Frequency Transformation
Frequency Transformation
Frequency Transformation
INTRODUCTION:
The standard procedure for analogue filter design is to first design (obtain the
transfer function) a low-pass normalized to unity, and then, by means of a
frequency transformation, obtain the transfer function of the desired filter, either
low-pass, high-pass, band-pass or band-stop with desired critical values.
The low-pass filter initially designed, before performing the frequency
transformation, is called the low-pass prototype filter.
Since the desired performance characteristics (design specifications) are for the
final, or actual, filter, it is necessary to determine what the corresponding
characteristics or specifications need to be for the low-pass prototype filter.
LOWPASS-TO-LOWPASS TRANSFORMATION
k 0
M M
k fM bk skk fM k k fN M bk skk fM k
k 0
N
N
k 0
[1]
k N
f a s k
k 0
k
k N k
f a s k
k 0
k
k N k
f
M
k fN M k sk
H LP s N
k 0
[2]
s
k 0
k
k
Page 1 of 21
M
K s k
H LPP s N
k 1
[3]
s
k 1
kk
Where the zeroes, k and the poles, k may be, and in general are complex.
When frequency-scaled, the transfer function becomes
M
K k N M s k f k
H LP s N
k 1
[4]
s k
k 1
f k
Hence, frequency scaling multiplies the poles and zeros by the frequency-scaling
factor k f .
Example 1: Given the 3rd order Butterworth lowpass prototype transfer function
1
H LPP s 3 , obtain its lowpass transfer function given that
s 2s 2s 1
2
c 1000 rad / s .
Solution:
1
H LP s H LPP s s s 3 2
s s
1000 s
2 2 1
1000 1000 1000
1
3 2
s s s
9
2 6 2 3 1
10 10 10
Multiplying the numerator and denominator by 10 9 , we get
109
s3 s2 s
9 2 6
2 3 1109
10 10 10
109
3
s 2 103 s 2 2 106 s 109
Page 2 of 21
LOWPASS-TO-HIGHPASS TRANSFORMATION
M
k f
b k bk f
k k
s bk
kf
H HP s H LPP s s k f
k 0 k 0 s
k 0 kf [5]
N k N k
k
ak k f s
N N
s k k
s
ak f ak
k 0
k 0 s
k 0 kf
For a Butterworth prototype filter, M 0 and there is symmetry to the ak
coefficients: a 0 a N , a1 a N 1 , etc. hence, equation [5] becomes
N
s
kf sN s
H HP s k
N H LPP [6]
N s kf kf
ak
kf
k 0
Solution:
1
H HP s H LPP s s 1000 3 2
s 1000 1000 1000
2 2 1
s s s
1 s3
9 3
10 2 10 2 10 1 s
6 3
3
s s2 s
s3
s 3 2 103 s 2 2 106 s 109
To confirm that this is a highpass filter, have that
j
3
H HP j or
j 2 10 j 2 106 j 109
3 3 2
3
H HP j
10 2000 2 2 2 106 2
9 2 2
Page 3 of 21
Hence,
H HP 0 0,
H HP 1 and
1000
3
H HP j1000
10 9
2000 1000
2 2
1000 2 106 1000
2
2 2
109
1 1018 1 1012
1
2
So H HP s is clearly a highpass filter with c 1000 rad / s .
s
k 1
k
M
K k
Where K HP N
k 1
k 1
k
Page 4 of 21
LOWPASS-TO-BANDPASS TRANSFORMATION
s 2 02
This is achieved by replacing every s in H LPP s by ,
B ps
where 0 center frequency of the bandpass filter
B p the passband bandwidth i.e. B p p2 p1 , where p2 is the upper
edge of the passband and p1 is the lower edge of the passband.
