You are on page 1of 9

TFY Signal Processing

Lecture 7A

Introduction to the Laplace transform (I)

T7A.pdf and VL7A.mp4


The Laplace transform – introduction (7.1 – 7.3) For using the Laplace transform in causal ‘initial
The Laplace transform is basically an extension of value problems’ we first express the Laplace
the Fourier transform but allowing the ‘imaginary transform as:
frequency’ being complex instead of purely 0 ∞
−𝑠𝑡
imaginary. 𝑠 = 𝜎 + 𝑖𝜔 ℒ𝑏 𝑓 𝑡 =න 𝑓 𝑡 ∙𝑒 𝑑𝑡 + න 𝑓 𝑡 ∙ 𝑒 −𝑠𝑡 𝑑𝑡
−∞ 0

ℒ𝑏 𝑓 𝑡 = 𝐹𝑏 𝑠 = න 𝑓 𝑡 ∙ 𝑒 −𝑠𝑡 𝑑𝑡 For the situation 𝑓 𝑡 = 0; 𝑡 < 0 we may define
−∞
the unilateral (or single-sided) Laplace transform:
‘b’ here denotes bilateral (or two-sided), of which
the meaning will become apparent in a moment. ∞
ℒ 𝑓 𝑡 = 𝐹 𝑠 = න 𝑓 𝑡 ∙ 𝑒 −𝑠𝑡 𝑑𝑡
0
The inverse transform follows the scheme of the
Fourier counterpart, but here we must make a The latter, we usually refer to as ‘the Laplace
contour integration in the complex plane. transform’. Both forms are of course linear
transformation suitable for analyzing LTI systems.
−1
1 𝑐+𝑖∞
𝑓 𝑡 =ℒ 𝐹 𝑠 = න 𝐹 𝑠 ∙ 𝑒 𝑠𝑡 𝑑𝑠
2𝜋𝑖 𝑐−𝑖∞ Region of convergence (ROC).
Consider the Laplace transform of 𝑢 𝑡
Im 𝒔 Fortunately, in practice we ∞
1 −𝑠𝑡 ∞
find the inverse transform ℒ 𝑢 𝑡 = න 1 ∙ 𝑒 −𝑠𝑡 𝑑𝑡 = − 𝑒 0 =
𝜎<0 𝜎>0 using other techniques than
0 𝑠
integration.
𝜔>0 𝜔>0 1 1
=− lim 𝑒 −𝑠𝑡 − 1 =
𝑠 𝑡→∞ 𝑠

𝜎<0 𝜎>0 Re 𝒔 This converges only if Re 𝑠 > 0


𝜔<0 𝜔<0 ROC will be important for the stability of a
system/signal.
The Laplace transform – introduction (7.1 – 7.3) Consider the LT of a general function:

Example 1 ℒ 𝑓 𝑡 = න 𝑓 𝑡 ∙ 𝑒 −𝑠𝑡 𝑑𝑡 = ?
0
So let’s find the LT of 𝑓 𝑡 = 𝑒 −𝑎𝑡 a decaying
Integration
exponential function. From the definition: 𝑏 𝑒 −𝑠𝑡 𝑓 𝑡 by parts
∞ ∞ න 𝑔′ 𝑥 ∙ 𝑓 𝑥 𝑑𝑥 =
−𝑎𝑡 −𝑠𝑡
ℒ 𝑢 𝑡 =න 𝑒 ∙𝑒 𝑑𝑡 = න 𝑒 − 𝑠+𝑎 𝑠𝑡
𝑑𝑡 𝑎
𝑏
𝑏
0 0 𝑔 𝑥 ∙𝑓 𝑥 𝑎 − ‫𝑓 ∙ 𝑥 𝑔 𝑎׬‬′ 𝑥 𝑑𝑥
1 𝑠+𝑎 𝑡 ∞
1
=− 𝑒− =− lim 𝑒 −(𝑠+𝑎)𝑡 − 1 =
𝑠+𝑎 0 𝑠+𝑎 𝑡→∞
ℒ 𝑓 𝑡 =
1 ∞ ∞
= 𝑒 −𝑠𝑡 𝑒 −𝑠𝑡 𝑑𝑓 𝑡
Provided Re 𝑠 + 𝑎 > 0 = − ∙𝑓 𝑡 −න − ∙ 𝑑𝑡 =
𝑠+𝑎 𝑠 𝑠 𝑑𝑡
0 0

