Ch13 (5) Handouts 4e PDF

Nyquist Filters Lth-Band Filters
• Under certain conditions, a lowpass filter • These filters, called the Nyquist filters or
can be designed to have a number of zero- Lth-band filters, are often used in single-rate
valued coefficients and multi-rate signal processing
• When used as interpolation filters these • Consider the factor-of-L interpolator shown
filters preserve the nonzero samples of the below
up-sampler output at the interpolator output xu [n]
x[n ] L H (z ) y[n ]
• Moreover, due to the presence of these
zero-valued coefficients, these filters are • The input-output relation of the interpolator
computationally more efficient than other in the z-domain is given by
lowpass filters of same order Y ( z) = H ( z) X ( z L )
1 2
Copyright © 2010, S. K. Mitra Copyright © 2010, S. K. Mitra
Lth-Band Filters Lth-Band Filters

• Then we can express Y(z) as
L −1
• If H(z) is realized in the L-band polyphase
Y ( z ) = α z −k X ( z L ) + ∑ z −l El ( z L ) X ( z L )
form, then we have l =0
H ( z ) = ∑iL=−01z −i Ei ( z L ) • As a result,
l≠k
• Assume that the k-th polyphase component y[ Ln + k ] = α x[n]

of H(z) is a constant, i.e., Ek (z ) = α:
• Thus, the input samples appear at the output
H ( z ) = E0 ( z L ) + z −1E1 ( z L ) + ... + z −( k −1) Ek −1 ( z L ) without any distortion for all values of n,
+ αz − k + z −( k +1) Ek +1 ( z L ) + ... + z −( L −1) EL −1( z L ) whereas, in-between ( L − 1) output samples
are determined by interpolation
3 4

• A filter with the above property is called a
Nyquist filter or an Lth-band filter • Figure below shows a typical impulse
response of a third-band filter (L = 3)
• Its impulse response has many zero-valued
h[n]
samples, making it computationally
attractive
• For example, the impulse response of an _3
n
_6 0 3 6
Lth-band filter for k = 0 satisfies the
following condition • Lth-band filters can be either FIR or IIR
α, n=0
h[Ln] = ⎧⎨ filters
5 ⎩ 0, otherwise Copyright © 2010, S. K. Mitra
6
Copyright © 2010, S. K. Mitra
1
Lth-Band Filters Half-Band Filters
• If the 0-th polyphase component of H(z) is a
constant, i.e., E0 ( z ) = α then it can be shown • An Lth-band filter for L = 2 is called a half-
that band filter
L −1
∑k =0 H ( zWLk ) = Lα = 1 (assuming α = 1/L) • The transfer function of a half-band filter is
• Since the frequency response of H ( zWLk ) is thus given by
the shifted version H (e j (ω− 2 πk / L ) ) of H (e jω) , H ( z ) = α + z −1E1 ( z 2 )
the sum of all of these L uniformly shifted
with its impulse response satisfying
versions of H (e jω) add up to a constant
α, n=0
H(z) H(zWL) H(zWL2)
_
H(zWL L 1) H(z)
h[2n] = ⎧⎨
ω
⎩ 0, otherwise
7 0 2π 8
Half-Band Filters Half-Band Filters

• The condition • H (e j ( π / 2−θ) ) and H (e j ( π / 2+θ) ) add up
H ( z ) = α + z −1E1 ( z 2 ) to 1 for all θ
reduces to • Or, in other words, H (e jω) exhibits a
H ( z ) + H (− z ) = 1 (assuming α = 0.5) symmetry with respect to the half-band
• If H(z) has real coefficients, then frequency π/2, hence the name “half-band
filter”
H (−e jω ) = H (e j ( π−ω) )
• Hence
9
H ( e j ω ) + H ( e j ( π − ω) ) = 1 10

• Figure below illustrates this symmetry for a
half-band lowpass filter for which passband • Attractive property: About 50% of the
and stopband ripples are equal, i.e., δ p = δs coefficients of h[n] are zero
and passband and stopband edges are • This reduces the number of multiplications
symmetric with respect to π/2, i.e., ω p + ωs = π required in its implementation significantly
H(e jω )
1+ δ
• For example, if N = 101, an arbitrary Type 1
1_ δ
FIR transfer function requires about 50
multipliers, whereas, a Type 1 half-band
filter requires only about 25 multipliers
δ ω
ωp π/2 ωs
11 π 12
2
• An FIR half-band filter can be designed • Then R is odd as a result of the condition
α, n=0
h[2n] = ⎧⎨
with linear phase
• However, there is a constraint on its length ⎩ 0, otherwise
• Consider a zero-phase half-band FIR filter • Therefore R = 2K+1 for some integer K
for which h[ n] = α * h[ − n] , with | α | = 1 • Thus the length of h[n] is restricted to be of
• Let the highest nonzero coefficient be h[R] the form 2R+1 = 4K+3 [unless H(z) is a
constant]
13 14
Design of Linear-Phase Design of Linear-Phase

