Chapter 3.
Sampling Theorem and Alias(取樣定理與假象)
§ 3.1. Continue_time Sinusoidal Signals
xa(t)=Acos(Ωt+θ), Ω=2πFa -∞<Ω<∞
xa(t), A----phyical unit Ω---radian/sec Fa ---cycle θ---radian
[Link] every fixed value of the frequency Fa, xa(t) is periodic
⇔ ∃Tp xa(t +Tp)= xa(t ), Tp=1/ Fa is the fundmental period.
[Link]-time sinusoidal signals with different freuencies are themselves distinct.
[Link] the frequency Fa result in the rate of oscillation of signal,in the same sense
That the more periods are included in a given interval.
⇔ when Tp→∞, then Fa=1/ Tp→0
§ 3.2. Discrete-Time Sinusoidal Signals----Find its period N
x(n)=Acos(2πfn+θ), -∞<n<∞
=Acos(ωn+θ) ω=2πf
n---sample, ω---radian/sample, f---cycle/sample, θ---radian
B1. A discrete_time sinusoidal is periodic only if its frequency is a rational number.
⇔{if ∃ k,N is in integers such f=k/N , then x(n) is periodic }
根據定義: 一個離時訊號是有週期 N,N>0⇔x(n+N)=x(n), ∀n
⇒ 若一個弦態訊號有數位頻率 f 是週期 N
⇒ cos(2πf(N+n)+ θ)=cos(2πfn+ θ)
⇒ ∃k, 2πfN=2πk ⇔ f=k/N, k和 N 是互質的
Example 1: x(n)=5sin(0.8πn)
ω=0.8π=2π*0.8π/(2π)=2π*2/5 →f=2/5, k=2,N=5
Example 2: x(n)=-2sin(0.055πn)
ω=0.055π=2π*0.055π/(2π)=2π*(0.055/2)=2π*55/2000=2π*11/400
f=k/N=11/400, k=11, N=400
Example 3: x(n)=2sin(0.05πn)+3sin(0.12πn)
→0.05π=2π*(0.05π/2π)=2π*(5/200)=2π*1/40,
→0.12π=2π(0.12π/2π)=2π(12/200)=2π*3/50
∴ N1=40, N2=50, LCM(N1,N2)= LCM (40,50)=200
B2. Discrete-time sinusoidals whose frequencies are separated by an integer multiple of 2π
are identical.
Cos([w0+2π]n+θ)=cos(w0n+2πn+θ)= cos(w0n+θ)
⇒When all sinusoidals xk(n)=Acos(wkn+θ),
wk=w0+2π*k, n=0,1,2,3…. is indistinguishable(i.e. is identical)
⇒any two sequences with frequencies in the range -π≦w≦π or –1/2≦f≦1/2 are distinct.
⇒analog signal Ω=2πFa -∞<Ω<∞ -∞< Fa <∞
1
w=2πf -π≦w≦π –1/2≦f≦1/2
B3. The highest rate of oscillation in discrete-time sinusoidal is attained
when w=π(=-π) Or f=1/2(=-1/2)
From w=0,π/8,π/4,π/2,π or f=0,1/16,1/8,1/4,1/2
⇒the period decrease as the frequency increase.
From w=π--Æ2π, x1(n)=Acos(w1n)
x2(n)=Acos(w2n)=Acos(2π- w1)n=Acos(- w1)n= Acos(w1n)
w2 is an alias of w1
§ 3.3 Sampling of analog signals
xa(t)=cos(Ωt)=cos(2πt)=cos(2πFatsn)=cos[2π(Fa/Fs)n]=cos(2πfn)=x(n)
Ω=2πFa , t=ts*n, ts=1/Fs
w=2πf, f=Fa/Fs, w=Ωts , Ω=w/ ts ,
xa(t)-analog signal,Fa-analog frequency(cycle/sec),t-analog time(sec)
x(n)-discrete signal,f-discrete frequency(cycle/sample),n-discrete time(sample)
ts---sample time(sec/sample) , Fs----sample frequency(sample/sec)
If the sinusoids is xa(t)=Acos(2πFkt+θ), where Fk= F0+k Fs, k=±1,±2,±3…….
