EE264 Nov 5, 2004

Fall 04–05 Supplemental Notes

There are two version of Laplace Tranforms, the one-sided, and the two-sided, Usually the one
sided is used when we are dealing with causal systems and signals. The Laplace transform of a
function f (t) is a function F (s) of the complex variable s = σ + jω. The transforms are defined
as follows:
Two sided: Z ∞
F (S) = f (t)e−st dt
−∞
One sided: Z ∞
F (s) = f (t)e−st dt
0
Of course this transformations are only defined for those values of s for which the integrals con-
verge. The conditions for convergence are:
Two sided: Z ∞
|f (t)e−σt |dt < ∞
−∞
One sided: Z ∞
|f (t)e−σt |dt < ∞
0
The set of values s for which the above conditions are met in each case is called the Region of
Convergence of the Laplace Transform. Notice that the ROC depends on whether the L.T. is one
sided or two sided. Since the condition for the convergence depends only on the real part of s: σ,
then the ROC’s are the intersections of halfplanes in the complex plane with parallel lines to the
imaginary axis as boundaries. Note that when f (t) is causal, its one sided L.T. is equal to its two
sided L.T.

Example
One sided laplace tranform:
∞ 1
Z
f (t) = u(t) ⇒ F (s) = e−st dt =
0 s
ROC: σ > 0 or Re{s} > 0

Example
One sided laplace tranform:
∞ 1
Z
f (t) = e−at u(t) ⇒ F (s) = e−at e−st dt =
0 s+a
ROC: σ > −a or Re{s} > −a

1
Properties

Obvously, the Laplace transform is linear, and in addition to it, it has several properties. Let
L{f (T )} = F (s) be the one-sided Laplace Transform of f (t), below are some useful properties
of it:
Differentition
¯
( )
df (t)
L = sF (s) − f (0− )
dt
Integration
¯
t F (s)
Z 
L f (t)dt =
0 s
Shift in time
¯
L {f (t − T )u(t − T )} = e−T s F (s)
Initial value theorem
¯
lim sF (s) = f (0+ )
s→∞

Final value theorem

lim sF (s) = lim f (t)

Shift in frequency
¯
n o
L e±at f (t) = F (s ∓ a)
Convolution
If¯ f (t) and g(t) are two causal time functions, then we have

L {f (t) ∗ g(t)} = F (s)G(s)

Multiplication by time
¯
dF (s)
L {tf (t)} = −
ds
Partial fraction expansion

Analogously to the z-transform, we usually work with time functions that have Laplace Trans-
forms that are rational functions of s, therefore, partial fraction expansion together with a table of
the most comon Laplace Transform pairs are extensively used to find the time domain functions
corresponding to a given Laplace Transform that is a rational Function of s. For a reference on

2
Partial Fraction expansion you can look in any Control Textbook.

Example
s+3 2 −2 −1
F (s) = = + +
(s + 1)(s + 2)2 s + 1 s + 2 (s + 2)2
By using tables and/or properties we get:

f (t) = (2e−t − 2e−2t − te−2t )u(t)

Remark:
In this example we considered the one-sided Laplace Transform and we didn’t need the ROC of
F (s); however, when the two sided L.T. is used, then we need to know the ROC in order to decide
which inverse Laplace transform to take, since a single F(s) can have several inverse L.T. depend-
ing on the ROC of it (Analogously to the Z-transform).

Relationship between Laplace and Z-transform

Let’s say we have a continuous time function f (t) with L.T. F (s). We can sample f (t) with a
sampling period T , multiplying it by a train of dirac delta functions to obtain
∞
f ∗ (t) =
X
f (nT )δ(t − nT )
n=−∞

If we know define the sequence f [n] = f (nT ), the we have the follwing important relationship:
∞ ∞
F ∗ (s) = L {f ∗ (t)} = f (nT )e−nsT = f [n](esT )−n = F (z)|z=esT
X X

n=−∞ n=−∞

There are some useful tables that shows a list of time-functions, their Laplace transforms, and the
z-transform of its samples obtained with a sampling period of T, that is F (z)|z=esT = F ∗ (s). It is
sometimes used the following notation: F (z) = Z{F (s)} to denote the z transform of the samples
of f (t) obtained with a sampling period of T .

Zero Order Hold

Consider a system as in the top of page 73 of Handout 15, i.e, feed a sequence f [n] to a Zero
Order Hold DAC as seen in class, then feed the output of the DAC into a LTI system with transfer
function H(s) and finally sample the ouput of it with a sampling period T equal to the pulse length
of the ZOH to obtain a sequence g[n].Then the transfer function that relates F (z) and G(z) is:
( )
G(z) H(s)
= (1 − z −1 )Z
F (z) s