if the spectrum of the
modulating signal is given by Fm (ω), then the spectrum of the product will be given by
1 1
M (ω) = Fm (ω − ωc ) + Fm (ω + ωc ) (12.7.14)
2 2
This is illustrated in Fig.12.9.
Figure 12.9: Spectra illustrating modulation. (a) Spectrum of modulating signal. (b) Spectrum of modu-
lated signal.
Thus, we note that the effect of amplitude modulation is to produce two sidebands of the carrier fre-
quency, as shown in Fig.12.9, and that the complete information is contained in either of the sidebands.
The procedure for recovering the information from the sidebands will be discussed in the section of demod-
ulation.
94 CHAPTER 12. THE FOURIER TRANSFORM
Frequency modulation
Let us consider the case when the amplitude of the modulated signal is held constant, say equal to A, but
its phase φ(t) varies with time according to the relationship
Z t
φ(t) = λ fm (τ )dτ
0
Phase modulation is obtained if we vary the phase phi according the modulating signal
Hence, the instantaneous frequency of the modulated signal varies sinusoidally between ωc + λB and
ωc − λB. Let us defineδω = λB as the frequency deviation. We obtain the following expression for the
modulated signal by substituting equation (12.7.18) into (12.7.15)
Z t
m(t) =A cos ωc t + λ B cos ωm (t)dt
0
(12.7.19)
δω
=A cos ωc t + sin ωm t
ωm
δω
Defining the factor ωm as the modulation index mi , we may rewrite and simplify equation (12.7.19) as
cos(mi sin ωm t) ≈ 1
and
sin(mi sin ωm t) ≈ mi sin ωm t
Consequently, equation (12.7.20) may be simplified to yield
This is shown that the modulated signal contains sinusoids at frequencies ωc , ωc − ωm and ωc + ωm .
The Fourier transform of m(t), shown in Fig.12.11, is seen to be similar to that obtained by amplitude
modulation with a sinusoidal modulating signal, but now the carrier differs in phase with the modulated
carrier by π/2.
The analysis presented above does not apply for wide-band FM, the case when mi is not small. The
spectrum of the modulated signal for this case depends on both the amplitude and spectrum of the mod-
ulating signal.
In general, the bandwidth of FM signals is larger than that of AM signals, but they are less affected by
noise. However, it is necessary that the frequency of the carrier be tightly controlled. this is often achieved
by using a phase-locked loop, which is a feedback method for locking the phase. The interested reader is
referred to a suitable book on communication systems to learn more about these methods.
12.7.2 Demodulation
Demodulation is the process by which the original information bearing signal is extracted from the modu-
lated signal. This is carried out at the receiver and can be regarded as bring the spectrum of the message
signal back down to its original low-frequency location. Thus, demodulation is the inverse operation of
modulation.
Demodulation can be accomplished in many ways and also depends on the method used for modulation.
For example, an amplitude modulated signal is easily demodulated if we multiply it at the receiving end
with the carrier signal and then pass it through a low-pass filter. This is known as coherent or synchronous
demodulation, since it requires that the carrier frequency signal used at the receiver should be locked in
96 CHAPTER 12. THE FOURIER TRANSFORM
phase with the carrier signal at the transmitter. The basic idea behind this method is evident immediately
by recalling that the spectrum of the amplitude modulated signal is, as shown in Fig.12.12a, centred around
the frequencies −ωc and ωc . Multiplication of this signal by a sinusoid at the frequency of the carrier will
again produce components at frequencies obtained by adding ±ωc to the frequencies of the modulated
carrier. The frequency spectrum of the resulting signal is show in Fig.12.12b. It is easy seen that it
has components at the modulating frequency as well as at frequencies obtained by adding ±2ωc to the
modulating frequency. The low-pass filter effectively removes the high-frequency components and recovers
the modulating signal.
Figure 12.12: Synchronous demodulation. (a) Spectrum of amplitude modulated signal. (b) Spectrum
after multiplication by cos ωc t.
