Professional Documents
Culture Documents
Active Filters Theory and Design. Pactitis S. A
Active Filters Theory and Design. Pactitis S. A
FILTERS
Theory and Design
\
ACTIVE
FILTERS
Theory and Design
S. A. PACTITIS
\
CRC Press
Taylor & Francis Group
6000 Broken Sound Parkway NW, Suite 300
Boca Raton, FL 33487-2742
This book contains information obtained from authentic and highly regarded sources. Reasonable efforts
have been made to publish reliable data and information, but the author and publisher cannot assume
responsibility for the validity of all materials or the consequences of their use. The authors and publishers
have attempted to trace the copyright holders of all material reproduced in this publication and apologize to
copyright holders if permission to publish in this form has not been obtained. If any copyright material has
not been acknowledged please write and let us know so we may rectify in any future reprint.
Except as permitted under U.S. Copyright Law, no part of this book may be reprinted, reproduced, transmit-
ted, or utilized in any form by any electronic, mechanical, or other means, now known or hereafter invented,
including photocopying, microfilming, and recording, or in any information storage or retrieval system,
without written permission from the publishers.
For permission to photocopy or use material electronically from this work, please access www.copyright.
com (http://www.copyright.com/) or contact the Copyright Clearance Center, Inc. (CCC), 222 Rosewood
Drive, Danvers, MA 01923, 978-750-8400. CCC is a not-for-profit organization that provides licenses and
registration for a variety of users. For organizations that have been granted a photocopy license by the CCC,
a separate system of payment has been arranged.
Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and are used
only for identification and explanation without intent to infringe.
Visit the Taylor & Francis Web site at
http://www.taylorandfrancis.com
Chapter 2
Sallen–Key Filters....................................................................................................21
2.1 Introduction ....................................................................................................21
2.2 Frequency Response Normalization ..............................................................21
2.3 First-Order Low-Pass Filter ...........................................................................22
2.3.1 Frequency Response ..........................................................................23
2.4 First-Order High-Pass Filter ..........................................................................25
2.4.1 Frequency Response ..........................................................................26
2.5 Second-Order Filters ......................................................................................28
2.6 Low-Pass Filters.............................................................................................31
2.6.1 Frequency Response ..........................................................................32
2.6.2 Design Procedure ...............................................................................34
2.7 High-Pass Filters ............................................................................................44
2.8 Higher-Order Filters.......................................................................................50
2.9 Wide-Band Filters ..........................................................................................60
2.10 Wide-Band Band-Reject Filters .....................................................................64
2.11 Comments on VCVS Filters ..........................................................................68
2.11.1 Low-Pass Filters ...............................................................................68
2.11.2 High-Pass Filters ..............................................................................69
Problems ..................................................................................................................70
\
Chapter 3
MultiFeedback Filters..............................................................................................73
3.1 Low-Pass Filters.............................................................................................73
3.2 High-Pass Filters ............................................................................................78
3.3 Higher-Order Filters.......................................................................................83
3.4 Band-Pass Filters............................................................................................90
3.4.1 Narrow-Band Band-Pass Filter ..........................................................92
3.4.1.1 Design Procedure................................................................94
3.4.1.2 Frequency Response ...........................................................95
3.4.2 Narrow-Band Band-Pass Filter with Two Op-Amps ......................102
3.4.2.1 Design Procedure..............................................................105
3.4.3 Deliyannis’s Band-Pass Filter ..........................................................107
3.4.3.1 Design Procedure..............................................................111
3.5 Band-Reject Filters ......................................................................................114
3.5.1 Wide-Band Band-Reject Filters .......................................................114
3.5.2 Narrow-Band Band-Reject Filter.....................................................119
3.5.2.1 Design Procedure..............................................................121
3.5.3 MFB Narrow-Band Band-Reject Filter ...........................................123
3.6 Comments on MFB Filters ..........................................................................126
3.6.1 Low-Pass Filters...............................................................................126
3.6.2 High-Pass Filters ..............................................................................127
3.6.3 Band-Pass Filters..............................................................................127
Problems ................................................................................................................127
Chapter 4
Filters with Three Op-Amps .................................................................................129
4.1 State-Variable Filter .....................................................................................129
4.1.1 Low-Pass Filter ................................................................................130
4.1.1.1 Design Procedure..............................................................131
4.1.2 High-Pass Filter................................................................................137
4.1.2.1 Design Procedure..............................................................138
4.1.3 Narrow-Band Band-Pass Filter ........................................................149
4.1.3.1 Design Procedure..............................................................150
4.2 Biquad Filters...............................................................................................154
4.2.1 Narrow-Band Band-Pass Filter ........................................................155
4.2.1.1 Design Procedure..............................................................157
4.2.2 Low-Pass Filter ................................................................................159
4.2.2.1 Design Procedure..............................................................161
Problems ................................................................................................................166
Chapter 5
Sensitivity ..............................................................................................................169
5.1 Introduction ..................................................................................................169
5.2 Some General Properties .............................................................................171
5.3 Magnitude and Phase Sensitivities ..............................................................174
5.4 Root Sensitivity............................................................................................176
Problems ................................................................................................................180
Chapter 6
Filters with GIC.....................................................................................................183
6.1 Introduction ..................................................................................................183
6.2 Generalized Impedance Converters .............................................................183
6.3 Low-Pass Filter Design................................................................................187
6.4 High-Pass Filter Design ...............................................................................191
6.5 Narrow-Band Band-Pass Filter Design .......................................................194
6.6 Narrow-Band Band-Reject Filter Design ....................................................197
Problems ................................................................................................................200
Chapter 7
OTA Filters ............................................................................................................203
7.1 Introduction ..................................................................................................203
7.2 Single OTA LP Filters with Three Passive Components ............................204
7.2.1 First-Order Low-Pass Filter .............................................................205
7.2.2 First-Order High-Pass Filter ............................................................206
7.3 Second-Order Low-Pass Filter.....................................................................207
7.4 Second-Order LP Filter with Four Passive Components ............................210
7.5 Second-Order Band-Pass Filter ...................................................................213
7.6 OTA-C Filter ................................................................................................216
7.7 Some nonIdeal Feature of the OTA.............................................................219
Problems ................................................................................................................220
Chapter 8
Switched Capacitor Filters ..................................................................... 225
8.1 Introduction ..................................................................................................225
8.2 The Switched Capacitor Resistors...............................................................225
8.3 The Switched Capacitor Integrator..............................................................226
8.4 Universal SC Filters.....................................................................................228
8.4.1 The LMF100 Universal SC Filter....................................................228
8.4.1.1 Modes of Operation..........................................................229
8.4.1.2 Low-Pass Filter .................................................................230
8.4.1.3 High-Pass Filter ................................................................232
8.4.1.4 Narrow-Band Band-Pass Filter.........................................235
8.5 Practical Limitations of SC Filters ..............................................................240
Problems ................................................................................................................240
Appendix A
Node Voltage Network Analysis ...........................................................................247
Appendix B
Filter Design Nomograph......................................................................................251
\
Appendix C
First- and Second-Order Factors of Denominator Polynomial.............................253
Appendix D
Formulas of Normalized Filters ............................................................................257
Appendix E
Element Values for Low-Pass LC Filters..............................................................261
Appendix F
Coefficients of Denominator Polynomial..............................................................265
Bibliography .........................................................................................................269
Index......................................................................................................................271
Preface
This book was primarily written to provide readers with a simplified approach to
the design of active filters. Filters of some sort are essential to the operation of most
electronic circuits. It is therefore in the interest of anyone involved in electronic
circuit design to have the ability to design filters capable of meeting a given set of
specifications.
Three basic active filter types are used throughout the book: Butterworth,
Chebyshev, and Bessel. Those three types of filters are implemented with the
Sallen–Key, infinite gain multiple feedback, state-variable, and Biquad circuits that
yield low-pass, high-pass, band-pass, and band-reject circuits.
Many examples of low-pass, high-pass, band-pass, and notch active filters are
illustrated in complete detail, including frequency normalizing and denormalizing
techniques.
It is felt that this book can be used for the following purposes:
S. A. Pactitis
\
1 Introduction
1.1 FILTERS AND SIGNALS
A filter is a circuit that is designed to pass a specified band of frequencies while
attenuating all signals outside this band. Filter networks may be either active or
passive. Passive filter networks contain only resistors, inductors, and capacitors.
Active filters, which are the only type covered in this text, employ operational
amplifiers (op-amps) as well as resistors and capacitors.
The output from most biological measuring systems is generally separable into signal
and noise. The signal is that part of the data in which the observer is interested; the rest
may be considered noise. This noise includes unwanted biological data and nonbiological
interference picked up by or generated in the measuring equipment. Ideally, we would
like to remove it while retaining the signal, and often this is possible by suitable filtration.
If the spectra of signal and noise occupy completely separate frequency ranges,
then a filter may be used to suppress the noise (Figure 1.1).
As filters are defined by their frequency-domain effects on signals, it makes
sense that the most useful analytical and graphical descriptions of filters also fall
under the frequency domain. Thus, curves of gain versus frequency and phase versus
frequency are commonly used to illustrate filter characteristics, and most widely
used mathematical tools are based on the frequency domain.
The frequency-domain behavior of a filter is described mathematically in terms
of its transfer function or network function. This is the ratio of the Laplace
transforms of its output and input signals. The voltage transfer function of a filter
can therefore be written as
V0 (s)
H (s ) = (1.1)
Vi (s)
1
2 Active Filters: Theory and Design
A A
Filter
Vi Vo
dB
dB
f1 f2 f f1 f2 f
V0 ( jω )
H ( jω ) = (1.2)
Vi ( jω )
or
A = 20 log H ( jω ) in dB (1.3)
V0 ( jω )
arg H ( jω ) = arg (1.4)
Vi ( jω )
1. The first type is the low-pass filter (LPF). As might be expected, an LPF
passes low-frequency signals, and rejects signals at frequencies above the
filter’s cutoff frequency (Figure 1.2.). The ideal filter has a rectangular shape,
indicating that the boundary between the passband and the stopband is abrupt
and that the rolloff slope is infinitely steep. This type of response is ideal
because it allows us to completely separate signals at different frequencies
from one another. Unfortunately, such an amplitude response curve is not
physically realizable. We will have to settle for the approximation that will
still meet our requirements for a given application. Deciding on the best
Introduction 3
A
dB
Ao
Amax
Amax Amin
Ideal Filter
Amin
f1 fs f
Passband Stopband
Transition Region
f0 = f1 f2 (1.5)
Ao
Amax
Amax
dB Amin
Ideal Filter
Amin
fs f2 f
Passband
For narrow filters, where the ratio of f2 to f1 is less than 1.1, the response
shape approaches arithmetic symmetry. fo can then be computed as the
average of the cutoff frequencies:
f1 + f2
f0 = (1.6)
2
Ao
Amax Pass-band
Amax
Amin
dB
Ideal Filter
Amin Stopband
fs1 f1 f2 fs2 f
Passband Stopband
Stopband Transition Region
Ideal Filter
Ao
Amax
Amax
dB Amin
Pass-band
Amin
f1 fo f2 f
Passband Passband
Band
Reject
Transition
Region
f0 f0
Q= = (1.7)
BW f2 − f1
N (s )
H (s ) = (1.8)
D(s)
Kb0
H (s ) = (1.9)
s + bn−1s
n n −1 + bn− 2s n− 2 + b1s + b0
6 Active Filters: Theory and Design
TABLE 1.1
Butterworth filters: s n + bn−1s n−1 … b1s + b0
n b0 b1 b2 b3 b4 b5 b6 b7
1 1.000
2 1.000 1.414
3 1.000 2.000 2.000
4 1.000 2.613 3.414 2.613
5 1.000 3.236 5.236 5.236 3.236
6 1.000 3.863 7.464 9.142 7.464 3.864
7 1.000 4.494 10.098 14.592 14.592 10.098 4.494
8 1.000 5.126 13.137 21.846 25.688 21.846 13.137 5.126
where b0 , b1 ,… bn−1 and K are calculated depending on the order of the transfer
function n. The Butterworth, Chebyshev, and Bessel filter circuits differ only by the
choice of the coefficients bi, which yield slightly different response curves. The
coefficients for the normalized Butterworth, Chebyshev, and Bessel cases are given
in Tables 1.1 to 1.7.
Another way of writing a filter’s transfer function is to factor the polynomials
in the denominator so that they take the form:
Kb0
H (s ) = (1.10)
(s − p0 )(s − p1 ) … (s − pn )
The roots of the denominator, p0 , p1 , … , pn, are called poles. All of the poles will
be either real roots or complex conjugate pairs.
