DIGITAL SIGNAL PROCESSING

Lectured by Assoc. Prof. Thuong Le-Tien

Tel: 0903 787 989
Email: Thuongle@hcmut.edu.vn
September 2011
1
Quantization process
and noise shaping
Quantization process and noise shaping
1. Quantization process.
2. Over sampling and Noise Shaping.
3. Digital to Analog conversion DAC.
4. Analog to Digital Conversion ADC.
2
1. Quantization Process
Analog to digital converter - ADC.
3

Analog
signal
x(t)
Sample & hole
Sampler & quantizer
x(nT)
Sampled
signal
A/D
converter
Quantized
signal x (nT)
Q
To DSP
B bits/sample
Quantized sample x
Q
(nT) represented by B bits take
only one of 2
B
possible value.
Quantization width or quantizer resolution Q
4
R is the full-scale range
B
R
Q
2
=
B
Q
R
2 =
R is in the symmetrical range:
Quantization error:
e(nT) = x
Q
(nT) x(nT)
In general case: e = x
Q
x
where, x
Q
is the quantized value
5
2 2
Q
e
Q
s s
2
) (
2
R
nT x
R
Q
< s
Root Mean Square error:
Quantization error e can be assumed as a random
variable which is distributed uniformly over the
range [-Q/2, Q/2] then having probability density:
6
Mean:
0
1
2 /
2 /
= =
}

Q
Q
ede
Q
e
12
1
2
2 /
2 /
2 2
Q
de e
Q
e
Q
Q
= =
}

12
2
Q
e e
rms
= =
7
}

=
2 /
2 /
) ( ] [
Q
Q
de e ep e E
SNR (Signal-to-noise ratio):
20log
10
(R/Q) = 20log
10
(2B) = 20Blog
10
(2)
1 ) (
2 /
2 /
=
}

Q
Q
de e p
}

=
2 /
2 /
2 2
) ( ] [
Q
Q
de e p e e E
-Q/2 Q/2 0
p(e)
e

s s
=
other
Q
e
Q
Q e p
, 0
2 2
,
1
) (
Normalization 1/Q needed to guarantee
The statistical expectation
B
Q
R
SNR 6 log 20
10
=
|
|
.
|

\
|
=
8
) (
2
k
e
o o
Assumed e(n) is white noise then the
autocorrelation function is the delta function
| |
12
) (
2
2 2
Q
n e E
e
= = o
The average power or variance of e(n)
Example: in digital audio application, signal sampled at
44kHz and each sample quantized using a ADC having
full scale of 10volts. Determine number of bits B if the
rms quantization error must be kept below 50 microvolts.
Then determine the actual rms error and bit rate
Sol:
Which is rounded to B=16
Then bit rate:
The dynamic range of the quantizer is 6B = 96 dB
10
2. Oversampling and noise shaping
Power spectrum of white quantization noise
Power spectrum density of e(n)
The noise power within at Nyquist sub-interval [f
a
, f
b
]
with Af = f
b
f
a
:
The noise power over the entire interval Af = f
s
Noise shaping quantizers reshape the spectrum of the
quantization noise into more convenient shape. This
accomplished by filtering the white noise sequence e(n)
by a noise shaping filter H
NS
(f).
x
Q
(n) = x(n) + c(n)
11
2
2
e s
s
e
f
f
o
o
=
12
Over Sampling ratio
Noise power within a given interval
Power spectral density
2
2
2
) ( ) ( ) ( ) ( f H
f
f S f H f S
NS
s
e
ee NS
o
cc
= =
} }
=
b
a
b
a
f
f
NS
s
e
f
f
df f H
f
df f S
2
2
) ( ) (
o
cc
s
s
f
f
L
'
=
Quantization noise powers
12
2
2
Q
e
= o
To maintain the same quality
required the power spectral
density remain the same
'
'
2 2
s
e
s
e
f f
o o
=
13
12
'
'
2
2
Q
e
= o
L f
f
e
s
e
s e
2 2
2
'
'
' o o
o = =
B B B
e
e
L
A
= = =
2 ) ' ( 2
2
2
2 2
'
o
o
AB = B-B, or AB = 0.5 log
2
L
14
The total noise power in the Nyquist interval:
|
.
|

\
|
+
=
+
A
1 2
2
B 2 -
1
1 2
2
p
p
L p
t
15
p
s
NS
f
f
f H
2
2
'
2
) (
|
|
.
|

\
|
=
t
2 / '
s
f f <<
|
.
|

\
|
+
=
|
|
.
|

\
|
+
=
|
|
.
|

\
|
+
=
|
|
.
|

\
|
=
+
+

+
}
1 2
2
2
1 2
2
2
2 /
2 /
1 2
2
2
2
2
2
1
1 2
'
' 1 2
'
' 1 2
'
'
2
'
'
p
p
e
p
s
s
p
e
f
f
p
s
s
p
e
p
s s
e
e
L p f
f
p
f
f
p
df
f
f
f
s
s
t
o
t
o
t
o
t o
o
2 2
' /
e e
o o
= 2
-2(B-B)
= 2
-2AB
|
.
|

