Chapter 2

Sampling and Reconstruction

Nguyen Khanh Loi, M.Eng.
[email protected]
 Content

v Sampling
q Sampling theorem
q Spectrum of sampling signals

v Anti-aliasing pre-filter
q Ideal pre-filter
q Practical pre-filter

v Analog reconstruction
q Ideal reconstructor
q Practical reconstructor

Digital Signal Processing 2 Sampling and Reconstruction

 1. Introduction

v A typical signal processing system includes 3 stages:

v The analog signal is sampled and each sample is quantized to a finite

number of bits (A/D converter).
v The digitalized samples are processed by a digital signal processor.
• The digital processor can be programmed to perform signal processing
operations such as filtering, spectrum estimation.
• Digital signal processor can be a general purpose computer, DSP chip or other
digital hardware.
v The resulting output samples are converted back into analog by an
analog reconstructor (D/A converter).
Digital Signal Processing 3 Sampling and Reconstruction
 2. Analog to digital conversion

v Analog to digital (A/D) conversion is a three-step process.

x(t) Sampler x(nT)≡x(n) xQ(n) 11010

Quantizer Coder
t=nT
A/D converter

x(t) x(n) 111 xQ(n)

110
101
100
011
t n 010 n
001
000

Digital Signal Processing 4 Sampling and Reconstruction

 3. Sampling

v Sampling is to convert a continuous time signal into a discrete time

signal. The analog signal is periodically measured at every T seconds

v x(n)≡x(nT)=x(t=nT), n=…-2, -1, 0, 1, 2, 3…

v T: sampling interval or sampling period (second);
v Fs=1/T: sampling rate or frequency (samples/second or Hz)

Digital Signal Processing 5 Sampling and Reconstruction

 Example 1

v The analog signal x(t)=2cos(2πt) with t(s) is sampled at the rate Fs=4
Hz. Find the discrete-time signal x(n) ?

Digital Signal Processing 6 Sampling and Reconstruction

 Example 2

v Consider the two analog sinusoidal signals

7 1
x1 (t ) = 2 cos(2p t ), x2 (t ) = 2 cos(2p t ); t ( s)
8 8
These signals are sampled at the sampling frequency Fs=1 Hz.
Find the discrete-time signals ?

Digital Signal Processing 8 Sampling and Reconstruction

 4. Aliasing of Sinusoids

v In general, the sampling of a continuous-time sinusoidal signal

x (t ) = A cos(2p F0t + q ) at a sampling rate Fs=1/T results in a
discrete-time signal x(n).
v The sinusoids xk (t ) = A cos(2p Fk t + q ) is sampled at Fs , resulting
in a discrete time signal xk(n).
v If Fk=F0+kFs, k=0, ±1, ±2, …., then x(n)=xk(n) .

Proof:

v Remarks: We can that the frequencies Fk=F0+kFs are

indistinguishable from the frequency F0 after sampling and hence
they are aliases of F0
Digital Signal Processing 12 Sampling and Reconstruction
 5. Spectrum Replication

Fourier Transform Discrete-time Discrete Fourier Fast Fourier

Fourier Transform Transform
Transform
#" #" 1!2
𝟐𝝅𝒌 𝟐𝝅𝒌
𝑋 𝑓 = $ 𝑥 𝑡 𝑒 !$%&'(
𝑑𝑡 𝑋) 𝑓 = * 𝑥 𝑛 𝑒 !$𝟐𝝅𝒇)𝑻 𝑋) = * 𝑥 𝑛 𝑒 !$ 𝑵 )
𝑵
Formulation !" )*!"
#"
)*0
1!2
𝟐𝝅𝒌
Same as DFT
𝑋) 𝜔 = * 𝑥 𝑛 𝑒 !$𝝎) 𝑋) 𝑘 = * 𝑥 𝑛 𝑒 !$
𝑵
)

)*!" )*0

Time Domain continuous discrete discrete Same as DFT

(can be infinite) (can be infinite) (finite)

Frequency continuous continuous discrete

Same as DFT
Domain

FT

Digital Signal Processing 15 Sampling and Reconstruction

 5. Spectrum Replication

Digital Signal Processing 16 Sampling and Reconstruction

 5. Spectrum Replication

Ideal sampler
𝑥 𝑡 𝑥 𝑛𝑇
analog signal sampled signal

Continuous-signal
Continuous-signal

Fourier transform Discrete-time Fourier Transform

Continuous-spectrum Continuous-spectrum

Digital Signal Processing 17 Sampling and Reconstruction

 Example 3

Consider the analog signal x(t) = cos(20πt), where (t: seconds), which is
sampled at a frequency of 40 Hz/15 Hz.
a) Plot the spectrum of the signal before sampling.
b) Plot the spectrum of the signal after sampling within the range [-40
to 40] Hz.