• A lowpass linear-phase Lth-band FIR filter • Now, the impulse response of an ideal Lth-
can be readily designed via the windowed band lowpass filter with a cutoff at ωc = π / L
Fourier series approach is given by
• In this approach, the impulse response sin( πn / L)
hLP [n] = , −∞ ≤ n ≤∞
coefficients of the lowpass filter are chosen πn
as h[n] = hLP [n] ⋅ w[ n] where hLP [n] is the • It can be seen from the above that
impulse response of an ideal lowpass filter
with a cutoff at π/L and w[n] is a suitable hLP [n] = 0 for n = ± L, ± 2 L, ...
15 window function 16

• Hence, the coefficient condition of the Lth- • There are many other candidates for Lth-
band filter band FIR filters
α, n=0
h[Ln] = ⎧⎨
• Program 13_8 can be used to design an Lth-
⎩ 0, otherwise band FIR filter using the windowed Fourier
is indeed satisfied series approach
• Hence, an Lth-band FIR filter can be • The program employs the Hamming
designed by applying a suitable window window
w[n] to hLP [n] • However, other windows can also be used
17 18
3
• Figure below shows the gain response of a • The filter coefficients are given by
half-band filter of length-23 designed using h[ −11] = h[11] = − 0.002315; h[ −10] = h[10] = 0;
Program 13_8 h[ −9] = h[9] = 0.005412; h[ −8] = h[8] = 0;
0
h[ −7 ] = h[ 7 ] = − 0.001586; h[ −6] = h[ 6] = 0;
-20 h[ −5] = h[5] = 0.003584; h[ −4] = h[ 4] = 0;
Gain, dB
-40
h[ −3] = h[3] = − 0.089258; h[ −2] = h[ 2] = 0;
-60
h[ −1] = h[1] = 0.3122379; h[ 0] = 0.5;
-80
0 0.2 0.4 0.6 0.8 1 • As expected, h[n] = 0 for
n = ± 2, ± 4, ± 6, ± 8, ±10
ω/π
19 20

Lth-Band Filters Half-Band Filters
• We show below the gain response of a • The problem of designing a real-coefficient
length 23 4th-band lowpass filter designed half-band FIR filter can be transformed into
using the M-file firnyquist with 2 the design of a single passband FIR filter
different roll-off factors with no stopband which can be easily
0
Gain response of 4th-band FIR filter
designed using the Parks-McClellan
-20
R = 0.1→ ← R = 0.4 algorithm
Gain, dB
-40
• An inverse transformation of the wideband
-60
-80
0 0.2 0.4 0.6 0.8 1
filter then yields the half-band FIR filter
ω/π
21 22

• Let the specifications of the real-coefficient • Now, consider the design of a wide-band
half-band filter G(z) of order N be as linear-phase FIR filter F(z) of degree N/2
follows: with a passband from 0 to 2ωp , a transition
• Passband edge at ωp , stopband edge at ωs , band from 2ωp to π, and a passband ripple
passband ripple of δp , and stopband ripple of 2δ
of δ s
• Since N/2 is odd, F(z) has a zero at z = −1
• Now for a half-band filter δ p = δ s = δ ,
ω p + ωs = π and the order N is even with • Let f[n] denote the impulse response of F(z)
N/2 odd
23 24
4
• Define • The plots below the magnitude response of an
G ( z ) = 1 [ z − N / 2 + F ( z 2 )] wide-band lowpass filter of degree 13 with a
2 passband from 0 to 0.85π and a transition band
• G(z) can be seen to be the transfer function from 0.9π to π and the magnitude response of the
of a causal half-band lowpass FIR filter derived half-band filter
with an impulse response Wide-Band Lowpass Filter Half-Band Lowpass Filter
⎧1 f [n / 2], n even 1 1
⎪⎪2 0.8 0.8
Magnitude
Magnitude
0.6 0.6
g[n] = ⎨ 0, n odd , n ≠ N 0.4 0.4
2
⎪ 1
0.2 0.2
= N 0 0
⎪⎩ 2 , n 0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1

ω/π ω/π
25 2 Copyright © 2010, S. K. Mitra 26
Design of Half-Band IIR Filters Design of Half-Band IIR Filters

• It follows from the power-symmetry
• Recall that an odd-order bounded real (BR)
condition that G(z) is a half-band lowpass
lowpass IIR transfer function G(z) =
transfer function
P(z)/D(z) with a symmetric numerator and
satisfying the power-symmetry condition • A Butterworth half-band lowpass IIR filter
G(z) can be designed by first designing an
G ( z )G ( z −1 ) + G (− z )G (− z −1 ) = 1
odd-order analog Butterworth lowpass filter
can be decomposed in the form with a 3-dB cutoff frequency at Ωc = 1 and
G ( z ) = 1 [A0 ( z 2 ) + z −1A1 ( z 2 )] then applying a bilinear transformation
2
where A0 (z ) and A1 (z ) are stable allpass • We next consider the design of an elliptic
27
functions 28 IIR half-band filter