1 +
1 ∞ 𝑑𝑓 𝑡 −𝑠𝑡
= 𝑓 0 + න 𝑒 𝑑𝑡 Multiply
𝑠 𝑠 0 𝑑𝑡
both sides
F(s) ∞
with s
+
𝑑𝑓 𝑡 −𝑠𝑡
𝑠ℒ 𝑓 𝑡 =𝑓 0 +න 𝑒 𝑑𝑡
Re(s) > 0 0 𝑑𝑡
𝑑𝑓 𝑡
ℒ = 𝑠ℒ 𝑓 𝑡 − 𝑓 0+
Re(s) > - a 𝑑𝑡

Differentiation theorem of LT. Take into account


So how to deal with the initial values and the initial value of the function to transform.
differentiation ?
The Laplace transform – introduction (7.1 – 7.3) Partial fraction expansion
Example 2 In order to find the inverse transform we start by
Let’s use FT to find the current and voltage of an utilizing the technique of partial fraction
initial value problem. Let’s assume a constant expansion (PFE). Let us assume:
voltage is applied in the following situation: 1 𝑎 𝑏
= +
𝑠 𝑅/𝐿 + 𝑠 𝑠 𝑅
𝑠+
𝐿
It follows, 𝑅
1 𝑎 𝑠+ + 𝑏𝑠
= 𝐿
𝑠 𝑅/𝐿 + 𝑠 𝑅
𝑠 𝑠+
𝑣𝑙 (𝑡) 𝐿
and by identification to powers in ‘s’,
𝑑𝑖 𝑡 𝑅 𝐿
𝑉∙𝑢 𝑡 =𝑅∙𝑖 𝑡 +𝐿 𝑠0: 1 = 𝑎 →𝑎=
𝐿 𝑅
𝑑𝑡
𝐿
Laplace transform using what we know… 𝑠1 : 0 = 𝑎𝑠 + 𝑏𝑠 → a + b = 0 → b = −
𝑅
=0
1
𝑉 ∙ = 𝑅 ∙ 𝐼 + 𝑠𝐿 𝐼 + 𝑖 0+ = 𝑅 + 𝑠𝐿 𝐼 So, these results we put back into I(s)
𝑠
𝑉 𝐿1 𝐿 1
𝐼 𝑠 = ∙ −
1 𝑉 1 𝐿 𝑅𝑠 𝑅 𝑠+𝑅
𝐼 𝑠 =𝑉∙ = ∙ 𝐿
𝑠 𝑅 + 𝑠𝐿 𝐿 𝑠 𝑅/𝐿 + 𝑠
We readily found the LT of the current I(s), so now and finally 𝑉 1 1
𝐼 𝑠 = ∙ −
we only need to find i(t) giving this LT. get, 𝑅 𝑠 𝑅
𝑠+
𝐿
The Laplace transform – introduction (7.1 – 7.3)
Using Table 7.1 we now easily can identify the
Example 2, cont time-domain functions of I(s) and VL(s)
𝑉 1 1
We had 𝐼 𝑠 = ∙ −
𝑅 𝑠 𝑅
𝑠+
𝐿

𝑣𝑙 (𝑡) 𝑉 𝑅
𝑖 𝑡 = ∙ 𝑢 𝑡 − 𝑒 −𝐿 𝑡 𝑢 𝑡
𝑅
𝑅
𝑣𝑙 𝑡 = 𝑉 𝑒 − 𝐿 𝑡 𝑢 𝑡
Now what about the voltage? Let’s examine the
voltage of the inductance vl(t). and for simplicity let V = R = L = 1 in magnitude
to plot the time profiles:
Ohm’s law for impedances:
L/R is here the
𝑉 1 1 time-constant t
𝑉𝐿 𝑠 = 𝑍𝐿 ∙ 𝐼 = 𝑠𝐿 −
𝑅 𝑠 𝑅
𝑠+ The time-constant 𝜏 of an
𝐿
exponential funtion 𝑒 −𝑡/𝜏 is
1
defined at amplitude 𝑒 = 0.3679
𝑅
𝐿 𝑠 𝐿 𝑠+ −𝑠 1
=𝑉 1− =𝑉 𝐿 =𝑉
𝑅 𝑅 𝑅 𝑅 𝑅
𝑠+ 𝑠+ 𝑠+
𝐿 𝐿 𝐿
The Laplace transform – introduction (7.1 – 7.3)
Example 3, Laplace transform of a function The Laplace Transform of most functions is
Let’s find the LT of a typical function: cos 𝑏𝑡 readily obtained by standard integration.