Example 3: Given the 3rd order Butterworth lowpass prototype transfer function
1
H LPP s 3 , obtain its bandpass transfer function.
s 2s 2s 1
2
Solution:
1
H BP s H LPP s s s2 02 3 2
B ps s 2 02 s 2 02 s 2 02
2 2 1
B p s B p s B p s
B p3s 3
s 6 2B p s 5 302 2B p2 s 4 B p 402 2B p2 s 3
02 302 2B p2 s 2 2B p04s 06
If the denominator is factored, then we get
K1 K 2s 3
H BP s 3
s 1s 2s 3 s 1s 2s 3
2 3 2
H LS s H HS s
From the above analysis, we can conclude that:
a) The order of the bandpass transfer function is twice that of the lowpass
prototype and therefore will be always be even,
b) The bandpass transfer function may be factored into the product of lowpass
transfer function and highpass transfer function, and
c) H BP 0 H BP 0, which is typical for a bandpass filter.
s
k 1
k
Then,
Page 5 of 21
CASE I: k is real
Then the transfer function for bandpass filter is
M 2
s 02
K k
B s
s N 2 p 2
k 1
H BP [8]
s 0
k 1 B ps
k
To determine corresponding bandpass filter poles, let
s 2 02
k 0 or
B ps
s 2 B p k s 02 0
Then the two poles corresponding to the one lowpass prototype pole, k , are
B p k
j
s1,2 4o2 B 2p k2
[9]
2 2
Therefore, each real pole in the H LPP s will yield a pair of complex conjugate
poles in H BP s , and one zero at the origin.
Page 6 of 21
Hence, the two zeros corresponding to the one lowpass prototype zero, j k
are
jB p k
j
s1,2 B 2p k2 4o2
[10]
2 2
Note that both zeros are imaginary, but not the same magnitude ( not
conjugates). However, if the lowpass prototype has a zero at j k , there will
be also be a zero at j k .
Hence, the four bandpass zeros corresponding to the two lowpass prototype
zeros at j k are:
jB p k j
s1,2 B 2p k2 4o2
2 2
jB p k
j
s3,4 B 2p k2 4o2
[11]
2 2
Note also that
s 2 02
j k
B ps
Indicates two poles at s 0 , cancelling two of the zeros introduced at the
origin by the lowpass-to-bandpass transformation of two real poles of H LPP s
or of one complex-conjugate pair of poles.
j
j3 36 100
2
136
j3 j
4
j 3 j 5.83
Hence, s1 j 8.83 and s2 j 2.83
jB p k j j2 3 j
s3,4
2
2
B 2p k2 4o2
2
2
22 32 4 52
j 3 j 5.83
Hence, s3 j 8.83 and s4 j 2.83
Page 7 of 21
CASE III: k k j k
To determine the bandpass filter poles, we set the denominator of equation
[8] to 0 and solve as follows:
s 2 02
k 0 or s 2 B p k s 02 0
B ps
Then the two poles corresponding to the one lowpass prototype pole
k k j k , are
B p k 1
s B 2p k2 4o2
2 2
B p k
2
Substituting the value of k in , we get
1 1 1
B 2p k j k 4o2 B 2p k j k k j k 4o2
2
B 2p k2 4o2
2 2 2
1
2
B 2p k2 j k k j k k k2 4o2
1
2
B 2p k2 k2 j 2k k 4o2
1
2
B 2p k2 k2 4o2 j 2B 2p k k
1
Rk jI k
2
Where Rk B 2p k2 k2 4o2 and I k 2B 2p k k
1 1 1 k
M k k 2 M k cos k j 2 M k sin 2
2 2 2
I
Where M k Rk2 I k2 and k tan1 k
RK
Therefore,
B B B
s p k p k Re j p k Im
2 2 2
B 1 B 1
p k M k cos k j p k j M k sin k
2 2 2 2 2 2
Therefore, the lowpass-to-bandpass transformation results in two pair of
complex-conjugate poles for every complex-conjugate pair of poles in the
lowpass prototype.
Page 8 of 21
In addition, two zeros at the origin are introduced for every complex-conjugate
pair of poles in the lowpass prototype.