Another way to arrange the terms in the network function expression is to
recognize that each complex conjugate pair is simply the factored form of a second-
order polynomial. By multiplying the complex conjugate pairs out, we can get rid
of the complex numbers and put the transfer function into a form that essentially
TABLE 1.2
0.1-dB Chebyshev filter
n b0 b1 b2 b3 b4 b5 b6 b7
1 6.552
2 3.313 2.372
3 1.638 2.630 1.939
4 0.829 2.026 2.627 1.804
5 0.410 1.436 2.397 2.771 1.744
6 0.207 0.902 2.048 2.779 2.996 1.712
7 0.102 0.562 1.483 2.705 3.169 3.184 1.693
8 0.052 0.326 1.067 2.159 3.419 3.565 3.413 1.681
Introduction 7
TABLE 1.3
0.5-dB Chebyshev filter
n b0 b1 b2 b3 b4 b5 b6 b7
1 2.863
2 1.516 1.426
3 0.716 1.535 1.253
4 0.379 1.025 1.717 1.197
5 0.179 0.753 1.310 1.937 1.172
6 0.095 0.432 1.172 1.590 2.172 1.159
7 0.045 0.282 0.756 1.648 1.869 2.413 1.151
8 0.024 0.153 0.574 1.149 2.184 2.149 2.657 1.146
K
H (s ) = (1.11)
(s 2 + a11s + a10 )(s 2 + a21s + a20 ) … (s 2 + an1s + an 0 )
K
H ( jω ) = 1/ 2
(1.12)
⎡ ⎛ s ⎞ 2n ⎤
⎢1 + ⎜ ⎟ ⎥
⎢ ⎝ ω1 ⎠ ⎥
⎣ ⎦
TABLE 1.4
1-dB Chebyshev filter
n b0 b1 b2 b3 b4 b5 b6 b7
1 1.965
2 1.103 1.098
3 0.491 1.238 0.988
4 0.276 0.743 1.454 0.953
5 0.123 0.581 0.974 1.689 0.937
6 0.069 0.307 0.939 1.202 1.931 0.928
7 0.031 0.214 0.549 1.358 1.429 2.176 0.923
8 0.017 0.107 0.448 0.847 1.837 1.655 2.423 0.920
8 Active Filters: Theory and Design
TABLE 1.5
2-dB Chebyshev filter
n b0 b1 b2 b3 b4 b5 b6 b7
1 1.308
2 0.823 0.804
3 0.327 1.022 0.738
4 0.206 0.517 1.256 0.716
5 0.082 0.459 0.693 1.450 0.706
6 0.051 0.210 0.771 0.867 1.746 0.701
7 0.020 0.166 0.383 1.144 1.039 1.994 0.698
8 0.013 0.070 0.360 0.598 1.580 1.212 2.242 0.696
where n is the order of the filter and can be a positive whole number, w1 is the -3
dB frequency of the filter, and K is the gain of the filter.
We see that | H (0) | = K and | H ( jω ) | is monotonically decreasing with w. In
addition, the 0.707 or -3 dB point is at w = 1 for all n; that is,
K
H ( jω ) = for all n (1.13)
2
⎛ 1 3 5 35 8 n ⎞
H ( jω ) = K ⎜ 1 − ω 2 n + ω 4 n − ω 6 n + ω + ⎟ (1.14)
⎝ 2 8 16 128 ⎠
TABLE 1.6
3-dB Chebyshev filter
n b0 b1 b2 b3 b4 b5 b6 b7
1 1.002
2 0.708 0.645
3 0.251 0.928 0.597
4 0.177 0.405 1.169 0.582
5 0.063 0.408 0.549 1.415 0.574
6 0.044 0.163 0.699 0.691 1.663 0.571
7 0.016 0.146 0.300 1.052 0.831 1.912 0.568
8 0.011 0.056 0.321 0.472 1.467 0.972 2.161 0.567
Introduction 9
TABLE 1.7
Bessel filter
n b0 b1 b2 b3 b4 b5 b6
1 1
2 3 3
3 15 15 6
4 105 105 45 10
5 945 945 420 105 15
6 10395 10395 4725 1260 210 21
7 135135 13 5135 62370 17325 3150 378 28
1
H ( jω ) ≅ ω >> 1 (1.15)
ωn
We observe that asymptotically, H( jw) falls off as w-n for Butterworth response. In
terms of dB, the asymptotic slope is obtained as
K
H ( jω ) = 1/ 2
(1.17)
⎡⎣1 + ε Cn2 (ω ) ⎤⎦
2
10 Active Filters: Theory and Design
−10
Magnitude/dB
−20
−30
n=6
n = 10
n=4
n=
2
−40
0.1 0.2 0.4 0.6 0.8 1 2 4 6 8 10
ω
FIGURE 1.6 Amplitude response curves for Butterworth filters of various orders (K = 1).
where
Cn (ω ) =
1.2
1.0
Normalize Step Response
1.8
n=2 3 4 5 6 7 8 9 10
1.6
1.4
1.2
0
0 4 8 12 16
t
FIGURE 1.7 Step response for Butterworth low-pass filters; w1 = 1 rad/s, and step amplitude
is 1.
Introduction 11
For n = 0 we have:
C0 (ω ) = 1 (1.20)
C1 (ω ) = ω (1.21)
C2 (ω ) = 2ω (ω ) − 1 = 2ω 2 − 1 (1.23)
1
Ripple = 1 − (1.24)
(1 + ε 2 )1/ 2
TABLE 1.8
Chebyshev polynomials Cn (ω ) = cos(n cos −1 ω )
n
0 1
1 w
2 2w 2 - 1
3 4w 3 - 3w
4 8w 4 - 8w 2 + 1
5 16w 5 - 20w 3 + 5w
6 32w 6 - 48w 4 + 18w 2 - 1
7 64w 7 - 112w 5 + 56w 3 - 7w
8 128w 8 - 256w 6 + 160w 4 - 32w 2 + 1
12 Active Filters: Theory and Design
|H(jw)|
1
1/(1 + e2)
0 1 ω
At w = 1, we have:
1
H ( j1) = (1.25)
(1 + ε 2 )1/ 2
1
H ( jω ) ≅ ω > ωp (1.26)
εCn (ω )
A = 20 log H ( jω )
A few different Chebyshev filter responses are shown in Figure 1.9 for various
values of n. Note that a filter of order n will have n - 1 peaks or dips in the
Introduction 13
1
Magnitude
1
1 + e2
n=2
n = 10
n=5
0
0 1 2 ω
FIGURE 1.9 Examples of Chebyshev amplitude response: (a) 3-dB ripple, (b) expanded view
of passband region showing form of responses below cutoff frequency.
passband response. Note also that the nominal gain of the filter (K = 1) is equal
to the filter’s maximum passband gain. An odd-order Chebyshev will have a dc
gain to the nominal gain, with “dips” in the amplitude response curve to the ripple
value. An even-order Chebyshev will have its dc gain equal to the nominal filter
minus the ripple values; the ripple in this case increases the gain to the nominal
value.
The addition of a passband ripple as a parameter makes the specification process
for a Chebyshev filter more complicated than for a Butterworth filter, but also
increases flexibility, because passband ripple can be treated for cutoff slope.
Figure 1.10 shows the step response of 3-dB ripple Chebyshev filters of various
orders. As with the Butterworth filters, the higher-order filters ring more.
1.2
1.0
Normalize Step Response
0.8
n=2 3 4 5 6 7 8 9 10
0.6
0.4
0.2
0
0 4 8 12 16
t
FIGURE 1.10 Step response for Chebyshev low-pass filters; ω 1 = 1 rad/s and step amplitude
is 1.
14 Active Filters: Theory and Design
H (s) = Ke − sT (1.29)
where K is a positive real constant. Then, the frequency response of the system can
be expressed as
H ( jω ) = Ke − jωT (1.30)
so that the amplitude response A(w) is a constant K, and the phase response
v0 (t ) = ℑ−1{V0 (s)} ∴
We see that the response v0 (t ) is simply the excitation delayed by a time T, and
multiplied by a constant. Thus, no signal distortion results from transmission through
a system described by H(s) in Equation (1.29). We note further that the delay T can
be obtained by differentiating the phase response ϕ (ω ) by w; that is,
d ϕ(ω )
delay = − =T (1.34)
dω
If ringing or overshoot must be avoided when pulses are filtered, the phase shift
between the input and output of a filter must be a linear function of frequency, i.e.,
the rate of change of the phase with respect to frequency must be constant. The net
Introduction 15
effect of a constant group delay in a filter is that all frequency components of a signal
transmitted through it are delayed by the same amount, i.e., there is no dispersion of
signals passing through the filter. Accordingly, because a pulse contains signals of
different frequencies, no dispersion takes place, i.e., its shape will be retained when
it is filtered by a network that has a linear phase response or constant group delay.
Just as the Butterworth filter is the best approximation to the ideal of “perfect
flatness of the amplitude response” in the filter passband, so the Bessel filter provides
the best approximation to the ideal of “perfect flatness of the group delay” in the
passband, because it has a maximally flat group delay response. However, this applies
only to low-pass filters because high-pass and band-pass Bessel filters do not have
the linear-phase property. Figure 1.11 compares the amplitude (a) and phase response
Butterworth
−2
Magnitude/dB
−4
Bessel
−6
−8
−10
0.1 0.2 0.4 0.6 0.8 1.0 2.0
ω
(a)
−90
Bessel
−180
Phase/degrees
Butterworth
−270
−360
−450
−540
0 1 2
ω
(b)
FIGURE 1.11 Response of Butterworth and Bessel: (a) amplitude, (b) phase.
16 Active Filters: Theory and Design
1.2
1.0
Bessel Chebyshev 0.1 dB ripple
0.8
Butterworth
0.6
0.4
0.2
0
0 2 4 6 8 10
t
FIGURE 1.12 Step response for Bessel low-pass filters; w1 = 1 rad/s and input step amplitude
is 1.
of a Bessel filter with that of a Butterworth filter of the same order. Bessel step
response is plotted in Figure 1.12 for various values of n.
To determine which basic filter is most suitable for a given application, it is
useful to have a side-by-side comparison of their amplitude, phase, and delay
characteristics. These characteristics are determined by the location, in the s plane,
of the n poles of H(s). Thus, the poles of a Butterworth filter lie on a semicircle in
the left-half s plane, those of a Chebyshev filter on an ellipse that becomes narrower
with increasing ripple, and those of a Bessel filter on a curve outside the Butterworth
semicircle. This is shown in Figure 1.13.
Im
Butterworth
Bessel
Chebyshev
Re
FIGURE 1.13 Comparison of pole location of Butterworth, Chebyshev, and Bessel filters.
Introduction 17
1
Magnitude
1
1 + e2
n=3
n=5 n=4
0
0 1 2 ω
1
| H ( jΩ) |2 = (1.35)
1 + ε Un2 (Ω, L )
2
where ε is the passband parameter and Un (Ω, L ) is the nth-order Jacobian elliptic
function. The parameter L contains information about relative heights of ripples in
the passband and stopband.
Relative Performance
Passive Filtering
Excellent performance,
Active Filtering
Sophisticated hardware
Beyond present
scale of the art
integrators, inverter, summing amplifier, and lossy integrators. Efficient and low-
cost active filter design, therefore, depends on the realization of a desired transfer
function into a circuit that uses the fewest components while maintaining all per-
formance requirements. Numerous circuits have evolved to meet this objective. Some
of the more common are Sallen–Key, multifeedback, state-variable, and biquad.
Each circuit has been designed to optimize specific performance aspects. Some are
easily tuned, others use a minimum number of components, and still others feature
a fixed bandwidth. Most designers would rather not “reinvent the wheel”; however,
they would like to develop a basic engineering understanding of the operation,
advantages, and restrictions of each type of filter design.
Once the concept of the active filter as a frequency-selective control system is
understood, active filter analysis becomes straightforward. In the next chapters we
will present some basic active filter techniques. Several examples will be used to
show the trade-offs and guidelines of active filter design and op-amp selection.
1.5.2 BIOFEEDBACK
The four commonly recognized brainwaves, delta, theta, alpha, and beta, can be
segregated using active band-pass filtering techniques. Various graphical techniques
may be employed to monitor transitions between wave-ground as a response to
stimuli. Careful attention must be paid to the amplifier’s low-frequency noise char-
acteristic. The OP-07 (PMI) has the lowest noise of any monolithic op-amp.
1.5.3 INSTRUMENTATION
Applications of active filters to instrumentation is the most diverse of all fields. The
low-pass filter is often used as a signal conditioner or noise filter. Because low-pass
filters are designated to operate down to dc, low op-amp offset voltages, such as those
offered by the OP-07, are of prime importance. In harmonic distortion measurements,
notch or bandstop filters allow harmonic interference to be accurately determined.
1.5.5 AUDIO
Electronic music and audio equalizers use a large number of active filters. Synthesizers
combine various low-pass, high-pass, and bandpass functions to generate waveforms
that have spectral densities similar to orchestra instruments. Symmetrical positive and
negative slew rates make the OP-11 (PMI) well suited for audio applications.
vl
vi R2 vo
R1
For a circuit operating in the noninverting mode, for an ideal op-amp, we find
from figure 1.16, the gain K.
From node v1, we have:
⎛ G ⎞
V0 = ⎜ 1 + 1 ⎟ Vi ∴
⎝ G2 ⎠
⎛ R ⎞
V0 = ⎜ 1 + 2 ⎟ Vi ∴
⎝ R1 ⎠
V0 R
K= = 1+ 2 (1.36)
V1 R1
ω1 2 π f1
FSF = = = 2 π f1 (2.1)
ωn 1
The FSF must be a dimensionless number. So, both the numerator and denominator
of Equation (2.1) must be expressed in the same units, usually rad/s.
