Active Filters :

Active-RC Filters :
Real frequency characteristics of (a) a low-pass and (b) a band-pass
filter.
Filter Specification : We use the low-pass response in figure to
illustrate the information that must be available to the designer of a
filter. At a minimum, the designer requires the following
specifications:
1.The cutoff frequency f
H
(or f
C
), that is, the range of
passband frequencies.
2.The stopband attenuation H
0
H
2
.
3.The stopband frequency range. That is, f
S
is specified.
4.The allowable passband ripple = H
0
H
1
. If no ripple is
permitted, = 0 and H
1
= H
0
.
Butterworth and Chebyshev Filter Functions : The general form
of the transfer function H(s) can be expressed as
) (
) (
) (
s B
s A
s H =
Where A(s) and B(s) are polynomials in the frequency variables.
Type of
characteristic
Form of transfer function H(s) = A(s)/B(s)
Low-pass
2 2
) / (
o o
s Q s
K
+ +
2 2
) / (
) (
o o
s Q s
z s K
+ +
+

High-pass
2 2
2
) / (
o o
s Q s
Ks
+ +
2 2
) / (
) (
o o
s Q s
z s Ks
+ +
+

Band-pass
2 2
) / (
o o
s Q s
Ks
+ +

Band-elimination
2 2
2 2
) / (
) (
o o
r
s Q s
s K

+ +
+




Biquadratic Transfer Functions
The Biquadratic Function : Consider H(s) to be of the form
0 1
2
0 1
2
2
) (
b s b s
a s a s a
s H
+ +
+ +
=
The expression for H(s) in the above equation is called a biquadratic
function or simply a biquad because both numerator and
denominator are quadratics in s.
The low-pass function in the left-hand column in Table 5-1 can be
written as
1 ) / )( / 1 ( ) / (
) (
0
2
0
2
0
+ +
=
s Q s
H
s H
2
0
/ ) ( K s H =
where
Butterworth Polynomials : The use of Butterworth polynomials is
a common all-pole approximation of the low-pass characteristic.
Hence H(s) = H
0
/B(s), where B(s) is a Butterworth polynomial
whose magnitude is given by
n
B
2
0
2
1 ) (

+ =

Filters that use Butterworth polynomials are called Butterworth

filters.
Normalized Butterworth Polynomials
n Factor of polynomial B
n
(s)
1 (s +1)
2 (s
2
+ 1.414s + 1)
3 (s + 1)(s
2
+ s + 1)
4 (s
2
+ 0.765s + 1)(s
2
+ 1.848s + 1)
5 (s + 1)(s
2
+ 0.618s + 1)( s
2
+ 1.618s + 1)
6 (s
2
+ 0.518s + 1)(s
2
+ 1.414s + 1)(s
2
+ 1.932s + 1)
7 (s +1)( s
2
+ 0.445s + 1)( s
2
+ 1.802s + 1)
8 (s
2
+ 0.390s + 1)( s
2
+ 1.111s + 1)( s
2
+ 1.663s + 1)( s
2
+ 1.962s +1)
Example
Determine the order of a low-pass Butterworth filter that is to
provide 40-dB attenuation at /
0
= 2.
Solution : The normalized magnitude of the filter transfer
function is
n
H
j H
2
0
2
0
) / ( 1
1 ) (

+
=
An attention of 40 dB corresponds to H(j)/H
0
= 0.01, and
hence
) )
n 2
2
2 1 / 1 010 . 0 + =
1 10 2
4 2
=
n
or
Solving for n by taking the logarithmof both sides gives
2 log
) 1 10 log(
2
4

= n
and n = 6.64
Since the order of the filter must be an integer, n = 7.
Chebyshev Filters : When specifications permit a small amount
of passband ripple, a frequently used all-pole approximation is the
Chebyshev filter. The transfer function is of the form
) / ( 1
) (
2 2
2
0
2
C n
C
H
j H
s