In summary, one pair of complex-conjugate poles in the lowpass prototype,
k j k results in the following poles in the bandpass transfer function:
1 k
s1,3 B pk M k cos j M k sin k B p k [12]
2 2 2
1 k
s2,4 B pk M k cos j M k sin k B p k [13]
2 2 2
Solution:
Rk B 2p k2 k2 4o2 22 12 12 4 52 4 0 100 100
I k 2B 2p k k 2 2 11 8
2
Then,
100
2
Mk 82 100.319
Ik 1 8
k tan1 tan 4.574 or 175.426
0 0
RK 100
1 k
s1,3 B pk M k cos j M k sin k B p k
2 2 2
1 175.426 175.426
2 1 100.319 cos j 100.319 sin 2 1
2 2 2
1
2 0.3997 j 10.008 2
2
1
1.6003 j 8.00795
2
0.8002 j 4.004
1 k
s2,4 B pk M k cos j M k sin k B p k
2 2 2
1 175.426 175.426
2 1 100.319 cos j 100.319 sin 2 1
2 2 2
Page 9 of 21
1
2 0.3997 j 10.008 2
2
1
2.3997 j12.00795
2
1.1996 j 6.004
Example 7: Given that the lowpass prototype is a 4th order Butterworth filter
with 0 100 rad / s and B p 20 rad / s . Find the corresponding bandpass
filter poles.
Solution:
The Butterworth poles are obtained as follows:
2k n 1 2k 3 2k n 1
j j j
s4 e 2n
e 8
e 2n
,k 1,2,3,4
5
j 5 5
For k 1, s1 e 8
cos j sin 0.3827 j 0.9239
8 8
5
j 5 5
For k 2, s2 e 8
cos j sin 0.9239 j 0.3827
8 8
9
j 9 9
For k 3, s3 e 8
cos j sin 0.9239 j 0.3827
8 8
11
11
j 11
For k 4, s4 e cos j sin
8
0.3827 j 0.9239
8 8
Hence, the four lowpass prototype poles are
Rk B 2p k2 k2 4o2 202 0.38272 0.92392 4 1002 40282.85277
Then,
40282.85277
2
Mk 282.8612442 40283.84587
Page 10 of 21
Ik 1 282.861244
k tan1 tan 0.40232 or 179.6
0 0
R
K 40282.85277
Hence,
1 k
s1,3 B pk M k cos j M k sin k B p k
2 2 2
1 179.6
20 0.3827 40283.84587 cos
2 2
179.6
j 40283.84587 sin 20 0.9239
2
1
2
7.654 0.7006 j 200.7071 18.478
1
6.9534 j 182.2291
2
3.477 j 91.11
1 k
s2,4 B pk M k cos j M k sin k B p k
2 2 2
1 179.6
20 0.3827 40283.84587 cos
2 2
179.6
j 40283.84587 sin 20 0.9239
2
1
7.654 0.7006 j 200.7071 18.478
2
1
8.3546 j 219.1851
2
4.18 j109.59
b) s2,4 0.9239 j 0.3827
Rk B 2p k2 k2 4o2 202 0.92392 0.38272 4 1002 39717.14723
Then,
39717.14723
2
Mk 282.8612442 39718.15447
Page 11 of 21
Ik 1 282.861244
k tan1 tan 0.40805 or 179.592
0 0
R
K 39717.14723
Hence,
1 k
s1,3 B pk M k cos j M k sin k B p k
2 2 2
1 179.592
20 0.9239 39718.15447 cos
2 2
179.592
j 39718.15447 sin 20 0.3827
2
1
2
18.478 0.6957 j 199.293 7.654
1
17.7823 j 91.639
2
8.89 j 95.82
1 k
s2,4 B pk M k cos j M k sin k B p k
2 2 2
1 179.592
20 0.9239 39718.15447 cos
2 2
179.592
j 39718.15447 sin 20 0.3827
2
1
18.4784 0.0.6957 j 199.293 7.654
2
1
19.478 j 206.947
2
9.59 j103.47
Page 12 of 21
LOWPASS-TO-BANDSTOP TRANSFORMATION
Example 8: For the given lowpass prototype filter transfer function, obtain its
bandstop equivalent.
1
H LPP s 3
s 2s 2s 1
2
Solution:
1
H BS s H LPP s s B p s 3 2
s 2 02 B ps B ps B ps
2 2 2 2 2
1
s 0 s 0 s 0
2 2 2
s 02
2 3
s 6 2B p s 5 302 2B p2 s 4 B p 402 2B p2 s 3
02 302 2B p2 s 2 2B p04s 06
NB: The denominator is the same as that for lowpass-to-bandpass
transformation. H BS s can be expressed as
K1 K 2s 3
H BS s 3
s 1s 2 2s 3 s 3 1s 2 2s 3
H LP s H HP s [14]
From above analysis, we have that
a) The order of bandstop transfer function is twice that of the lowpass prototype
and therefore will always be even.
b) The bandstop transfer function may be separated into a sum of a lowpass
transfer and a highpass transfer function and
c) H BS 0 H BS 1 , which is a typical bandpass filter.
