“sued wiopuen uno t4-r amdha
, mm ivivivivi
7 ore ebre
Po eee THT
r+ nek
nek
seb =u
on
; (are
~ ton
som
wm
Unron OQ x-0%
yeu on
F Pe a =
Pn boom
-
ure= (He t-
a Poon1
{
28 re rount axsron chop. 2
domain and the total energy computed in o, the radian frequency domain.
For example, Parseval's Theorem (to be derived in Chapter 4) states.
JO at = 2xat [7 |H(@)/* do (2-46)
tthe energy computed in is required to be equal to the energy computed
ing, then a, = 1/./2x. However, ifthe requirement is made that the Laplace
transform, universally defined as
LNA) = f° koeae = F Koei any
shall reduce to the Fourier transform when the real part of s is set to 2er0,
then & comparison of Eqs. (2-44) and (2-47) requires ag = I, ie = V2
which isin contradiction to the previous hypothesis,
A logical way to resolve this conflict is to define the Fourier transform
pair as follows:
HS) = [me ade (2-48)
mo
With this definition Parsev
"HN ap 2-49)
's Theorem becomes
S_Pod=f impre
and Eq. (2-48) is consistent with the definition of the Laplace transform.
Note that as long as integration is with respect to f, the scale factor |/2n
never appears. For this reason, the latter definition of the Fourier transform
pair was chosen for this book.
25 FOURIER TRANSFORM PAIRS
A pictorial table of Fourier transform pairs is given in Fig, 2-11. This
‘graphical and analytical dictionary is by no means complete but does contain
the most frequently encountered transform pairs,
PROBLEMS
2A, Determine the real and imaginary parts of the Fourier transform of each of
the following function
aE ce
bk ibe
wwo=[E 120
Size
“Par FounuER TRaNeroRN 29
<1 1<0
caon fe" 150
|
acoraniy 10
wef
o 10
A acecbiab>o
. w-ff
5
dewin Ga) 120
Ho ={f aD
KO = F[ +) + 50) + 3(6 + $) +(e $)]
22 Determine the amplitude spectrum |(J)] and phase 6(/) of the Fourier
transform of KO:
eMnT me .
For example, Papoulis (2, page 44 gives
mO- 3 u—nTy Heo) ein
By the frequency scaling elationsip (1-16) we know that
3S aaa(s— 9) FE a(s- G8)36 rune reason PnorenTes
2S aay,
Jao
a
Chap.
0
non
eat
vA 7
r
bey
® e ‘
ou ws
° sna
By 7
— iN
Figure 3.3. Frequency sealing property.
See 36 unin Tunsront rrorenras 37
and (-17) can be rewritten in txms ofthe frequency varable /
wo= 3 de- ary EP HN=
whichis Eq, 240),
#) ow
3 TIME-SHIFTING
If H(@) is shifted by a constant ‘4 then by substituting = — 4, the
Foutier transform becomes
[mem iectede = J aera
tn Foie as
OH) 20)
‘The time-shifted Fourier transform pair is
men) OQ Mpen em
ExaMrns 3.6
‘A pictorial description of this pair is illustrated in Fig. 34, As shown, time=
shifting results in a change in the phase angle (f) = tan I/(//R(/)}. Note that
time-shifting does not alter the magnitude of the Fourier transform, This follows
Hi fet = H(f eos (2x fe) — Jin 2a fie)
land hence the magnitude is given by
[He Bt6| = ARTIC ORT + TRIN = VT) (3-22)
where H(J) has been assumed to be real for simplicity. These results are easily
extended tothe case of H(/), a complex function.
36 FREQUENCY SHIFTING
If H(J) is shifted by a constant f, its inverse transform is multiplied by
oem ED NF-H) eon)
‘This Fourier transform pair is established by substituting s = f— fg into
the inverse Fourier transform-defining relationship
Jae —so0%n ar = [meyers ds
= ere ("Hise ds
mt Ke) om‘38 oun raanaronn prorenris hap. 3 See. 36 ounien reoserona enorenres 39
Exar 3-7
‘Fo illustrate the effect of frequency-shiftng let us assume that the frequency
function H(i real. For this case, frequency shifting results ina multiplication of|
the time function A) by a cosine whose frequency is determined by the frequency
shift f (Fig. 3-5). This process is commonly known as moculaton.
n~7 I YA
men ni 19
[ae
es
ee
mt 24)
=
Figure 34. Time shifting propery
Figure 3.5. Frequency siting property