+
=
where C
n
(/
C
) are Chebyshev polynomials defined by

C C
n
n C

1
cos cos 1 0 e e
C

=

C
n

1
cosh cosh 1 u
C

Normalized Polynomials for Chebyshev Filters

n Factor of chebyshev filter polynomials
0.5-dB ripple (s=0.3493)
1 s + 2.863
2 s
2
+ 1.425s + 1.516
3 (s + 0.626)(s
2
+ 0.626s + 1.142)
4 (s
2
+ 0.351s + 1.064)(s
2
+ 0.845s + 0.356)
5 (s + 0.362)(s
2
+ 0.224s + 1.036)( s
2
+ 0.586s + 0.477)
6 (s
2
+ 0.1554s + 1.024)(s
2
+ 0.4142s + 0.5475)(s
2
+ 0.5796s + 0.157)
7 (s + 0.2562)( s
2
+ 0.1014s + 1.015)( s
2
+ 0.3194s + 0.6657)( s
2
+ 0.4616s +
0.2539)
8 (s
2
+ 0.0872s + 1.012)( s
2
+ 0.248s + 0.7413)( s
2
+ 0.3718s + 0.3872)( s
2
+
0.4386s +0.08805)
0.5-dB ripple (s=0.3493)
1 s + 1.965
2 s
2
+ 1.098s + 1.103
3 (s + 0.494)(s
2
+ 0.494s + 0.994)
4 (s
2
+ 0.279s + 0.987)(s
2
+ 0.674s + 0.279)
5 (s + 0.289)(s
2
+ 0.179s + 0.988)( s
2
+ 0.468s + 0.429)
6 (s
2
+ 0.1244s + 0.9907)(s
2
+ 0.3398s + 0.5577)(s
2
+ 0.4642s +
0.1247)
7 (s + 0.2054)( s
2
+ 0.0914s + 0.9927)( s
2
+ 0.2562s + 0.6535)( s
2
+
0.3702s + 0.2304)
8 (s
2
+ 0.07s + 0.9942)( s
2
+ 0.1994s + 0.7236)( s
2
+ 0.2994s +
0.3408)( s
2
+ 0.3518s + 0.0702)
The parameter s is related to the passband ripple in decibels by
1 10
10 2
=

s
For 0.5-dB ripple, s = 0.3493 and s = 0.5089 if = 1 dB. The
frequency f
C
=
C
/2x is the ripple bandwidth. The 3-dB
frequency f
H
is related to f
C
by

=

s
1
cosh
1
cosh
1
n
f f
C H
Example (a) Determine the order of a 1-dB ripple Chebyshev
filter that gives a 40-dB attenuation at /
C
= 2.(b) Determine the
3-dB bandwidth of the filter.
Solution : (a) The normalized response is,
) / ( 1
1 ) (
2 2 2
0
2
C n
C H
j H
s

+
=
A 40-dB attenuation corresponds to H(j)/H
0
= 0.01. Thus
) 2 ( ) 5089 . 0 ( 1
1
) 01 . 0 (
2 2
2
n
C +
=
4
2
4
2
10 861 . 3
) 5089 . 0 (
1 10
) 2 ( v =

=
n
C
and
) 5 . 196 2 =
n
C
or
Use of Chebyshev polynomials yields 196.5 = cosh(ncosh
-1
2).
Solution for n given n = 4.536 and n = 5 is selected.
(b) The value of f
H
/f
C
is obtained with n = 5 as
034 . 1
5089 . 0
1
cosh
5
1
cosh
1
= =

C
H
f
f
Comparison of the results in Examples 1 and 2 reveals that if
passband ripple is permitted, a lower-order filter can be used.
That is, a Chebyshev filter requiring fewer stages (lower order) is
needed to obtain the same attenuation as a Butterworth filter.
Single-Amplifier Biquad Sections :
Sallen and Key low-pass section using (a) a noninverting
amplifier (positive-feedback) and (b) an inverting amplifier
(negative feedback)
Low-Pass Sections : The Sallen and Key circuit in figire uses a
noninverting Op-Amp stage to provide positive feedback. The
transfer function of this circuit can be expressed as
1 )] 1 ( ) ( [
) (
1 1 2 1 2
2
2 1 2 1
+ + + +
=
V
V
A C R R R C s s C C R R
A
s H
where A
V
= 1 + R
b
/R
a
is the gain of the Op-Amp stage.
Comparison of the above equation with the low-pass function gives
2 1 2 1
0
1
C C R R
=
) ( ) 1 (
2 1 2 1 1
2 1 2 1
R R C A C R
C C R R
Q
V
+ +
=
Clearly, the five circuit parameters R
1
, R
2
, C
1
, C
2
and A
V
provide
more degrees of freedom than are needed to specify
0
and Q.
Often, to simplify fabrication, C
1
= C
2
= C and R
1
= R
2
= R.
The equations for these conditions reduce to
1 ) 3 (
) (
2 2 2
+ +
=
V
V
A RCs s C R
A
s H
RC
1
0
=
V
A
Q