Appendix A
Mathematical Formulas
1 1
1 t
s s√2
n! 1 π
tn (n = 1, 2, 3, · · · ) t2 3
sn+1 2s 2
r
1 π Γ(α + 1)
t− 2 tα
s sα
1 tn e−at 1
e−at
s+a n! (s + a)n+1
1 1 − e−at a
te−at t−
(s + a)2 a 2
s (s + a)
b−a s
e−at − e−bt (1 − at)e−at
(s + a)(s + b) (s + a)2
ω s
sin ωt cos ωt
s2 + ω 2 s2 + ω 2
ω s+a
e−at sin ωt e−at cos ωt
(s + a)2 + ω 2 (s + a)2 + ω 2
2ωs s2 − ω 2
t sin ωt t cos ωt
(s + ω 2 )2
2 (s2 + ω 2 )2
ω cos θ + s sin θ s cos θ − ω sin θ
sin(ωt + θ) cos(ωt + θ)
s2 + ω 2 s2 + ω 2
a s
sinh at cosh at
s 2 − a2 s 2 − a2
δ(t) unit impulse 1
δ(t − a) delayed unit impulse e−as
1
u(t) unit step
s
e−as
u(t − a) delayed unit step
s
1 −as
(t − a)u(t − a) e
s2
• Definition Z ∞
L[f (t)] = F (s) = f (t)e−st dt
0
97
98 APPENDIX A. MATHEMATICAL FORMULAS
• Linearity
L[αf (t) ± βg(t)] = αL[f (t)] ± βL[g(t)]
• First derivative
df (t)
L = sF (s) − f (0)
dt
• Second derivative
d2 f (t)
L = s2 F (s) − sf (0) − f ′ (0)
dt2
• nth derivative
n
dn f (t)
X
n
L = s F (s) − sn−i f i−1 (0)
dtn i=1
• Integration Z t
1
L f (λ)dλ = F (s)
0 s
• Multiplication by time
dF (s)
L[tf (t)] =
ds
• Time shift
L[f (t − a)u(t − a)] = e−as F (s)
• Complex shift
L[e−at f (t)] = F (s + a)
• Time scaling
t
L f = aF (as)
a
• Convolution
L[f (t) ∗ g(t)] = F (s)G(s)
z z
1 k
z−1 (z − 1)2
z(z + 1) z
k2 ak
(z − 1)3 z−a
az(z + a) ak a
k 2 ak ez
(z − a)3 k
z az
ek kak
z−e (z − a)2
z sin b z(z − cos b)
sin bk 2
cos bk 2
z − 2z cos b + 1 z − 2z cos b + 1
ze−a sin b z 2 − ze−a cos b
e−ak sin bk e−ak cos bk
z 2 − 2ze−a cos b + e−2a z 2 − 2ze−a cos b + e−2a
z sinh b z(z − cosh b)
sinh bk 2
cosh bk 2
z − 2z cosh b + 1 z − 2z cosh b + 1
z 1
u(k) u(k − 1)
z−1 z−1
1
u(k) − u(k − 1) 1 u(k − n)
z n−1 (z − 1)
1 1 z
ak−1 u(k − 1) u(k − 1) log
z−a k z−1
1 1
ez δ(k) 1
k!
δ(k − m) z −m
[3] A. Croft, R. Davison & M. Hargreaves, Engineering Mathematics, Third Edition, Pearson, 2001.
[4] R. Finney, M. Weir & F. Giordano, Thomas’ Calculus, Addison Wesley, 2003.
[6] G. James, Modern Engineering Mathematics, Third Edition, Prentice Hall, 2001.
1. http://www.lboro.ac.uk/research/helm/
2. http://www.mathcentre.ac.uk/
3. http://www.math.fsu.edu/Virtual/index.php
4. http://www.mathworks.co.uk/
5. http://tutorial.math.lamar.edu/
101