Analog Active Filter Design – Li Tan, Jean Jiang, DeVry University, Atlanta, Georgia 1

Analog Active Filter Design

This topic first studies some simple active operational amplifier circuits, then introduces filter notations
and design specifications. With the filter specifications, the topic shows how to obtain the mathematical
filter transfer function using the Butterworth or Chebyshev function and use MATLAB to verify the
filter frequency responses. The topic develops the circuit topologies such as non-inverting and inverting
topologies. Finally, the topic investigates realizations of different filters such as lowpass, highpass,
bandpass, and bandstop filters using the Sallen-Key and multi-feedback circuits and cascading
realization using second-order filter sections to achieve a higher-order filter. The topic is organized
below:

Outline:

1.1 Operational Amplifiers

1.2 Filter Notations and Specifications
1.3 Lowpass Prototype Approximations
1.4 Analog Lowpass Prototype Transformation
1.5 First-Order Circuit Topology and Filter Circuit Realization
1.6 Second-Order Non-inverting Topology and Sallen-Key Filters
1.7 Second-Order Inverting Topology and Multi-Feedback Filters
1.8 Higher-Order filter by Cascading Filter Sections
Analog Active Filter Design – Li Tan, Jean Jiang, DeVry University, Atlanta, Georgia 2

1.1 Operational Amplifiers

In this topic, analog active filter design is described. We first study a simplified analytic model of the
operational amplifier since it is one of key elements in active filters. Then we determine an analog filter
transfer function based on filter specifications and realize the transfer function using the active circuit.
Specifically, we focus on the Butterworth and Chebyshev filters. The design in general can be concluded
into the following steps. 1. Develop filter notations and specifications; 2. Obtain the transfer function
from the mathematical approximation such as the Butterworth function or Chebyshev function; 3. Select
the circuit topology; 4. Acquire the circuit transfer function; 1. Realize the analog filter by coefficient
matching between the mathematical transfer function and the circuit transfer function; 6. Finally,
perform the circuit simulation, hardware construction, and testing. Figure 1.1.1 shows the design steps.

1. Filter specifications 3. Circuit topology

2. Transfer function 4. Transfer function

from the mathematical from the circuit
approximation H(s) G(s)

5. Realization by
coefficient matching
H(s)=G(s)
Find R's and C's

6. Simulation,
construction, and
testing

Figure 1.1.1 General procedure for analog active filter design.

In this section, let us start with the basic principles of operational amplifiers. For active filter
design, the operational amplifier is one of key elements. It is a directly-coupled amplifier with a high
gain. The operational amplifier has five-terminals, two inputs, one output, and two power supply
terminals as described in Figure 1.1.2a. The power supply terminals are omitted for simplicity of
representation as shown in Figure 1.1.2b. The terminal marked as V − is called the inverting input
terminal while the terminal marked as V + is called the non-inverting input terminal.

+Power supply

V− − V− −
I− Vo I− Vo
V+ + + V+ + +
I+ I+

-Power supply

(a) Five terminals (b) Simplified representation

Figure 1.1.2 Simplified representation for an operational amplifier.
Analog Active Filter Design – Li Tan, Jean Jiang, DeVry University, Atlanta, Georgia 3

We can represent the operational amplifier (Op-Amp) by the following equivalent circuit for
analysis as shown in Figure 1.1.3. The characteristics are listed below:
ri = input impedance in Mega ohms range
ro = output impedance (50 ohms to 100 ohms range)
A = 100000 open loop gain, Vo = A(V + − V − )
operating frequency range: 10 Hz to 1 MHz
Power supply = ± (Vo max + 2 ) volts

V− − ro Rules for analysis:

Vo V + =V −
ri +
A(V + − V − ) +
V+ I+ = I− = 0
−
+

Figure 1.1.3 Analytical model for an operational amplifier.

To allow a simplified analysis, we make the following approximations to get an ideal Op-Amp.
Infinite voltage gain, A = ∞
Infinite input impedance, ri = ∞
Zero output impedance, ro = 0
Infinite bandwidth
From the open loop gain, we know that (V + − V − ) = V0 / A . When applying the first assumption, A = ∞ ,
we see that
(V + − V − ) = V0 / A → 0 .
Hence, we have V + ≈ V − . The potentials on the positive terminal and negative terminal are the same,
thus, the current flowing through ri will be approximately zero. This means that I + ≈ I − ≈ 0 . Therefore,
the governing rules for the Op-Amp analysis are given by
V + =V − (1.1.1)
I+ = I− = 0 (1.1.2)
The basic electrical KCL (Kirchhoff’s current law) is usually applied to solve for the circuit transfer
function.
Analog Active Filter Design – Li Tan, Jean Jiang, DeVry University, Atlanta, Georgia 4

1.1.1 Inverting Amplifier

Let us look at an inverting amplifier shown in Figure 1.1.4, where

Z i = input Laplace impedance

Z f = feedback Laplace impedance

Zf
I2
Vin Zi
I− − Vo
I1 +

Figure 1.1.4 Inverting amplifier.

Based on KCL applied at the negative terminal we have

I1 + I 2 = I − (1.1.3)
− + − +
Applying the operational amplifier rules, it follows that I = I = 0 and V = V = 0 since the
positive terminal is grounded. We have
Vin Vo
+ =0 (1.1.4)
Zi Z f
The transfer function is then solved to be
V Zf
G ( s) = o = − (1.1.5)
Vin Zi
Notice that the transfer function carries a 180 degree phase shift.
Substituting the input impedance and feedback impedance with the real passive elements, we
obtain the basic circuits such as the inverting voltage scaler (phase shifter), integrator, and differentiator
whose transfer functions are given correspondingly.

R2 C
Vin R1 Vin R
− Vo − Vo
+ +

R2 1
(a) Inverting voltage scaler G ( s) = − (b) Integrator G ( s) = −
R1 RCs
R

Vin C
− Vo
+

(c) Differentiator G ( s) = − RCs

Figure 1.1.5 Inverting voltage scaler, integrator, and differentiator using operational amplifiers.
Analog Active Filter Design – Li Tan, Jean Jiang, DeVry University, Atlanta, Georgia 5

Example 1.1.1 For an operational amplifier circuit shown in Figure 1.1.6, determine the transfer
function and identify the circuit type.

0.2 F
Vin 100 Ω
− Vo
+

Figure 1.1.6 Operational amplifier circuit for in Example 1.1.1.

Solution:
We first convert the circuit to the Laplace circuit shown below

5/ s
Vin 100
− Vo
+

Figure 1.1.7 Laplace circuit in Example 1.1.1.

5
Since Z i = 100 and Z f = , applying Equation (1.1.5) yields the transfer function as
s
5
 
G ( s) = −   = −
s 0.05
100 s
This circuit is an integrator.

1.2.2 Non-Inverting Amplifier

Another type of the circuit configuration, is called the non-inverting amplifier, is shown in Figure 1.1.8.

Vin = V + = V −
Vo
+
I− = 0 −
Vin
Zf I
I Zi

Figure 1.1.8 Non-inverting operational amplifier.

Analog Active Filter Design – Li Tan, Jean Jiang, DeVry University, Atlanta, Georgia 6

Based on the analytical rules used for the operational amplifier, the same current I flows
through Z f and Z i . Using Vin = V + = V − and Ohms’ law, it follows that
Vin Vo
= (1.1.6)
Z i Zi + Z f
Hence the transfer function is solved as
V Zi + Z f Zf
G ( s) = o = = 1+ (1.1.7)
Vin Zi Zi
Let Z f = R2 and Z i = R1 as shown in Figure 1.1.9. We obtain a non-inverting voltage amplifier whose
transfer function is
R
G ( s) = 1 + 2 (1.1.8)
R1

Vin Vo
+
−

R2
R1
R2
G ( s) = 1 +
R1

Figure 1.1.9 Non-inverting voltage amplifier.

Example 1.1.2 In an analog signal processing application, we want to amplify the sensor output voltage
by a gain of 16. Design the voltage amplifier.
Solution:
A non-inverting amplifier is suitable to be used since there is no phase shift which exits in the inverting
amplifier. Let
R
G ( s) = 1 + 2 = 16
R1
R2
This requires that = 16 − 1 = 15 . To maintain the ratio requirement, we choose
R1
R1 = 1 k Ω and R2 = 15 k Ω
Figure 1.1.10 shows the designed circuit.

Vin Vo
+
−

15 kΩ
1 kΩ

Figure 1.1.10 Designed voltage amplifier in Example 1.1.2.

Analog Active Filter Design – Li Tan, Jean Jiang, DeVry University, Atlanta, Georgia 7

Next, we look at a complicated circuit below.

Example 1.1.3 Determine the transfer function for the following circuit.

50 Ω

Vin 100 Ω
001
. F Vo
−
+

Figure 1.1.11 Active circuit in Example 1.1.3.

Solution:
Transforming the circuit to the Laplace domain, we have the Laplace circuit below

50

Vin 100 100 / s Vo

−
+

Figure 1.1.12 Laplace circuit in Example 1.1.3.

Since
Z i = 100
 100 
50  
 100   s  = 100
Z f = 50 //   =
 s  50 +  100  s + 2
 
 s 
Applying Equation (1.1.5), we obtain the transfer function as
100
G ( s) = − s + 2 = −
1
100 s+2

The operational amplifiers are common devices used for analog active filter design. Its low
output impedance provides good isolation between stages of a multistage filter and yields each stage
design with minimum effort.
Analog Active Filter Design – Li Tan, Jean Jiang, DeVry University, Atlanta, Georgia 8

1.2 Filter Notations and Specifications

Now let us study the specifications of the ideal filter frequency response for each filter type
shown in Figure 1.2.1.

H ( jω ) H ( jω )
1 1
Ideal Ideal
(a) lowpass (b) highpass
response response
ω ω
ωc ωc

H ( jω ) H ( jω )
1 1
Ideal Ideal
(c)
bandpass (d) bandstop
response response
ω ω
ω L ω0 ω H ω L ω0 ω H

Figure 1.2.1 Ideal filter frequency response. a. Ideal lowpass filter. b. Ideal highpass filter. c.
Ideal bandpass filter. d. Ideal bandstop filter. ωc = cutoff frequency; ω L = low cutoff frequency; ω H =
high cutoff frequency; ω0 = ωLωH =geometric center frequency.

Notice that an ideal filter has a sharp boundary between passband where all the frequencies are
passed and stopband in which all the frequencies are attenuated. But a real filter has a transition between
the two bands, which is controlled by the order of the filter. The higher the filter order, the sharper
transition of the filter is obtained. However, the higher-order analog filter requires complex hardware
components. In practice, the filter order is chosen to meet the minimum design requirement.

1.3 Lowpass Prototype Approximations

There are many mathematical functions which can be used to achieve the transfer functions of different
filters. In this section, we focus on two most popular ones: Butterworth function and Chebyshev
function.

1.3.1 Butterworth Lowpass Prototype Function

The normalized Butterworth magnitude response function H P ( jv ) with an order of n is defined as

1
H P ( jv ) = , n ≥1 (1.3.1a)
1 + ε 2 v2 n

where ν is the normalized frequency, ε the ripple specification controlling the gain fluctuation in the
passband, and ν p the cutoff frequency at 1 radian (passband frequency edge). The normalized
magnitude response function is plotted in Figure 1.3.1a.
Analog Active Filter Design – Li Tan, Jean Jiang, DeVry University, Atlanta, Georgia 9

HP ( jv)
1
1
1+ ε 2

n=1
n=3 n=2 v
0 vp = 1

Figure 1.3.1a Normalized Butterworth magnitude response function.