Digital Signal Processing 18 Sampling and Reconstruction

 5. Spectrum Replication

Fourier Transform Discrete-time Discrete Fourier Fast Fourier

Fourier Transform Transform
Transform
#" #" 1!2
𝟐𝝅𝒌 𝟐𝝅𝒌
𝑋 𝑓 = $ 𝑥 𝑡 𝑒 !$%&'(
𝑑𝑡 𝑋) 𝑓 = * 𝑥 𝑛 𝑒 !$𝟐𝝅𝒇)𝑻 𝑋) = * 𝑥 𝑛 𝑒 !$ 𝑵 )
𝑵
Formulation !" )*!"
#"
)*0
1!2
𝟐𝝅𝒌
Same as DFT
𝑋) 𝜔 = * 𝑥 𝑛 𝑒 !$𝝎) 𝑋) 𝑘 = * 𝑥 𝑛 𝑒 !$ 𝑵
)

)*!" )*0

Time Domain continuous discrete discrete Same as DFT

(can be infinite) (can be infinite) (finite)

Frequency continuous continuous discrete

Same as DFT
Domain

DTFT

Digital Signal Processing 20 Sampling and Reconstruction

 5. Spectrum Replication

Fourier Transform Discrete-time Discrete Fourier Fast Fourier

Fourier Transform Transform
Transform
#" #" 1!2
𝟐𝝅𝒌 !$
𝟐𝝅𝒌
)
𝑋 𝑓 = $ 𝑥 𝑡 𝑒 !$%&'(
𝑑𝑡 𝑋) 𝑓 = * 𝑥 𝑛 𝑒 !$𝟐𝝅𝒇)𝑻
𝑋) = *𝑥 𝑛 𝑒 𝑵
𝑵
Formulation !" )*!"
#"
)*0
1!2
𝟐𝝅𝒌
Same as DFT
𝑋) 𝜔 = * 𝑥 𝑛 𝑒 !$𝝎) 𝑋) 𝑘 = * 𝑥 𝑛 𝑒 !$ 𝑵
)

)*!" )*0

Time Domain continuous discrete discrete Same as DFT

(can be infinite) (can be infinite) (finite)

Frequency continuous continuous discrete

Same as DFT
Domain

DTFT DFT

Digital Signal Processing 21 Sampling and Reconstruction

 5. Spectrum Replication

Fourier Transform Discrete-time Discrete Fourier Fast Fourier

Fourier Transform Transform
Transform
#" #" 1!2
𝟐𝝅𝒌 𝟐𝝅𝒌
𝑋 𝑓 = $ 𝑥 𝑡 𝑒 !$%&'(
𝑑𝑡 𝑋) 𝑓 = * 𝑥 𝑛 𝑒 !$𝟐𝝅𝒇)𝑻
𝑋) = * 𝑥 𝑛 𝑒 !$ 𝑵 )
𝑵
Formulation !" )*!"
#"
)*0
1!2
𝟐𝝅𝒌
Same as DFT
𝑋) 𝜔 = * 𝑥 𝑛 𝑒 !$𝝎) 𝑋) 𝑘 = * 𝑥 𝑛 𝑒 !$ 𝑵
)

)*!" )*0

Time Domain continuous discrete discrete Same as DFT

(can be infinite) (can be infinite) (finite)

Frequency continuous continuous discrete

Same as DFT
Domain

DTFT FFT

• Gives same results as

DFT
• Compute more efficient

Digital Signal Processing 22 Sampling and Reconstruction

 6. Sampling Theorem

v Fs/2 ≥ Fmax

Fig: Spectrum replication caused by sampling

Fig: Typical badlimited spectrum

v Fs/2 < Fmax

Fig: Aliasing caused by overlapping spectral replicas

Digital Signal Processing 23 Sampling and Reconstruction
 6. Sampling Theorem
v For accurate representation of a signal x(t) by its time samples x(nT),
two conditions must be met:
1) The signal x(t) must be band-limited, i.e., its frequency spectrum must
be limited to Fmax .
2) The sampling rate Fs must be chosen at least twice the maximum
frequency Fmax.