\
|
+
=
|
|
.
|

\
|
+
=
|
|
.
|

\
|
+
=
|
|
.
|

\
|
=
+
+

+
}
1 2
2
2
1 2
2
2
2 /
2 /
1 2
2
2
2
2
2
1
1 2
'
' 1 2
'
' 1 2
'
'
2
'
'
p
p
e
p
s
s
p
e
f
f
p
s
s
p
e
p
s s
e
e
L p f
f
p
f
f
p
df
f
f
f
s
s
t
o
t
o
t
o
t o
o
|
.
|

\
|
+
=
|
|
.
|

\
|
+
=
|
|
.
|

\
|
+
=
|
|
.
|

\
|
=
+
+

+
}
1 2
2
2
1 2
2
2
2 /
2 /
1 2
2
2
2
2
2
1
1 2
'
' 1 2
'
' 1 2
'
'
2
'
'
p
p
e
p
s
s
p
e
f
f
p
s
s
p
e
p
s s
e
e
L p f
f
p
f
f
p
df
f
f
f
s
s
t
o
t
o
t
o
t o
o
2 2
' /
e e
o o
|
|
.
|

\
|
+
+ = A
1 2
log 5 . 0 log ) 5 . 0 (
2
2 2
p
L p B
p
t
16
Oversampling and noise shaping system
3. Digital to Analog Converter DAC
B bit 0 and 1 at input, b = [b
1
, b
2
,, b
B
],
(a) unipolar natural binary,
(b) bipolar offset binary,
(c) bipolar 2s complement.
17
Unipolar natural binary
x
Q
= R(b
1
2-
1
+ b
2
2
-2
+ + b
B
2
-B
)
x
Q
= R2
-B
(b
1
2
B-1
+ b
2
2
B-2
+ + b
B-1
2
1
+ b
B
)
Bipolar offset binary
x
Q
= R(b
1
2
-1
+ b
2
2
-2
+ + b
B
2
-B
0.5)
Twos complement
x
Q
= R(b
1
2
-1
+ b
2
2
-2
+ + b
B
2
-B
0.5)
18
19
Converter code for B=4bits, R=10volts
20
4. Analog to Digital Converter (ADC)
Example: A sampled sinusoid x(n)=Acos(2pfn), A=3volts
And f=0.04 cycles/sample. The sinusoid is evaluated at the ten
Sampling times n=0,1,29 and x(n) is quantized using a 4-bit
ADC with R=10volts. The following table shows the sampled
and quantized values and its codes
22
5. Analog and Digital Dither
Dither is a low-level white noise signal added to the
input before quantization for eliminating granulation
or quantization distortion and making the total
quantization error behave like white noise
Analog dither
23
Digital dither can be added to a digital prior to a
requantization operation that reduces the number
of bits representing the signal.
Nonsubtractive dither process and quantization
(Analog and digital dithers)
v(n) is dither noise
24
y(n) = x(n) + v(n)
Quantization error: e(n) = y
Q
(n) y(n)
Total error resulting from dithering and quantization:
c(n) = y
Q
(n) x(n)
c(n) = (y(n) + e(n)) x(n) = x(n) + v(n) + e(n) x(n)
or
c(n) = yQ(n) x(n) = e(n) + v(n)
Total error noise power
2 2 2 2 2
12
1
v v e
Q o o o o
c
+ = + =
The two common Rectangular and triangular dither probability densities
25
10log2 = 3 dB
10log3 = 4.8 dB
10log4 = 6 dB
Total error variance (the noise penalty in using dither)
Subtractive dither
Total error
c(n) = y
out
(n) x(n) = (y
Q
(n) v(n)) x(n) = y
Q
(n) (x(n) + v(n))
c(n) = y
Q
(n) y(n) = e(n)
26
0 20 40 60 80 100 120
-1,5
-1,0
-0.5
0
0.5
1,0
1,5
0 0,1 0,2 0,3 0,4 0,5
60
120
180
240
Undilhered Quantization Undilhered Spectrum
M
a
g
n
i
t
u
d
e
(
U
n
i
t
s

o
f

Q
)
Quantized
Original
Dithered Quantization Dithered Spectrum
-1,5
-1,0
-0.5
0
0.5
1,0
1,5
(
U
n
i
t
s

o
f

Q
)
60
120
180
240
M
a
g
n
i
t
u
d
e
0 20 40 60 80 100 120 0 0,1 0,2 0,3 0,4
Quantized
Dithered original
0,5