Sampling of Continuous-Time Signals

Dr. Sujit Kumar Sahoo

Assistant Professor,SES
IIT Goa

School of Electrical Sciences

IIT Goa

October 11, 2023

Dr. Sujit Kumar Sahoo (IIT GOA) Sampling of Continuous-Time Signals October 11, 2023 1 / 61
Sampling of Continuous-Time Signals
Periodic Sampling:
x[n] = xc (nT ), −∞ < n < ∞

Figure: Block diagram representation of an ideal continuous-to-discrete-time (C/D)

converter
∞
X
Impulse train: s(t) = δ(t − nT )
n=−∞
Sampled signal: ∞
X
xs (t) = xc (t)s(t) = xc (t) δ(t − nT )
n=−∞
∞
X ∞
X
= xc (t)δ(t − nT ) = xc (nT )δ(t − nT )
n=−∞ n=−∞

Dr. Sujit Kumar Sahoo (IIT GOA) Sampling of Continuous-Time Signals October 11, 2023 2 / 61
Sampling of Continuous-Time Signals
s(t)

Dr. Sujit Kumar Sahoo (IIT GOA) Sampling of Continuous-Time Signals October 11, 2023 3 / 61
Frequency domain Representation of Sampling
P∞
What is the Fourier transform of impulse train s(t) = n=−∞ δ(t − nT )?
It is a periodic signal with period T . The Fourier series coefficients of s(t) are
1 T /2
Z
ck = s(t)e −j2πkt/T dt
T −T /2
∞
1 T /2 X
Z
= δ(t − nT )e −j2πkt/T dt
T −T /2 n=−∞

1 T /2
Z
1
= δ(t)e −j2πkt/T dt =
T −T /2 T
So, the periodic impulse train can also be expressed as
∞ ∞ ∞
X X 1 X j2πkt/T
s(t) = δ(t − nT ) = ck e j2πkt/T = e
n=−∞
T
k=−∞ k=−∞

and its Continuous Time Fourier Transform (CTFT) is

∞ ∞
2π X 2πk 2π X
F [s(t)] = S(jΩ) = δ(Ω − )= δ(Ω − kΩs ),
T T T
k=−∞ k=−∞
2π
where Ωs = T is the sampling frequency in radians/sec.
Dr. Sujit Kumar Sahoo (IIT GOA) Sampling of Continuous-Time Signals October 11, 2023 4 / 61
Frequency domain Representation of Sampling

CTFT 1
xs (t) = xc (t)s(t) ↔ Xs (jΩ) =
Xc (jΩ) ∗ S(jΩ)
2π
where ∗ denotes the operation of continuous variable convolution. It results
∞ ∞
!
1 2π X 1 X
Xs (jΩ) = Xc (jΩ) ∗ δ(Ω − kΩs ) = Xc (j(Ω − kΩs ))
2π T T
k=−∞ k=−∞

Dr. Sujit Kumar Sahoo (IIT GOA) Sampling of Continuous-Time Signals October 11, 2023 5 / 61
Frequency domain Representation of Sampling

Dr. Sujit Kumar Sahoo (IIT GOA) Sampling of Continuous-Time Signals October 11, 2023 6 / 61
Frequency domain Representation of Sampling

How do you reconstruct back the signal?

where Hr (jΩ) is a continuous time system that help us reconstructs the

continuous time signal.
Xr (jΩ) = Hr (jΩ)Xs (jΩ),
where Xr (jΩ) is its Fourier transform of the reconstructed continuous time signal,
xr (t) ≃ xc (t).

Dr. Sujit Kumar Sahoo (IIT GOA) Sampling of Continuous-Time Signals October 11, 2023 7 / 61
Frequency domain Representation of Sampling
If the sampled signal
spectrum, Xs (jΩ), satisfies

ΩN ≤ Ωc ≤ (Ωs − ΩN ),

and we have Hr (jΩ) as

shown in the figure.