Fs ³ 2 Fmax

• Fs=2Fmax is called Nyquist rate;

• Fs/2 is called Nyquist frequency
• [-Fs/2, Fs/2] is Nyquist interval. Fig: Typical band-limited spectrum

Digital Signal Processing 24 Sampling and Reconstruction

 Example 4

Consider the analog signal x(t) = 4 + cos(2πt) + 6cos(8πt), where

(t:ms). Determine the appropriate value of fs?.

Digital Signal Processing 25 Sampling and Reconstruction

 6. Sampling Theorem

v The values of Fmax and Fs depend on the application

Application Fmax Fs
Speech 4 KHz 8 KHz
Audio 20 KHz 40 KHz
Video 4 MHz 8 MHz

Digital Signal Processing 27 Sampling and Reconstruction

 7. Ideal analog reconstruction

Fig: Ideal reconstructor as a lowpass filter

v An ideal reconstructor acts as a lowpass filter with cutoff frequency
equal to the Nyquist frequency Fs/2.
ìT F Î [ - Fs / 2, Fs / 2]
v An ideal reconstructor (lowpass filter) H ( F ) = í
î0 otherwise
Then  
X a (F ) = X (F )H (F ) = X (F )

Digital Signal Processing 28 Sampling and Reconstruction

 7. Ideal analog reconstruction
s(t)=Ʃб(t-nT)

xc(t) xs(t) xr(t)

X Hr(f)

Xc(f)
1

-fc fc f
Hr(f) Xs(f)
1/T

-fs -fc fc fs f
Xr(f)
1

-fc fc f

Digital Signal Processing 29 Sampling and Reconstruction

 Example 5

v The analog signal x(t)=cos(20πt) [t:s] is sampled at the sampling

frequency Fs=40 Hz.
a) Plot the spectrum of signal x(t) ?
b) Find the discrete time signal x(n) ?
c) Plot the spectrum of signal x(n) ?
d) The signal x(n) is an input of the ideal reconstructor, find the
reconstructed signal xa(t) ?

Digital Signal Processing 30 Sampling and Reconstruction

 Example 6

v The analog signal x(t)=cos(100πt) is sampled at the sampling

frequency Fs=40 Hz.
a) Plot the spectrum of signal x(t) ?
b) Find the discrete time signal x(n) ?
c) Plot the spectrum of signal x(n) ?
d) The signal x(n) is an input of the ideal reconstructor, find the
reconstructed signal xa(t) ?

Digital Signal Processing 31 Sampling and Reconstruction

 7. Ideal analog reconstruction

v Remarks: xa(t) contains only the frequency components that lie in the
Nyquist interval (NI) [-Fs/2, Fs/2].

sampling at Fs ideal reconstructor

v x(t), F0 Î NI ------------------> x(n) ----------------------> xa(t), Fa=F0

sampling at Fs ideal reconstructor

v xk(t), Fk=F0+kFs-----------------> x(n) ---------------------> xa(t), Fa=F0

v The frequency Fa of reconstructed signal xa(t) is obtained by adding

to or substracting from F0 (Fk) enough multiples of Fs until it lies
within the Nyquist interval [-Fs/2, Fs/2]. That is
Fa = F mod( Fs )

Digital Signal Processing 32 Sampling and Reconstruction

 Example 7

v The analog signal x(t)=10sin(4πt)+6sin(16πt) is sampled at the rate 20

Hz. Find the reconstructed signal xa(t) ?

Digital Signal Processing 33 Sampling and Reconstruction

 Example 8

v Let x(t) be the sum of sinusoidal signals

x(t)=4+3cos(πt)+2cos(2πt)+cos(3πt) where t is in milliseconds.
a) Determine the minimum sampling rate that will not cause any
aliasing effects ?
b) To observe aliasing effects, suppose this signal is sampled at half its
Nyquist rate. Determine the signal xa(t) that would be aliased with
x(t) ? Plot the spectrum of signal x(n) for this sampling rate?