Frequency-scaling a filter has the effect of multiplying all points on the frequency
axis of the response curve by the FSF. Therefore, a normalized response curve can
be directly used to predict the attenuation of the denormalized filter.
Any linear active or passive network maintains its transfer function if all
resistors are multiplied by an impedance-scaling factor (ISF) and all capacitors
are divided by the same factor ISF. This occurs because the ISFs cancel one
another out in the transfer function. Impedance scaling can be mathematically
21
22 Active Filters: Theory and Design
expressed as
R = ISF × Rn (2.2)
Cn
C= (2.3)
ISF
R = ISF × Rn (2.4)
Cn
C= (2.5)
ISF × FSF
G
V1 (s) = V
G + sC i
where
Rb
V0 = KV1 and K = 1 + \
Ra
KG
V0 (s) = V ∴
G + sC i
R V1 = Vo/K
+
−
vi C
Rb
v1 vo
Ra
V0 (s) KG
H (s ) = =
Vi (s) G + sC
K ω1 K
H (s ) = = (2.6)
s + ω1 s
1+
ω1
where
s = jω
1 (2.7)
and b = ω1 =
RC
ω1 = 2 π f1, f1 is the cutoff frequency of the filter.
For the normalized filter Gn = 1 S ( Rn = 1 Ω).
Hence
1
Cn = (2.8)
b
For Ran = 1 Ω ∴
Rbn = K − 1 Ω (2.9)
A = 20 log H ( j ω ) = 20 log K dB
s
2. For >> 1 ∴
ω1
−1
K ⎛ω⎞
H ( jω ) = = K⎜ ⎟ ∴
ω ⎝ ω1 ⎠
ω1
−1
⎛ω⎞ ⎛ω⎞
A = 20 log H ( jω ) = 20 log K + 20 log ⎜ ⎟ = 20 log K − 20 log ⎜ ⎟ dB
⎝ ω1 ⎠ ⎝ ω1 ⎠
24 Active Filters: Theory and Design
ω
For = 10 \ slope = −20 dB/dec
ω1
ω
For =2 ∴ slope = −6 dB/oct
ω1
A = 20 log K − 20 log10 dB = 20 log K − 20 dB
s
3. For =1 ∴
ω1
K
H ( jω ) = ∴
1+ j
K
H ( jω ) = ∴
2
K
A = 20 log = 20 log K − 20 log 2 = 20 log K − 3 dB
2
Figure 2.2 shows the frequency response of the filter.
EXAMPLE 2.1
A first-order LP Butterworth filter must be designed with gain of 5 at a cutoff
frequency of 1 kHz.
Solution
From the Butterworth coefficients of Appendix C, we have (n = 1):
b = 1, hence, from Equation (2.9), we have:
Cn = 1F
A
dB
20 log K −20 dB/dec
20 log K –3 or
−6 dB/oct
ideal
20 dB
0.1 f1 f1 10 f1 log f
a decade
A
dB
R 20 a decade
+ 11
10 K 14
− 10
vi C Rb
15.9 nf
vo 20 dB
40 K 0
0.1 1 10
Ra 10 K −6 log f (kHz)
−10
−20 dB/dec
−20
(a) (b)
Denormalization
ISF = 10 4
ω1 2 π f1
FSF = = = 2 π × 10 3 ∴
ωn 1
Cn 1
C= = = 15.9 nF
ISF × FSF 2 π × 10 7
R = ISF × Rn = 10 4 × 1 Ω = 10 kΩ
Ra = ISF × Ran = 10 × 1 kΩ = 10 kΩ
Rb = ISF × Rbn = 10 × 4 kΩ = 40 kΩ
Figure 2.3 shows the designed filter with its frequency response.
Rb
K = 1+
Ra
V0
− sCVi + (G + sC ) =0 ∴
K
V0 KsC Ks Ks
H (s ) = = = = (2.10)
Vi G + sC 1 s + ω2
s+
RC
26 Active Filters: Theory and Design
C v/K
+
−
R
vi Rb Rb
vo K = 1 +
Ra
Ra
where
1
ω2 = =b (2.11)
RC
1
Rn = (2.12)
b
Rbn = K − 1 (2.13)
⎛ s ⎞
K⎜ ⎟
⎝ ω2 ⎠
H (s ) = (2.14)
⎛ s ⎞
1+ ⎜ ⎟
⎝ ω2 ⎠
⎛ ω⎞
A = 20 log H ( jω ) = 20 log K + 20 log ⎜ ⎟ dB
⎝ ω2 ⎠
Sallen–Key Filters 27
ω
For = 10 ∴ slope = 20 dB / dec.
ω2
ω
For =2 ∴ slope = 6 dB / oct. ∴
ω2
A = 20 log K + 20 dB
s
2. For >> 1 ∴ H ( jω ) = K \
ω2
A = 20 log H ( jω ) = 20 log K dB
K K
H ( jω ) = ∴ H ( jω ) = ∴
1+ j 2
K
A = 20 log = 20 log K − 20 log 2 = 20 log K − 3 dB
2
EXAMPLE 2.2
A first-order HP Butterworth filter must be designed with gain of 5 at a cutoff
frequency of 100 Hz.
Solution
From the Butterworth coefficients of Appendix C, we have (n = 1):
b = 1, hence:
A
dB
20 log K
20 log K − 3
ideal
20 dB
0
0.1 f2 f2 10 f2 log f (Hz)
20 dB/dec or 6 dB/oct
a decade
159 nf
+ A
− 20 11
10 K 40 K 14
vi
vo 10
dB 10
10 K
0
−6 100 1K 10 K log f (Hz)
−10 20 dB/dec
(a) (b)
Cn = 1 F Rn = 1 Ω
Denormalization
ISF = 10 4
ω 2 2 π f2
FSF = = = 2 π × 100 = 200 π \
ωn 1
Cn 1
C= = = 159.2 nF
ISF × FSF 2 π × 10 6
R = ISF × Rn = 10 4 × 1 Ω = 10 kΩ
Rb = ISF × Rbn = 10 × 1 kΩ
Rb = ISF × Rbn = 10 × 4 kΩ = 40 kΩ
Figure 2.6 shows the designed filter with its frequency response.
R1 R2
+ K
vi C1 C2 vo vi vo
(a) (b)
1 ⎛ 1 1 ⎞ 1
− Vi + ⎜ + + sC1 ⎟ V1 − V2 − sC1V0 = 0 (2.15)
R1 ⎝ R1 R2 ⎠ R2
node v2
1 ⎛ 1 ⎞
− V1 + ⎜ + sC2 ⎟ V2 = 0 (2.16)
R2 ⎝ R2 ⎠
but
V0
V2 = (2.17)
K
jω
C1 K=1 j K=2
j0.5
R1 R2 v2 K=0 −1.5 0.5
+K
v1 C2 1 1.5 σ
vi vo −j0.5
K=1 −j K = 2
(a) (b)
FIGURE 2.8 (a) An RC passive network and an op-amp are combined to form a general
active filter. (b) When the gain K is varied throughout the s-plane, the response with respect
to pole position is obtained.
30 Active Filters: Theory and Design
1 ⎛ 1 ⎞V
V1 = ⎜ + sC2 ⎟ 0 ∴
R2 ⎝ R2 ⎠K
V0
V1 = (1 + sR2C2 ) (2.18)
K
V0
( R1 + R2 + sR1R2C1 )V1 − R1 − sR1R2C1V0 = R2Vi ∴
K
V0 V
(1 + sR2C2 )( R1 + R2 + sR1R2C1 ) − R1 0 − sR1R2C1V0 = R2Vi ∴
K K
R2 [s 2 R1R2C1C2 + s( R1C1 + R1C2 + R2C2 − R1C1K ) + 1]V0 = KR2Vi ∴
V0 K
H (s ) = =
Vi s 2 R1R2C1C2 + s[C2 ( R1 + R2 ) + R1C1 (1 − K )] + 1
K
H (s ) = (2.19)
as 2 + bs + 1
where
− b ± b 2 − 4 ac
s=
2a
where
behavior in the right half-plane for larger values of K is then analogous to that in
the left half-plane, with the right half-plane characteristics representing unstable
network behavior.
Of interest are the network characteristics at K = 1. In this case, the poles are
located at s = − 0.707 ± j 0.707, which is the same as that of a two-pole Butterworth
low-pass filter.
Rb
K = 1+
Ra
node v1
V0
−GVi + (G1 + G2 + sC2 )V1 − G2 − sC2V0 = 0 (2.21)
K
node v2 (v0 /K )
V0
−G2V1 + (G2 + sC1 ) =0 (2.22)
K
\
(G2 + sC1 )V0
V1 = (2.23)
KG2
C2
R1 R2 v2 = v0/K
+ Rb
v1 K=1+
− Ra
C1
vi Rb
vo
Ra
G1G2
K
C1C2
H (s ) = (2.24)
⎡ G + G2 (1 − K )G2 ⎤ G1G2
s2 + s ⎢ 1 + ⎥+
⎢⎣ C2 C1 ⎥⎦ C1C2
Kb
H (s ) = (2.25)
s + as + b
2
where
G1G2
ω12 = b = (2.26)
C1C2
and
G1 + G2 (1 − K )G2
a= + (2.27)
C2 C1
For R1n = R2 n = Rn = 1 Ω ∴
1
b= (2.28)
C1C2
and
2 1− K
a= + (2.29)
C2 C1
K ω12
H (s ) = ∴
⎡ G + G2 (1 − K )G2 ⎤
s +s⎢ 1
2 + ⎥ + ω12
⎢⎣ C2 C1 ⎥⎦
K
H (s ) = 2
∴
⎛ s ⎞ 1 ⎡ G1 + G2 (1 − K )G2 ⎤ s
⎜ ⎟ + ⎢ + ⎥ +1
⎝ ω1 ⎠ ω1 ⎢⎣ C2 C2 ⎥⎦ ω1
Sallen–Key Filters 33
K
H (s ) = 2
(2.30)
⎛ s ⎞ ⎛ s ⎞
⎜ ⎟ + a⎜ ⎟ +1
⎝ ω1 ⎠ ⎝ ω1 ⎠
s
1. For << 1, we have:
ω1
H ( jω ) ≅ K
A = 20 log H ( jω ) = 20 log K dB
s
2. For >> 1, we have:
ω1
−2
K ⎛ω⎞
H ( j ω) ≅ = K⎜ ⎟ ∴
⎝ ω1 ⎠
2
⎛ω⎞
⎜ ⎟
⎝ ω1 ⎠
−2
⎛ω⎞ ⎛ω⎞
A = 20 log H ( jω ) = 20 log K ⎜ ⎟ = 20 log K − 40 log ⎜ ⎟ dB
⎝ ω1 ⎠ ⎝ ω1 ⎠
ω
For = 10 ∴ slope = −40 dB / dec
ω1
ω
For =2 ∴ slope = −12 dB / oct
ω1
A = 20 log K − 40 dB
s
3. For =1 ∴
ω1
K K
H ( jω ) = ∴ H ( jω ) =
1+ j 2
K
A = 20 log = 20 log K − 20 log 2 = 20 log K − 3 dB
2
A
dB −40 dB/dec
20 log K or
20 log K − 3 −12 dB/oct
ideal
0
0.1 f1 f1 10 f1 log f
40 dB
a decade
1
C2 = (2.31)
bC1
1− K
2bC1 + =a ∴ 2bC12 − aC1 + 1 − K = 0 ∴
C1
a ± a 2 − 8b( K − 1)
C1 = (2.32)
4b
The capacitance C1 could have two values that satisfy Equation (2.31) because plus
and minus signs appear outside the radical. However, we will use the plus sign on
the radical for our solutions to active filter design, hence:
a + a 2 + 8b( K − 1)
C1n = (2.33)
4b
4
C2 n = (2.34)
a + a + 8b( K − 1)
2
Sallen–Key Filters 35
a2
K ≥ 1− (2.35)
4b
a
C1n = (2.36)
2b
2
C2 n = (2.37)
a
EXAMPLE 2.3
A second-order low-pass Butterworth filter must be designed with gain 10 and
f1 = 1 kHz.
Solution
From Butterworth coefficients in Appendix C, for n = 2, we have:
a = 1.414, b = 1.000
1
C2 = = 0.399 F
2.504
Finally, we denormalize the resistance and capacitance values. We find the frequency-
normalizing factor FSF and the impedance-scaling factor ISF from Equations (2.4)
and (2.5), respectively.
ISF = 10 4 and
ω1 2 π f1
FSF = = = 2 π × 10 3 ∴
ωn 1
36 Active Filters: Theory and Design
R = ISF × Rn = 10 4 × 1 Ω = 10 kΩ
C1n 2.504
C1 = = ∴ C1 = 39.9 nF
ISF × FSF 2 π × 10 7
C2 n 0.399
C2 = = ∴ C2 = 6.4 nF
ISF × FSF 2 π × 10 7
Rb = ( K − 1) Ra
For Ra = 1 kΩ ∴ Rb = 9 Ra or Rb = 9 kΩ
The active filter circuit designed in the example is shown in Figure 2.11a and its
frequency response in Figure 2.11b.