As shown in Figure 1.3.1a, the passband gain fluctuation is between 1 and 1/ 1 + ε 2 . We

usually specify the ripple in terms of attenuation in dB
 1 
Ap = −20 × log10   dB (1.3.1b)
 2 
 1+ ε 
If we choose ε = 1 , then the normalized Butterworth magnitude response function becomes to be
1
H P ( jv ) = , n ≥1 (1.3.1c)
1 + v2 n
and the corresponding ripple attenuation is
 1 
Ap = −20 × log10   = 3 dB
 2 
 1+1 
The Butterworth prototype function in the factored form (cascade form) with 3-dB ripple specification is
given in Table 1.3.1a. For other ripple specifications, we can use the formulas in Appendix B to
compute the polynomial coefficients.

Table 1.3.1a 3-dB Butterworth lowpass prototype transfer functions in the cascade form
n H P ( s)
1 1
s +1
2 1
s + 2s + 1
2

3 1
( s + 1)( s 2 + s + 1)
4 1
( s + 0.7654s + 1)( s 2 + 1.8478s + 1)
2

5 1
( s + 1)( s + 0.6180s + 1)( s 2 + 1.6180s + 1)
2

6 1
( s + 0.5176s + 1)( s + 1.4142s + 1)( s 2 + 1.9319 s + 1)
2 2
Analog Active Filter Design – Li Tan, Jean Jiang, DeVry University, Atlanta, Georgia 10

H P ( jv)
1
1
1
H P ( jv) =
1+ ε 2
1 + ε 2 v 2n

As
v
0 v p = 1 vs
Figure 1.3.1b Specifications of the Normalized Butterworth magnitude response function.

Besides the passband ripple specification Ap in dB and the normalized passband frequency edge v p in
radians, if the stopband attenuation As in dB and the normalized stop frequency vs in radians are
required as shown in Figure 1.3.1b, the following two equations must be satisfied simultaneously,
 1   1 
Ap = −20 × log10   and As = −20 × log10  
 2   1 + ε 2v 2n 
 1+ ε   s 
Solving for the above two equations achieves the order of the Butterworth prototype function as
 100.1 As − 1 
log10  
 ε2 
ε 2 = 10 p − 1 , n ≥
0.1 A
(1.3.1d)
[ 2 ⋅ log10 (vs )]
where
ε = the absolute ripple specification.
AP = passband attenuation in dB
As = stopband attenuation in dB
vs = normalized stop frequency (radians) obtained using Table 1.3.1b below

Table 1.3.1b Conversion from practical analog filter specifications to lowpass prototype specifications
ωap = passband frequency edge; ωas =stopband frequency edge;
ωapl = lower cutoff frequency in passband; ωaph = upper cutoff frequency in passband;
ωasl = lower cutoff frequency in stopband; ωash = upper cutoff frequency in stopband;
ω0 = geometric center frequency.
Analog Filter Specifications Converted Lowpass Prototype Specifications
Lowpass: ωap , ωas v p = 1 , vs = ωas / ωap
Highpass: ωap , ωas v p = 1 , vs = ωap / ωas
Bandpass: ωapl , ωaph , ωasl , ωash , ωash − ωasl
v p = 1 , vs =
ω0 = ωapl ωaph , ω0 = ωaslωash ωaph − ωapl

Bandstop: ωapl , ωaph , ωasl , ωash , ωaph − ωapl

v p = 1 , vs =
ω0 = ωapl ωaph , ω0 = ωaslωash ωash − ωasl

Frequency edges for the lowpass and bandpass filters defined in Table 1.3.1b are illustrated in Figure
1.3.1c. Similarly, we can define the frequency edges for the highpass filter and bandstop filters.
Analog Active Filter Design – Li Tan, Jean Jiang, DeVry University, Atlanta, Georgia 11

Lowpass filter Bandpass filter ω 0 = ω aplω aph

H LP ( jω ) H BP ( jω ) W = ω aph − ω apl
AP AP
W

As As
0 ω ap ω as ω 0 ω asl ω apl ω 0 ω aph ω ash ω

Figure 1.3.1c Frequency edge specifications for lowpass and bandpass filters.

Example 1.3.1 The filter specifications are given below:

Butterworth type
Filter order = 2
Find the lowpass prototype function.
Solution:
From Table 1.3.1a, we get
1
H P ( s) = 2
s + 1.4142 s + 1

Example 1.3.2 Specifications of an analog filter are given below:

Butterworth lowpass filter
3-dB cutoff frequency at 1000 radians/second
20 dB attenuation at the stop frequency of 20000 radians/second
Determine the lowpass prototype function.
Solution:
The order of the prototype function is not given. Using the passband frequency edge and stopband
frequency edge as
ω p = 1000 radians/second and ωs = 20000 radians/second
we determine the normalized frequency as
vs = ω s / ω p = 20000 /1000 = 20 radians/second
Noting that Ap = 3 dB and As = 20 dB, then the filter order is computed as
ε 2 = 100.1×3 − 1 = 1
 100.1×20 − 1 
log10  
 1  = 0.7669
n≥
[ 2 ⋅ log10 (20) ]
We choose the order of the prototype as n = 1 . From Table 1.3.1a, we get
1
H P ( s) =
s +1
Analog Active Filter Design – Li Tan, Jean Jiang, DeVry University, Atlanta, Georgia 12

1.3.2 Chebyshev Lowpass Prototype Function

The Chebyshev magnitude response function with an order of n and the normalized cutoff frequency
v p = 1 radian (passband frequency edge) is given by
1
H P ( jv ) = , n ≥1 (1.3.2)
1 + ε 2 Cn2 (v)

where the function Cn (v) is defined as

 cos  n cos −1 (v )  v ≤1
  
Cn (v ) =  (1.3.3)
−1
cosh  n cosh (v)  v > 1

ε is the ripple specification on the filter passband and

cosh −1 ( x ) = ln( x + x 2 − 1) (1.3.4)
The Chebyshev magnitude response function is plotted in Figure 1.3.2a, where the magnitude responses
for Chebyshev filters are different in terms of their orders. If the Chebyshev filter to be designed has a
filter order of the odd number, the magnitude response begins with the filter gain of 1. On the other
hand, if the Chebyshev filter has a filter order of the even number, the magnitude starts at the filter gain
of 1/ 1 + ε 2 . The filter gain at the normalized cutoff frequency (passband frequency edge) is
1/ 1 + ε 2 for both cases.

H P ( jv ) H P ( jv )
n (odd number) n (even number)
1 1
1 1
1+ε 2 1+ ε 2

v v
0 vp = 1 0 vp = 1

Figure 1.3.2a Normalized Chebyshev magnitude response function.

The Chebyshev prototype functions in the factored form (cascade form) are given in Table 1.3.2 for 0.5-
dB ripple specification and Table 1.3.3 for 1-dB ripple specification. The functions for other ripple
specifications can be calculated using the derived formulas in Appendix B.
Analog Active Filter Design – Li Tan, Jean Jiang, DeVry University, Atlanta, Georgia 13

Table 1.3.2 Chebyshev lowpass prototype functions in the cascade form with 0.5 dB ripple
( ε = 0.3493 )
n H P ( s) 0.5 dB Ripple ( ε = 0.3493 )
1 2.8628
s + 2.8628
2 1.4314
s + 1.4652s + 1.5162
2

3 0.7157
( s + 0.6265)( s + 0.6265s + 1.1425)
2

4 0.3579
( s + 0.3507 s + 1.0635)( s 2 + 0.8467 s + 0.3564)
2

5 0.1789
( s + 0.3623)( s + 0.2239s + 1.0358)( s 2 + 0.5862s + 0.4768)
2

6 0.0895
( s + 0.1553s + 1.0230)( s + 0.4243s + 0.5900)( s 2 + 0.5796 s + 0.1570)
2 2

Table 1.3.3 Chebyshev lowpass prototype functions in the cascade form with 1 dB ripple ( ε = 0.5088)
n H P ( s) 1 dB Ripple ( ε = 0.5088)
1 1.9652
s + 1.9652
2 0.9826
s + 1.0977 s + 1.1025
2

3 0.4913
( s + 0.4942)( s + 0.4942 s + 0.9942)
2

4 0.2456
( s + 0.2791s + 0.9865)( s 2 + 0.6737 s + 0.2794)
2

5 0.1228
( s + 0.2895)( s + 0.1789s + 0.9883)( s 2 + 0.4684s + 0.4293)
2

6 0.0614
( s + 0.1244s + 0.9907)( s + 0.3398s + 0.5577)( s 2 + 0.4641s + 0.1247)
2 2

H P ( jv) n (odd) n (even)

1
1 H P ( jv) =
1 1 + ε 2Cn2 (v)
1+ ε 2
 cos  n cos−1(v)  v ≤ 1
  
C n (v ) = 
cosh  n cosh −1(v)  v > 1
As
v   
0 vp = 1 vs

Figure 1.3.2b Specifications of the normalized Chebyshev magnitude response function.

Analog Active Filter Design – Li Tan, Jean Jiang, DeVry University, Atlanta, Georgia 14

Figure 1.3.2b shows that the attenuation As in dB at the normalized stop frequency vs in radians is
specified. Similar to the Butterworth prototype function, we can determine the order of the Chebyshev
prototype function as
 −1  
0.5
−1  10
0.1 As
cosh    
 ε 2
 
ε 2 = 10 p − 1 , n ≥
0.1 A  (1.3.5)
cosh −1 ( vs )
where
ε = the absolute ripple specification.
AP = passband attenuation in dB
As = stopband attenuation in dB
vs = normalized stop frequency (radians) obtained using Table 1.3.1b.

