Chapter 14 Analog Filters
14.1 General Considerations
14.2 First-Order Filters
14.3 Second-Order Filters
14.4 Active Filters
14.5 Approximation of Filter Response
Outline of the Chapter
Why We Need Filters
In order to eliminate the unwanted interference that accompanies a signal, a filter is needed.
Filter Characteristics
Ideally, a filter needs to have a flat pass band and a sharp roll- off in its transition band.
Realistically, it has a rippling pass/stop band and a transition band.
Example 14.1: Filter I
Solution: A filter with stopband attenuation of 40 dB
Problem : Adjacent channel interference is 25 dB above the
signal. Determine the required stopband attenuation if Signal to Interference ratio must exceed 15 dB.
Example 14.2: Filter II
Problem: Adjacent 60-Hz channel interference is 40 dB above the signal. Determine the required stopband attenuation
To ensure that the signal level remains 20dB above the interferer level.
Solution: A high-pass filter with stopband attenuation of 60 dB at 60Hz.
Example 14.3: Filter III
A bandpass filter around 1.5 GHz is required to reject the adjacent Cellular and PCS signals.
Classification of Filters I
Classification of Filters II
Continuous-time Discrete-time
Classification of Filters III
Passive Active
Summary of Filter Classifications
Filter Transfer Function
Filter (a) has a transfer function with -20dB/dec roll-off.
Filter (b) has a transfer function with -40dB/dec roll-off and provides a higher selectivity.
(a) (b)
General Transfer Function
pk = pole frequencies zk = zero frequencies
1 2
1 2
( )
mn
s z s z s z
H s s p s p s p
Example 14.4 : Pole-Zero Diagram
1 1
( ) 1
a 1
H s R C s
1 2
1 1 2
( ) 1
( ) 1
b
R C s H s R C C s
1 2
1 1 1 1 1
c( ) H s C s
R L C s L s R
Impulse response contains
Example 14.5: Position of the poles
Poles on the RHP Unstable (no good)
Poles on the jω axis Oscillatory
(no good)
Poles on the LHP Decaying
(good)
exp( p t
k) exp(
kt ) exp( j
kt )
Transfer Function
The order of the numerator m ≤ The order of the denominator n Otherwise, H(s)→ as s→
For a physically-realizable transfer function, complex zeros or poles occur in conjugate pairs.
If a zero is located on the jω axis, z1,2=± jω1 , H(s) drops to zero at ω1.
1 2
1 2
( )
mn
s z s z s z
H s s p s p s p
Imaginary Zeros
Imaginary zero is used to create a null at certain frequency.
For this reason, imaginary zeros are placed only in the stop band.
Sensitivity
P C
dP dC S P C
Sensitivity indicates the variation of a filter parameter due to variation of a component value.
P=Filter Parameter
C=Component Value
Example 14.6: Sensitivity
1 1 /
1
0 1
1 1 0
0
1 2 1 1
0
1 1 0
S
RR dR d
C R dR
d
C R
Problem: Determine the sensitivity of ω
0with respect to R
1.
For example, a +5% change in R1translates to a -5% error in
ω
0.First-Order Filters
1 1
( ) s z
H s s p
First-order filters are represented by the transfer function shown above.
Low/high pass filters can be realized by changing the relative positions of poles and zeros.
Example 14.8: First-Order Filter I
R2C2 < R1C1 R2C2 > R1C1
2 1 1
1 2 1 2 1 2
( 1)
( ) ( )
out in
V R R C s
V s R R C C s R R
1
1 (
1 1),
z R C p
1 [( C
1 C R
2)
1 R
2]
1.
Example 14.9: First-Order Filter II
R2C2 < R1C1 R2C2 > R1C1
2
2
1
1
2 1 1
1 2 2
( 1 )
( ) 1
1 1
out in
V R C s
V s R
C s R R C s R R C s
2
2 2
( )
n
n
s s
H s
s s
Q
1,2 2
1 1
2 4
n
p j n
Q Q
Second-Order Filters
Second-order filters are characterized by the “biquadratic” equation with two complex poles shown above.
When Q increases, the real part decreases while the imaginary part approaches ± ωn.
=> the poles look very imaginary thereby bringing the circuit closer to instability.
Second-Order Low-Pass Filter
2 2
2 2
2 2
( )
n n
H j
Q
α = β = 0
Peak magnitude normalized to the passband magnitude: Q 1 (4 Q2)1
.
Example 14.10: Second-Order LPF
2
2
3
/ 1 1/(4 ) 3 1 1/(2 )
n n
Q
Q Q
Q
Problem: Q of a second-order LPF = 3.
Estimate the magnitude and frequency of the peak in the frequency response.