=
3
1
In the above equation, observe that making both resistors and both
capacitors equal results in the time constant RC establishing
0
and the gain A
V
determining Q. Furthermore, note A
V
u3, then, the
s-coefficient is e 0 indicating H(s) has right-half-plane poles.
Example
Design a low-pass filter having no more than 1-dB ripple from dc
to 1 kHz and that gives a minimum of 40-dB attenuation at 2 kHz.
Solution :
In the example it is shown that with 1-dB ripple, an attenuation of
40 dB obtained with a fifth-order Chebyshev filter. The normalized
characteristic polynomial, obtained fromTable, is
) ) ) ) 429 . 0 468 . 0 988 . 0 179 . 0 289 . 0
2 2
+ + + + + = s s s s s s B
Since H(s) = H
0
/B(s) has three factors, a three-section filter is
used. These section are
289 . 0
) (
01
1
+
'
=
s
H
s H
988 . 0 179 . 0
) (
2
02
2
+ +
'
=
s s
H
s H
429 . 0 468 . 0
) (
2
03
3
+ +
'
=
s s
H
s H
Each function given above is normalized to the cutoff angular
frequency
c
= 1 rad/s. The cutoff frequency desired is f
c
= 1 kHz or

c
= 2x x 10
3
rad/s. The unnormalized functions, written as
1 ) 289 . 0 / (
) (
01
1
+
=
C
s
H
s H

1 ) 988 . 0 / 179 . 0 ( ) 988 . 0 / (

) (
2 2
02
2
+ +
=
C C
s s
H
s H

1 ) 429 . 0 468 . 0 ( ) 429 . 0 / (
) (
2 2
03
3
+ + +
=
C C
s s
H
s H

Both H
2
and H
3
can be realized with Sallen and Key low-pass
sections (Fig. 5.2.2a) with R
1
= R
2
and C
1
= C
2
. Comparison of H
2
and H
3
gives
s rad
C C
/ 10 2 994 . 0 994 . 0 988 . 0
3 2
02
v v = = = x
988 . 0
179 . 0 1
2
=
Q
55 . 5
2
= Q
and
s rad
C C
/ 10 2 655 . 0 655 . 0 429 . 0
3 2
03
v v = = = x
429 . 0
468 . 0 1
3
=
Q
4 . 1
3
= Q
and
The equation relate Q and
0
to the circuit parameters; note,
however, that specification of
0
does not permit unique
determination of R and C but only their product. In filter
fabrication, particularly in hybrid technology, it is sometimes
convenient to use the same value of capacitance in all sections. We
choose C = 0.05F. Hence
O =
v v v v
= =

k
C
R 20 . 3
10 05 . 0 10 2 994 . 0
1 1
6 3
02
2
x
O =
v v v v
= =

k
C
R 86 . 4
10 05 . 0 10 2 655 . 0
1 1
6 3
03
3
x
Two circuits which realize a zero on the negative real axis : (a)
noninverting (b) inverting.
Solution for A
V
gives A
V
= 3 1/Q. Hence
82 . 2
55 . 5
1
3
2
= =
V
A 285 . 2
4 . 1
1
3
3
= =
V
A
Since A
v
>> 1 for a commercial Op-Amp, A
v
= 1 + R
b
/R
a
. To
minimize both the number of resistance values used and the
spread of element values (the ratio of the largest to the smallest
resistance), we choose R
a
= 4.86 kO for both selections. Then
) ) O = = = k R A R
a v b
84 . 8 86 . 4 1 82 . 2 1
2 2 2
) ) O = = = k R A R
a v b
4 . 6 86 . 4 1 285 . 2 1
3 3 3
The remaining section, H
1
(s), contains a real pole of s = -0.289
C
=
0.289 x 2x x 10
3
rad/s. The simple voltage follower driving an RC
circuit as
1
1
1
21
+
=
RCs V
V
and is used to realize the real pole in H
1
(s). Comparison of the
transfer functions yields
C
RC
289 . 0
1
=
Choice of C = 0.05 F results in
O = v v v v =