Example 1.3.3 The filter specifications are given as follows:

Chebyshev type
Filter order = 2
Passband ripple= 1 dB
Find the lowpass prototype function.
Solution:
From Table 1.3.3, it follows that
0.9826
H P ( s) = 2
s + 1.0977 s + 1.1025

Example 1.3.4 Given the following filter specifications:

Butterworth bandpass filter
10-dB stopband attenuation at 100 and 900 radians/sec
1-dB passband frequency edges at 400 and 600 radians/sec
Determine the prototype function.
Solution:
ωsL = 100 radians/second, ωsH = 900 radians/second
ω pL = 400 radians/second, ω pH = 600 radians/second
ω0 = ω pH ω pL = 489.90 radians/second, ω pH − ω pL = 200 radians/second
We adjust the specifications to satisfy the requirement of the geometric center frequency as follows.
Choosing ωsL = 100 , we get
ω ' sH = ω02 / ωsL = 2400 and ω ' sH − ωsL = 2400 radians/second
Choosing ωsH = 900 , we obatin
ω 'sL = ω02 / ωsH = 266.67 and ωsH − ω ' sL = 633.33 radians/second
Choosing a smaller bandwidth for aggressive rejection, we have
vs = (ωsH − ω ' sL ) /(ωH − ω L ) = 633.33/ 200 = 3.1667 radians
Ap = 1 and As = 10 dB
The filter order is then computed as
ε 2 = 100.1×1 − 1 = 0.2589
Analog Active Filter Design – Li Tan, Jean Jiang, DeVry University, Atlanta, Georgia 15

  100.1×10 − 1 
 0.5

cosh −1   
  0.2589   ln(5.8960 + 5.89602 − 1)
n≥  = = 1.3518
cosh −1 ( 3.1667 ) ln(3.1667 + 3.1667 2 − 1)
We finally select n = 2 . From Table 1.3.3, we obtain the lowpass prototype function as
0.9826
H P ( s) = 2
s + 1.0977 s + 1.1025

1.4 Analog Lowpass Prototype Transformation

In this section, we learn how to acquire an analog filter transfer function using a prototype
transformation. The prototype transformation is a method that converts the analog lowpass filter with a
cutoff frequency of 1 radian per second called the lowpass prototype, or normalized lowpass filter, to
the lowpass, or highpass, or bandpass, or bandstop filter with the filter frequency specifications,
respectively. Here, we describe each transformation without proof.
Let H P ( s) be a transfer function of the analog lowpass prototype with the order of n , the
lowpass prototype to the lowpass filter transformation is described below

H P ( jv ) H LP ( jω )
s
s=
ωc

0 1 v 0 ωc ω
Order n Order n
H LP ( s) = H P ( s) s = s /ω
c

Figure 1.4.1 Analog lowpass prototype transformation to the lowpass filter.

As shown in Figure 1.4.1, H LP ( s ) designates the lowpass filter transfer function with a cutoff
frequency of ωc radians/second. The lowpass prototype to lowpass filter transformation is to substitute
s in the lowpass prototype function H P ( s) by s / ωc . Notice that v is the normalized frequency of the
lowpass prototype (left plot) in Figure 1.4.1 and ωc the cutoff frequency of the lowpass filter to be
designed. Let us consider the first-order lowpass prototype below
1
H P ( s) = (1.4.1)
s +1
Its frequency response is obtained by substituting s = jv to Equation (1.4.1), that is,
1
H P ( jν ) =
jv + 1
with the magnitude gain given in Equation (1.4.2) below:
1
H P ( jν ) = (1.4.2)
1 +ν 2
Analog Active Filter Design – Li Tan, Jean Jiang, DeVry University, Atlanta, Georgia 16

We compute gains at v = 0 , v = 1 , v = 100 , and v = 10000 to obtain 1, 1/ 2 , 0.0995 , and 0.01

respectively. The cutoff frequency gain at v = 1 equals 1/ 2 , which is equivalent to -3 dB, the DC
gain is 1. The gain approaches zero when the frequency goes to v = +∞ . This verifies that the lowpass
prototype is a normalized lowpass filter with a normalized cutoff frequency of 1. After applying the
prototype transformation s = s / ωc in Figure 1.4.1, we get an analog lowpass filter with a cutoff
frequency of ωc as
1 ωc
H ( s) = =
s / ωc + 1 s + ωc
We can obtain the analog frequency response by substituting s = jω to the above equation, that is,
1
H ( jω ) =
jω / ωc + 1
The magnitude response is determined by
1
H ( jω ) = (1.4.3)
2
ω 
1+  
 ωc 
Similarly, we verify gains at ω = 0 , ω = ωc , ω = 100ωc , and ω = 10000ωc to be 1, 1/ 2 , 0.0995 ,
and 0.01 respectively. The filter gain at the cutoff frequency of ωc equals 1/ 2 and the DC gain is 1.
The gain approaches zero when ω = +∞ . We also notice that the shape of the filter gain does not
change but the filter frequency is scaled up by a factor of ωc . This means that the prototype
transformation transforms a lowpass prototype to an analog lowpass filter with the specified cutoff
frequency of ωc without affecting the shape of the filter gain. This first-order prototype function is used
for an illustrative purpose. We can obtain general lowpass prototype functions such as Butterworth and
Chebyshev lowpass prototypes with different orders in Tables 1.3.1 to 1.3.3.
The transformation from the lowpass prototype to the highpass filter H HP ( s) with a cutoff
frequency of ωc radians/second is given by

H P ( jv ) H HP ( jω )
ωc
s=
s

0 1 v 0 ωc ω
Order n Order n
H HP ( s) = H P ( s) s =ω
c /s

Figure 1.4.2 Analog lowpass prototype transformation to the highpass filter.

Next we examine the transformation from the lowpass prototype transfer function to the bandpass filter
with a center frequency of ω0 , lower cutoff frequency of ωl , and upper cutoff frequency of ωh in the
passband depicted in Figure 1.4.3.
Analog Active Filter Design – Li Tan, Jean Jiang, DeVry University, Atlanta, Georgia 17

H P ( jv ) H BP ( jω ) ω 0 = ω lω h
s 2 + ω 20 W = ωh −ωl
s=
sW W

0 1 v 0 ωl ω0 ωh ω
H BP ( s) = H P ( s) s 2 +ω 20
s=
sW

Figure 1.4.3 Analog lowpass prototype transformation to the bandpass filter.

As shown in Figure 1.4.3, ω0 is the geometric center frequency, which is defined as ω0 = ωl ωh while
the passband bandwidth is given by W = ωh − ωl .
Similarly, the lowpass prototype to the bandstop (bandreject) filter transformation is illustrated
in Figure 1.4.4.

H P ( jv) H BS ( jω )

s=
sW ω 0 = ω lω h
s + ω 20
2
W = ωh −ωl

0 1 v 0 ωl ω0 ωh ω
H BS ( s) = H P ( s) sW
s=
s 2 +ω 20

Figure 1.4.4 Analog lowpass prototype transformation to the bandstop filter.

Finally, the prototype transformations are summarized in Table 1.4.1 below:

Table 1.4.1 Analog lowpass prototype transformation

Filter Type Prototype Transformation
Lowpass s
, ωc is the cutoff frequency
ωc
Highpass ωc
, ωc is the cut off frequency
s
Bandpass s 2 + ω02
, ω0 = ωl ωh , W = ωh − ωl
sW
Bandstop sW
, ω0 = ωl ωh , W = ωh − ωl
s + ω02
2

Table 1.4.2 lists MATLAB functions for the lowpass prototype transformations and MATLAB function
bode() for plotting the filter frequency responses for verification.
Analog Active Filter Design – Li Tan, Jean Jiang, DeVry University, Atlanta, Georgia 18

Table 1.4.2 MATLAB functions for the lowpass prototype transformation

>>format short eng
Lowpass to lowpass: H ( s) = H P ( s) s = s
ωc
>>[B,A]=lp2lp(Bp,Ap,wa)
Lowpass to highpass: H ( s) = H P ( s) s = ωc
s
>>[B,A]=lp2hp(Bp,Ap,wa)
Lowpass to bandpass: H ( s) = H P ( s) s = s 2 +ω02
sW
>>[B,A]=lp2bp(Bp,Ap,w0,W)
Lowpass to bandstop: H ( s) = H P ( s) s = sW
s 2 +ω02
>>[B,A]=lp2bs(Bp,Ap,w0,W)
Plot of the magnitude and phase frequency response:
>>bode(B,A);

Definitions of design parameters:

Bp: vector containing the numerator coefficients of the lowpass prototype;
Ap: vector containing the denominator coefficients of the lowpass prototype;
wc: cutoff frequency of the lowpass or highpass analog filter (radians/sec);
w0: center frequency of the bandpass or bandstop analog filter (radians/sec);
W: bandwidth of the bandpass or bandstop analog filter (radians/sec);
B: vector containing the numerator coefficients of the analog filter;
A: vector containing the denominator coefficients of the analog filter.

Example 1.4.1 Determine the transfer function for a first-order lowpass filter with a cutoff frequency of
100 radians per second and passband gain of 1.
a. 3-dB Butterworth type.
b. Chebyshev type with 0.5 dB ripple.
Solution:
a. From Table 1.3.1a, we get the lowpass prototype function for the Butterworth type as
1
H P ( s) =
s +1
Applying lowpass to lowpass transformation with a cutoff frequency of ωc = 100 radians/second yields
1 ωc 100
H ( s) = = =
 s  s + ωc s + 100
  +1
 ωc 
b. Table 1.3.2 gives the first-order Chebyshev lowpass prototype function
2.8628
H P ( s) =
s + 2.8628
2.8628 286.28
Hence H ( s) = ==
 s  s + 286.28
  + 2.8628
 100 
The MATLAB verification for both filter frequency responses are shown in Figure 1.4.1.

>> [B,A]=lp2lp([1],[1 1],100)

Analog Active Filter Design – Li Tan, Jean Jiang, DeVry University, Atlanta, Georgia 19

B=
100.0000e+000
A=
1.0000e+000 100.0000e+000
>> bode(B,A);grid
>> subplot(2,1,1); bode(B,A);grid
>> [B,A]=lp2lp([2.8628],[1 2.8628],100)
B=
286.2800e+000
A=
1.0000e+000 286.2800e+000
>> subplot(2,1,2); bode(B,A);grid

Butterworth filter Bode Diagram

Phase (deg) Magnitude (dB)

-20 -3 dB

-40
0

-45

-90
0 1 2 3 4
10 10 10 10 10
Frequency (rad/sec)

Chebyshev filter Bode Diagram

Phase (deg)Magnitude (dB)

-20 -0.5 dB

-40
0

-45

-90
0 1 2 3 4
10 10 10 10 10
Frequency (rad/sec)

Figure 1.4.5 Frequency responses for Butterworth filter and Chebyshev filter in Example 1.4.1.

Example 1.4.2 Determine the transfer function for a second-order bandpass filter with following
specifications:
3-dB Butterworth type
Lower cutoff frequency = 100 radian/second
Upper cutoff frequency = 144 radians/second
Passband gain =2
Solution:
We choose the first-order prototype function since after lowpass to bandpass transformation, the order
of the resultant transfer function will be doubled.
2
H P ( s) =
s +1
Since W = 144 − 100 = 44 radians/second and ω0 = ωLωH = 100 × 144 = 120 radians/second,
applying lowpass to bandpass transformation, we have
Analog Active Filter Design – Li Tan, Jean Jiang, DeVry University, Atlanta, Georgia 20

2 2 × Ws 2 × 44s
H ( s) = = = 2
 s +ω 
2 2
s + Ws + ω0 s + 44s + 14400
2 2

 +1
0

 sW 

We can verify the filter frequency responses using the MATLAB function as shown in Figure 1.4.6.

>> [B,A]=lp2bp([2],[1 1],120, 44)

B=
88.0000e+000 1.2506e-012
A=
1.0000e+000 44.0000e+000 14.4000e+003
>> bode(B,A);grid

Bode Diagram
10

6 dB
0
Magnitude (dB)

3 dB
-10

-20

-30
90

45
Phase (deg)

-45

-90
1 2 3
10 10 10
Frequency (rad/sec)

Figure 1.4.6 Frequency responses for the bandpass filter in Example 1.4.2.