Digital Signal Processing 34 Sampling and Reconstruction

 8. Ideal antialiasing prefilter

v The signals in practice may not band-limited, thus they must be

filtered by a lowpass filter

Fig: Ideal antialiasing prefilter

Digital Signal Processing 37 Sampling and Reconstruction

 9. Practical antialiasing prefilter

v A lowpass filter: [-Fpass, Fpass] is the frequency range of interest for

the application (Fmax=Fpass)
v The Nyquist frequency Fs/2 is in the middle of transition region.
v The stopband frequency Fstop and the minimum stopband
attenuation Astop dB must be chosen appropriately to minimize the
aliasing effects.

Fs = Fpass + Fstop

Fig: Practical antialiasing lowpass prefilter

Digital Signal Processing 38 Sampling and Reconstruction
 9. Practical antialiasing prefilter

v The attenuation of the filter in

decibels is defined as
H (F )
A( F ) = -20log10 ( dB )
H ( F0 )

where f0 is a convenient reference frequency, typically taken to be at

DC for a lowpass filter.
v α10 =A(10F)-A(F) (dB/decade): the increase in attenuation when F is
changed by a factor of ten.
v α2 =A(2F)-A(F) (dB/octave): the increase in attenuation when F is
changed by a factor of two.

Digital Signal Processing 40 Sampling and Reconstruction

 Example 9

v A sound wave has the form

x(t ) = 2 A cos(10p t ) + 2 B cos(30p t ) + 2C cos(50p t )
+ 2 D cos(60p t ) + 2 E cos(90p t ) + 2 F cos(125p t )

where t is in milliseconds. What is the frequency content of this

signal ? Which parts of it are audible and why ?
This signal is prefilter by an anlog prefilter H(f). Then, the output y(t)
of the prefilter is sampled at a rate of 40KHz and immediately
reconstructed by an ideal analog reconstructor, resulting into the final
analog output ya(t), as shown below:

Digital Signal Processing 41 Sampling and Reconstruction

 Example 10

Determine the output signal y(t) and ya(t) in the following cases:
a)When there is no prefilter, that is, H(F)=1 for all F.
b)When H(F) is the ideal prefilter with cutoff Fs/2=20 KHz.
c)When H(F) is a practical prefilter with specifications as shown
below:

The filter’s phase response is assumed to be ignored in this example.

Digital Signal Processing 42 Sampling and Reconstruction

 10. Practical analog reconstructors

sin(p Fs t)
v The ideal reconstructor has the impulse response: h (t ) =
which is not realizable since its impulse response is not casual p Fs t

v It is practical to use a
staircase reconstructor

Digital Signal Processing 43 Sampling and Reconstruction

 10. Practical analog reconstructors

Fig: Frequency response of staircase reconstructor

Digital Signal Processing 44 Sampling and Reconstruction

 11. Anti-image postfilter

v An analog lowpass postfilter whose cutoff is Nyquist frequency Fs/2

is used to remove the surviving spectral replicas.