Second-Order High-Pass Filter
2
2 2
( )
n
n
H s s
s s
Q
β = γ = 01 1 (2 2)
n Q
Frequency of the peak:
Peak magnitude normalized to the passband magnitude: Q 1 (4 Q2)1
Second-Order Band-Pass Filter
2 2
( )
n
n
H s s
s s
Q
α = γ = 0Example 14.2: -3-dB Bandwidth
Problem: Determine the -3dB bandwidth of a band-pass response.
2 2
( )
(
n n)
H s s
s s
Q
1,2 0 2
1 1
1 4 Q 2 Q
2 2 2 2
2 2 2 2
2
2(
n) (
n)
nQ Q
BW
0Q
LC Realization of Second-Order Filters
1
1 1 2
1 1 1
( ) || 1
1 Z L s L s
C s L C s
An LC tank realizes a second-order band-pass filter with two imaginary poles at ±j/(L1C1)1/2 , which implies infinite
impedance at ω=1/(L C )1/2.
Example 14.13: LC Tank
At ω=0, the inductor acts as a short.
At ω=, the capacitor acts as a short.
1
1 1 2
1 1 1
( ) || 1
1 Z L s L s
C s L C s
RLC Realization of Second-Order Filters
1 1 1
2 1 2 2
1 1 1 1 1 1 1
1 1 1 1
2 2 2
1 1 1 1 1 1
1 1 1 1
|| 1
1 1
( ) (
n n)
L s R L s
Z R
L C s R L C s L s R
R L s R L s
R L C s s R L C s s
R C L C Q
1,2 2
1 2
1 1 1 1 1 1
1 1
2 4
1 1
2 1 4
n n
p j
Q Q
j L
R C L C R C
With a resistor, the poles are no longer pure imaginary which implies there will be no infinite impedance at any ω.
1 1 1 1 1
1 ,
n
Q R C L C L
Voltage Divider Using General Impedances
out ( ) P
in S P
V Z
V s Z Z
Low-pass High-pass Band-pass
Low-pass Filter Implementation with Voltage Divider
2 1
1 1 1 1 1
out in
V R
V s R C L s L s R
1
as
Z
S L s s
1 1
1 0 as
Z
PR s
C s
Example 14.14: Frequency Peaking
2 11 1 1 1 1
o u t in
V R
V s R C L s L s R
22 2 2 2
1 1 1 1 1
Let D R R C L L
Voltage gain greater than unity (peaking) occurs when a solution exists for
2
2 2
1 1 1 1 1 1 1 1
2 2( )( )
( )
0
d D R C L R R C L L
d
1 1
1
Thus, when 1 ,
2 peaking occurs.
Q R C
L
Example 14.15: Low-pass Circuit Comparison
The circuit (a) has a -40dB/dec roll-off at high frequency.
However, the circuit (b) exhibits only a -20dB/dec roll-off since the parallel combination of L1 and R1 is dominated by R1
because L1ω→, thereby reduces the circuit to R1 and C1.
Good Bad
(a) (b)
High-pass Filter Implementation with Voltage Divider
2
1 1 1 1 1
2
1 1 1 1 1
1 1
1
( ) ||
( ) || 1
o u t i n
V L s R L C R s
V s L s R R C L s L s R
C s
Band-pass Filter Implementation with Voltage Divider
1
1 1
2
1 1 1 1 1
1 1
1
( ) || 1
( ) || 1
out in
V L s C s L s
V s L s R R C L s L s R C s
Why Active Filter?
Passive filters constrain the type of transfer function.
They may require bulky inductors.
Sallen and Key (SK) Filter: Low-Pass
1 2 1 2 2
1 2
21
1
out in
V s
V R R C C s R R C s
1 2 1
1 2 2
1 C
Q R R
R R C
1 2 1 21
n
R R C C
Sallen and Key filters are examples of active filters. This
particular filter implements a low-pass, second-order transfer
Example 14.16: SK Filter with Voltage Gain
3 4
2 3
1 2 1 2 1 2 2 2 1 1
4
1
1
out in
R
V R
V s R
R R C C s R C R C R C s R
Example 14.17: SK Filter Poles
The poles begin with real, equal values for and become complex for .
3 4
2 3
1 2 1 2 1 2 2 2 1 1
4
1 ( )
( ) 1
out in
R
V R
s R
V R R C C s R C R C R C s R
1 2 2 2 1 1 3
2 1 1 1 2 2 4
1 R C R C R C R
Q R C R C R C R
3 4
1 2
Q R
R
3 4
0
R R 0
R R
Problem: Assuming R1=R2, C1=C2, Does such a filter contain complex poles?
