Tutorial 6 - CT Fourier Transform (Solutions) PDF
Tutorial 6 - CT Fourier Transform (Solutions) PDF
Faculty of Engineering
Department of Electrical and Electronics Engineering
conceptual questions
1. Define the CT Fourier Transform (CTFT). Is the CTFT an operation that changes the
signal x (t) in time or an equivalent representation of the signal x (t) in the frequency
domain? Justify your answer
The CT Fourier Transform of a signal x (t) is the representation of x (t) in the frequency
domain, where it is reconstructed using complex exponential functions of the form
e− jωt ∀ ω ∈ (−∞, ∞).
3. Do the signals x1 (t) = sin (ω0 t) and x2 (t) = u (t) fulfill the Dirichlet conditions? If
yes/no, justify your answer. Why do have their CT Fourier Transforms in common?
No. Neither of them is absolutely integrable from −∞ to ∞. Their CT Fourier Transforms
are given by
F {sin (ω0 ) t} = πj [δ (ω + ω0 ) − δ (ω − ω0 )]
1
F {u (t)} = + πδ (ω )
jω
Both Fourier Transforms have singularity functions in the frequency domain. They ap-
pear because of the fact that both signals do not satisfy the first Dirichlet condition.
4. (True or False) Every practical signal has a complex-valued CTFT. Justify your answer
True. Most practical signals are positive-time (i.e., they are defined for t ≥ 0 where t = 0
is the instant at which a signal starts being studied). Under this condition, only cosine
signals will have a real-valued spectrum (see Exercise 3). And these particular wave-
form does not represent the broad variety of signals that arise in practical applications.
Therefore, it is worth to assume that every practical signal has a complex-valued CTFT.
5. Are the CT Fourier Series (CTFS) more general than the CTFT? Or is the CTFT more
general than the CTFS? Justify your answer
The CTFT is more general than the CTFS. A CTFS representation is limited just to peri-
odic signals. The CTFT can be applied both to periodic and aperiodic signals. There are
another points that are worth to mention:
• The value of the CTFS coefficients do not change as the periodic signal is com-
pressed or expanded, even when the signal itself is different. In such cases, the
corresponding CT Fourier Transform is a different and unique function.
• While the CTFT is a function of the frequency ω, the CTFS coefficients are not
explicitly a function of the frequency. They only make sense when used within the
synthesis equation.
exercises
Thus,
F {δ (t)} = 1
b) x (t) = e jω0 t (Hint: use the result of the previous exercise and the synthesis equa-
tion)
By following the hint, we have that
Z ∞
1
F −1 { 1 } = (1) e jωt dω = δ (t)
2π −∞
Then, Z ∞
1
e jωt dω = δ (t)
2π −∞
n o Z ∞
F e jω0 t = e jω0 t e− jωt dt
−∞
Z ∞
= e j(ω0 −ω )t dt
−∞
c) x (t) = e− jω0 t
If we apply the previous result with ω0 = −ω0 we get
n o
F e− jω0 t = 2πδ (ω + ω0 )
d) x (t) = sin (ω0 t) (Hint: make use of the linearity property1 )
To start with, we write sin (ω0 t) in terms of complex exponentials
1 jω0 t
sin (ω0 t) = e − e− jω0 t
2j
Hence,
F {sin (ω0 t)} = πj [δ (ω + ω0 ) − δ (ω − ω0 )]
1 jω0 t
cos (ω0 t) = e + e− jω0 t
2
Therefore,
1 jω0 t
F {cos (ω0 t)} = F e + e− jω0 t
2
1 n jω0 t o 1 n − jω0 t o
F e + F e
2 2
1
[2πδ (ω − ω0 ) + 2πδ (ω + ω0 )]
2
π [δ (ω + ω0 ) + δ (ω − ω0 )]
1 :∞
= (∞
− 1)
2 − jω
F
1 Linearity Property: For all ᾱ1,2 ∈ C, we have that ᾱ1 x1 (t) + ᾱ2 x2 (t) ↔ ᾱ1 X1 ( jω ) + ᾱ2 X2 ( jω )
Z ∞
− at
e− at u (t) e− jωt dt
F e u (t) =
−∞
Z ∞
= e−( jω + a)t dt
0
−1 ∞
= e−( jω +a)t
jω + a t =0
−1 :0
−
at − jωt
−1
= lim e e
jω + a t→∞
1
=
jω + a
Consequently,
1
F e−at u (t) =
jω + a
2. Consider the periodic pulse train x̃ (t) shown in Figure 1. Compute the CT Fourier
Transform of x̃ (t)
The CT Fourier Transform of this signal is computed based on its CTFS coefficients. The
period of the signal is T0 = T = 3 s. Then, the CTFS of x̃ (t) are given by
1
Z
c̄k = x̃ (t) e− jkω0 t dt
T <T >
1 1 − jk 2π t
Z
= e 3 dt
3 0
1 3 2π t =1
= e− jk 3 t
3 − jk2π t =0
−1 2π
= e− jk 3 − 1
j2πk
−1 − jk π − jk π
e 3 e 3 − e jk 3 e− jk 3
π π
=
j2πk
−1 − jk π − jk π π
= e 3 e 3 − e jk 3
j2πk | {z }
−2j sin(k π3 )
−1 − jk π π
= e 3 −2j sin k
j2πk 3
1 − jk π π
= e 3 sin k
πk 3
Consequently, the CT Fourier Transform of x̃ (t) is
∞
X̃ ( jω ) = ∑ 2π c̄k δ (ω − kω0 )
k =−∞
c̄
}|k
∞
z
{
1 − jk π π 2π
= ∑ 2π e 3 sin k δ ω−k
k =−∞
πk 3 3
∞
e− jk 3
π
π 2π
=2 ∑ sin k δ ω−k
k =−∞
k 3 3
3. Show that if x (t) is real and even, then X ( jω ) is real and even. Can the spectrum of
a positive-time signal (i.e., x (t) = 0 for t < 0) be real?