6.4 n
C2
10 K 10 K
+ LM741
R1 R2 −
40 n
C1 9K
Rb
1K Ra
(a)
30.00
20.00
10.00
Gain dB
0.00
–10.00
–20.00
100 1K 10 K
Frequency in Hz
(b)
FIGURE 2.11 (a) The second-order Butterworth VCVS LPF, f1 = 1 kHz, K = 10; (b) its
frequency response.
Sallen–Key Filters 37
EXAMPLE 2.4
A second-order LP Butterworth filter must be designed with gain 1 and f1 = 750 Hz.
Solution
From the Butterworth coefficients in Appendix C, for n = 2, we have:
a = 1.414, b = 1.000
a 1.414
C1 = = = 0.707
2b 2
2 2
C2 = = = 1.414
a 1.414
Denormalization
ISF = 10 4
ω1 2 π f1
FSF = = = 2 π × 750 = 1.5π × 10 3
ωn 1
C1n 0.707
C1 = = = 15 nF
ISF × FSF 1.5π × 10 7
C2 n 1.414
C2 = = = 30 nF
ISF × FSF 1.5π × 10 7
R = ISF × Rn = 10 4 × 1 Ω = 10 kΩ
The final circuit looks like the one shown below in Figure 2.12a, with its frequency
response in Figure 2.12b.
EXAMPLE 2.5
Design a 300-Hz LP Chebyshev filter with gain 5 and ripple width 3 dB.
Solution
From Chebyshev 3-dB coefficients in Appendix C, we find:
a = 0.645, b = 0.708
38 Active Filters: Theory and Design
22.5 n
C2
10 K 10 K
+ LM741
R1 R2 −
11.3 n
C1
(a)
10.00
0.00
–10.00
Gain dB
–20.00
–30.00
–40.00
100 1K 10 K
Frequency in Hz
(b)
FIGURE 2.12 A 750-Hz LP Butterworth VCVS (a); its frequency response (b).
Hence,
1 1 1
C2 = = = = 0.734 F
bC1 0.708 × 1.924 1.362
Denormalization
ISF = 10 4
ω1 2 π f1
FSF = = = 2 π × 300 = 600 π
ωn 1
Sallen–Key Filters 39
C1n 1.924
C1 = = = 102 nF
ISF × FSF 6 π × 10 6
C2 n 0.734
C2 = = = 38.9 nF
ISF × FSF 6 π × 10 6
R = ISF × Rn = 10 4 × 1 Ω = 10 kΩ
Rb = ( K − 1) Ra ∴ Ra = 5 kΩ ∴ Rb = 20 kΩ
Figure 2.13a shows the designed filter, with its frequency response in Figure 2.13b.
39 n
C2
10 K 10 K
+ LM741
R1 R2 −
102 n
C1 20 K
Rb
Ra 5K
(a)
20.00
10.00
0.00
Gain dB
–10.00
–20.00
–30.00
10 100 1K 10 K
Frequency in Hz
(b)
FIGURE 2.13 (a) Second-order LP Chebyshev 3-db VCVS, f1 = 300 Hz, K = 5; (b) its
frequency response.
40 Active Filters: Theory and Design
EXAMPLE 2.6
Design a 1-kHz Bessel filter with gain 10.
Solution
From Bessel coefficients in Appendix C, we find:
a = 3.000, b = 3.000
Hence:
a + a 2 + 8b( K − 1) 3 + 9 + 8 × 3 × 9
C1 = = = 1.5 F
4b 4×3
1 1
C2 = = = 0.222 F
bC1 3 × 1.475
Denormalization
ISF = 10 4
ω1
FSF = = 2 π f1 = 2 π × 10 3
ωn
C1n 1.475
C1 = = = 23.8 nF
ISF × FSF 2 π × 10 7
C2 n 0.226
C2 = = = 3.5 nF
ISF × FSF 2 π × 10 7
For Ra = 2.2 kΩ ∴
Rb = ( K − 1) Ra = 9 × 2.2 kΩ ≅ 20 kΩ
Figure 2.14a shows the designed filter, with its phase response in Figure 2.14b.
C1n = C2 n = Cn
1
C1n = C2 n = Cn = (2.38)
b
Sallen–Key Filters 41
3.5 n
C2
10 K 10 K
+ LM741
R1 R2 −
24 n
C1 20 K
Rb
Ra 2.2 K
(a)
30.00 0.0
20.00 –36.0
10.00 –72.0
Gain dB
Phase Deg
0.00 –108.0
–10.00 –144.0
–20.00 –180.0
100 1K 10 K
Frequency in Hz
(b)
FIGURE 2.14 (a) Second-order LP Bessel filter, f1 = 1 kHz, K = 10; (b) its phase response.
3− K
=a ∴
a
a
K = 3− (2.39)
b
EXAMPLE 2.7
Design a 1-kHz “equal-component” Butterworth low-pass filter.
Solution
Cn = 1 F, Rn = 1 Ω
a = 1.414, b = 1.000
Hence:
a
K = 3− = 3 − 1.414 = 1.586
b
Denormalization
ISF = 10 4
ω1
FSF = = 2 π f1 = 2 π × 10 3
ωn
Hence:
R = ISF × Rn = 10 4 × 1 Ω = 10 kΩ
Ra = ISF × Ran = 10 4 × 1 Ω = 10 kΩ
Cn 1
C= = = 15.9 nF
ISF × FSF 2 π × 10 7
Figure 2.15a shows the designed filter, with its frequency response in Figure 2.15b.
EXAMPLE 2.8
Design a 1-kHz LP Chebyshev 1-dB “equal component” filter.
Solution
From Chebyshev 1-dB coefficients in Appendix C, we find:
16 n
C2
10 K 10 K
+ LM741
R1 R2 −
16 n
5.9 K
C1
Rb
Ra 10 K
(a)
10.00
0.00
–10.00
Gain dB
–20.00
–30.00
–40.00
100 1K 10 K
Frequency in Hz
(b)
FIGURE 2.15 (a) Second-order LP Butterworth filter; (b) its frequency response.
Hence:
1 1
Cn = = = 0.950 F
b 1.103
a 1.098
K = 3− = 3− = 1.954
b 1.103
Denormalization
ISF = 10 4
ω1
FSF = = 2 π f1 = 2 π × 10 3
ωn
44 Active Filters: Theory and Design
16 n
C2
10 K 10 K
+ LM741
R1 R2 −
16 n
9.54 K
C1
Rb
Ra 10 K
(a)
10.00
0.00
–10.00
Gain dB
–20.00
–30.00
–40.00
100 1K 10 K
Frequency in Hz
(b)
FIGURE 2.16 (a) Second-order LP Chebyshev 1-dB LPF; (b) its frequency response.
R = ISF × Rn = 10 4 × 1 Ω = 10 kΩ
Cn 1
C= = = 15.9 nF
ISF × FSF 2 π × 10 7
Figure 2.16a shows the designed filter, and its frequency response is shown in Figure 2.16b.
resistor by a capacitor having the reciprocal value and vice versa as follows:
1
CHP = (2.40)
RLP
1
RHP = (2.41)
CLP
It is important to recognize that only the resistors that are part of the low-pass RC
networks are transformed into capacitors by Equation (2.39). Feedback resistors that
strictly determine operational amplifier gain, such as Ra and Rb, in Figure 2.9, are
omitted from the transformation.
After the normalized low-pass configuration is transformed into a high-pass
filter, the circuit is frequency- and impedance-scaled in the same manner as in the
design of low-pass filters.
EXAMPLE 2.9
Design a 100-Hz HP Butterworth with gain 10.
Solution
From Butterworth coefficients in Appendix C, we have:
a = 1.414, b = 1.000
1 1
C2 n = = = 0.399 F
bC1n 2.504
Hence:
1 1
R1n = = = 0.399 Ω
C1n 2.504
1 1
R2 n = = = 2.504 Ω
C2 n 0.399
C1n = C2 n = Cn = 1 F
Denormalization
ISF = 10 4
46 Active Filters: Theory and Design
ω2
FSF = = 2 π f2 = 2 π × 10 2
ωn
Cn 1
C1 = C2 = C = = = 159.2 nF
ISF × FSF 2 π × 10 6
Ra = 10 ζΩ ∴ Rb = ( K − 1) Ran = 9 × 10 kΩ = 90 kΩ
Figure 2.17a shows the designed filter, with its frequency response in Figure 2.17b.
25 K
R2
159 n 159 n
+ LM741
C C −
4K Rb
R1
90 K
10 K Ra
(a)
40.00
20.00
0.00
Gain dB
–20.00
–40.00
–60.00
1 10 100 1K 10 K
Frequency in Hz
(b)
FIGURE 2.17 (a) HP Butterworth filter, with f2 = 100 Hz, K = 10; (b) frequency response.
Sallen–Key Filters 47
EXAMPLE 2.10
A second-order HP Chebyshev 0.5-dB filter must be designed with a gain of 1 at a
cutoff frequency of 100 Hz.
Solution
From Chebyshev 0.5-dB coefficients in Appendix C, we find:
a = 1.426 and b = 1.516
a 1.426
C1n = = = 0.470 F
2b 2 × 1.516
2 2
C2 n = = = 1.403 F
a 1.426
Hence:
1 1
R1n = = = 2.126 Ω
C1n 0.470
1 1
R2 n = = = 0.713 Ω
C2 n 1.403
Denormalization
ISF = 10 4
ω2
FSF = = 2 π f2 = 2 π × 10 2
ωn
Cn 1
C1 = C2 = C = = = 159.2 nF
ISF × FSF 2 π × 10 6
Figure 2.18a shows the designed filter, with its frequency response in Figure 2.18 b.
EXAMPLE 2.11
Design an HP Chebyshev 3-dB “equal component” filter at a cutoff frequency of 200 Hz.
Solution
From Chebyshev 3-dB coefficients in Appendix C, we find:
a = 0.645, b = 0.708
48 Active Filters: Theory and Design
7K
R2
159 n 159 n
+ LM741
C C −
21 K
R1
(a)
10.00
0.00
–10.00
Gain dB
–20.00
–30.00
–40.00
10 100 1K
Frequency in Hz
(b)
FIGURE 2.18 (a) HP Chebyshev 0.5 dB, f2 = 100 Hz, K = 1; (b) its frequency response.
R1n = R2 n = Rn = 1 Ω
1 1
C1n = C2 n = Cn = = = 1.188 F ∴
b 0.708
a 0.645
K = 3− = 3− = 3 − 0.767 = 2.233
b 0.708
1 1
R1n = R2 n = Rn = = = 0.841 F
Cn 1.188
C1n = C2 n = Cn = 1 F
Sallen–Key Filters 49
Denormalization
ISF = 10 4
ω2
FSF = = 2 π f2 = 2 π × 200 = 4 π × 10 2
ωn
Cn 1
C1 = C2 = C = = = 79.6 nF
ISF × FSF 4 π × 10 6
R1 = R2 = R = ISF × Rn = 10 × 0.841 Ω = 8.4 kΩ
Ra = 10 kΩ ∴ Rb = ( K − 1) Ra = (2.233 − 1) × 10 kΩ
Figure 2.19a shows the designed filter, with its frequency response in Figure 2.19b.
8.4 K
R2
79.6 n 79.6 n
+ LM741
C C −
8.4 K Rb
R1
12.3 K
10 K Ra
(a)
20.00
10.00
0.00
Gain dB
–10.00
–20.00
–30.00
10 100 1K 10 K
Frequency in Hz
(b)
FIGURE 2.19 (a) HP Chebyshev 3-dB, “equal component,” f2 = 200 Hz; (b) its frequency response.
50 Active Filters: Theory and Design
V0 (s) K
H (s ) = = n (2.42)
Vi (s) s + bn−1s + . . . + b1s + b0
n −1
Such a function has a magnitude characteristic that decreases at the rate of 20n
dB/dec or 6n dB/oct outside the passband. If n is even, Equation (2.42) may be
written in the form
n/2
∏s
Ki
H (s ) = (2.43)
i =1
2 + ai1s + bi 0
n −1
2
∏s
1 Ki
H (s ) = (2.44)
s+b i =1
2 + ai1s + bi 0
The quadratic factors are realized as before, and the first-order filter may be realized
by the circuit shown in Figure 2.1 with unity gain.
Sallen–Key Filters 51
⎛ 10 0.1 Amin − 1 ⎞
log ⎜ 0.1 A ⎟
⎝ 10 max − 1⎠
n= (2.45)
⎛f ⎞
2 log ⎜ s ⎟
⎝ f1 ⎠
When n has been determined, the attenuation in dB at fs can be found from the expression:
⎡ ⎛ f ⎞ ⎤
2n
Next, f3dB must be found. Note that f3dB, is the nominal cutoff frequency. f3dB can
be found from the expression:
1/ 2 n
⎛ 10 0.3 − 1 ⎞
f3dB = f1 ⎜ 0.1 A ⎟ (2.47)
⎝ 10 max − 1⎠
Consider a low-pass filter with f1 = 300 Hz, Amax = 1 dB, fs = 500 Hz , and Amin = 20 dB.