Sensitivity in Low-Pass SK Filter
1 2 1 2
1 2
n n n n
R R C C
S
S
S
S
1 2
2 2 1 1
1 2
Q Q
R R
S S Q R C
R C
1 2
2 2 1 2
1 1 2 1
1 2
Q Q
C C
R C R C
S S Q
R C R C
1 1 2 2 Q
K
S QK R C
R C
3 4
1
K R R
Example 14.18: SK Filter Sensitivity I
1 2
1 2
1 1
2 3
1 2
2 3 3
Q Q
R R
Q Q
C C
Q K
S S
K
S S
K S K
K
Problem: Determine the Q sensitivities of the SK filter for the common choice R1=R2=R, C1=C2=C.
1 2
With =1, 0
1
Q Q
R R
Q Q Q
C C K
K
S S
S S S
Integrator-Based Biquads
2
2 2
out in n
n
V s
V s s s
Q
2 2
.1
n n
out in out out
V s V s V s V s
Q s s
It is possible to use integrators to implement biquadratic transfer functions.
KHN (Kerwin, Huelsman, and Newcomb) Biquads
2 2
Comparing with Vout s Vin s n .1Vout s n Vout s
Q s s
5 6
4 5 3
R 1 R
R R R
현재 이 이미지를 표시할 수 없습니다 .
4 6
4 5 1 1 3
. 1 . 1
n R R
Q R R R C R
2 6
3 1 2 1 2
. 1
n
R
R R R C C
2
1 1 2 2 1 2 1 2
1 1 1
X out, Y X out
V V V V V
R C s R C s R R C C s
5 4 6 6
4 5 3 3
in X 1
out Y
V R V R R R
V V
R R R R
Versatility of KHN Biquads
2
2 2
High-pass: out
in n
n
V s
V s s s
Q
2
2 2 1 1
Band-pass: X . 1
in n
n
V s
V s s s R C s
Q
2
2 2 1 2 1 2 2
Low-pass: Y . 1
in n
n
V s
V s s s R R C C s
Q
Sensitivity in KHN Biquads
1n, 2, 1, 2, 4, 5, 3, 6
0.5
R R C C R R R R
S
1 2 1 2 4 5
3 6
5
, , , ,
4 5
3 6 2 2
, 5 3 6 1 1
0.5, 1,
2 1
Q Q
R R C C R R
Q R R
S S R
R R
R R R C
S Q
R R R R C
Tow-Thomas Biquad
3
2 2 4 1 1
1 1
out in
out
V V
R V
R C s R R sC
Tow-Thomas Biquad
2 3 4 2
2
1 2 3 4 1 2 2 4 2 3
Band-pass:
out.
in
V R R R C s
V R R R R C C s R R C s R
3 4
2
1 2 3 4 1 2 2 4 2 3
Low-pass:
Y. 1
in
V R R
V R R R R C C s R R C s R
Differential Tow-Thomas Biquads
An important advantage of this topology over the KHN biquad is accrued in integrated circuit design, where differential
integrators obviate the need for the inverting stage in the loop.
Example 14.20: Tow-Thomas Biquad
2 4 1 2
1
n R R C C
Adjusted by R2 or R4
1 2 4 2
3 1
1 R R C
Q R C
Adjusted by R3
Note that n and Q of the Tow-Thomas filter can be adjusted (tuned) independently.
Antoniou General Impedance Converter
1 3
5 2 4 in
Z Z Z Z
Z Z
It is possible to simulate the behavior of an inductor by using active circuits in feedback with properly chosen passive
elements.
1 3 5 X
V V V V
4 4
5 X
X
V V Z V
Z
4 3
3
3 Z
V V
I Z
4
5 3
VX Z Z Z
2 3 2 Z3
V V Z I
4 2
5 3
X X
V Z
V Z
Z Z
2 1 X X
V V
I Z
2 4 1 3 5 X
V Z Z
Z Z Z
Simulated Inductor
By proper choices of Z1-Z4, Zin has become an impedance that increases with frequency, simulating inductive effect.
Thus,
in X Y
eq X Y
Z R R Cs
L R R C
High-Pass Filter with SI
With the inductor simulated at the output, the transfer function resembles a second-order high-pass filter.
2 1 2
1 1 1 1 1
out in
V L s
V s R C L s L s R
Example 14.22: High-Pass Filter with SI
4 out 1 Y
X
V V R
R
Node 4 can also serve as an output.
V4 is better than Vout since the output impedance is lower.
Low-Pass Filter with Super Capacitor
1
in
1
X
Z Cs R Cs
1
2 2
1 1
1
1
out in
in in
X
V Z
V Z R
R R C s R Cs
How to build a floating inductor to derive a low-pass filter?
Not possible. So use a super capacitor.