k R 0 . 11 10 05 . 0 10 2
289 . 0
1
6 3
x
The final circuit is displayed in Figure.
Negative-feedback circuits employing inverting Op-Amp stages
are also used to obtain low-pass transfer functions. The Sallen and
Key circuits in the figure is low-pass biquad section. Real pole can
be realized by using the circuit shown in the figure. The transfer
functions for these circuits are
1
) (
1
2
2
0
+ +
=
s a s a
H
s H
Circuit diagram for 5-pole 1-db Chebyshev low-pass filter
described in Example
) 1 (
) (
1 3
2 2 2 1 1 1 3
1
v
A R R
C R C R C R R
a
+ +
+ +
= where
) 1 (
1 3
2 1 3 2 1
2
v
A R R
C C R R R
a
+ +
=
) 1 (
1 3
3
0
v
V
A R R
R A
H
+ +
=
1
1
1
2
1
2
+
v =
s
RC R
R
V
V
a b v
R R A / =
where
Circuit diagram for 5-pole 1-dB Chebyshev low-pass filter
described in Example
High-Pass Sections : The Sallen and Key circuits in the figure is
easily converted to high-pass sections by interchanging the resistors
and capacitors as displayed. Similarly, interchanging the position of
R and C in the next figure results in circuits having transfer
functions of the form H(s) = A
v
RC
s
/RC
s
+ 1).
Band-Pass Sections :
1
2
) (
V
V
s H =
1
] ) 1 ( ) ( [
) /(
) (
3 1
1 3 1 2 1 2 2 1 3 2
3 1
2 1 3 2 1
2
3 1 2 3 2
+
+
+ + +
+
+
+
=
R R
R R C A R R C R R R C s
R R
C C R R R s
R R s C R R A
s H
V
V
Sallen and Key high-pass sections using (a) positive and
(b) negative feedback.
1 ) (
) (
2 1 1
2
2 1 2 1
1 2
1
2
+ + +

= =
C C sR s C C R R
s C R
V
V
s H
The addition of positive feedback (R
a
and R
b
in the figure) to the
circuit in the figure results in improved performance. The
combination of positive and negative feedback allows practical
fabrication of the higher-Q circuits required in high-order filters. It
can be shown that the transfer function of this circuit is given by
1 )] 1 /( ) ( [
) 1 /(
) (
1 2 2 1 1
2
2 1 2 1
1 2
1
2
+ + +

= =
V
V V
A C R C C R s s C C R R
A s C R A
V
V
s H
a b v
R R A / 1+ =
where
(a) Sallen and Key noninverting band-pass section.
(b) An inverting band-pass circuit.
The Delyiannis band-pass circuit using both positive and
negative feed back.
Band-Reject Sections :
1 ) 2 ( 2
) 1 (
) (
2 2 2
2 2 2
1
2
+ +
+
= =
s A RC s C R
s C R A
V
V
s H
V
V
In the above equation we observe that
o
=
r
= 1/RC and the Q
of the notch is determined by the gain A
V
= 1 + R
b
/R
a
of the Op-
Amp stage. Inspection of the equation also indicates that stability
requires A
V
<2.
(a) A low-pass-high-pass cascade to form a broad-band band-pass circuit. (b)
The frequency-response characteristics of the low-pass and high-pass networks
result in the (c) band-pass response of the cascade.
(a) A parallel-channel band-reject circuit. (b) If
L
<
H
for the high-pass and
low pass characteristics in the figure, the band-reject response is obtained.
A notch response
A twin-tee network used to obtain a notch. If Y = 0, the notch
frequency and pole frequencies are equal. Inclusion of Y { 0
converts the circuit to a high-pass or low-pass notch.
Resistance Simulation :
) ( ) (
2 1 2 1
V V Cf V V
T
C
T
Q
I
s eq
= = =
The current I
eq
is the same that exists in R
eq
in Figure. Thus
) (
2 1
2 1
V V Cf
R
V V
s
eq
=

s
eq
Cf
R
1
=
or
(a) Switched capacitor and (b) its equivalent resistance.
(a) Switched-capacitor integrator and (b) its continuos-time
equivalent.
(a) The circuit in the figure with clocked MOSFETs are used as the
switched. (b) The nonoverlapping two-phase clock waveforms.
(a) An impractical switched-capacitor inverting Op-Amp gain stage
and (b) a practical circuit.
Switched-capacitor realizations of the circuits.
Switched-capacitor realization of the band-pass circuit.