are sampled at rate Fs, it is clear that the frequency Fk is outside the fundamental
frequency range −Fs/2 ≦F≦+Fs/2
x(n)= xa(t)= Acos[2π(( F0+k Fs)/ Fs)n+θ]= Acos[2πnF0/Fs+θ+2πkn]=Acos[2πnf0+θ]
Sampling Theorem
If the highest frequency contined in the an analog signal xa(t) is Fmax=B
and the signal is sampled at a rate Fs>2 Fmax=2B,then xa(t) can be recovered
from its sample values using the interpolation function g(t)=(sin2πBt)/(2πBt)
Thus xa(t) can be expressed as
⎛n⎞ ⎛
∞
n⎞
x a (t) = ∑x
⎜⎜ ⎟⎟g⎜⎜ t − ⎟⎟
a
Fs ⎠
n = −∞ ⎝ Fs ⎠ ⎝
Where xa(n/Fs)= xa(nts)=x(n) are samples of xa(t)
Nyquist Rate FN=2B=2 Fmax
the folding frequency =Fs/2
=the maximum frequency that can be represented uniquely by sample frequency Fs
2
Example1. Consider the analog signal xa(t)=3cos(100πt)
(a) the analog frequency(Fa),minmum Fs ?
Fa=Ω/2π=100π/2π=50, minimum Fs=2 Fa=100(sample/sec)
(b) when Fs=200(sample/sec), x(n)=?
f=Fa/Fs=(50 cycle/sec)/( 200sample/sec)=1/4 cycle/sample < 1/2, is unique
x(n)=3cos(2πfn)= 3cos[(2π1/4)n]= 3cos[(π/2)n]
(c)when Fs=75Hz,x(n)=?
f=Fa/Fs=(50 cycle/sec)/( 75 sample/sec)=2/3 cycle/sample >1/2, alias
x(n)=3cos(2πfn)= 3cos[(2π2/3)n]= 3cos{2π[(2/3) –1]n}
= 3cos{2π[-(1/3)]n}=3cos{2π[(1/3)]n}=3cos[(2π/3)n]
(d) when Fs=75Hz,what signal is unique after sampled (i.e. Fa<Fs/2) ?
Because f=Fa/Fs, so f=Fa/Fs=Fa/75= 1/3< ½, Fa=25 Hz=25 cycle/sec
xa(t)=3cos[2π(25)t]=3cos(50πt), Hence Fa=50Hz is an alias of Fa=20Hz under
Fs=75
Example2. Consider the xa(t)=3cos(50πt)+10sin(300πt)-cos(100πt)
What is the Nyquist rate for this signal ?
(a) Analog frequency Fa1=50π/(2π)=25Hz, Fa2=300π/(2π)=150Hz,
Fa3=100π/(2π)=50Hz
(b) Nyquist rate, FN=2B=2 Fmax=2*150=300
Example3. Consider the xa(t)=3cos(2000πt)+5sin(6000πt)+10cos(12000πt)
(a) What is the Nyquist rate for the signal?
(b) Assume the Sample rate Fs=5000 samples/sec
What is the discrete_time signal after sampling?
(c) What is the analog signal ya(t) we can reconstruct from the samples if we
use ideal interpolation?