Example 14.24: Poor Low-Pass Filter
4
1 2
out X out out X
X
V V Cs R V V R Cs
R
Node 4 is no longer a scaled version of the Vout. Therefore the output can only be sensed at node 1, suffering from a high impedance.
Frequency Response Template
With all the specifications on pass/stop band ripples and
transition band slope, one can create a filter template that will lend itself to transfer function approximation.
Butterworth Response
2
0
( ) 1
1 H j
n
ω-3dB=ω0, for all n.
The Butterworth response completely avoids ripples in the pass/stop bands at the expense of the transition band slope.
Poles of the Butterworth Response
0
2 1
exp exp , 1, 2, ,
2 2
k
j k
p j k n
n
2
nd-Order nth-Order
Example 14.24: Order of Butterworth Filter
The minimum order of the Butterworth filter is three.
2 2 1
64.2 f n
f
2
2
1f f
Specification: passband flatness of 0.45 dB for f < f1=1 MHz, stopband attenuation of 9 dB at f2=2 MHz.
( 1 1MHz) 0 95 H f
2 2
1 0
1 0 95
1 2
f n
3, 0 2 (1 45MHz) n
( 2 2MHz) 0 355 H f
2 2
2 0
1 0 355
1 2
f n
Example 14.25: Butterworth Response
1
2 2
2 *(1.45 ) * cos sin
3 3
p MHz j
3
2 2
2 * (1.45 ) * cos sin
3 3
p MHz j
2 2 *(1.45 )
p MHz RC section
2nd-order SK Using a Sallen and Key topology, design a Butterworth filter for the
response derived in Example 14.24.
Example 14.25: Butterworth Response (cont’d)
2
1 3
2 2
1 3
( )( ) [2 (1 45MHz)]
( ) ( )( ) 4 (1 45MHz) cos(2 3) [2 (1 45MHz)]
SK
p p H s
s p s p s s
2 (1 45MHz) and 1 / 2 cos 2 1
n Q 3
1 2 1k , 2 54.9pF, and 1 4 2
R R C C C
3 3
3 3
1 2 (1 45MHz) R 1k and C 109.8pF
R C
Chebyshev Response
2 2
0
1
1 n
H j
C
The Chebyshev response provides an “equiripple” pass/stop band response.
Chebyshev Polynomial
Chebyshev Polynomial
Chebyshev polynomial for n=1,2,3
Resulting transfer function for n=2,3
1
0
0 0
1
cos cos ,
Cn n
Chebyshev Response
2 2 1
0
( ) | 1
1 cos cos
HPB j
n
2 2 1
0
( ) | 1
1 cosh cosh
HSB j
n
2
Peak-to-peak Ripple dB 20 log 1
Example 14.26: Chebyshev Response
2 (2MHz)
0.116 18.7dBH j
ω0=2π (1MHz)
A third-order Chebyshev response provides an attenuation of -18.7 dB a 2MHz.
Suppose the filter required in
Example 14.24 is realized with third- order Chebyshev response.
Determine the attenuation at 2MHz.
2
1 0 95 0 329
1
3 2 2
0 0
1
1 4 3
H j
Example 14.27: Order of Chebyshev Filter
Specification:
Passband ripple: 1 dB Bandwidth: 5 MHz
Attenuation at 10 MHz: 30 dB What’s the order?
1 dB 20 log 1
2 0 509
Attenuation at 2
0 10 MHz: 30 dB
2 2 1
1 0 0316
1 0 509 cosh ( cosh 2) n
cosh (1 317 )
2 n 3862 n 3 66 n 4
Example 14.28: Chebyshev Filter Design
1
10 0
2 1 1 1 2 1 1 1
sin sinh sinh cos cosh sinh
2 2
k
k k
p j
n n n n
2,3
0.337
00.407
0p j
SK2
1,4 0.140 0 0.983 0
p
j
SK1
Using two SK stages, design a filter that satisfies the requirements in Example 14.27.
Example 14.28: Chebyshev Filter Design (cont’d)
1 4
1
1 4
2 0
2 2
0 0
( )( )
( ) ( )( )
0 986
0 28 0 986
SK
p p
H s
s p s p
s s
1 0 993 0 2 (4 965MHz)
n
1 3 55
Q
1 2 1 k , 1 50 4 2
R R C C
1 2
1 2 (4 965MHz) 50 4
4.52 pF, 227.8 pF R C
C C
2 3
2
2 3
2 0
2 2
0 0
( )( )
( ) ( )( )
0 279
0 674 0 279
SK
p p
H s
s p s p
s s
2 0 528 0 2 (2 64MHz)
n
2 0 783 Q
1 2
2 1
1 2 (2 64 MHz) 2 45
38 5 pF, C 94.3 pF R C
C
1 2 1 k , 1 2 45 2
R R C C