Fig: Analog anti-image postfilter

Fig: Spectrum after postfilter

Digital Signal Processing 45 Sampling and Reconstruction
 Homework 1

Digital Signal Processing 49 Sampling and Reconstruction

 Homework 2

Digital Signal Processing 50 Sampling and Reconstruction

 Homework 3

Digital Signal Processing 51 Sampling and Reconstruction

 Homework 4

Digital Signal Processing 52 Sampling and Reconstruction

 Homework 5

Digital Signal Processing 53 Sampling and Reconstruction

 Homework 6

Digital Signal Processing 54 Sampling and Reconstruction

 Homework 7

Digital Signal Processing 55 Sampling and Reconstruction

 Homework 8

v Cho tín hiệu ngõ vào tương tự x(t) = 3cos103πt – 4sin104πt (t: s) đi qua
hệ thống lấy mẫu và khôi phục lý tưởng với tần số lấy mẫu Fs = 8 KHz.
a) Viết biểu thức của tín hiệu sau lấy mẫu x[n]? Xác định giá trị mẫu
x[n=2] của tín hiệu sau lấy mẫu.
b) Có hay không 1 tần số lấy mẫu khác (Fsb ≠ 8 KHz) cho cùng kết quả
tín hiệu sau lấy mẫu x[n]? Nếu không, hãy chứng minh. Nếu có, hãy
chỉ ra 1 tần số lấy mẫu khác đó.
c) Vẽ phổ biên độ của tín hiệu sau lấy mẫu trong phạm vi tần số từ 0 đến
10 KHz.
d) Xác định biểu thức của tín hiệu sau khôi phục.
e) Xác định biểu thức của tín hiệu sau khôi phục trong trường hơp dùng
thêm bộ tiền lọc thông thấp thực tế có biên độ phẳng trong tầm [-4 4]
KHz và suy giảm với tốc độ -1@0dB/decade bên ngoài dải thông.

Digital Signal Processing 56 Sampling and Reconstruction

 Homework 9

v Cho tín hiệu ngõ vào tương tự x(t) = 2 – 4cos6πt + 8cos10πt

(t:ms) đi qua hệ thống lấy mẫu và khôi phục lý tưởng với tần số
lấy mẫu lựa chọn Fs = 7 KHz.
a) Vẽ phổ biên độ của tín hiệu ngõ vào x(t).
b) Vẽ phổ biên độ của một tín hiệu chồng lấn (aliased signal) với
x(t).
c) Vẽ phổ biên độ của tín hiệu sau lấy mẫu trong phạm vi tần số từ
0 đến 10 KHz.
d) Tìm giá trị mẫu x[n=2] của tín hiệu sau lấy mẫu.
e) Xác định biểu thức (theo thời gian) của tín hiệu sau khôi phục lý
tưởng
f) Tìm điều kiện của tần số lấy mẫu để khôi phục đúng tín hiệu ngõ
vào x(t).
Digital Signal Processing 57 Sampling and Reconstruction
 Homework 10

v Cho tín hiệu ngõ vào tương tự x(t) = 14sin23pt + 3sin14pt (t: ms) đi qua
hệ thống lấy mẫu và khôi phục lý tưởng với tần số lấy mẫu Fs = 8 KHz.
a) Tìm giá trị mẫu x[n=4] của tín hiệu sau lấy mẫu?
b) Xác định biểu thức của 1 tín hiệu chồng lấn (aliased signal) với tín
hiệu ban đầu x(t)?
c) Vẽ phổ biên độ của tín hiệu sau lấy mẫu trong phạm vi tần số từ 0 đến
8 KHz?
d) Xác định biểu thức của tín hiệu sau khôi phục?
e) Xác định biểu thức của tín hiệu sau khôi phục trong trường hơp dùng
thêm bộ tiền lọc thông thấp thực tế có biên độ phẳng trong tầm 4 KHz
và suy giảm với tốc độ -40dB/decade bên ngoài dải thông?
f) Xác định 1 tập giá trị thích hợp (A, B, FA ≠ FB) của tín hiệu ngõ vào
x(t) = AsinFAt + BsinFBt (t: ms) để tín hiệu sau khôi phục (khi
không dùng thêm bộ tiền lọc) y(t) = 2sin2pt (t: ms)?
Digital Signal Processing 58 Sampling and Reconstruction
 Homework 11

v Cho tín hiệu ngõ vào tương tự x(t) = 1 – 2cos6pt + 3sin14pt (t: ms) đi qua hệ
thống lấy mẫu và khôi phục lý tưởng với tần số lấy mẫu Fs = 8 KHz.
a) Tìm giá trị mẫu x[n=2] của tín hiệu sau lấy mẫu?
b) Xác định biểu thức (theo thời gian) của 1 tín hiệu chồng lấn (aliased signal) với
tín hiệu ban đầu x(t)?
c) Vẽ phổ biên độ của tín hiệu sau lấy mẫu trong phạm vi tần số từ 0 đến 8 KHz?
d) Xác định biểu thức (theo thời gian) của tín hiệu sau khôi phục?
e) Xác định biểu thức (theo thời gian) của tín hiệu sau khôi phục trong trường hơp
dùng thêm bộ tiền lọc thông thấp thực tế có biên độ phẳng trong tầm 4 KHz và
suy giảm với tốc độ -60dB/decade bên ngoài dải thông?

Digital Signal Processing 59 Sampling and Reconstruction

 Homework 12

Digital Signal Processing 60 Sampling and Reconstruction