Hence
⎛ 10 2 − 1 ⎞
log ⎜ 0.1 ⎟
⎝ 10 − 1⎠
n= = 5.82
⎛ 500 ⎞
2 log ⎜ ⎟
⎝ 300 ⎠
⎡ ⎛ 500 ⎞ ⎤⎥
12
1/12
⎛ 10 0.3 − 1⎞
f3dB = 300 ⎜ 0.1 ⎟ = 300 × 1.119 Hz = 335.7 Hz
⎝ 10 − 1⎠
The order n for a Chebyshev low-pass filter can be found from the expression:
ln( x + x 2 − 1)
n= (2.48)
ln( y + y 2 − 1)
where
10 0.1 Amin − 1
x= (2.49)
10 0.1 Amax − 1
and
fs
y= (2.50)
f1
where
⎛ 2 ⎞
⎛ f⎞
z = n ln ⎜⎜ + ⎜ ⎟ − 1⎟⎟
f
(2.52)
⎜⎝ f1 ⎝ f1 ⎠ ⎟⎠
Consider a low-pass Chebyshev filter with f1 = 1 kHz, fs = 3 kHz, Amax = 0.2 dB, and
Amin = 50 dB. Hence
10 5 − 1
x= = 1456.65
10 0.02 − 1
ln(1456.65 + 2121834 )
n= = 4.53 ∴ n=5
ln(3 + 8 )
z = 5 ln(3 + 9 − 1) = 8.8137
Sallen–Key Filters 53
A (3 kHz) = 57.3 dB
Another way to determine n is to use the nomographs in Appendix B. They are used
to determine the order n necessary to meet a given set of specifications. To use a
nomograph, Amax, Amin, and the ratio fs / f1 or f2 / fs (for HPF) must be known. Draw a
straight line (1) from the desired value of Amax through Amin to the point of intersection
with the left edge, corresponding to fs / f1 = 1, of the nomograph. Then draw a vertical
line (2) corresponding to the desired value of fs / f1. Finally, draw a horizontal line (3)
from the end of line (1) to line (2). The point of intersection of line (2) and line (3)
should be between two of the numbered nomograph curves, and the required order n
will be equal to the higher of the curves. Figure 2.20 illustrates the procedure.
EXAMPLE 2.12
Design an LP Butterworth filter to realize the following specifications: f1 = 3 kHz,
fs = 9 kHz, Amax = 3 dB, Amin = 40 dB, and K = 9.
Solution
From the Butterworth nomographs we find the order of the filter, n = 5. From the
Butterworth coefficients in Appendix C, we find:
First stage (first order)
b0 = 1, hence:
1
Cn = = 1 F, Rn = 1 Ω
b0
Amax Amin
n
(1)
(3)
n–1
(2)
fs/f1
a = 1.618, b = 1, K1 = K = 9 = 3
hence:
1.618 + 1.6182 + 8 × 2
C1n = = 1.483 F
4
1
C2 n = = 0.674 F
1.483
R1n = R2 n = 1 Ω
a = 0.618, b = 1, K 2 = K = 9 = 3 ∴
0.618 + 0.6182 + 16
C1n = = 1.166 F
4
1
C2 n = = 0.857 F, R1n = R2 n = 1 Ω
1.166
Denormalization
ISF = 10 4
ω1
FSF = = 2 π f1 = 2 π × 3 × 10 3 = 6 π × 10 3
ωn
First stage
R = ISF × Rn = 10 4 × 1 Ω = 10 kΩ
Cn 1
C= = = 5.31 nF
ISF × FSF 6 π × 10 7
Second stage
R1 = R2 = R = ISF × Rn = 10 4 × 1 Ω = 10 k Ω
Sallen–Key Filters 55
C1n 1.483
C1 = = = 7.9 nF
ISF × FSF 6 π × 10 7
C2 n 0.674
C2 = = = 3.6 nF
ISF × FSF 6 π × 10 7
Ra = 10 kΩ ∴ Rb = ( K − 1) Ra = 2 × 10 kΩ = 20 kΩ
Third stage
R1 = R2 = R = ISF × Rn = 10 4 × 1 Ω = 10 kΩ
C1n 1.166
C1 = = = 6.2 nF
ISF × FSF 6 π × 10 7
C2 n 0.857
C2 = = = 4.5 nF
ISF × FSF 6 π × 10 7
Ra = 10 kΩ, Rb = 20 kΩ
Figure 2.21a shows the designed filter, and its frequency response is shown in Figure 2.21b.
EXAMPLE 2.13
Design an HP Chebyshev 3-dB filter to realize the following specification: Gain 1,
f2 = 1 kHz, fs = 333 Hz, and Amax = 30 dB.
Solution
From the Chebyshev nomographs in Appendix B, we find n = 3. From Chebyshev
3-dB coefficients in Appendix C, we have:
First stage (first order)
b = 0.299
3.6 n 4.5 n
10 K 10 K 10 K 10 K 10 K
+ LM741 + LM741 + LM741
− − −
5.3 n 7.9 n 6.2 n
20 K 20 K
10 K 10 K
(a)
FIGURE 2.21 (a) LP Butterworth filter, n = 5, f1 = 3 kHz, K = 9; (b) its frequency response.
56 Active Filters: Theory and Design
40.00
20.00
0.00
Gain dB
–20.00
–40.00
–60.00
10 100 1K 10 K 100 K
Frequency in Hz
(b)
Low-pass filter:
1 1
Rn = 1 Ω, Cn = = = 3.334 F
b 0.299
High-pass filter:
1 1
Cn = 1 F, Rn = = = 0.299 Ω
Cn 3.334
a = 0.299, b = 0.839
Low-pass filter:
0.299
C1n = = 0.178 F
2 × 0.839
2
C2 n = = 6.689 F
0.299
Rn = 1 Ω
Sallen–Key Filters 57
High-pass filter:
1 1
R1n = = = 5.618 Ω
C1n 0.178
1 1
R2 n = = = 0.150 Ω
C2 n 6.689
C1n = C2 n = Cn = 1 F
Denormalization
ISF = 10 4
ω2
FSF = = 2 π f2 = 2 π × 10 3
ωn
First stage
Cn 1
C= = = 15.9 nF
ISF × FSF 2 π × 10 7
Second stage
Cn 1
C= = = 15.9 nF
ISF × FSF 2 π × 10 7
Figure 2.22a shows the designed filter, and its frequency response is shown in Figure 2.22b.
1.5 K
(a)
FIGURE 2.22 (a) HP Chebyshev 3-dB filter, f2 = 1 kHz, K = 1; (b) its frequency response.
58 Active Filters: Theory and Design
20.00
0.00
–20.00
Gain dB
–40.00
–60.00
–80.00
100 1K 10 K 100 K
Frequency in Hz
(b)
EXAMPLE 2.14
Design an LP Bessel filter with the following characteristics: “equal component,”
third order, and cutoff frequency of 500 Hz.
Solution
From Bessel coefficients of Appendix C, we have:
First stage (first order)
b = 2.322 ∴
1 1
Cn = = = 0.431 F, Rn = 1 Ω
b 2.322
Second stage (second order)
a = 3.678, b = 6.459
1 1
Cn = = = 0.393 Ω, Rn = 1 Ω
b 6.459
a 3.678
K = 3− = 3− = 1.553 or 3.8 dB
b 6.459
Denormalization
ISF = 10 4
ω1
FSF = = 2 π f1 = 2 π × 500 = π × 10 3
ωn
Sallen–Key Filters 59
First stage
R = ISF × Rn = 10 4 × 1 Ω = 10 kΩ
Cn 0.431
C= = = 13.7 nF
ISF × FSF π × 10 7
Second stage
R = ISF × Rn = 10 4 × 1 Ω = 10 kΩ
Cn 0.393
C= = = 12.5 nF
ISF × FSF π × 10 7
Ra = 10 kΩ ∴ Rb = ( K − 1) Ra = 0.553 × 10 kΩ = 5.53 kΩ
Figure 2.23a shows the designed filter, and its frequency response is shown in Figure 2.23b.
12.5 n
10 K 10 K 10 K
+ LM741 + LM741
− −
13.7 n 12.5 n
5.53 K
10 K
(a)
20.00 0.0
0.00 –54.0
–20.00 –108.0
Phase Deg
Gain dB
–40.00 –162.0
–60.00 –216.0
–80.00 –270.0
100 1K 10 K
Frequency in Hz
(b)
FIGURE 2.23 (a) Bessel filter, “equal component,” f1 = 500 Hz; (b) frequency and phase response.
60 Active Filters: Theory and Design
EXAMPLE 2.15
Design a 100–1000 Hz, Butterworth wide-band filter with the following specifica-
tions: Amax = 3 dB, Amin = 30 dB at fs = 40 Hz, and 2500 Hz, with K = 9.
Solution
(a) High-pass filter
For Amax = 3 dB, Amin = 30 dB, f2 = 100 Hz, and fs2 = 40 Hz, we find n = 4 ( f2/fs2 =
100/40 = 2.5).
From the Butterworth coefficients in Appendix C, we find:
First stage
a = 1.848, b = 1.000
K HP = K = 9 = 3 ∴ K1 = K 2 = 3 = 1.732
A
dB
20 log K
20 log K – 3
0
f1 f2 log f
1 1
R2 n = = = 1.223 Ω
C2 n 0.818
a = 0.765, b = 1.000
1 1
R2 n = = = 0.826 Ω
C2 n 1.211
Denormalization
ISF = 10 4
ω2
FSF = = 2 π f2 = 2 π × 100
ωn
First stage
R2 = ISF × R2 n = 10 4 × 1.223 Ω ≅ 12 kΩ
Cn 1
C1 = C2 = C = = ≅ 159 nF
ISF × FSF 2 π × 10 6
Ra = 10 kΩ ∴ Rb = ( K − 1) Ra = 0.732 × 10 kΩ ≅ 7.3 kΩ
62 Active Filters: Theory and Design
Second stage
Cn 1
C= = ≅ 159 nF
ISF × FSF 2 π × 10 6
Ra = 10 kΩ ∴ Rb ≅ 7.3 kΩ
a = 1.848, b = 1.000
K HP = K = 9 ∴ K1 = K 2 = 3 = 1.732
R1n = R2 n = Rn = 1 Ω
Second stage
a = 0.765, b = 1.000
Denormalization
ω1
ISF = 10 4 , FSF = = 2 π f1 = 2 π × 10 3 , ISF × FSF = 2 π × 10 7
ωn
Sallen–Key Filters 63
12 K 8.2 K
10 K 10 K
13 n 19.5 n
10 K 10 K 10 K 10 K
+ LM741 + LM741
− −
19.5 n 13 n
7.3 K 7.3 K
10 K 10 K
(a)
40.00
20.00
0.00
Gain dB
–20.00
–40.00
–60.00
10 100 1K 10 K
Frequency in Hz
(b)
First stage
R1 = R2 = R = ISF × Rn = 10 4 × 1 Ω = 10 kΩ
C1n 1.223
C1 = = = 19.5 nF
ISF × FSF 2 π × 10 7
C2 n 0.818
C2 = = = 13 nF
ISF × FSF 2 π × 10 7
Ra = 10 kΩ ∴ Rb = ( K − 1) Ra = 7.3 kΩ
Second stage
R1 = R2 = R = 10 4 × 1 Ω = 10 kΩ
C1n 0.826
C1 = = ≅ 13 nF
ISF × FSF 2 π × 10 7
C2 n 1.211
C2 = = ≅ 19.5 nF
ISF × FSF 2 π × 10 7
Ra = 10 kΩ ∴ Rb = 7.3 kΩ
Figure 2.25a shows the designed filter, and its frequency response is shown in
Figure 2.25b.
EXAMPLE 2.16
Design a band-reject Butterworth filter having 3-dB points at 100 and 400 Hz and
greater than 30 dB of attenuation between 180 and 222 Hz with gain 1.
Sallen–Key Filters 65
R
LPF KR
−
+
vi R
HPF vo
Solution
Because the ratio of upper cutoff to lower is well in excess of an octave, a wide-
band approach can be used. First, separate the specification into individual low-pass
and high-pass requirements.