ℒ𝑏 cos 𝑏𝑡 = න cos 𝑏𝑡 ∙ 𝑒 −𝑠𝑡 𝑑𝑡 It is also possible to make numerical
−∞ integration if the function does not have an
To get the unilateral transform multiply with 𝑢 𝑡 analytical form (say some experimental
input signal). →
∞ ∞
−𝑠𝑡
ℒ cos 𝑏𝑡 = න cos 𝑏𝑡 𝑢 𝑡 𝑒 𝑑𝑡 = න cos 𝑏𝑡 𝑒 −𝑠𝑡 𝑑𝑡
−∞ 0


1 𝑖𝑏𝑡
=න 𝑒 + 𝑒 −𝑖𝑏𝑡 𝑒 −𝑠𝑡 𝑑𝑡 =
0 2 Euler formula → sin
1 ∞ − 𝑠−𝑖𝑏 𝑡 1 ∞ − 𝑠−𝑖𝑏 𝑡
= න 𝑒 + 𝑒− 𝑠+𝑖𝑏 𝑡
𝑑𝑡 = ℒ sin 𝑏𝑡 = න 𝑒 − 𝑒 − 𝑠+𝑖𝑏 𝑡
𝑑𝑡
2 0 2𝑖 0
∞ ∞
1 −1 1 1 −1 1
= 𝑒 − 𝑠−𝑖𝑏 𝑡 − 𝑒 − 𝑠+𝑖𝑏 𝑡 = 𝑒− 𝑠−𝑖𝑏 𝑡
+ 𝑒− 𝑠+𝑖𝑏 𝑡
2 𝑠 − 𝑖𝑏 𝑠 + 𝑖𝑏 2𝑖 𝑠 − 𝑖𝑏 𝑠 + 𝑖𝑏 0
0

As previously, this converges only if Re 𝑠 > 0 As previously, this converges only if Re 𝑠 > 0

1 1 1 1 𝑠 + 𝑖𝑏 + 𝑠 − 𝑖𝑏 1 1 1 1 𝑠 + 𝑖𝑏 − 𝑠 − 𝑖𝑏
= + = = − =
2 𝑠 − 𝑖𝑏 𝑠 + 𝑖𝑏 2 𝑠2 + 𝑏2 2𝑖 𝑠 − 𝑖𝑏 𝑠 + 𝑖𝑏 2𝑖 𝑠2 + 𝑏2

∞ ∞
−𝑠𝑡
𝑠 𝑏
ℒ cos 𝑏𝑡 = න cos 𝑏𝑡 𝑒 𝑑𝑡 = 2 ℒ sin 𝑏𝑡 = න sin 𝑏𝑡 𝑒 −𝑠𝑡 𝑑𝑡 =
0 𝑠 + 𝑏2 0 𝑠2 + 𝑏2
Script for calculating LT of cos bt and plot in complex plane using both numerical
integration and directly from formulas as s/(s^2+b^2) in the Laplace transform table
clear all; close all;
t=[0:0.01:300]; % time vector
sigma=[0.05:0.05:2]; % real part of s, n.b. Real(s) > 0 𝑠 = 𝜎 + 𝑖𝜔
omega=[-2:0.05:2]; % imaginary part of s
H_n=zeros(length(sigma),length(omega)); % allocate memory for Hn
H_a=H_n; % allocate memory for H_a & H_b We must integrate for
H_b=H_n; all possible s and w in
b=1; % set frequency for cosine the complex plane
for m=1:length(sigma)
for n=1:length(omega)

s=sigma(m)+sqrt(-1)*omega(n); % assigning s value
𝑭 𝒔 = න 𝒇 𝒕 ∙ 𝒆−𝒔𝒕 𝒅𝒕
f=cos(b*t).*exp(-s*t); % calculating integrand 𝟎
H_n(m,n)=trapz(t,f); % integral for LT 𝒔
H_a(m,n)=s/(s*s+b*b); %calculating 'analogue' response from table formula 𝒔𝟐 + 𝒃𝟐
H_b(m,n)=0.5*(1/(s-sqrt(-1)*b)+1/(s+sqrt(-1)*b)); % calc from partial fraction intermediate
end 𝟏 𝟏 𝟏
end +
𝟐 𝒔 − 𝒊𝒃 𝒔 + 𝒊𝒃
figure
surf(omega,sigma,abs(H_n)) % 3D plot of real(H) vs. real and imag part of s
ylabel('\sigma');xlabel('\omega');zlabel('abs[H(s)]');
box on;
s
Direct integration
s 2 + b2

1 1 1  Plot of Laplace Transform of cos bt in the complex


+
2  s − ib s + ib 
s-plane. b = 1
N.b. 1: ROC; s > 0 (Re(s)) for the numerical
integral to converge.
N.b. 2: Along the imaginary axis (±w) we get
something like the FT.
So the Laplace Transform can be understood as
a Fourier transform extended over the whole
complex plane.
The Laplace transform –
introduction (7.1 – 7.3)

Next time: more on LT


properties (7.4-7.7)

You might also like