⇒ Xr (jΩ) = Xc (jΩ)

therefore, xr (t) = xc (t)

Dr. Sujit Kumar Sahoo (IIT GOA) Sampling of Continuous-Time Signals October 11, 2023 8 / 61
Example:

Dr. Sujit Kumar Sahoo (IIT GOA) Sampling of Continuous-Time Signals October 11, 2023 9 / 61
Nyquist Shannon Sampling Theorem

Deinition: Let xc (t) be a bandlimited signal with

Xc (jΩ) = 0 for |Ω| ≥ ΩN

then xc (t) is uniquely determined by its samples x[n] = xc (nT ), n = 0, ±1, ±2, ....
if
2π
Ωs = ≥ 2ΩN
T

Dr. Sujit Kumar Sahoo (IIT GOA) Sampling of Continuous-Time Signals October 11, 2023 10 / 61
Continuous time FT (CTFT) & Discrete time FT (DTFT)
Let’s take the CTFT of the sampled signal xs (t), which a series of impulses.
" ∞ #
X
F [xs (t)] = F xc (nT )δ(t − nT )
n=−∞
∞ Z
X ∞
Xs (jΩ) = xc (nT )δ(t − nT )e −jΩt dt
n=−∞ −∞

X∞ ∞
X
Xs (jΩ) = xc (nT )e −jΩTn = [n]e −jΩTn [∵ xc (nT ) = x[n]]
n=−∞ n=−∞

Now, let’s look at the DTFT of x[n]

∞
X
X (e jω ) = x[n]e −jωn
n=−∞

From above expressions of Xs (jΩ) and X (e jω ) we can draw the flowing relation.
Xs (jΩ) = X (e jω )|ω=ΩT = X (e jΩT )

Dr. Sujit Kumar Sahoo (IIT GOA) Sampling of Continuous-Time Signals October 11, 2023 11 / 61
Continuous time FT (CTFT) & Discrete time FT (DTFT)
Another from of writing the CTFT of xs (t) is the periodic sum representation,
∞
1 X
Xs (jΩ) = Xc (j(Ω − kΩs ))
T
k=−∞

Using the derived relation X (e jω ) = Xs (jΩ)|Ω= jω = Xs ( Tω ), the DTFT of x[n] can

T
be expressed as
∞
jω ω 1 X  ω 
X (e ) = Xs ( ) = Xc j − kΩs
T T T
k=−∞
∞   
1 X ω 2π
= Xc j − k
T T T
k=−∞

If Xc (Ω) is band limited to the frequency Ωs ,it will not cause any aliasing in the
spectrum. Thus we can write,

1  ω
X (e jω ) = Xc j for |ω| < π
T T
Dr. Sujit Kumar Sahoo (IIT GOA) Sampling of Continuous-Time Signals October 11, 2023 12 / 61
Continuous time FT (CTFT) & Discrete time FT (DTFT)
Example:
If we sample the continuous time signal xc (t) = cos(4000πt) with sampling
period T = 1/6000, we obtain x[n] = xc (nT ) = cos(4000πTn) = cos(ω0 n),
where ω0 = 4000πT = 2π 2π
3 . In this case, Ω = T = 12000π, and the highest
frequency of the signal is Ω0 = 4000π, so the conditions of the Nyquist sampling
theorem are satisfied.

Dr. Sujit Kumar Sahoo (IIT GOA) Sampling of Continuous-Time Signals October 11, 2023 13 / 61
Reconstruction of a Band limited signal from its samples

The following the internal structure of Hr (jΩ) = CTFT {hr (t)}

the D/C block.

The time domain view of the ideal

reconstruction filter hr (t) = sin(πt/T
πt/T .
)

Dr. Sujit Kumar Sahoo (IIT GOA) Sampling of Continuous-Time Signals October 11, 2023 14 / 61
Reconstruction of a Band limited signal from its samples

∞
X
xs (t) = x[n]δ(t − nT )
n=−∞

xr (t) = xs (t) ∗ hr (t)

X ∞
= x[n]hr (t − nT )
n=−∞
∞
X sin[π(t − nT )/T ]
= x[n]
n=−∞
π(t − nT )/T

It’s because δ(t − τ ) ∗ hr (t) = hr (t − τ )