(a) Low-pass filter
a = 1.848, b = 1.000 ∴
1.848
C1n = = 0.924 F
2
2
C2 n = = 1.082 F
1.848
Rn = 1 Ω
Second stage
a = 0.765, b = 1.000
0.765
C1n = = 0.383 F
2
2
C2 n = = 2.614 F
0.765
Rn = 1 Ω
Denormalization
ω1
ISF = 10 4 , FSF = = 2 π f1 = 2 π × 10 2
ωn
66 Active Filters: Theory and Design
First stage
R = ISF × Rn = 10 4 × 1 Ω = 10 kΩ
C1n 0.924
C1 = = ≅ 147 nF
ISF × FSF 2 π × 10 6
C2 n 1.082
C2 = = ≅ 172 nF
ISF × FSF 2 π × 10 6
Second stage
R = ISF × Rn = 10 4 × 1 Ω = 10 kΩ
C1n 0.383
C1 = = ≅ 61 nF
ISF × FSF 2 π × 10 6
C2 n 2.614
C2 = = = 416 nF
ISF × FSF 2 π × 10 6
a = 1.848, b = 1.000
1.848
C1n = = 0.924 F
2
2
C2 n = = 1.082 F ∴
1.848
1 1
R1n = = = 1.082 Ω
C1n 0.924
1 1
R2 n = = = 0.924 Ω
C2 n 1.082
Cn = 1 F
Second stage
a = 0.765, b = 1.000
0.765
C1n = = 0.378 F
2
Sallen–Key Filters 67
172 n 416 n 10 K
10 K 10 K 10 K 10 K 10 K
+ LM741 + LM741 −
− − +
147 n 61 n
9.2 K 3.8 K
10 K
39.8 n 39.8 n 39.8 n 39.8 n
+ LM741 + LM741
− −
11 K 26 K
(a)
10.00
0.00
–10.00
Gain dB
–20.00
–30.00
–40.00
10 100 1K 10 K
Frequency in Hz
(b)
FIGURE 2.27 (a) Wide-band band-reject filter; (b) its frequency response.
2
C2 n = = 2.614 F ∴
0.765
1 1
R1n = = = 2.614 Ω
C1n 0.383
1 1
R2 n = = = 0.383 Ω
C2 n 2.614
Cn = 1 F
68 Active Filters: Theory and Design
Denormalization
ω2
ISF = 10 4 , FSF = = 2 π f2 = 2 π × 400 = 8 π × 10 2
ωn
Cn 1
C= = = 39.8 nF
ISF × FSF 8 π × 10 6
First stage
Second stage
Figure 2.27a shows the designed wide-band band-reject filter, and its frequency
response is shown in Figure 2.27b.
6. In the low-pass filter section, maximum gain peaking is very nearly equal
to Q at f1. So, as a rule of thumb:
(a) The op-amp bandwidth (BW) should be at least
BW = 100 × K × Q 3 × f1 (Q = 1 / a)
BW = 50 × f1
(b) For adequate full-power response, the slew rate (SR) of the op-amp must be
SR > π × V0 pp × BW f
EXAMPLE 2.17
A unity gain 20-kHz 5-pole Butterworth filter would require:
BW3 = 50 × 20 × 10 3 = 1 MHz
in which case trimming is generally required. In the case of the lower orders
(n ≤ 4), 10% tolerance capacitors are quite often satisfactory.
5. The gain of each stage of the filters is 1+ Rb / Ra , which can be adjusted to
the corrected value by using a potentiometer in lieu of resistors Ra and Rb.
This is accomplished by connecting the center tap of the potentiometer
to the inverting input of the op-amp. These gain adjustments are very
useful in tuning the overall response of the filter.
PROBLEMS
2.1 Design an “equal component” Butterworth band-pass filter having 3 dB
at 300 and 3000 Hz and attenuation greater than 30 dB between 50 and
18,000 Hz.
2.2 Design an “equal component” Chebyshev 2-dB band-reject filter at 250
and 1500 Hz and attenuation greater than 30 dB between 500 and 750 Hz.
2.3 For the high-pass filter:
a. Find the transfer function.
b. Find the formulas for the normalized filter, i.e., C = 1 F and w2 = 1 rad/s.
R2
C C
+
−
vi R1
Rb
vo
Ra
R C
−
+
vi R C vo
Rb
Ra
R2 C2
C1
+
−
R1
vi Rb vo
Ra
node v2
V2 ≅ 0 (3.3)
sC2
V1 = − V (3.4)
G3 0
V0 G1G3
H (s ) = =− 2 ∴
Vi s C1C2 + sC2 (G1 + G2 + G3 ) + G2G3
G1G3
C1C2
H (s) = − (3.5)
G + G2 + G3 GG
s2 + 1 s+ 2 3
C1 C1C2
73
74 Active Filters: Theory and Design
Y2
Y5
Y1 Y3
−
+
vi Y4
vo
For s = 0, we have:
G1 R
K = H (0) = − =− 2 (3.6)
G2 R1
K
H (s ) = − (3.7)
s 2 + as + b
where
G1 + G2 + G3
a= (3.8)
C1
and
G2G3
b = ω12 = (3.9)
C1C2
R2
C2
R1 v R3 v
1 2
−
+
vi C1 vo
If we set
R1 = R3 = R = 1 Ω (3.10)
R2 = K (3.11)
2 + G2
=a (3.12)
C1
G2
=b (3.13)
C1C2
1
2+
KR = a ∴
C1
2K + 1
C1 = (3.14)
aK
G2 1 1
C2 = = = ∴
bC1 bR2C1 bKC1
a
C2 = (3.15)
(2 K + 1)b
EXAMPLE 3.1
Design a second-order LP Butterworth filter with a gain of 10 at a cutoff frequency
of 1 kHz.
Solution
From the Butterworth coefficients in Appendix C, for n = 2, we find:
a = 1.414 b = 1.000
76 Active Filters: Theory and Design
Hence,
2K + 1 21
C1n = = = 1.485 F
aK 14.14
a 1.414
C2 n = = = 0.067 F
(2 K + 1)b 21
Denormalization
ω1
ISF = 10 4 , FSF = = 2π f1 = 2π × 10 3
ωn
C1n 1.485
C1 = = = 23.6 nF
ISF × FSF 2π × 10 7
C2 n 0.067
C2 = = = 1.1 nF
ISF × FSF 2π × 10 7
R = ISF × Rn = 10 4 × 1 Ω = 10 kΩ
R2 = KR = 10 × 10 4 Ω = 100 kΩ
Figure 3.3a shows the designed filter, and its frequency response is shown in
Figure 3.3b.
EXAMPLE 3.2
Design a second-order LP Chebyshev 3-dB filter with a gain of 1 and cutoff fre-
quency of 1 kHz.
Solution
From Chebyshev 3-dB coefficients in Appendix C, for n = 2, we find:
a = 0.645 b = 0.708
Hence,
Rn = 1 Ω, R2 n = KRn = 1 × 1 Ω = 1 Ω
2K + 1 3
C1n = = = 4.651 F
aK 0.645
a 0.645
C2 n = = = 0.304 F
(2 K + 1)b 3 × 0.708
MultiFeedback Filters 77
100 K
1.1 n
10 K 10 K
−
23.6 n + LM318
(a)
40.00
20.00
0.00
Gain dB
–20.00
–40.00
–60.00
10 100 1K 10 K 100 K
Frequency in Hz
(b)
FIGURE 3.3 (a) Second-order LP Butterworth filter, f1 = 1 kHz, K = 10; (b) its frequency
response.
Denormalization
ISF = 10 4
ω1 2π f1
FSF = = = 2π × 10 3
ωn 1
R = ISF × Rn = 10 4 × 1 Ω = 10 kΩ
R2 = ISF × R2 n = 10 4 × 1 Ω = 10 kΩ
C1n 4.651
C1 = = = 74 nF
ISF × FSF 2π × 10 7
C2 n 0.304
C2 = = = 4.8 nF
ISF × FSF 2π × 10 7
78 Active Filters: Theory and Design
10 K
4.8 n
10 K 10 K
−
74 n + LM741
(a)
10.00
0.00
–10.00
Gain dB
–20.00
–30.00
–40.00
10 100 1K 10 K
Frequency in Hz
(b)
Figure 3.4a shows the designed filter, with its frequency response in Figure 3.4b
[From mCap III].
C1 = C3 = C = 1 F (3.16)
MultiFeedback Filters 79
R2 C2
C2 1/K
R2
R1 R3 C1 C3 1/C2
− −
1Ω 1Ω + 1F 1F +
vi C1 vi R1 1/C1
vo vo
(a) (b)
1
ωC2 C1
K= = ∴
1 C2
ωC1
C1 1
C2 = = (3.17)
K K
EXAMPLE 3.3
Design a second-order HP Butterworth filter with a gain of 5 and a cutoff frequency
of 100 Hz.
Solution
(a) Low-pass filter
From Butterworth coefficients of Appendix C, for n = 2, we have:
a = 1.414 b = 1.000
Hence:
2K + 1 11
C1n = = = 1.556 F
aK 1.414 × 5
a 1.414
C2 n = = = 0.129 F
(2 K + 1)b 11 × 1
80 Active Filters: Theory and Design
1 1
R1n = = = 0.647 Ω
C1n 1.556
1 1
R2 n = = = 7.752 Ω
C2 n 0.129
C1n = C3n = Cn = 1 F
Cn 1
C2 n = = = 0.2 F
K 5
Denormalization
ISF = 10 4
ω2
FSF = = 2π f2 = 2π × 10 2 ∴
ωn
Cn 1
C1 = C3 = = = 159.2 nF
ISF × FSF 2π × 10 6
C2 n 0.2
C2 = = = 31.8 nF
ISF × FSF 2π × 10 6
Figure 3.6a shows the designed filter, and its frequency response is shown in
Figure 3.6b.
EXAMPLE 3.4
Design a second-order HP Chebyshev 3-dB filter with a gain of 10 and a cutoff
frequency of 100 Hz.
Solution
From Chebyshev 3-dB coefficients of Appendix C, for n = 2, we have:
a = 0.645 b = 0.708
MultiFeedback Filters 81
C2
32 nf R2
C1 C3 77.5 K
−
159 nf 159 nf +
vi R1 6.4 K vo
(a)
20.00
10.00
0.00
Gain dB
–10.00
–20.00
–30.00
10 100 K 1K
Frequency in Hz
(b)
FIGURE 3.6 (a) Second-order HP Butterworth filter, f2 = 100 Hz, K = 5; (b) frequency
response.
2K + 1 21
C1n = = = 3.256 F
aK 0.645 × 10
a 0.645
C2 n = = = 0.043 F
(2 K + 1)b 21 × 0.708
1 1
R1n = = = 0.307 Ω
C1n 3.256
1 1
R2 n = = = 23.051 Ω
C2 n 0.043
82 Active Filters: Theory and Design
C2
15.9 nf R2
C1 C3 231 K
−
159 nf 159 nf +
vi R1 3.1 K vo
(a)
30.00
20.00
10.00
Gain dB
0.00
–10.00
–20.00
10 100 1K 10 K
Frequency in Hz
(b)
FIGURE 3.7 (a) Second-order HP Chebyshev 3 dB, f2 = 100 Hz, K = 10; (b) frequency
response.
C1n = C3n = Cn = 1 F
C2 n 1
C2 n = = = 0.1 F
K 10
Denormalization
ω2
ISF = 10 4 , FSF = = 2π f2 = 2π × 10 2 ∴
ωn
Cn 1
C1 = C3 = C = = = 159.2 nF
ISF × FSF 2π × 10 6
MultiFeedback Filters 83
C2 n 0.1
C2 = = = 15.9 nF
ISF × FSF 2π × 10 6
R1 = ISF × R1n = 10 4 × 0.307 Ω ≅ 3.1 kΩ
Figure 3.7a shows the designed filter, and its frequency response is shown in Figure 3.7b.
EXAMPLE 3.5
Design an LP Butterworth filter 3 dB at 1 kHz and attenuation 35 dB at 4 kHz with
a gain of 5.
Solution
From Butterworth nomographs for Amax = 3 dB, Amin = 35 dB, and fs / f1 = 4, we find
n = 3, hence n = 4 (only even number).
K1 = K 2 = K = 5 = 2.236
a = 1.848 b = 1.000
R1n = R3n = 1 Ω
2 K + 1 2 × 2.236 + 1 5.472
C1n = = = = 1.324 F
aK 1.848 × 2.236 4.132
a 1.848 1.848
C2 n = = = = 0.338 F
(2 K + 1)b (2 × 2.236 + 1) × 1 5.472
Second stage
a = 0.765 b = 1.000
R1n = R3n = 1 Ω
84 Active Filters: Theory and Design
2 K + 1 2 × 2.236 + 1 5.472
C1n = = = = 3.199 F
aK 0.765 × 2.236 1.711
a 0.765 0.765
C2 n = = = = 0.140 F
(2 K + 1)b (2 × 2.236 + 1) 5.472
Denormalization
ISF = 10 4
ω1
FSF = = 2π f1 = 2π × 10 3
ωn
R1 = R3 = ISF × Rn = 10 4 × 1 Ω = 10 kΩ
First stage
C1n 1.324
C1 = = = 21.1 nF
ISF × FSF 2π × 10 7
C2 n 0.388
C2 = = = 5.4 nF
ISF × FSF 2π × 10 7
Second stage
C1n 3.199
C1 = = = 50.9 nF
ISF × FSF 2π × 10 7
C2 n 0.140
C2 = = = 2.2 nF
ISF × FSF 2π × 10 7
Figure 3.8a shows the designed filter, and its frequency response is shown in Figure 3.8b.