1.5 First-Order Circuit Topology and Filter Realization

Circuit topology is a form of a generic circuit, where each component is represented by the Laplace
impedance or Laplace admittance which is the reciprocal of the Laplace impedance. A generic transfer
function can be developed based on the topology. By changing arrangement of components, the Laplace
impedance or Laplace admittance would change correspondingly to produce different transfer functions.
Hence, the generic transfer function could generate the transfer function such as the lowpass filter,
highpass, bandpass, and bandstop filter without resolving the Laplace circuit.
Due to the complexity of algebra, we focus on the first-order non-inverting and the first-order
inverting topologies. The second-order circuit topologies are discussed briefly.
Analog Active Filter Design – Li Tan, Jean Jiang, DeVry University, Atlanta, Georgia 21

1.1.1 First-Order Circuit Topology

Let us study the first-order circuit topology in Figure 1.1.1 where Yi is the Laplace admittance is given
by
1
Yi = (1.1.1)
Zi
Using the admittance could help us to apply KCL since the Laplace current passing through the element
Yi is the product of the admittance and voltage across the component; that is,
I i = YV
i i (1.1.2)
We first examine the first-order non-inverting topology as shown in Figure 1.1.1a.

Y2
Vin Y1
V1 Vo
+ Y1
Vin Vo
− −
Y2 V2 +
R2
R1

(a) Non-inverting topology (b) Inverting topology

Figure 1.1.1 First-order non-inverting and inverting topologies.

Applying the KCL at nodes V1 and V2 , respectively, we have

(Vin − V1 )Y1 − V1Y2 = 0 (1.1.3)
1 1
(V0 − V2 ) − V2 =0 (1.1.4)
R2 R1
Based on the analysis rules of the operational amplifier, we have V2 = V1 . Solving Equations (1.1.3) and
(1.1.4) gives
 Y1 
V1 =   ⋅ Vin (1.1.5)
 Y1 + Y2 
 R   R 
V0 =  1 + 2  ⋅ V2 = 1 + 2  ⋅ V1 (1.1.6)
 R1   R1 
Substituting Equation (1.1.5) to Equation (1.1.6), we find the transfer function as

V0  R2  Y1 
G ( s) = = 1 +   (1.1.7)
Vin  R1  Y1 + Y2 

Next, we study the first-order inverting topology shown in Figure 1.1.1b.

Applying the KCL at node V − and noting that V − = V + = 0 , it follows that
VinY1 + V0Y2 = 0 (1.1.8)
The transfer function can be easily found to be
V Y
G ( s) = 0 = − 1 (1.1.9)
Vin Y2
Analog Active Filter Design – Li Tan, Jean Jiang, DeVry University, Atlanta, Georgia 22

1.1.2 Circuit Realization by Coefficient Matching

With the circuit topology, we can develop a desired circuit transfer function by selecting each
admittance component with a resistor or capacitor. By matching the circuit transfer function G ( s) with
the mathematical transfer function H ( s ) ; that is,
H ( s) = G ( s) (1.1.10)
the component values could be determined and hence the circuit design could be completed.

Example 1.1.1 A mathematical transfer function is given below

400
H ( s) =
s + 100
Determine the circuit using the non-inverting topology.
Solution: Let Y1 = 1/ R and Y2 = sC . We have the circuit transfer function as
 1 
 R  
G ( s) = 1 + 2   R 
 R1  1
+ sC 
R 
We put it into the following format
 1 
 R2   RC 
G ( s) = 1 +  
 R1  s + 1 
 
 RC 
Via coefficient matching as follows:
 1 
4 × 100  R2   RC 
= 1 +  
s + 100  R1  s + 1 

 RC 
it requires that
1 R
= 100 and 1 + 2 = 4
RC R1
So we can choose
C = 1 uF, R = 10 k Ω , R1 = 10 k Ω , and R2 = 30 k Ω
Note that there are many other options since the circuit realization is not unique. Figure 1.1.2 shows the
realized circuit.

Vin 10 kΩ
Vo
+
−
1 uF
30 kΩ
10 kΩ

Figure 1.1.2 Realized circuit in Example 1.1.1.

Analog Active Filter Design – Li Tan, Jean Jiang, DeVry University, Atlanta, Georgia 23

We next examine a simple inverting amplifier realization.

Example 1.1.2 A transfer function is shown below

100
H ( s) =
s
Determine the circuit using the inverting topology.
Solution:
Let Y1 = 1/ R and Y2 = sC . We have
1
Y1 1/ R
G ( s) = − = − = − RC
Y2 sC s
Coefficient matching leads
1
100 RC
=
s s
which requires that
1
= 100
RC
Thus, we choose C = 1 uF and find R = 10 k Ω .
Figure 1.1.3 shows the realized circuit.

1 uF

10 k Ω
Vin Vo
−
+

Figure 1.1.3 Realized circuit in Example 1.1.2.

Analog Active Filter Design – Li Tan, Jean Jiang, DeVry University, Atlanta, Georgia 24

1.1.3 Impedance Scaling

As we have seen in Section 1.1.2, the circuit realization is not unique and its solution is yielded by
selecting the capacitor value(s) and solving for the resistor value(s) or vice versa. For many situations,
we may end up with impractical element values. For example, we could get an impractical value for
R = 2 Ω in a circuit realization. To accommodate feasible value(s), we could raise the resistor value
and inductor value by a factor while reduce the capacitor value by the same factor without affecting the
filter frequency response, in which the shape of frequency response and the cutoff frequency remain
unchanged. We refer the procedure to accomplish this as impedance scaling. To develop the impedance
scaling procedure, let us verify the following two circuits. One is an original circuit while the other is its
impedance scaled circuit.

R L k⋅R k⋅L
V0 V0

Vi ~ C Vi ~ C /k

Original Network G ( s ) Impedance scaled Network G ' ( s )

Figure 1.1.4 Circuit impedance scaling.

Using the voltage divider rule, we can easily obtain the transfer function for the first circuit on the left
hand side as
(1/(sC ) ) 1
G ( s) = = 2 (1.1.10)
R + sL + 1/( sC ) s LC + sRC + 1
Now, let us investigate the second circuit on the right hand side, where the resistor and inductor are
scaled up by a factor of k (scale factor), respectively, while the capacitor is scaled down by the same
factor value. Solving for the transfer function results in
1/ ( sC / k )
G ' ( s) = (1.1.11)
kR + ( skL ) + 1/ ( sC / k )
Simplifying Equation (1.1.11) leads to
1/ 1
G ' ( s) = = 2 (1.1.12)
( sC / k ) kR + ( sC / k )( skL ) + ( sC / k ) ×1/ ( sC / k ) s LC + sRC + 1
We verify that Equations (1.1.10) and (1.1.12) achieve the same transfer function. The illustration can
be extended to a general case for passive and active circuits. Hence, we summarize the impedance
scaling in Figure 1.1.5.
Analog Active Filter Design – Li Tan, Jean Jiang, DeVry University, Atlanta, Georgia 25

Impedance scaling
Original value
by a factor of k
R k⋅R

L k⋅L

C C /k

Figure 1.1.5 Impedance scaling.

Example 1.1.3 In Example 1.1.1, perform impedance scaling to scale the resistor R by a scale factor of
k = 10 . Note that R1 and R2 are unrelated to scaling and they are determined based on the passband
 R 
gain  1 + 2  .
 R1 
Solution: From Figure 1.1.5, we can determine the resistor value as
R = kR = 10 × 10 k Ω = 100 k Ω
and the required capacitor value as
C = C / k = 1 uF /10 = 0.1 uF
The realized circuit is shown in Figure 1.1.6.

Vin 100 k Ω
Vo
+
−
0.1 uF
30 k Ω
10 k Ω

Figure 1.1.6 The realized circuit in Example 1.1.3.

Analog Active Filter Design – Li Tan, Jean Jiang, DeVry University, Atlanta, Georgia 26

1.6 Second-Order Non-inverting Topology and Sallen-Key Filters

The first-order topology often cannot satisfy the practical design requirement due to its limited filter
types (lowpass filter and highpass filter only) and poor rolloff rate approximately being -20 dB/decade.
The second-order topology could be employed for performance improvement. The second-order non-
inverting topology called the Sallen-Key filter is given in Figure 1.6.1.

Y2

Vin Y1
Vo
+
Y3 −
Y5 Y4
R4
R3

Sallen-Key topology

Figure 1.6.1 Second-order non-inverting topology.

Although the derivation of the generic circuit transfer function is omitted here, Equation (1.6.1) gives its
generic transfer function.

 R  Y1Y3
G ( s) = 1 + 4  (1.6.1)
 R3    R 
(Y1 + Y2 + Y5 )(Y3 + Y4 ) + Y3 Y4 − Y2 1 + 4  
  R3  
1. Lowpass filter
By selecting the following components, we obtain the Sallen-Key second-order lowpass filter as shown
in Figure 1.6.2.

Y1 = 1/ R1 , Y2 = sC2 , Y3 = 1/ R2 , Y4 = sC1 , and Y5 = 0 (open branch)

C2

Vin
R1 R2 Vo
+
−
C1
R4
R3

Figure 1.6.2 Second-order Sallen-Key lowpass filter.

Analog Active Filter Design – Li Tan, Jean Jiang, DeVry University, Atlanta, Georgia 27

The circuit transfer function is found as

Vout Gb0
G ( s) = = 2 (1.6.2)
Vin s + b1 s + b0

R4
G =1+ = passband gain (1.6.3)
R3
1
b0 = (1.6.4)
R1 R2 C1C2
1 1 R4
b1 = + − (1.6.5)
R1C2 R2 C2 R2 R3C1

where b0 , b1 , and G in Equations (1.6.2) to (1.6.5) can be identified as the coefficients from the
mathematical transfer function H ( s ) acquired in Section 1.4.
There are many design solutions for the above set of equations since there are six unknown
circuit elements to be solved based on three equations, that is, Equations (1.6.3) to (1.6.4). Here we
provide a simple solution from coefficient matching between the circuit transfer function G ( s) and
mathematical transfer function H ( s ) , that is,
Gb0
G ( s) = 2 = H (s) (1.6.6)
s + b1s + b0
Assuming that C1 and C2 are given, and R1 = R2 , matching coefficients leads the following design
equations.

1
R1 = R2 = (1.6.7)
C1C2 b0
R4 2C1
= − b1C1 R2 (1.6.8)
R3 C2

2. Highpass filter
After selecting the following components in Figure 1.6.1, we then obtain the Sallen-Key second-order
highpass filter as shown in Figure 1.6.3.
Y1 = sC1 , Y2 = 1/ R2 , Y3 = sC2 , Y4 = 1/ R2 , and Y5 = 0
R1

Vin C1 C2
Vo
+
−
R2
R4
R3

Figure 1.6.3 Second-order Sallen-Key highpass filter.

Analog Active Filter Design – Li Tan, Jean Jiang, DeVry University, Atlanta, Georgia 28

The circuit transfer function is given by

Vout Gs 2
G ( s) = = 2 (1.6.9)
Vin s + b1 s + b0

R4
G =1+ = passband gain (1.6.10)
R3
1
b0 = (1.6.11)
R1 R2 C1C2
1 1 R4
b1 = + − (1.6.12)
R2 C1 R2 C2 R1 R3C1

Gs 2
Assuming R1 = R2 and matching G ( s) = = H ( s ) , we get the following solution
s 2 + b1s + b0

1
R1 = R2 = (1.6.13)
C1C2 b0
R4 C
= 1 + 1 − b1C1 R1 (1.6.14)
R3 C2

3. Bandpass filter
The Sallen-Key second-order bandpass filter shown in Figure 1.6.4 is achieved by choosing the
following components:

Y1 = 1/ R1 , Y2 = 1/ R2 , Y3 = sC1 , Y4 = 1/ R0 , and Y5 = sC2

R2

Vin R1 C1
Vo
+
−
C2 R0
R4
R3

Figure 1.6.4 Second-order Sallen-Key bandpass filter.