NB: If the lowpass prototype filter, has a ripple in either the passband or the
stopband, then H BS 0 and H BS may not be unity.
Page 13 of 21
Poles and Zeros
If H LPP s is factored as:
M
K s k
H LPP s N
k 1
s
k 1
k
Then,
CASE I: k is real
Then the transfer function for bandpass filter is
M
B s
K 2 p 2 k
k 1 s 0
H BS s N [15]
B ps
2
k 1 s 0
2
k
To determine corresponding bandstop filter poles, let
B ps
k 0 or
s 02
2
B ps k s 2 02 0 or
Bp
s2 s 02 0
k
Hence, the two poles corresponding to the one lowpass prototype, k , are
2
Bp1 Bp
s1,2 4o2 [16]
2k 2 k
2
B 2p
In practical bandstop filters, 4 2
o then equation [16] becomes
2k
Bpj B p2
s1,2 4o 2
2
[17]
2k 2 k
NB: Since numerator and denominator of the transfer function is multiplied
by s 2 02 in order to form the above quadratic equation, it follows that two
transfer function zeros are introduced at j o .
Therefore, each real pole in H LPP s will yield a pair of complex conjugate poles
in H BS s and two zeros at j o .
Page 14 of 21
Example 8: Given k 1, o 5, and B p 2, find the bandstop poles and
zeros.
Solution:
2
Bp j B 2 j 22
s1,2 4o2 2p 4 52 2
2k 2 k 2 1 2 1
j
1 100 4
2
1 j 4.9
In addition, the resultant zeros will be at s j 5.
2
jB p j Bp
s 4o2 [17]
2k 2 k
2
2
jB p j Bp
s3,4 4o2 [19]
2k 2 k2
In summary, each pair of conjugate imaginary zeros in H LPP s will yield two
pair of conjugate imaginary zeros in H BS s , and two pair of conjugate
imaginary poles at j o .
Page 15 of 21
Example 9: If o 5, o 5, and B p 2, and there are two lowpass prototype
zeros located at j 3. Find the poles and zeros bandstop filter.
Solution:
2
jB p
j Bp j2 j 22
s1,2 4 2
4 52
2 3 2 3
o
2k 2 k
2 3
j j 4
100
3 2 9
j
j 5.011
3
j 4.68 or j 5.33
2
jB p j Bp j2 j 22
s3,4 4 2
4 52
2 3 2 33
o
2k 2 k2
j j 4
100
3 2 9
j
j 5.011
3
j 5.33 or j 4.68
There are two conjugate imaginary poles introduced at j 5.
CASE III: k k j k
To determine the bandpass filter poles, we set the denominator of equation
[15] to 0 and solve as follows:
B ps Bp
k 0 or s 2 s 02 0
s
2 2
0 k
Then the two poles corresponding to the one lowpass prototype pole
k k j k , are
2
Bp
1 Bp
s 4o2
2k 2 k2
Bp
2k
Substituting the value of k in , we get
Page 16 of 21
2
1 Bp 1 B p2 1 B p2
4o2 4 2
4o2
2 k
2
2 k j k
2
2
o
k j k k j k
1 B p2
4o2
2 k
2
jk k jk k k 2
1 B p2
4o2
2 k
2
k j 2k k
2
1
B p2 k2 k2 j 2k k 4o2
2 k
2
k j 2k k
2
2
k
k j 2k k
2
2 2
1 B p k k j 2B pk k
2 2
4o2
2
2 k2 k2 2k k
2
2
2 2
2
k
k2 k k k4 2 k2 k2 k4 4 k2 k2
k4 2 k2 k2 k4
2
k2 k2
Hence,
1 B p k k j 2B pk k
2 2 2 2
4o2
2
2 2 2 k k
j 2B 4
B p2 k2 k2 2
p k k 2
2 2 o
2 2 2 2
k k k k
B
2
p
2
j 2B
2 2
4
k k 2 p k k
2 o 2
2 2 2 2