EXAMPLE 3.6
Design an HP Butterworth filter with the following specifications: 3 dB at 100 Hz
and attenuation 30 dB at 50 Hz, with a gain of 5.
Solution
From the Butterworth nomographs for Amax = 3 dB, Amin = 30 dB, and for
f2 / fs 2 = 100 /50 = 2 we find n = 6.
K1 = K 2 = K3 = K 1/ 3 = 51/ 3 = 1.71
MultiFeedback Filters 85
R2 R2
22.4 K C 22.4 K C
2 2
R1 R3 5.4 nf 2.2 nf
− R1 R3
10 K 10 K −
+ 10 K 10 K +
vi C1 21 nf C1 51 nf vo
(a)
20.00
0.00
–20.00
Gain dB
–40.00
–60.00
–80.00
100 1K 10 K
Frequency in Hz
(b)
a = 1.932 b = 1.000 ∴
2 K + 1 2 × 1.71 + 1 4.420
C1n = = = = 1.338 F
aK 1.932 × 1.71 3.304
a 1.932
C2 n = = = 0.437 F
(2 K + 1)b 4.420
86 Active Filters: Theory and Design
Second stage
a = 1.414 b = 1.000
2K + 1 4.420 4.4420
C1n = = = = 1.828 F
aK 1.414 × 1.71 2.418
a 1.414
C2 n = = = 0.320 F
(2 K + 1)b 4.420
Third stage
a = 0.518 b = 1.000
2K + 1 4.420
C1n = = = 4.999 F
aK 0.518 × 1.71
a 0.518
C2 n = = = 0.117 F
(2 K + 1)b 4.420
1 1
R1n = = = 0.747 Ω
C1n 1.338
1 1
R2 n = = = 2.288 Ω
C2 n 0.437
Second stage
1 1
R1n = = = 0.547 Ω
C1n 1.828
1 1
R2 n = = = 3.125 Ω
C2 n 0.320
Third stage
1 1
R1n = = = 0.200 Ω
C1n 4.99
1 1
R2 n = = = 8.547 Ω
C2 n 0.177
C1n = C3n = Cn = 1 F
Cn 1
C2 n = = = 0.585 F
K 1.71
MultiFeedback Filters 87
Denormalization
ω2
ISF = 10 4 , FSF = = 2πf2 = 2π × 10 2
ωn
Cn 1
C1 = C3 = = ≅ 159 nF
ISF × FSF 2π × 10 6
C2 n 0.585
C2 = = = 93 nF
ISF × FSF 2π × 10 6
First stage
Second stage
Third stage
Figure 3.9a shows the designer filter, and its frequency response is shown in Figure 3.9b.
EXAMPLE 3.7
Design an HP Chebyshev with the following specifications: 2 dB at 500 Hz and
attenuation 40 dB at 200 Hz, with a gain of 1.
Solution
From the Chebyshev nomographs for Amax = 2 dB, Amin = 40 dB, and f2 / fs 2 =
500 / 200 = 2.5, we find n = 4.
(a) Low-pass filter
First stage
From Chebyshev 2-dB coefficients, we find:
a = 0.506 b = 0.222 ∴
2K + 1 3
C1n = = = 5.926 F
aK 0.506
a 0.506
C2 n = = = 0.760 F
(2 K + 1)b 3 × 0.222
88 Active Filters: Theory and Design
93 n 93 n 93 n
23 K 31 K 85.5 K
(a)
40.00
20.00
0.00
Gain dB
–20.00
–40.00
–60.00
10 100 1K 10 K
Frequency in Hz
(b)
FIGURE 3.9 (a) Sixth-order HP Butterworth filter, f2 = 100 Hz, K = 5; (b) frequency response.
Second stage
a = 0.210 b = 0.929 ∴
2K + 1 3
C1n = = = 14.286 F
aK 0.210
a 0.210
C2 n = = = 0.075 F
(2 K + 1)b 3 × 0.929
1 1
R2 n = = = 1.316 Ω
C2 n 0.760
MultiFeedback Filters 89
Second stage
1 1
R1n = = = 0.070 Ω
C1n 14.286
1 1
R2 n = = = 13.271 Ω
C2 n 0.075
C1n = C3n = Cn = 1 F
Cn 1
C2 n = = =1 F
K 1
31.8 n 31.8 n
13.2 K 132.7 K
(a)
10.00
0.00
–10.00
Gain dB
–20.00
–30.00
–40.00
100 1K 10 K 100 K
Frequency in Hz
(b)
FIGURE 3.10 (a) Fourth-order HP Chebyshev 2 dB, f2 = 500 Hz, K = 1; (b) frequency
response.
90 Active Filters: Theory and Design
Denormalization
ω2
ISF = 10 4 , FSF = = 2πf2 = 2π × 500 = π × 10 3
ωn
First stage
Second stage
Cn 1
C1 = C2 = C3 = = = 31.8 nF
ISF × FSF π × 10 7
Figure 3.10a shows the designed filter, and its frequency response is shown in
Figure 3.10b.
EXAMPLE 3.8
Design a Butterworth band-pass filter with the following specifications: 3 dB at 200
to 800 Hz and attenuation 20 dB below 50 Hz and above 3200 Hz, with gain 1.
Solution
Because the ratio of upper cutoff frequency to lower cutoff frequency is well in
excess of an octave, the design will be treated as a cascade of low-pass and high-
pass filters. The frequency response requirements can be restated as the following
set of individual low-pass and high-pass specifications:
a = 1.414 b = 1.000
Hence,
R1n 1
R1n = 1 Ω R2 n = = =1 Ω
K 1
2K + 1 3
C1n = = = 2.122 F
aK 1.414
a 1.414
C2 n = = = 0.471 F
(2 K + 1)b 3
Denormalization
ω1
ISF = 10 4 , FSF = = 2 π f1 = 2 π × 800 = 1.6 π × 10 3
ωn
R1 = R2 = ISF × R1n = 10 4 × 1 Ω = 10 kΩ
C1n 2.122
C1 = = = 42.2 nF
ISF × FSF 1.6π × 10 7
C2 n 0.471
C2 = = = 9.4 nF
ISF × FSF 1.6π × 10 7
1 1
R1n = = = 0.471 Ω
C1n 2.122
1 1
R2 n = = = 2.122 Ω
C2 n 0.471
C1n = C2 n = C3n = 1 F
Denormalization
ω2
ISF = 10 4 , FSF = = 2π f2 = 2 π × 200 = 400 π
ωn
Cn 1
C1 = C2 = C3 = C = = = 79.6 nF
ISF × FSF 4 π × 10 6
92 Active Filters: Theory and Design
79.6 n 10 K
21.2 K 9.4 n
79.6 n 79.6 n 10 K 10 K
− −
+ LM741 + LM741
4.7 K 42 n
(a)
10.00
0.00
–10.00
Gain dB
–20.00
–30.00
–40.00
10 100 1K 10 K
Frequency in Hz
(a)
FIGURE 3.11 (a) Wide-band band-pass Butterworth filter 200 to 800 Hz, K = 1; (b) frequency
response.
Figure 3.11a shows the designed filter, and its frequency response is shown in
Figure 3.11b.
C2
R2
R1 v C1
1 v2
−
+
vi R3
vo
node v2
V2 ≅ 0 (3.20)
G2
V1 = − V (3.21)
sC1 0
⎧⎪ G2 [G1 + G3 + s(C1 + C2 )] ⎫⎪
⎨ + sC2 ⎬ V0 = − G1Vi ∴
⎪⎩ sC1 ⎪⎭
{ [G + G
1 3 + s(C1 + C2 )] G2 + s 2C1C2 }V0 = − sC1G1Vi ∴
V0 sC1G1
H (s ) = =− 2 (3.22)
Vi s C1C2 + s(C1 + C2 )G2 + (G1 + G3 )G2
For C1 = C2 = C , we have:
sCG1
H (s ) = −
s 2C 2 + 2sCG2 + (G1 + G3 )G2
G1 (3.23)
s
H (s ) = − C
2G (G + G )G
s2 + 2 s + 1 2 3 2
C C
94 Active Filters: Theory and Design
Hence,
G1
s
H (s ) = − 2 C (3.24)
s + as + ω 20
where
(G1 + G3 )G2
ω 20 = (3.25)
C2
and
1 2G2 2
a= = = (3.26)
Q C R2C
For s = jω 0 ∴
G1
jω 0 QG1
K = H ( jω 0 ) = − C =− ∴
1 C
−ω 20 + j ω 0 + ω 20
Q
C G1 R
K=− ⋅ =− 2 (3.27)
2G2 C 2 R1
R2 = 2Q (3.28)
Q
R1 = (3.29)
K
(G1 + G3 )G2 = 1 ∴
1 1
+ = R2 ∴
R1 R3
MultiFeedback Filters 95
1 1
= R2 − (3.30)
R3 R1
1 K
= 2Q − ∴
R3 Q
1 2Q 2 − K
= ∴
R3 Q
Q
R3 = (3.31)
2Q 2 − K
2Q 2 > K (3.32)
K = 2Q 2 (3.33)
hence
G1 G1 ⎛ s ⎞
s ⎜ ⎟
C ω 0C ⎝ ω 0 ⎠
H (s ) = − =− 2
∴
2G
s 2 + 2 s + ω 20 ⎛ s ⎞ 2G2 ⎛ s ⎞
C ⎜ ⎟ + ⎜ ⎟ +1
⎝ ω0 ⎠ ω 0C ⎝ ω 0 ⎠
⎛ s ⎞
K0 ⎜
⎝ ω 0 ⎟⎠
H (s ) = − 2
(3.35)
⎛ s ⎞ 1⎛ s ⎞
⎜⎝ ω ⎟⎠ + Q ⎜⎝ ω ⎟⎠ + 1
0 0
96 Active Filters: Theory and Design
s
(1) For << 1 ∴
ω0
⎛ s ⎞ ⎛ ω⎞
H (s ) ≅ − K 0 ⎜ ∴ H ( jω 0 ) = − K 0 ⎜ j ∴
⎝ ω 0 ⎟⎠ ⎝ ω 0 ⎟⎠
⎛ ω⎞
A = 20 log H ( jω 0 ) = 20 log K 0 ⎜ ∴
⎝ ω 0 ⎟⎠
⎛ ω⎞
A = 20 log K 0 + 20 log ⎜
⎝ ω 0 ⎟⎠
Hence, we have:
slope = 20 dB/dec
s
(2) For >> 1 ∴
ω0
⎛ s ⎞
K0 ⎜
⎝ ω 0 ⎟⎠
−1
⎛ s ⎞
H (s ) ≅ − = − K0 ⎜ ∴
⎛ s ⎞
2
⎝ ω 0 ⎟⎠
⎜⎝ ω ⎟⎠
0
−1
⎛ ω⎞
H ( jω ) = − K 0 ⎜ j ∴
⎝ ω 0 ⎟⎠
−1
⎛ ω⎞
A = 20 log H ( jω ) = 20 log K 0 ⎜
⎝ ω 0 ⎟⎠
⎛ ω⎞
A = 20 log K 0 − 20 log ⎜
⎝ ω 0 ⎟⎠
s
(3) For =1 ∴
ω0
jK 0
H ( jω ) = − = − QK 0 = K ∴
1
−1 + j + 1
Q
K
K0 = (3.36)
Q
A
dB
20 log K
20 log K0
K
20 log
Q Slope = 20 dB/dec
Slope = –20 dB/dec
0
0.1 f0 f0 10 f0 log f
EXAMPLE 3.9
Design a narrow-band band-pass filter with f0 = 1 kHz, Q = 7, and K = 10.
Solution
Q 7
R1n = = = 0.7 Ω
K 10
R2 n = 2Q = 2 × 7 = 14 Ω
Q 7 7
R3n = = = = 0.08 Ω
2Q − K 2 × 7 − 10 98 − 10
2 2
Denormalization
ω
ISF = 10 4 , FSF = = 2π f0 = 2π × 10 3
ω0
Cn 1
C= = = 15.92 nF
ISF × FSF 2π × 10 7
R2 = ISF × R2 n = 10 4 × 14 Ω = 140 kΩ
C2
16 nf
R2
R1 C1 140 K
−
7K 16 nf +
vi R3 0.8 K vo
(a)
A
dB
20
10
3.0
0
10 100 1K 10 K f (Hz)
–10
–17
–20
(b)
FIGURE 3.14 (a) Band-pass filter, f0 = 1 kHz, Q = 7, K = 10; (b) frequency response.
Figure 3.14a shows the designed filter, and its frequency response is shown in
Figure 3.14b.
A much sharper band-pass filter may be obtained by cascading two or more
identical band-pass second-order filters. If Q1 is the quality factor of a single stage
and there are n stages, the Q of the filter is expressed by the following equation:
Q1
Q= (3.37)
n
2 −1
EXAMPLE 3.10
Design a narrow-band band-pass filter with f0 = 1 kHz, Q = 10, and with gain as a
free parameter.