Analog Active Filter Design – Li Tan, Jean Jiang, DeVry University, Atlanta, Georgia 29

The circuit transfer function is given by

 R4  1 
1 +   b1 s
Vout  R3   R1C1b1 
G ( s) = = (1.6.15)
Vin s 2 + b1 s + b0

 R  1 
G =  1 + 4   = passband gain
 R3   R1C1b1 
(1.6.16)
1  1 1 
b0 =  +  (1.6.17)
R0 C1C2  R1 R2 
1 1 1 R4
b1 = + + − (1.6.18)
R1C2 R0 C2 R0C1 R2 R3C2

Gb1 s
After matching G ( s) = = H ( s ) with R0 = R1 = R2 , it follows that
s + b1s + b0
2

2
R0 = R1 = R2 = (1.6.19)
C1C2b0
R4 C
= 2 + 2 − b1C2 R2 (1.6.20)
R3 C1

Let us practice several design examples.

Example 1.6.1 Determine the transfer function for a second-order lowpass filter with a cutoff frequency
of 3400 Hz and passband gain of H 0 . Realize the Butterworth filter and Chebyshev filter, respectively.
a. 3-dB Butterworth type.
b. Chebyshev type with 1 dB ripple.
Solution:
1. Let us develop the transfer functions for the Butterworth and Chebyshev filters, respectively.
a. From Table 1.3.1a, the Butterworth filter prototype is found to be
H0
H P ( s) = 2
s + 1.4142 s + 1
Since ωc = 2π fc = 2π × 3400 = 6800π radians/second, applying the lowpass prototype transformation
yields
H0 H 0ωc2
H ( s) = =
 s 
2
 s  s 2 + 1.4142ωc s + ωc2
  + 1.4142   + 1
 ωc   ωc 
Substituting ωc = 6800π , we get the Butterworth filter transfer function as
H 0 4.5637 × 108
H ( s) =
s 2 + 30211.31s + 4.5637 × 108
Analog Active Filter Design – Li Tan, Jean Jiang, DeVry University, Atlanta, Georgia 30

b. From Table 1.3.2, the Chebyshev filter prototype is determined as

H 0 0.9826
H P ( s) = 2
s + 1.0977 s + 1.1025
Since ωc = 2π fc = 2π × 3400 = 6800π radians/second, applying the lowpass prototype transformation
yields
H 0 0.9826 H 0 0.9826ωc2
H ( s) = =
 s 
2
 s  s 2 + 1.0977ωc s + 1.1025ωc2
  + 1.0977   + 1.1025
 ωc   ωc 
Substituting ωc = 6800π , we obtain the Chebyshev filter transfer function as
H 0 4.4843 × 108
H ( s) =
s 2 + 23449.98s + 5.03148 × 108

2. We then perform their circuit realizations, respectively. The realization of the Butterworth filter is
shown below:
H 0 4.5637 × 108
a. Since H ( s) = 2 ,
s + 30211.31s + 4.5637 × 108
we identify that b0 = 4.5637 × 108 , b1 = 30207.0425 .
We select C1 = C2 = 0.01 uF, then
1
R1 = R2 = = 4681 Ω
0.01 × 10 × 0.01 × 10−6 × 4.5637 × 106
−6

R4 2 × 0.01 × 10−6
= −6
− 30211.31 × 0.01 × 10−6 × 4681 = 0.586
R3 0.01 × 10
Selecting R3 = 10000 Ω , we have R4 = 5860 Ω
R
The DC gain is G = H 0 = 1 + 4 = 1 + 0.586 = 1.586
R3
The frequency responses using MATLAB are plotted in Figure 1.6.5a while the circuit is shown in
Figure 1.6.5b.
>> [B,A]=lp2lp([1],[1 1.4142 1], 2*pi*3400)
B=
456.3705e+006
A=
1.0000e+000 30.2113e+003 456.3705e+006
>> bode(1.586*B,A);grid
Analog Active Filter Design – Li Tan, Jean Jiang, DeVry University, Atlanta, Georgia 31

Bode Diagram
20

Magnitude (dB)
-20

-40

-60
0

-45
Phase (deg)

-90

-135

-180
0 1 2 3 4 5
10 10 10 10 10 10
Frequency (Hz)

Figure 1.6.5a Frequency response plot of the Butterworth lowpass filter in Example 1.6.1a.

0.01 uF
Vin
4.68 k Ω 4.68 k Ω Vo
+
−
0.01 uF
5.86 k Ω
10 k Ω

Figure 1.6.5b Circuit of the Butterworth lowpass filter in Example 1.6.1a.

b. Next, we realize the Chebyshev filter. Since

H 0 4.4843 × 108
H ( s) = 2
s + 23449.98s + 5.03148 × 108
we could identify that b0 = 5.03148 × 108 , b1 = 23449.98 , and H 0 4.4843 × 108 = G ⋅ 5.03148 × 108
Similarly, we select C1 = C2 = 0.01 uF, it follows that
1
R1 = R2 = = 4458 Ω
0.01× 10 × 0.01× 10−6 × 5.03148 × 108
−6

R4 2 × 0.01 × 10−6
= −6
− 23449.98 × 0.01× 10−6 × 4458 = 0.9546
R3 0.01 × 10
Selecting R3 = 10000 Ω , we have R4 = 9546 Ω
R
The DC gain is G = 1 + 4 = 1 + 0.9546 = 1.9546
R3
Analog Active Filter Design – Li Tan, Jean Jiang, DeVry University, Atlanta, Georgia 32

Matching the coefficient in the numerator, it yields

H 0 4.4843 × 108 = 1.9546 × 5.03148 × 108
H 0 = 2.1931
Figure 1.6.6a shows the frequency response plot using MATLAB while Figure 1.6.6b is the designed
circuit.

[B,A]=lp2lp([0.9826],[1 1.0977 1.1025], 2*pi*3400)

B=
448.4297e+006
A=
1.0000e+000 23.4500e+003 503.1485e+006
>> bode(2.1931*B,A);grid

Bode Diagram
20

0
Magnitude (dB)

-20

-40

-60
0

-45
Phase (deg)

-90

-135

-180
0 1 2 3 4 5
10 10 10 10 10 10
Frequency (Hz)

Figure 1.6.6a Frequency response plot of the Chebyshev lowpass filter in Example 1.6.1b.

0.01 uF
Vin
4.458 k Ω 4.458 k Ω Vo
+
−
0.01 uF
9.546 k Ω
10 k Ω

Figure 1.6.6b Circuit of the Chebyshev lowpass filter in Example 1.6.1b.

Analog Active Filter Design – Li Tan, Jean Jiang, DeVry University, Atlanta, Georgia 33

Example 1.6.2 The filter design specifications are given below:

Cutoff frequency =3400 Hz
Butterworth highpass filter
50 dB attenuation at 100 Hz in the stopband
Determine the transfer function and realize it using the Sallen-Key circuit topology.
Solution:
From specifications, we determine
ωc = 2π f c = 2π × 3400 radians/second
ωs = 2π f s = 2π × 100 radians/second
vs = ωc / ωs = 6800π /(200π ) = 34 radians
Ap = 3 dB and As = 50 dB
The filter order is computed as
ε 2 = 100.1×3 − 1 = 1
log10 (100.1×50 − 1)
n= = 1.94
2 ⋅ log10 (34)
We choose n = 2 . To develop the filter transfer function, the second-order lowpass prototype is chosen
as
H0
H P ( s) = 2
s + 1.4142 s + 1
Applying the lowpass to highpass transformation, it follows that
ωc
s=
s
H0 H0s2
and H ( s) = =
 ωc 
2
 ωc  s 2 + 1.4142ωc s + ωc2
  + 1.4142   + 1
 s   s 
Substituting ωc = 2π fc = 2π × 3400 = 21362.83 gives the highpass filter transfer function as
H 0 s2
H ( s) =
s 2 + 30211.31s + 4.5637 × 108
To design the circuit, selecting C1 = C2 = 0.01 uF yields
1
R1 = R2 = = 4681 Ω
C1C2 b0
R4 C
= 1 + 1 − b1C1 R1 = 0.586
R3 C2
R
G = H 0 = 1 + 4 = 1.586
R3
We choose R3 = 10 k Ω and R4 = 5.86 k Ω

The following MATLAB codes are used to produce the frequency responses in Figure 1.6.7a. Figure
1.6.7b shows the circuit realization.

[B,A]=lp2hp([1],[1 1.4142 1],2*pi*3400)

B=
1.0000e+000 251.1919e-015 -21.3337e-009
Analog Active Filter Design – Li Tan, Jean Jiang, DeVry University, Atlanta, Georgia 34

A=
1.0000e+000 30.2113e+003 456.3705
>> bode(1.586*B,A);grid

Bode Diagram
50

0
Magnitude (dB)

-50

-100

-150
180

135
Phase (deg)

90

45

0
0 1 2 3 4 5
10 10 10 10 10 10
Frequency (Hz)

Figure 1.6.7a Frequency responses for the highpass filter in Example 1.6.2.

4.681 k Ω
Vin 0.01 uF 0.01 uF Vo
+
−
4.681 k Ω
5.86 k Ω
10 k Ω

Figure 1.6.7b Circuit of the highpass filter in Example 1.6.2.

Example 1.6.3 Design a Butterworth bandpass filter for the following specifications:
Bandwidth: 2000 radians/se
Center frequency: 7746 radians/sec
10-dB attenuation at 2746 radians/sec and 12746 radians/sec
Determine the transfer function and realize it using the Sallen-Key circuit topology.
Solution:
W = 2000 radians/sec, ω0 = 7746 radians/sec
ωsL = 2746 radians/sec and ωsH = 12746 radians/sec
To meet the requirement of the geometric center frequency, we adjust the specifications as follows.
Choosing ωsL = 2746 , we achieve
Analog Active Filter Design – Li Tan, Jean Jiang, DeVry University, Atlanta, Georgia 35

ω ' sH = ω02 / ωsL = (7746)2 /(2746) = 21850 and Ws = ω sH

'
− ωsL = 19104
Choosing ωsH = 12746 , it follows that
ω ' sL = ω02 / ωsH = (7746) 2 /(12746) = 4707 and Ws = ω sH − ω ' sL = 8039
Selecting a smaller passband bandwidth for aggressive rejection, we have
ω ' sL = 4707 and ωsH = 12746
Next, we can determine
vs = (ωsH − ω ' sL ) / W = 8039 / 2000 = 4.0195 radians
Ap = 3 dB and As = 20 dB
The filter order is computed as
ε 2 = 100.1×3 − 1 = 1
log10 (100.1×10 − 1)
n= = 0.79
2 ⋅ log10 (4.0195)
We select n = 1 .
The first-order low pass prototype is used as
H
H P ( s) = 0
s +1
Applying the lowpass prototype transformation leads
H0 H Ws
H ( s) = 2 = 2 0
s + ω0 2
s + Ws + ω02
+1
sW
Substituting the numerical values yields the transfer function as
H 0 × 2000s H 0 × 2000s
H ( s) = 2 = 2
s + 2000s + 7746 2
s + 2000s + 6 × 107
The calculations for circuit realization are listed below
C1 = C2 = 0.01 uF
2
R0 = R1 = R2 = = 18.26 k Ω
C1C2 b0
R4 C
= 2 + 2 − b1C2 R2 = 2.635
R3 C1
 R  1  1
G =  1 + 4   = 3.6348 × = 9.9529
 R3   R1C1b1  18260 × 0.01× 10−6 × 2000
We choose R3 = 10 k Ω and R4 = 26.35 k Ω
Figure 1.6.8a shows the filter frequency responses using the following MATLAB codes. The circuit is
shown in Figure 1.6.8b.