• It can be shown that any odd-order elliptic • It can be shown that the poles of the elliptic
lowpass half-band filter G(z) with a lowpass half-band filter lie on the imaginary
frequency response specification given by axis
1 − 2δ p ≤ G (e jω ) ≤ 1, for 0 ≤ ω ≤ ω p • Using the pole-interlacing property, we can
G ( e jω ) ≤ δ s , for ωs ≤ ω ≤ π readily identify the expressions for A0 ( z )
and satisfying the conditions and A1 ( z )
ω p + ωs = π, δ2s = 4δ p (1 − δ p )
is a power-symmetric transfer function
29 30
5
tan(ω p / 2)
• Design Steps: • Define r=
tan(ωs / 2)
• Since ω p + ωs = π, δ2s = 4δ p (1 − δ p ) , only
one of the bandedges and one of the ripples r' = 1 − r2
(1 − r ')
can be specified q0 =
• Let the specified stopband edge and 2(1 + r ')
and compute
stopband ripple be ωs and δs , respectively
q = q0 + 2 q05 + 15q09 + 150 q13
0
• Then ωp and δ p are determined using the 2
⎛ 1 − δ2s ⎞
equations at the top of the slide D = ⎜⎜ 2 ⎟⎟
31
32 ⎝ δs ⎠ Copyright © 2010, S. K. Mitra

• Next, the estimate of the order of G(z) is • As a result, the corresponding value of δs
determined by choosing the smallest odd will be smaller than the original specified
integer satisfying value
log (16 D)
N ≥ 10 • To determine the actual value of δs , the
log10 (1 / q ) actual value of the parameter D is first
• Now the integer value of N is almost always computed from
higher than the quantity on the RHS of the 10 N log10 (1 / q )
D=
above equation 16
33 34

• From the actual value of D, the actual value • Next, the poles of the two allpass filters are
of δs is computed by solving computed as follows:
2
⎛ 1 − δ2 ⎞ 2 q1 / 4 ∑i∞=0 (−1)i qi (i +1) sin((2i + 1)kπ / N )
D = ⎜⎜ 2 s ⎟⎟ λk =
⎝ δs ⎠
2
1 + 2 ∑i∞=0 (−1)i qi cos(2πki / N )
• From the new value of δs , the actual value
bk = (1 − rλ2k )⎛⎜1 − k ⎞⎟
λ2
of δ p is obtained from ⎝ r ⎠
δ2s = 4δ p (1 − δ p ) ck = 2bk /(1 + λ k )
2
35 36 α k −1 = (2 − ck ) /(2 + ck )
6
• In general, the two infinite sums in the • Using the pole-interlacing property, then
expression for converge after the poles of A0 (z ) and A1 (z ) are selected
addition of 5 or 6 terms • Their corresponding zeros are at the mirror-
• The poles of the two allpass filters are on image locations
the imaginary axis at z = ± j α k and are • Example
inside the unit circle, as the parameters α k • We consider the design of an elliptic half-
are distinct with magnitudes less than 1 band lowpass filter
37 38

• The specifications are: • The pole-zero plot and the magnitude
ωs = 0.6 π δ s = 0.016 response of the designed elliptic lowpass
• The transfer functions of the two allpass half-band filter are shown below:
sections of the half-band filter designed 1
The pole-zero plot Half-Band IIR Lowpass Filter
1
using Program 13_9 are given by 0.5
Imaginary Part
0.8
Magnitude
0 0.6
−1 −1
0.23647 + z 0.71454 + z 0.4
A0 ( z ) = , A1 ( z ) =
-0.5
1 + 0.23647 z −1 1 + 0.71454 z −1
0.2
-1
0
-1.5 -1 -0.5 0 0.5 1 1.5 0 0.2 0.4 0.6 0.8 1
Real Part ω/ π
39 40

Ch13 (5) Handouts 4e PDF

Încărcat de

Informații document

Titlu original

Drepturi de autor

Formate disponibile

Partajați acest document

Partajați sau inserați document

Opțiuni de partajare

Vi se pare util acest document?

Este necorespunzător acest conținut?

Drepturi de autor:

Formate disponibile

Ch13 (5) Handouts 4e PDF

Încărcat de

Drepturi de autor:

Formate disponibile

Nyquist Filters Lth-Band Filters

Lth-Band Filters Lth-Band Filters

• Assume that the k-th polyphase component y[ Ln + k ] = α x[n]

Lth-Band Filters Lth-Band Filters

Half-Band Filters Half-Band Filters

Half-Band Filters Half-Band Filters

Design of Linear-Phase Design of Linear-Phase

Design of Linear-Phase Design of Linear-Phase

Design of Linear-Phase Design of Linear-Phase

Design of Linear-Phase Design of Linear-Phase

⎪⎪2 0.8 0.8

g[n] = ⎨ 0, n odd , n ≠ N 0.4 0.4

⎪⎩ 2 , n 0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1

Design of Half-Band IIR Filters Design of Half-Band IIR Filters

Design of Half-Band IIR Filters Design of Half-Band IIR Filters

Design of Half-Band IIR Filters Design of Half-Band IIR Filters

Design of Half-Band IIR Filters Design of Half-Band IIR Filters

Design of Half-Band IIR Filters Design of Half-Band IIR Filters

S-ar putea să vă placă și