(a) Analog frequency Fa1=2000π/(2π)=1000Hz=1kHz,
Fa2=6000π/(2π)=3000Hz=3kHz,
Fa3=12000π/(2π)=6000Hz=6kHz,BW= Fmax=6000Hz,
Nyquist rate, FN=2B=2Fmax=2*6K=12KHz
(b) When sampling frequency Fs=5000 samples/sec,
f1= Fa1/Fs=1k/5k=1/5< ½, unique
f2= Fa2/Fs=3k/5k=3/5> ½, alias of f=2/5<1/2,
f3= Fa3/Fs=6k/5k=6/5> ½, alias of f=1/5<1/2
x(n)=3cos[2π(1/5)n]+5sin[2π(3/5)n]+10cos[2π(6/5)n]
=3cos[2π(1/5)n]+ 5sin{2π[(3/5)-1]n}+10cos{2π[(6/5)-1]n}
=3cos[2π(1/5)n]+ 5sin{2π[(3/5)-1]n}+10cos{2π[1/5]n}
=13cos[2π(1/5)n]+ 5sin{2π[-2/5]n}=13cos[2π(1/5)n]-5sin[2π(2/5)n]
(c)The recover signal is ya(t)=13cos[2π(1000)n]-5sin[2π(2000)n]
3
Example 4: x(n)=1.5cos(ω0n)
4
Example analyze the diagram
(1) x(t)=cos(2π(0.2)t),
Fa1=0.2, T=1/0.2=5
(2) Fa1=0.2, Fs=1,
f=Fa1/Fs=0.2=1/5<1/2
(3) x(t)=cos(2π(0.8)t),
Fa2=0.8, T=1/0.8=1.25
(4) Fa2=0.8, Fs=1,
f=Fa1/Fs=0.8=4/5>1/2, alias
Example analyze the diagram
(1)x(t)=cos(2π(0.2)t),
T=1/0.2=5
(2)Fa1=0.2, Fs=1,
f=Fa1/Fs=0.2=1/5<1/2
(3)x(t)=cos(2π(1.2)t),
T=1/1.2=1.25
(4)Fa2=1.2, Fs=1,
f=Fa1/Fs=1.2=6/5>1/2, alias
Example analyze the diagram
(1)Fa1=0.1, T=1/0.1=10
(2)f=Fa1/Fs=0.1/0.5=1/5<1/2
unique
(3)Fa2=0.4, T=1/0.4=2.5
(4)
f=Fa2/Fs=0.4/0.5=4/5>1/2, alias
5
※
6
※ 程式 alias7.m
7
1. An analog elecocardiogram (ECG) signal contains useful frequencies up to 100Hz
(a) What is the Nyquist rate for this signal ?
(b) Sppose that we sample this signal at a rate of 250 sample/sec. What is the highest
frequency
(c) that can be represented uniquely at this sample rate ?
2. An analog signal xa(t)=sin(400πt)+3sin(720πt) is sampled 600 times per second.
(a) Determine the Nyquist Rate for xa(t).
(b) Determine the folding frequency.
(c) What are frequencies, in radians, in the discrete signal x(n) ?
(d) If x(n) is passed through an ideal D/A converter, what is the reconstructed signal ya(t)
3. A digital communication link carries binary-coded word representing samples of an input signal
xa(t)=3cos(600πt)+2cos(1800πt)
The linkis operated at 10,000 bits/sec and each sample is quantized into 1024 different voltages.
(a) What is the sample frequency and folding frequency ?
(b) What is the Nyquist rate for the signal xa(t) ?
(c) What are the frquencies in the resulting discrete –time signal x(n) ?
(d) What is the resolution Δ ?
4. Consider the xa1(t)=cos(6πt), xa2(t)=cos(14πt), xa3(t)=cos(26πt)
(a) What is the Bandwidth, Nyquist Rate of these analog signals ?
(b) If sampling rate Fs=10Hz, what are the discrete frequency of xa1(t), xa2(t), xa3(t) ?
5. Assume there is a analog continuous signals .
xa(t)=6cos(60πt) +3sin(300πt)+2cos(340πt) +4cos(500πt) +10sin(660πt)
(a) What is the Bandwidth, Nyquist Rate of these analog signals ?
(b) If sampling time ts=0.005sec, what is the discrete signal x(n) of ?
(c) If H(z)=1(i.e. y(n)=x(n)), what is the reconstruction output analog signal ya(t) ?