>> [B,A]=lp2bp([1],[1 1],7746, 2000)

B=
2.0000e+003 0.0000e-003
A=
1.0000e+000 2.0000e+003 60.0005e+006
>> bode(9.9529*B,A);grid
Analog Active Filter Design – Li Tan, Jean Jiang, DeVry University, Atlanta, Georgia 36

Bode Diagram
20

10

Magnitude (dB)
0

-10

-20
90

45
Phase (deg)

-45

-90
2 3 4
10 10 10
Frequency (Hz)

Figure 1.6.8a Frequency responses for the bandpass filter in Example 1.6.3.

18.26 k Ω
Vin 18.26 k Ω 0.01 uF Vo
+
−
0.01 uF
18.26 k Ω 26.35 k Ω
10 k Ω

Figure 1.6.8b Frequency responses for the bandpass filter in Example 1.6.3.

4. The unit gain Sallen-Key lowpass filter

We now develop a Sallen-Key lowpass filter with a unit gain, which is applied to attenuate the high
frequency components while maintaining the passband gain to be 1. In order to satisfy such a
requirement, we need further to solve the lowpass Sallen-Key filter, which is described in Figure 1.6.2.
If we open resistor R3 ; that is, R3 = ∞ then the unit gain can be achieved since
R R
G =1+ 4 = 1+ 4 =1
R3 ∞
The resultant circuit is therefore simplified below:
Analog Active Filter Design – Li Tan, Jean Jiang, DeVry University, Atlanta, Georgia 37

C2

Vin
R1 R2 Vo
+
−
C1
R4

Figure 1.6.9 Second-order unit gain Sallen-Key lowpass filter.

The circuit transfer function then becomes

Gb0
G ( s) = (1.6.21)
s + b1 s + b0
2

G = 1 = passband gain (1.6.22)

1
b0 = (1.6.23)
R1 R2 C1C2
1 1
b1 = + (1.6.24)
R1C2 R2 C2
Assuming R1 = R2 , applying coefficient matching leads the following solution

2
R1 = R2 = (1.6.25)
C2b1
1
C1 = (1.6.26)
R1 R2C2b0

Example 1.6.4 Realize the Butterworth filter in Example 1.6.1 with a unit gain
Solution:
4.5637 × 108
Since H ( s) = 2
s + 30211.31s + 4.5637 × 108
Selecting C2 = 0.01 uF, we use Equations (1.6.25) and (1.6.26) to get
2
R1 = R2 = −6
= 6621 Ω
0.01 × 10 × 30207.0425
1
C1 = = 0.005 uF
6621 × 6621 × 0.01× 10−6 × 4.5637 × 108
We can choose R4 = 0 k Ω .

The frequency responses are generated by MATLAB codes, which are listed below. Figure 1.6.10a
shows the frequency response plots and Figure 1.6.10b shows the realized circuit.
>> [B,A]=lp2lp([1],[1 1.4142 1], 2*pi*3400)
B=
456.3705e+006
Analog Active Filter Design – Li Tan, Jean Jiang, DeVry University, Atlanta, Georgia 38

A=
1.0000e+000 30.2113e+003 456.3705e+006
>> bode(B,A);grid

Bode Diagram
0

Magnitude (dB) -10

-20

-30

-40

-50

-60
0

-45
Phase (deg)

-90

-135

-180
0 1 2 3 4 5
10 10 10 10 10 10
Frequency (Hz)

Figure 1.6.10a Frequency response plot of the unit gain filter in Example 1.6.4.

0.01 uF
Vin
6621 Ω 6621 Ω Vo
+
−
0.005 uF

Figure 1.6.10b Realized circuit of the unit gain filter in Example 1.6.4.

4 Gain Constant Adjustment

In the previous discussion, we design active Sallen-Key filters by matching the coefficients of circuit
transfer function to that of filter mathematical transfer function, leaving the gain constant floating. The
resultant gain can be adjusted by using an inverting amplifier or non-inverting amplifier. It is costive
and redundant to do so. By scaling the input voltage level or feedback voltage level, we could achieve
realization without introducing an additional operational amplifier. There are two situations to be
discussed below:
Analog Active Filter Design – Li Tan, Jean Jiang, DeVry University, Atlanta, Georgia 39

A. Gain attenuation.
Assuming that the output voltage of an active filter is larger than that of the desired one, we could apply
input attenuation as depicted in Figure 1.6.11. Considering the original input network with an input
voltage vin and the potential divider network with an input voltage α vin , where α < 1 is an attenuation
factor.
R R1

+ + α vin
vin vin R2
R
− −

a. Original network b. Potential divider network

Figure 1.6.11 Gain attenuation using the input attenuation.

Using the voltage divider rule and the condition of the output resistance equal to that of the original
output resistance, we have
R2 RR
α= and R = 1 2 (1.6.14)
R1 + R2 R1 + R2
Solving for R1 and R2 leads to
R R
R1 = and R2 = (1.6.15)
α 1−α
We illustrate the application in the following example.

Example 1.6.5 Continue to realize the circuit in Example 1.6.1 to obtain a unit gain without introducing
an additional operation amplifier.
Solution:
Since the DC gain of the lowpass filter is G = 1.586 , we determine the attenuation factor as
α = 1/ G = 0.631
Using the first resistor R = 4.68 k Ω , we compute
4.68 4.68
R1 = = 7.42 k Ω and R2 = = 12.68 k Ω
0.631 1 − 0.631
The final modified realization is shown in Figure 1.6.12.

0.01 uF
Vin
7.42 k Ω 4.68 k Ω Vo
+
−
0.01 uF
12.68 k Ω
5.86 k Ω
10 k Ω

Figure 1.6.12 Circuit realization with a DC gain equal to 1 in Example 1.6.1.

Analog Active Filter Design – Li Tan, Jean Jiang, DeVry University, Atlanta, Georgia 40

B. Gain amplification.
Assuming that the output voltage of an active filter is smaller than that of the desired one, we could
achieve gain amplification using the attenuated filter output as feedback. Considering the following
modification shown in Figure 1.6.13, where Vβ is the feedback voltage, which is a fraction of the output
V0 = β Vβ (1.6.16)
where β is the amplification factor, that is, β > 1 . We assume that Rb is small compared to the
impedance of the passive terminal looking into the left of Rb . Thus, the branch containing Ra and Rb
dominates the current delivered by the output of the operational amplifier. The amplification factor β
can be determined by using the voltage divider rule
R + Rb R
β≈ a =1+ a (1.6.17)
Rb Rb
Clearly, choosing a smaller Rb gives a better approximation.

Vin
RC Network + Vo
−
Ra
Vβ
Rb

Figure 1.6.13 Gain amplification.

Example 1.6.6 Continue to realize the circuit in Example 1.6.1 to have a DC gain of 5 without
introducing an additional operation amplifier.
Solution:
Since the original DC gain is G = 1.586 , we can determine the amplification factor as
β = 5 /1.586 = 3.153
From (1.6.17), it follows that
Ra
= 3.153 − 1 = 2.153
Rb
Choosing Rb = 200 Ω (very small as compared to 10 k Ω + 5.86 k Ω ), we then yield
Ra = 431 Ω
The modified realization is shown in Figure 1.6.14
Analog Active Filter Design – Li Tan, Jean Jiang, DeVry University, Atlanta, Georgia 41

0.01 uF
Vin
4.68 k Ω 4.68 k Ω
+ Vo
−
431 Ω
0.01 uF
5.86 k Ω
10 k Ω
200 Ω

Figure 1.6.14 Circuit realization with a DC gain of 5 in Example 1.6.6.

1.7 Second-Order Inverting Topology and Multi-Feedback Filters

In this section, we continue to study another second-order inverting topology.

Y4 Y5
Vin Y1
− Vo
Y3 +
Y2

Figure 1.7.1 Second-order inverting topology (Multiple-feedback filter)

From the Laplace circuit analysis, the generic transfer function is given by

Vo ( s ) −Y1Y3
G ( s) = = (1.7.1)
Vin ( s ) Y5 (Y1 + Y2 + Y3 + Y4 ) + Y3Y4

By selecting different components, we are able to get the circuit filter transfer function. Then by
matching the mathematical transfer function H ( s ) to the circuit transfer function G ( s) , we can obtain
the component values for resistors and capacitors in the circuits. The results for the lowpass filter,
highpass filter, and bandpass filter are listed below:

1. Lowpass filter circuit and design formulas:

By setting the following circuit elements, we get the MFB lowpass filter as
1 1 1
Y1 = , Y2 = sC1 , Y3 = , Y4 = , Y5 = sC2
R1 R3 R2
Analog Active Filter Design – Li Tan, Jean Jiang, DeVry University, Atlanta, Georgia 42

R2 C2
Vin R1
− Vo
R3 +
C1

Figure 1.7.2 MFB lowpass topology.

Gb0
Matching G ( s) = − = − H ( s ) leads the following solution
s + b1s + b0
2

4C b (1 + G )
C1 = 2 0 2 (1.7.2)
b1
b
R2 = 1 (1.7.3)
2b0 C2
R
R1 = 2 (1.7.4)
G
R2
R3 = (1.7.5)
(1 + G )

2. Highpass filter circuit and design formulas:

The MFB highpass filter is obtained by setting the following circuit elements:
1 1
Y1 = sC1 , Y2 = , Y3 = sC3 , Y4 = sC2 , Y5 =
R2 R1

C2 R1
Vin C1
− Vo
C3 +
R2

Figure 1.7.3 MFB highpass topology.

Gs 2
Matching G ( s) = − = − H ( s ) , we have
s 2 + b1s + b0
Analog Active Filter Design – Li Tan, Jean Jiang, DeVry University, Atlanta, Georgia 43

C1 = C3 (1.7.6)
C
C2 = 1 (1.7.7)
G
( 2G + 1)
R1 = (1.7.8)
b1C1
b1
R2 = (1.7.9)
 1
b0C1  2 + 
 G

3. Bandpass filter circuit and design formulas:

By setting the following circuit elements:
1 1 1
Y1 = , Y2 = , Y3 = sC , Y4 = sC , Y5 =
R1 R3 R2
we obtain the MFB bandpass filter shown in Figure 1.7.4.

C R2
Vin R1
− Vo
C +
R3

Figure 1.7.4 MFB bandpass topology.