Hypothesis: x (t) is real and even
The integral can be rewritten as two parts: one ranging from −∞ to zero and the other
from 0 to ∞. That is,
Z ∞ Z 0 Z ∞
X ( jω ) = x (t) e− jωt dt = x (t) e− jωt dt + x (t) e− jωt dt
−∞ −∞ 0
t → −∞ ⇒ τ → ∞
t→0⇒τ→0
Correspondingly,
Z 0 Z ∞
X ( jω ) = − x (−τ ) e jωτ dτ + x (t) e− jωt dt
∞ 0
By reversing the order of the integration limits of the first term, we get
Z ∞ Z ∞
X ( jω ) = x (−τ ) e jωτ + x (t) e− jωt dt
0 0
Lastly, we must verify whether X ( jω ) is even. We can prove this as follows. Recall the
last expression for X ( jω )
Z ∞
X ( jω ) = 2 x (t) cos (ωt) dt
0
Now, we evaluate X (− jω )
Z ∞
X (− jω ) = 2 x (t) cos (−ωt) dt
0
Thus,
X ( jω ) = X (− jω )
Thereby, we conclude that if x (t) is real and even, then X ( jω ) is also real and even.
Q.E.D.
If we consider all real-world signals to be positive-time, will we encounter real-valued
spectra in practice?
If x (t) is a positive-time signal, then
x (t) = 0 t < 0
We will rarely found an x (t) that satisfies this condition (i.e., orthogonal to sin (ωt) from
0 to ∞). An example is cos (kωt) k ∈ R. Nevertheless, most practical signals do not have
this specific form. Thus, we will rarely encounter real-valued spectra in practice.
4. Show that the steady-state response y (t) to an input x (t) = sin (ω0 t) of an LTI, BIBO
stable and causal system with impulse response h (t) = F −1 { H ( jω )} equals
Note that this relationship is the basis of steady-state phasor analysis in Linear Circuit
Theory and Power System Analysis
Hypothesis: the LTI system H is BIBO stable. This implies that its impulse response
h (t) is absolutely integrable and its Fourier Transform exists because the first Dirichlet
condition is fulfilled.
Since the system is LTI, the response to the input x (t) = sin (ω0 t) can be computed
through the convolution integral as
The last equation can be rewritten by means of the convolution commutative property.
Then,
Z ∞
y (t) = x (τ ) h (t − τ ) dτ
−∞
Z ∞
h (τ ) x (t − τ ) dτ
−∞
Now, after using the trig id sin (u − v) = sin u cos v − sin v cos u, we have
Z t
y (t) = h (τ ) [sin (ω0 t) cos (ω0 τ ) − sin (ω0 τ ) cos (ω0 t)] dτ
−∞
Correspondingly,
∞ ∞
Z Z
yss = lim y (t) = sin (ω0 t) h (τ ) cos (ω0 τ ) dτ − cos (ω0 t) h (τ ) sin (ω0 τ ) dτ
t→∞ −∞ −∞
Due to the fact that h (t) is absolutely integrable, its CTFT exists and is given by
Z ∞
H ( jω ) = h (t) e− jωt dt
−∞
By employing Euler’s formula, the last integral can be expressed in rectangular form as
Z ∞
H ( jω ) = h (t) [cos (ωt) − j sin (ωt)] dt
−∞
Z ∞ Z ∞
= h (t) cos (ωt) dt − j h (t) sin (ωt) dt
| −∞ {z } | −∞ {z }
Re{ H ( jω )} −Im{ H ( jω )}
Now, we substitute this result into the steady-state expression for y (t). Therefore, we
have
Z ∞ Z ∞
yss = lim y (t) = sin (ω0 t) h (τ ) cos (ω0 τ ) dτ − cos (ω0 t) h (τ ) sin (ω0 τ ) dτ
t→∞ −∞ −∞
| {z } | {z }
Re{ H ( jω0 )} −Im{ H ( jω0 )}
That is,
where q
| H ( jω0 )| = (Re { H ( jω0 )})2 + (Im { H ( jω0 )})2
Im { H ( jω0 )}
∠ H ( jω0 ) = tan−1
Re { H ( jω0 )}
Q.E.D.