review of discrete-time signal processing
DESCRIPTION
Review of Discrete-Time Signal Processing. Sampling. CT. DT. Zero-Order Hold (ZOH). Sampling. Impulse train. Sampling. δ (t-nT). Spectrum of Sampled Signal ( Ω s >2 Ω N ). The spectrum of the sampled signal is periodic in Ω s =2 π /T. Spectrum of Sampled Signal ( Ω sTRANSCRIPT
– 1 –
Data Converters Discrete-Time Signal ProcessingProfessor Y. Chiu
EECT 7327Fall 2014
Discrete-Time Signal Processing(A Review)
Sampling
– 2 –
Data Converters Discrete-Time Signal ProcessingProfessor Y. Chiu
EECT 7327Fall 2014
CT
fclk
xc(t) x(n)C-D
xc(t)
t0
x(n)
n0 1 2 3 4
xc(nT)
T 2T 3T 4T T
cx n =x t=nT
∞ ?jω -n
cn=-∞
X z =e = x n z ↔ X jΩ
DT FT
c cx t ⇔ X jΩ
Zero-Order Hold (ZOH)
– 3 –
Data Converters Discrete-Time Signal ProcessingProfessor Y. Chiu
EECT 7327Fall 2014
xc(t)
t0
xSH(t)
t0 T 2T 3T 4T T
xc(nT)T
xc(nT)
t0 T 2T 3T 4T
t0
u(t) - u(t-T)T
T
1/T
T
w
∞
SH cn=-∞
1x t = x nT u t-nT -u t-nT -TT
Sampling
– 4 –
Data Converters Discrete-Time Signal ProcessingProfessor Y. Chiu
EECT 7327Fall 2014
SH sw→0∞
cw→0 n=-∞∞
-s nT+w-snTcw→0 n=-∞
-sw ∞-snT
cw→0 n=-∞∞ ∞
-snT -nc
n=-∞ n=-∞
FT limx t =FT x t
1=FT lim x nT u t-nT -u t-nT - ww1 1 1=lim x nT e - ew s s1-e=lim x nT esw
= x nT e = x n z
xSH(t)
t0 w→0
Area fixed
Impulse train
∞
jω -ns
n=-∞X z =e = x n z ↔ X jΩ
xs(t)
t0 T 2T 3T 4T T
xc(t)·δ(t-nT)
jωCT →Ω, DT →ω,
s =jΩ, z =e , ω=ΩT.
Sampling
– 5 –
Data Converters Discrete-Time Signal ProcessingProfessor Y. Chiu
EECT 7327Fall 2014
s c c
s c
FTs s
k
s c sk
x t =x t s t =x t δ t-nT1X jΩ = X jΩ S jΩ2π
2π 2πs t ⇔ δ Ω-kΩ , Ω =T T1X jΩ = X Ω-kΩT
∞jω -n
n=-∞
ωs c sΩ= ωT k Ω=T
X e = x n z
1=X jΩ = X Ω-kΩT
xc(t)
t0
s(t)
t0 T
δ(t-nT)
xs(t)
t0 T 2T 3T 4T T
xc(t)·δ(t-nT)
Spectrum of Sampled Signal (Ωs>2ΩN)
– 6 –
Data Converters Discrete-Time Signal ProcessingProfessor Y. Chiu
EECT 7327Fall 2014
The spectrum of the sampled signal is periodic in Ωs=2π/T.
Xc(jΩ)
Xs(jΩ)
Ω
Ω
Ω
ΩN-ΩN
Ωs-Ωs-2Ωs-3Ωs 2Ωs 3Ωs0
Ωs-Ωs-2Ωs-3Ωs 2Ωs 3Ωs0
S(jΩ)
Spectrum of Sampled Signal (Ωs<2ΩN)
– 7 –
Data Converters Discrete-Time Signal ProcessingProfessor Y. Chiu
EECT 7327Fall 2014
• Aliasing (folding) results in irreversible signal distortion.• Can only be avoided by using sufficiently high sample rate, or band-
limit the input signal with a coarse, continuous-time filter – AAF.
Xc(jΩ)
Ω
Ω
ΩN-ΩN
Ωs-Ωs-2Ωs-3Ωs 2Ωs 3Ωs0
Xs(jΩ)
ΩΩs-Ωs-2Ωs-3Ωs 2Ωs 3Ωs0
S(jΩ)
Reconstruction Filter (Nyquist)
– 8 –
Data Converters Discrete-Time Signal ProcessingProfessor Y. Chiu
EECT 7327Fall 2014
∞
r rn=-∞
∞
n=-∞
x t = x n h t-nT
sin π t-nT /T= x n π t-nT /T
r
sin πt/Th t = πt/T
Reconstruction filter = “smoothing” filter = “interpolation” filter
Xr(jΩ)
Xs(jΩ)
Ω
Ω
ΩN-ΩN
Ωs-Ωs-2Ωs-3Ωs 2Ωs 3Ωs0
Hr(jΩ)
ΩΩs/2-Ωs/2