Dr. Sujit Kumar Sahoo (IIT GOA) Sampling of Continuous-Time Signals October 11, 2023 15 / 61
Discrete time Processing of Continuous time Signals

∞
sin[π(t − nT )/T ]
X
yr (t) = y [n]
n=−∞
π(t − nT )/T
(
jΩT TY (e jΩT ), |Ω| < π/T
Yr (jΩ) = Hr (jΩ)Y (e )=
0, otherwise

Dr. Sujit Kumar Sahoo (IIT GOA) Sampling of Continuous-Time Signals October 11, 2023 16 / 61
Discrete-Time LTI Processing of Continuous-Time Signals
Frequency domain representation of discrete system
Y (e jω ) = H(e jω )X (e jω )
Effective analogue output
Yr (jΩ) = Hr (jΩ) H(e jΩT )X (e jΩT )
| {z }
discrete system
∞
(   )
jΩT 1 X 2πk
= Hr (jΩ) H(e ) Xc j Ω −
T T
k=−∞
(
jΩT π
H(e )Xc (jΩ), |Ω| < T
Yr (jΩ) = π
If Xc (jΩ) = 0 for |Ω| ≥ π/T
0, |Ω| ≥ T
= Heff (jΩ)Xc (jΩ).
where
(
π
H(e jΩT ), |Ω| < T
Heff (jΩ) = π
0, |Ω| ≥ T

Dr. Sujit Kumar Sahoo (IIT GOA) Sampling of Continuous-Time Signals October 11, 2023 17 / 61
Example

Ideal Continuous-Time Lowpass Filtering Using a Discrete-Time Lowpass

Filter
(
jω 1, |ω| < ωc
H(e ) =
0, ωc < |ω| ≤ π

1
Rω sinωc n
h[n] = 2π −ωc
1.e jωn dω = πn

(
1, |Ω| < ωc /T
Heff (jΩ) =
0, |Ω| ≥ ωc /T

Dr. Sujit Kumar Sahoo (IIT GOA) Sampling of Continuous-Time Signals October 11, 2023 18 / 61
Example

Dr. Sujit Kumar Sahoo (IIT GOA) Sampling of Continuous-Time Signals October 11, 2023 19 / 61
Example
Discrete-Time implementation of an Ideal Continuous-Time Bandlimited
Differentiator
Continuous-time domain system
d
yc (t) = [xc (t)]
dt
Continuous-time frequency response

Hc (jΩ) = jΩ

Effective frequency response

(
jΩ, |Ω| < π/T
Heff (jΩ) =
0, |Ω| ≥ π/T

Corresponding discrete frequency response

jω
H(e jω ) = , |ω| < π
T
Dr. Sujit Kumar Sahoo (IIT GOA) Sampling of Continuous-Time Signals October 11, 2023 20 / 61
Example

Z π  
1 jω πncosπn − sinπn
h[n] = e jωn dω = , −∞ < n < ∞
2π −π T πn2 T
(
0, n = 0,
= cosπn
nT , n ̸= 0

Dr. Sujit Kumar Sahoo (IIT GOA) Sampling of Continuous-Time Signals October 11, 2023 21 / 61
Impulse Invariance

Definition
It is a technique for designing discrete-time filters from continuous-time filters in
which the impulse response of the continuous-time system is sampled to produce
the impulse response of the discrete-time system. In particular, h[n] = Thc (nT );

The sampling process h[n] = Thc (nT ) will result in the spectrum
∞
X 2π
H(e jΩT ) = Hc (j(Ω − k ))
T
k=−∞

Dr. Sujit Kumar Sahoo (IIT GOA) Sampling of Continuous-Time Signals October 11, 2023 22 / 61
Impulse Invariance

For the bandlimited Hc (jΩ) we can state that

ω π
H e jω = Hc (j ),


|ω| < π; Hc (jΩ) = 0, |Ω| ≥
T T

Dr. Sujit Kumar Sahoo (IIT GOA) Sampling of Continuous-Time Signals October 11, 2023 23 / 61