k k k k
1
Rk jI k
2
Where Rk
B p2 k2 k2 4 2
and I k
2B p2k k
2 o 2
k
2
k2 2
k
k2
I
Then M k Rk2 I k2 and k tan1 k
RK
Therefore,
Page 17 of 21
B B
s1 2 p k 2 M k cos k j 2 p k 2 M k sin k [20]
k k 2 k k 2
B B
s2 2 p k 2 M k cos k j 2 p k 2 M k sin k [21]
k k 2 k k 2
If there is a pole in the lowpass prototype at k j k there will also be one at
k j k resulting in the following two poles in the bandstop transfer
function:
B B
s3 2 p k 2 M k cos k j 2 p k 2 M k sin k [22]
k k 2 k k 2
B B p k k
s4 2 p k 2 M k cos k j M k sin 2 [23]
k k k k
2 2
2
Solution:
Rk
B p2 k2 k2 4 2
22 12 12
4 52 100
1
2 o
2 2
2
k
k2 2
1
2B p2k k 2 22 11 8
Ik 2
1
2 2 2
2
k 2 2
1 4
k
100 2 100.02
2 2
Mk
Ik 1 2
k tan1 tan 1.15 or 178.85
0 0
R
K 100
B B
s1,3 2 p k 2 M k cos k j 2 p k 2 M k sin k
k k 2 k k 2
2 1 178.85 2 1 178.85
2 2 100.02 cos j 2 2 100.02 sin
1 1 2 1 1 2
1
1 0.1004 j 1 10.0005
2
0.4498 j 5.5
Page 18 of 21
B B
s2,4 2 p k 2 M k cos k j 2 p k 2 M k sin k
k k 2 k k 2
2 1 178.85 2 1 178.85
2 2 100.02 cos j 2 2 100.02 sin
1 1 2 1 1 2
1
1 0.1004 j 1 10.0005
2
0.5502 j 4.5
Example 11: Given that the lowpass prototype is a 4th order Butterworth
filter, o 100, and B p 20. Find the poles and zeros of the corresponding
bandstop filter.
Solution:
From the previous question, the Butterworth poles are
s1,4 0.3827 j 0.9239 and
s2,3 0.9239 j 0.3827
Then, the bandstop poles are obtained as follows:
Rk
B p2 k2 k2 4 2
202 0.38272 0.92392
4 1002 40282.8242
0.3827
2 o
2 2
2
k
k 2 2
0.9239
k k
Ik 1 282.8327
k tan1 tan 0.4023 or 179.6
0 0
RK 40282.8242
B B
s1,3 2 p k 2 M k cos k j 2 p k 2 M k sin k
k k 2 k k 2
Page 19 of 21
20 0.3827 179.6
40283.8171cos
0.3827 0.9239
2 2
2
20 0.9239 179.6
j 40283.8171sin
0.3827 0.9239
2 2
2
1
6.95301 j 219.072
2
3.452 j109.536
B B
s2,4 2 p k 2 M k cos k j 2 p k 2 M k sin k
k k 2 k k 2
20 0.3827 179.6
40283.8171cos
0.3827 0.9239
2 2
2
20 0.9239 179.6
j 40283.8171sin
0.3827 0.9239
2 2
2
1
8.3538 j182.118
2
4.177 j 91.059
Rk
B p2 k2 k2 4 2
202 0.92392 0.38272
4 1002 39717.1758
0.9239
2 o
2 2
2
k
k 2 2
0.3827
Ik 1 282.8327
k tan1 tan 0.408 or 179.592
0 0
RK 39717.1758
B B
s1,3 2 p k 2 M k cos k j 2 p k 2 M k sin k
k k 2 k k 2
Page 20 of 21
20 0.9239 179.592
39718.1828 cos
0.9239 0.3827
2 2
2
20 0.3827 179.592
j 39718.1828 sin
0.9239 0.3827
2 2
2
1
17.7675 j 206.947
2
8.884 j103.473
B B
s2,4 2 p k 2 M k cos k j 2 p k 2 M k sin k
k k 2 k k 2
20 0.9239 179.592
39718.1828 cos
0.9239 0.3827
2 2
2
20 0.9239 179.592
j 39718.1828 sin
0.9239 0.3827
2 2
2
1
19.1866 j191.6393
2
9.593 j 95.820
The transform will also introduce four pair of conjugate imaginary zeros at
j100.
Page 21 of 21