Gb1 s
We develop the design equations by matching G ( s ) = − = − H ( s ) as
s + b1s + b0
2

1
R1 = (1.7.10)
Gb1C
R2 = 2 R1G (1.7.11)
b1
R3 =
( 2b )
(1.7.12)
0 − Gb12 C

4. Bandstop filter circuit and design formula:

The bandstop filter is designed based on the following principles. 1. Design the corresponding bandpass
filter with a passband gain of G1 (usually, choose a unit gain; that is, G1 = 1 ). 2. Design a subtractor to
subtract the bandpass filter output from its input to convert a bandpass filter to a bandstop filter. Figure
1.7.5 shows the circuit realization, where the bandpass filter has a passband gain of G1 while the final
notch filter is designed to have a final passband gain of G .
Analog Active Filter Design – Li Tan, Jean Jiang, DeVry University, Atlanta, Georgia 44

R5 Bandstop filter with a passband gain G RF

C R2
Vin R1
VBPF
− V0
−
C +
R3
R4 +
G RF
=
G1 R4
R
Bandpass filter with a passband gain G1 G= F
R5

Figure 1.7.5 MFB bandstop topology.

By matching the circuit transfer function G ( s) with the mathematical transfer function H ( s ) as shown
below
 RF RF
G ( s) = −  −
G1b1s  (
G s 2 + b0 )
 = − 2 = − H (s)
 R5 R4 s + b1s + b0  s + b1s + b0
2

we finally have the design solution as

1
R1 = (1.7.13)
G1b1C
R2 = 2G1 R1 (1.7.14)
b1
R3 =
( 2b )
(1.7.15)
0 − G1b12 C
RF = GR5 (1.7.16)
= control of the passband gain
R4 = G1 R5 (1.7.17)

Now, let us work on the following design examples.

Example 1.7.1 Design the analog filter with following specifications:

Passband gain: 1
Cutoff frequency: 3400 Hz
Second-order Butterworth highpass filter
MFB circuit topology
Solution:
From Example 1.6.2 and with H 0 = 1 , the transfer function is given as
s2
H ( s) =
s 2 + 30211.31s + 4.5637 × 108
The circuit realization via the formulas from the coefficient matching solution is carried out as follows:
b0 = 4.5637 × 108
b0 = 30211.31
G =1
Analog Active Filter Design – Li Tan, Jean Jiang, DeVry University, Atlanta, Georgia 45

Select C1 = C3 = 0.01 uF, we get

C1
C2 = = 0.01 uF
G
( 2G + 1) 2 ×1 + 1
R1 = = = 9930 Ω
b1C1 30211.31× 0.01 × 10−6
b1 30211.31
R2 = = = 2214 Ω
 1  4.5637 × 10 × 0.01× 10−6 ( 2 + 1/1)
8
b0C1  2 + 
 G

The frequency responses are generated by the MATLAB codes listed below. Figure 1.7.6a and Figure
1.7.6b show the frequency response plot and designed circuit.

>> [B,A]=lp2hp([1],[1 1.4142 1],2*pi*3400)

B=
1.0000e+000 -6.3130e-012 46.9866e-009
A=
1.0000e+000 30.2113e+003 456.3705e+006
>> bode(B,A);grid

Bode Diagram
0
Magnitude (dB)

-50

-100

-150
180

135
Phase (deg)

90

45

0
0 1 2 3 4 5
10 10 10 10 10 10
Frequency (Hz)

Figure 1.7.6a Frequency response plot of the highpass filter in Example 1.7.1.
Analog Active Filter Design – Li Tan, Jean Jiang, DeVry University, Atlanta, Georgia 46

0.01 uF 9930 Ω
Vin 0.01 uF
− Vo
0.01 uF +
2214 Ω

Figure 1.7.6b Designed circuit of the highpass filter in Example 1.7.1.

Example 1.7.2 Design a second-order Butterworth bandpass filter with the following specifications:
Bandwidth: 2000 radians/second
Passband gain: 5
Center frequency: 7746 radians/second
MFB circuit topology
Solution:
With the required passband gain of H 0 = 5 , we follow Example 1.6.3 to find the transfer function for
the bandpass filter as
5 × 2000 s 5 × 2000 s
H ( s) = 2 = 2
s + 2000s + 7746 2
s + 2000s + 6 × 107
After matching coefficients with the MFB bandpass filter topology, we have
G = 5 , b1 = 2000 , and b0 = 6 × 107
Applying the design formulas, we finally get
C = 0.01 uF
1 1
R1 = = = 10 k Ω
Gb1C 5 × 2000 × 0.01 × 10−6
R2 = 2 R1G = 2 × 10000 × 5 = 100 K Ω
2000
R3 = = 2 kΩ
( )
2 × 6 × 10 − 20002 × 5 × 0.01 × 10−6
7

The following shows MATLAB verification for the frequency responses. The frequency responses and
designed circuit are shown in Figure 1.7.7a and Figure 1.7.7b, respectively.

>> [B,A]=lp2bp([5],[1 1], 7746, 2000)

B=
10.0000e+003 0.0000e-003
A=
1.0000e+000 2.0000e+003 60.0005e+006
>> bode(B,A);grid
Analog Active Filter Design – Li Tan, Jean Jiang, DeVry University, Atlanta, Georgia 47

Bode Diagram
20

10

Magnitude (dB)
0

-10

-20
90

45
Phase (deg)

-45

-90
2 3 4
10 10 10
Frequency (Hz)

Figure 1.7.7a Frequency response plot of the bandpass filter in Example 1.7.2.

0.01 uF 100 k Ω
Vin 10 k Ω
− Vo
0.01 uF +
2 kΩ

Figure 1.7.7b Designed circuit of the bandpass filter in Example 1.7.2.

The following example demonstrates the design of a bandstop (notch) filter.

Example 1.7.3 To cancel a 1500 Hz tone interference, a notch filter is required with following
specifications:
Passband gain: 2
Center frequency: 1500 Hz
Bandwidth: 150 Hz
Butterworth notch filter
16-dB attenuation at frequency edges of 1490 Hz and 1510 Hz, respectively.
MFB circuit topology
Design the notch filter.
Solution:
W = 2 × π × 150 = 300π radians/sec, ω0 = 2π × 500 = 3000π radians/sec
ωsL = 2π × 1490 = 2980π radians/sec, ωsH = 2π × 1510 = 3020π radians/sec
To satisfy the requirement of the geometric center frequency, we adjust the frequency specifications
below.
Analog Active Filter Design – Li Tan, Jean Jiang, DeVry University, Atlanta, Georgia 48

Choosing ωsL = 2980π , we have

ω ' sH = ω02 / ωsL = (3000π ) 2 /(2980π ) = 3020.13π and Ws = ω sH
'
− ωsL = 40.13π
Choosing ωsH = 3020π , we get
ω ' sL = ω02 / ωsH = (3000π ) 2 /(3020π ) = 2980.13π and Ws = ω sH − ω 'sL = 39.87π
For a larger rejection bandwidth (aggressive frequency rejection), we choose ωsL = 2980π and
ω ' sH = 3020.13π to determine
vs = (ω H − ω L ) /(ω ' sH − ωsL ) / = 300π /(40.13π ) = 7.48 radians
Ap = 3 and As = 16 dB
The filter order is computed as
ε 2 = 100.1×3 − 1 = 1
log10 (100.1×16 − 1)
n= = 0.91
2 ⋅ log10 (7.48)
We select n = 1 .
Then
2
H P ( s) =
s +1
2 2( s 2 + ω02 )
H ( s) = = 2
 sW  s + Ws + ω02
 2 2 
+1
 s + ω0 
Substituting W = 300π radians/sec and ω0 = 3000π radians/sec, we have
2( s 2 + 8.8826 × 107 )
H ( s) =
s 2 + 942.48s + 8.8826 × 107
Then we identify b0 = 8.8826 × 107 and b1 = 942.48
Applying design formulas for the unit gain bandpass filter, that is, G1 = 1 , we have
C = 0.02 uF
1
R1 = = 53.05 k Ω
942.48 × 0.02 × 10−6
R2 = 2 R1 = 106.10 k Ω
942.48
R3 = = 266.55 Ω
( )
2 × 8.8826 × 10 − 942.482 × 0.02 × 10−6
7

Selecting R5 = 10 k Ω and adding the subtractor and passband gain G = 2 , we have

RF = GR5 = 2 × 10 k Ω = 20 k Ω
R4 = G1 R5 = 1× 10 k Ω = 10 k Ω
MATLAB verification for the frequency responses is given in Figure 1.7.8a. Figure 1.7.8b shows the
designed circuit.

>> [B,A]=lp2bs([2],[1 1], 3000*pi, 300*pi)

B=
2.0000e+000 0 1.7765e+008
A=
1.0000e+000 9.4248e+002 8.8826e+007
>> bode(B,A);grid
Analog Active Filter Design – Li Tan, Jean Jiang, DeVry University, Atlanta, Georgia 49

Bode Diagram
100

Magnitude (dB) -100

-200

-300
450

405
Phase (deg)

360

315

270
2 3 4
10 10 10
Frequency (Hz)

Figure 1.78a Frequency response plot of the notch filter in Example 1.7.3.

10 k Ω 20 k Ω

0.02 uF 106.10 k Ω
Vin 53.05 k Ω
− 10 k Ω
V0
−
0.02 uF +
266.55 Ω +

Figure 1.7.8b Frequency response plot of the notch filter in Example 1.7.3.
Analog Active Filter Design – Li Tan, Jean Jiang, DeVry University, Atlanta, Georgia 50

1.8 Higher-Order Filter by Cascading Filter Sections

In some applications, the higher-order filter is required. The higher-order filter can actually be broken
into cascaded sections where each section could be the first-order or second-order filter. In fact, each
cascaded section can be designed independently using the method and formulas as we discussed before.
The following example illustrates the procedure.

Example 1.8.1 Customer filter requirements are given below:

4th-order lowpass filter Butterworth filter
Cutoff frequency at 3 dB: 2000 Hz
Passband gain: 5
MFB circuit topology
Design the filter.
Solution:
ωc = 2π × 2000 = 4000π radians/second
The lowpass prototype is the 4th-order and its cascaded lowpass prototype (see Table 1.3.1) is given by
5
H P ( s) = 2
( s + 0.7654 s + 1)( s 2 + 1.8478s + 1)
After lowpass to lowpass prototype transformation, we have
5
H ( s) =
 s  2
 s    s 
2
 s  
  + 0.7654   + 1   + 1.8478   + 1
 ωc   ωc    ωc   ωc  
and
 1.58 × 108  5 × 1.58 × 108 
H ( s) =  2  8 
= H1 ( s ) H 2 ( s )
 s + 9618.30s + 1.58 × 10  s + 23220.14 s + 1.58 × 10 
8 2

We can break down the transfer function to two second-order sections. The design for each section is
carried out below.
Stage 1 realization: G = 1 , C2 = 0.01 uF
1.58 × 108
H1 ( s ) =
s 2 + 9618.30 s + 1.58 × 108
we apply the lowpass filter matching formulas to determine
4C b (1 + G ) 4 × 0.01uF × 1.58 × 108 × (1 + 1)
C1 = 2 0 2 = = 0.1366 uF
b1 9618.302
b 9618.30
R2 = 1 = = 3045 Ω
2b0 C2 2 × 1.58 × 108 × 0.01 × 10−6
R 3045
R1 = 2 = = 3045 Ω
G 1
R2 3045
R3 = = = 1523 Ω
(1 + G ) (1 + 1)
Stage 2 realization: G = 5 , C2 = 0.01 uF
5 × 1.58 × 108
H 2 ( s) =
s 2 + 23220.14s + 1.58 × 108
Similarly, after applying the lowpass filter matching formulas, we have
C1 = 0.07 uF, R2 = 7348 Ω , R1 = 1470 Ω , R3 = 1225 Ω
Analog Active Filter Design – Li Tan, Jean Jiang, DeVry University, Atlanta, Georgia 51

Finally, the cascaded two-stage circuit is shown in Figure 1.8.1

3045 Ω
0.01 uF 7348 Ω
Vin 3045 Ω 0.01 uF
1470 Ω
− Vo
−
1523 Ω +
01336
. uF 1225 Ω +
0.07 uF

Figure 1.8.1 4th-order Butterworth filter circuit in Example 1.8.1.

The MATLAB verification of the frequency responses is shown below

>> [B1,A1]=lp2lp([1],[1 0.7654 1], 2000*2*pi)

B1 =
157.9137e+006
A1 =
1.0000e+000 9.6183e+003 157.9137e+006
>> [B2,A2]=lp2lp([5],[1 1.8478 1], 2000*2*pi)
B2 =
789.5684e+006
A2 =
1.0000e+000 23.2201e+003 157.9137e+006
>> bode(conv(B1,B2),conv(A1,A2));grid

Bode Diagram
50

0
Magnitude (dB)

-50

-100

-150
0

-90
Phase (deg)

-180

-270

-360
0 1 2 3 4 5
10 10 10 10 10 10
Frequency (Hz)

Figure 1.8.2 Frequency response plot of the 4th-order lowpass filter in Example 1.8.1.
Analog Active Filter Design – Li Tan, Jean Jiang, DeVry University, Atlanta, Georgia 52

1.9 Summary

1. The active filter design includes the following steps: developing filter specifications, determining the
mathematical transfer function, selecting the circuit topology, obtaining the circuit transfer function by
choosing each circuit element in the selected topology, and realizing the circuit by solving values for
resistor (s) and capacitor(s).
2. The operational amplifier rules can be applied to analyze active operational amplifier circuits with
electrical laws such as Ohm’s law and KCL.
3. The topic reviews the filter design specifications such as lowpass, highpass, bandpass, and bandstop
filters.
4. The lowpass prototype transformation transforms the Butterworth or Chebyshev function to different
type transfer functions, such as lowpass, highpass, bandpass, and bandstop filters. The MATALB
function bode() can be used to effectively verify the filter frequency responses.
1. The circuit topology is a generic circuit. The non-inverting topology offers no phase shift in the
generic transfer function, while the inverting topology produces a 180-degree phase shift in the generic
transfer function.
6. The lowpass, highpass, bandpass, and bandstop filters can be realized by the Sallen-Key filters and
multi-feedback filters.
7. Finally, cascading second-order filter sections achieves a higher-order filter.

1.10 Problems

1. Given the following circuit, determine its transfer function and circuit type.

0.01 F
Vin 1000 Ω
− Vo
+

Figure P1

2. Given the following circuit, determine its transfer function and circuit type.

Vin Vo
+
−

10 kΩ
1 kΩ

Figure P2
Analog Active Filter Design – Li Tan, Jean Jiang, DeVry University, Atlanta, Georgia 53

3. Given the following circuit, determine its transfer function and circuit type.
100 Ω
0.2 F
Vin
− Vo
+

Figure P3

4. Determine the transfer function for the following circuit.

100 Ω

Vin 50 Ω
0.01 F Vo
−
+

Figure P4

5. Determine the transfer function for a first-order lowpass filter with a cutoff frequency of 1000 radians
per second and passband gain of 1 using the following filter types:
a. 3-dB Butterworth type.
b. Chebyshev type with 1-dB ripples on the passband
Use MATLAB to verify their frequency responses.

6. Determine the transfer function for a first-order high filter with a cutoff frequency of 1000 radians per
second and passband gain of 1 using the following types:
a. 3-dB Butterworth type.
b. Chebyshev type with 1 dB ripples on the passband
Use MATLAB to verify their frequency responses.

7. Determine the transfer function for a second-order bandpass filter with following specifications:
Chebyshev type with 1-dB ripples on the passband
Lower cutoff frequency = 100 radian/second
Upper cutoff frequency = 169 radians/second
Passband gain =1
Use MATLAB to verify their frequency responses.

8. Determine the transfer function for a second-order bandstop filter with following specifications:
3-dB Butterworth type
Lower cutoff frequency = 100 radian/second
Upper cutoff frequency = 169 radians/second
Passband gain =1
Use MATLAB to verify their frequency responses.

9. A filter transfer function is given below:

1000
H ( s) =
s + 1000
Determine the circuit using the non-inverting topology.
Analog Active Filter Design – Li Tan, Jean Jiang, DeVry University, Atlanta, Georgia 54

10. A differentiator is given below

H ( s) = −0.01s
Determine the circuit by using the inverting topology.

11. Determine the transfer function of the second-order (2 poles) lowpass Butterworth filter with a 3-dB
cutoff frequency of 5 kHz. Design it using the Sallen-Key lowpass filter configuration, find all the
values of the circuit elements, and draw the circuit. What is the DC gain?

12. A filter has the following design specifications:

Cutoff frequency 1500 Hz
Second-order Butterworth highpass filter
Sallen-Key circuit topology
Determine the transfer function and realize it using the Sallen-Key circuit topology.

13. Design a second-order Butterworth bandpass filter with the following specifications:
Bandwidth: 400 Hz
Center frequency: 2000 Hz
Determine the transfer function and realize it using the Sallen-Key circuit topology.

14. Realize the Butterworth filter in Problem 11 using a unit gain.

15. From the specifications in Problem 11, realize the transfer function with a DC gain of 1 using the
Multiple Feedback filter configuration.

16. Determine the transfer function of a second-order (2 poles) highpass Chebyshev filter with a 1-dB
ripples on the passband, cutoff frequency of 5 kHz, and the passband gain=1. Design it using the
Multiple Feedback filter configuration, find all the values of the circuit elements, and draw the circuit.

17. Determine the transfer function of a second-order (2 poles) bandpass Butterworth filter with a center
frequency of 5 kHz, passband gain=1, and a 3-dB bandwidth of 400 Hz. Design it using the Multiple
Feedback filter configuration, find all the values of the circuit components, and draw the circuit.

18. Determine the transfer function of a second-order (2 poles) Notch Butterworth filter with a center
frequency of 5 kHz, passband gain=1, and a bandwidth of 400 Hz. Design it using the Multiple
Feedback filter configuration, find all the values of the circuit components, and draw the circuit.

19. Determine the transfer function of a 4th-order (4 poles) lowpass Butterworth filter with a 3-dB
cutoff frequency of 3.4 kHz and passband gain=1. Design the circuit using the MFB topology (hint:
design two sections each with an order of 2).

20. Specifications of an analog filter are given below:

Chebyshev lowpass filter
Cutoff frequency at 500 radians/second
Passband ripple = 0.5 dB
Absolute DC gain =1
40 dB attenuation at the stop frequency of 2500 radians/second
Determine the lowpass prototype function.

21. Specifications of an analog filter are given below:

Butterworth highpass filter
Absolute passband gain =1
3-dB cutoff frequency at 4000 radians/second
Analog Active Filter Design – Li Tan, Jean Jiang, DeVry University, Atlanta, Georgia 55

40 dB attenuation at the stop frequency of 800 radians/second

Determine the lowpass prototype function.

22. The filter specifications are given below:

Chebyshev highpass filter
Cutoff frequency = 1500 Hz
Passband ripple = 0.5 dB
20 dB attenuation at the stop frequency of 500 Hz
Determine the order of the lowpass prototype function.

23. A Butterworth bandstop filter has the following specifications:

Bandwidth: 800 radians/sec
Center frequency: 10000 radians/sec
20-dB attenuation of the passband gains at 8000 and 12000 radians/sec
Determine the order of the lowpass prototype function.

24. A bandpass filter is required with following specifications:

Center frequency: 1500 Hz
Bandwidth: 200 Hz
Chebyshev filter with a passband ripple of 1 dB
20-dB attenuation of the stopband gains at 1200 Hz and 1800 Hz
Determine the order of the lowpass prototype function.

25. For the designed circuit in Problem 11, adjust the DC gain to be a unit DC gain.

26. For the designed circuit in Problem 13, adjust the passband gain to be a unit gain.

27. For the designed circuit in Problem 11, adjust the DC gain to be 10.

28. For the designed circuit in Problem 12, adjust the passband gain to be 1.

29. For the designed circuit in Problem 13, adjust the passband gain to be 10.
Analog Active Filter Design – Li Tan, Jean Jiang, DeVry University, Atlanta, Georgia 56

ANSWERS TO SELECTED PROBLEMS

1.
0.1
G ( s) = − integrator
s
3.
G ( s) = −20 s , differentiator
4.
2
G ( s) = −
s +1
5.
1000
a. G ( s) =
s + 1000

6.
s
b. G ( s) =
s + 508.85

7.
135.60 s
G ( s) =
s + 135.6s + 16900
2

9. Hint:
Vin 10 k Ω
Vo
+
−
01
. uF

11. Hint:
0.01 uF
Vin
3.18 k Ω 3.18 k Ω Vo
+
−
0.01 uF
5.86 k Ω
10 k Ω

DC gain =1.586
Analog Active Filter Design – Li Tan, Jean Jiang, DeVry University, Atlanta, Georgia 57

13. Hint:

11.25 k Ω
Vin 11.25 k Ω 0.01 uF Vo
+
−
0.01 uF
11.25 k Ω 27.17 k Ω
10 k Ω

Passband gain = 13.16

15. Hint:

2.25 k Ω 0.01 µF
Vin 2.25 k Ω
− Vo
1125 Ω +
0.04 uF

17. Hint:

0.01 uF 79.58 k Ω
Vin 39.79 k Ω
− Vo

127.73 Ω 0.01 uF +

19. Hint:

1791 Ω 0.01 uF
Vin 1791 Ω 4325 Ω 0.01 uF
4325 Ω
− Vo
896 Ω −
01366
. uF + 2162 Ω +
0.0234 uF
Analog Active Filter Design – Li Tan, Jean Jiang, DeVry University, Atlanta, Georgia 58

21.
s3
H ( s) =
( s + 4000)( s 2 + 4000 s + 16000000)

22. Hint:
The order of the lowpass prototype: n = 3

23. Hint:
The order of the lowpass prototype: n = 2

26. Hint:

11.25 k Ω
Vin
148.05 k Ω 0.01 uF Vo
+
−
0.01 uF
12.18 k Ω
11.25 k Ω 27.17 k Ω
10 k Ω

27. Hint:
0.01 uF
Vin
3.18 k Ω 3.18 k Ω
+ Vo
−
531 Ω
0.01 uF
5.86 k Ω
10 k Ω
100 Ω