MultiFeedback Filters 99
TABLE 3.1
Bandwidth (BW) and Q for identical cascaded second-order
band-pass filter
n Q BW
1 Q1 BW1
2 1.554Q1 0.664BW1
3 1.961Q1 0.510BW1
4 2.299Q1 0.435BW1
5 2.593Q1 0.386BW1
6 2.858Q1 0.350BW1
7 3.100Q1 0.323BW1
Solution
K = 2Q 2 = 2 × 10 2 = 200 or 46 dB
Q 10
R1n = = = 0.05 Ω
K 200
R2 n = 2Q = 2 × 10 = 20 Ω
Denormalization
ω0
ISF = 10 4 , FSF = = 2 π f0 = 2 π × 10 3
ωn
Hence,
Cn 1
C= = = 15.9 nF
ISF × FSF 2π × 10 7
R1 = ISF × R1n = 10 4 × 0.05 Ω = 0.5 kΩ
R2 = ISF × R2 n = 10 4 × 20 Ω = 200 kΩ
Figure 3.15a shows the designed filter, and its frequency response is shown in Fig-
ure 3.15b.
EXAMPLE 3.11
Design a sixth-order (three-pole) narrow-band band-pass filter with f0 = 750 Hz,
Q = 8.53, and K = 6.
100 Active Filters: Theory and Design
15.9 n
200 K
0.5 K 15.9 n
−
+ LM741
(a)
60.00
40.00
20.00
Gain dB
0.00
–20.00
–40.00
10 100 1K 10 K 100 K
Frequency in Hz
(b)
FIGURE 3.15 (a) Band-pass filter, f0 = 1 kHz, Q = 10, K: free parameter; (b) frequency
response.
Solution
We will use three identical stages.
K1 = K 1/ 3 = 61/ 3 = 1.82
Q1 4.35
R1n = = = 2.39 Ω
K1 1.82
R2 n = 2Q = 2 × 4.35 Ω = 8.7 Ω
Denormalization
ω0
ISF = 10 4 , FSF = = 2 π f0 = 2 π × 750 = 1.5π × 10 3
ωn
Cn 1
C= = = 21.2 nF
ISF × FSF 1.5π × 10 7
R1 = ISF × R1n = 10 4 × 2.39 Ω = 23.9 kΩ
R2 = ISF × R2 n = 10 4 × 8.7 Ω = 87 kΩ
Figure 3.16a shows the designed filter, and its frequency response is shown in
Figure 3.16b.
87 K 87 K 87 K
(a)
40.00
0.00
–40.00
Gain dB
–80.00
–120.00
–160.00
10 100 1K 10 K 100 K
Frequency in Hz
(b)
FIGURE 3.16 (a) Sixth-order band-pass filter, f0 = 750 Hz, Q = 8.53, K = 6; (b) frequency
response.
102 Active Filters: Theory and Design
node v2
V0
V01 = (3.40)
K
R4
K= (3.41)
R1
V2 ≅ 0 (3.42)
G
V1 = V (3.43)
sCK 0
R2
R
R4
R v1 C v2
− v01 R3
+ −
vi R1 +
vo
G sC
(G + G1 + G2 + 2sC ) V + V − G2V0 = GVi ∴
sCK 0 K 0
V0 sCGK
H (s ) = = 2 2 ∴
Vi s C + (2G − KG2 )sC + G (G + G1 + G2 )
KG
s
H (s ) = C ∴
2G − KG2 G (G + G1 + G2 )
s2 + s +
C C2
KG
s
H (s ) = C ∴
2G − KG
s2 + 2
s + ω 20
C
where
G (G + G1 + G2 )
ω 20 = (3.44)
C2
Hence,
KG ⎛ s ⎞
ω 0C ⎜⎝ ω 0 ⎟⎠
H (s ) = 2
∴
⎛ s ⎞ 2G − KG2 ⎛ s ⎞
⎜⎝ ω ⎟⎠ + ω C ⎜⎝ ω ⎟⎠ + 1
0 0 0
KG ⎛ s ⎞
ω 0C ⎜⎝ ω 0 ⎟⎠
H (s ) = 2
(3.45)
⎛ s ⎞ 1⎛ s ⎞
⎜⎝ ω ⎟⎠ + Q ⎜⎝ ω ⎟⎠ + 1
0 0
where
1 2G − KG2
a= = (3.46)
Q ω 0C
For s = jω 0 , we have:
KG
j
ω 0C KQG KG ω 0C
K 0 = H ( jω 0 ) = = = ⋅ ∴
1 ω 0C ω 0C 2G − KG2
−1 + j + 1
Q
104 Active Filters: Theory and Design
KG
K0 = (3.47)
2G − KG2
ω 0C
KG2 = 2G − ∴
Q
K 2 ω 0C
= − ∴
R2 R Q
KQR
R2 = (3.48)
2Q − ω 0 RC
We put
ω 0 RC = Q (3.49)
KQR KQR
R2 = = ∴
2Q − Q ⎛ 2Q ⎞
Q⎜ − 1⎟
⎜⎝ Q ⎟⎠
K QR
R2 = (3.50)
2 Q −1
1⎛1 1 1⎞
(ω 0C )2 = ⎜ + + ⎟ ∴
R ⎝ R R1 R2 ⎠
1 1 1
(ω 0C )2 R = + + ∴
R R1 R2
1 ω 20 R 2C 2 − 1 1
= − ∴
R1 R R2
1 ω 20 R 2C 2 − 1 2 Q − 1
= − ∴
R1 R KR Q
1 1⎛ 2 1 ⎞
= ⎜ Q − 1− + ⎟ (3.51)
R1 R ⎝ K K Q⎠
MultiFeedback Filters 105
Cn = 1 F, R3n = 1 Ω (3.52)
R= Q (3.53)
1 1⎛ 2 1 ⎞
= ⎜ Q − 1− + ⎟ (3.54)
R1 R ⎝ K K Q⎠
KQ
R2 = (3.55)
2 Q −1
EXAMPLE 3.12
Design a second-order narrow-band BPF with two op-amps with a center frequency
of 1 kHz, Q = 40, and K = 5.
Solution
Rn = Q = 40 = 6.325 Ω
R3n = 1 Ω
R4 n = K = 5 Ω
5 40 × 6.325 200
R2 n = = = 17.2 Ω
2 × 6.325 − 1 11.65
1 1 ⎛ 2 1 ⎞
= 40 − 1 − − = 0.164 Ω
R1n 6.325 ⎜⎝ 5 5 × 6.325 ⎟⎠
Denormalization
ω0
ISF = 10 4 , FSF = = 2π f0 = 2π × 10 3
ωn
R3 = ISF × R3n = 10 4 × 1 Ω = 10 kΩ
R4 = ISF × R4 n = 10 4 × 5 Ω = 50 kΩ
Ω
K K 5
K0 = = = = 31.3 or 29.9 dB
G R 63.3
2− K 2 2− K 2−5
G R2 172
1
C= = 15.92 nF
2π × 10 7
Figure 3.18 shows the designed filter with its frequency response.
172 K
15.9 n
63.3 K 50 K
63.3 K 15.9 n 10 K
− −
+ +
1.64 K LM741 LM741
(a)
40.00
20.00
0.00
Gain dB
–20.00
–40.00
–60.00
10 100 1K 10 K 100 K
Frequency in Hz
(b)
FIGURE 3.18 (a) Second-order narrow-band BPF, f0 = 1 kHz, Q = 40, K = 5; (b) frequency
response.
MultiFeedback Filters 107
1 ⎛1 ⎞
− V+ + 2sC ⎟ V1 − sCV2 − sCV0 = 0 (3.56)
R i ⎜⎝ R ⎠
node v2
⎛ 1 ⎞ 1
− sCV1 + ⎜ + sC ⎟ V2 − V0 = 0 (3.57)
⎝ kR ⎠ kR
node v3
Ra
V3 = V = β V0 (3.58)
Ra + Rb 0
where
Ra r 1
β= = = (3.59)
Ra + Rb r + Mr 1 + M
V3 − V2 = 0 ∴
V2 = V3 = β V0 (3.60)
kR
R v1 C v2
−
+
Rb
vi v3 vo
Mr
Ra r
⎛ 1 ⎞ 1
sCV1 = ⎜ + sC ⎟ βV0 − V0 ∴
⎝ kR ⎠ kR
(1 + skRC )β − 1
V1 = V0 ∴
skRC
(1 − β) − βkRCs
V1 = − V0 (3.61)
skRC
⎡ (1 − β) − βkRCs ⎤
(1 + 2sRC ) ⎢ − ⎥ V0 − sRCβV0 − sRCV0 = Vi ∴
⎣ skRC ⎦
V0 skRC
H (s ) = = ∴
Vi s 2 R 2C 2k (1 − β) + RC [2(1 − β) − k β] s + (1 − β)
s
−
(1 − β) RC
H (s) = ∴
2(1 − β) − β k 1
s +
2 s+ 2 2
(1 − β)kRC kR C
s
−
(1 − β) RC
H (s) = (3.62)
2(1 − β) − βk
s2 + s + ω 20
(1 − β)kRC
where
1
ω 20 = ∴
kR 2C 2
1
ω0 = (3.63)
RC k
MultiFeedback Filters 109
Hence
1 ⎛ s ⎞
−
(1 − β)ω 0 RC ⎜⎝ ω 0 ⎟⎠
H (s ) = 2
∴
⎛ s ⎞ 2(1 − β) − βk ⎛ s ⎞
⎜⎝ ω ⎟⎠ + (1 − β)kRCω ⎜ ω ⎟ + 1
0 0 ⎝ 0⎠
1 ⎛ s ⎞
(1 − β) ω 0 RC ⎜⎝ ω 0 ⎟⎠
H (s ) = − 2
(3.64)
⎛ s ⎞ 1⎛ s ⎞
⎜⎝ ω ⎟⎠ + Q ⎝⎜ ω ⎟⎠ + 1
0 0
where
(1 − β)kRC ω 0
Q= (3.65)
2(1 − β) − β k
2
β max = (3.66)
2+k
1
β= (3.67)
1+ M
⎛ 1 ⎞
⎜1 − 1 + M ⎟ k
⎝ ⎠
Q= ∴
⎛ 1 ⎞ 1
2 ⎜1 − ⎟− k
⎝ 1+ M ⎠ 1+ M
M k
Q= ∴
2M − k
kQ
M= (3.68)
2Q − k
2(1 − β)Q − β kQ = (1 − β) k ∴
2Q − k
β= (3.69)
2Q + kQ − k
1 1 1
− j
(1 − β)ω 0 RC (1 − β) ω 0 RC (1 − β) ω 0 RC
K = H ( jω 0 ) = =− =− ∴
1 1 2(1 − β) − β k
−1 + j + 1
Q Q (1 − β)kRC ω 0
k
K=− (3.70)
2(1 − β) − β k
1 R3
= (3.71)
K R1 + R2
kR
R1 = R C
−
+
Rb
vi R3 vo
Mr
Ra r
FIGURE 3.20 Deliyannis’s band-pass filter with attenuator input R1, R3.
MultiFeedback Filters 111
R1R3
R= (3.72)
R1 + R3
R1 = KR (3.73)
KR1
R3 = (3.74)
K −1
1
R= (3.75)
k
k k
K= = ∴
2(1 − β) − β k ⎛ 1 ⎞ k
2 ⎜1 − ⎟ −
⎝ 1+ M ⎠ 1+ M
k (1 + M )
K= ∴
2M − k
⎛ kQ ⎞ 2Q − k + kQ
k ⎜1 + ⎟ k
⎝ 2Q − k ⎠ 2Q − k
K= = ∴
2kQ
−k 2kQ − 2kQ − k k
2Q − k 2Q − k
⎛ kQ ⎞
(2Q − k ) ⎜1 + ⎟
2Q − k + kQ ⎝ 2Q − k ⎠
K= = ∴
k k
112 Active Filters: Theory and Design
Q(1 + M ) k
K= (3.76)
M
EXAMPLE 3.13
Design a filter with f0 = 200 Hz and Q = 12.
Solution
We accept k = 25 ( k = 5).
1 1
Rn = = = 0.200 Ω
k 5
kQ 25 × 12 300
M= = = = 15.8
2Q − k 24 − 5 19
Q(1 + M ) k 12 × 16.8 × 5
K= = = 63.8 or 36 dB
M 15.8
R1n = KRn = 63.8 × 0.200 = 12.8 Ω
R2 n = k Rn = 25 × 0.200 = 5 Ω
Denormalization
ω 0 2 π f0
ISF = 10 4 , FSF = = = 2π × 200 = 4 π × 10 2 ∴
ωn 1
Cn 1
C= = = 79.6 nF
ISF × FSF 4 π × 10 6
R = ISF × Rn = 10 4 × 0.2 Ω = 2 kΩ
R2 = ISF × R2 n = 10 4 × 5 Ω = 50 kΩ
MultiFeedback Filters 113
79.6 n
50 K
128 K 79.6 n
−
2K + LM741
158 K
10 K
(a)
0.00
–10.00
–20.00
Gain DB
–30.00
–40.00
–50.00
10 100 1K 10 K
Frequency in Hz
(b)
FIGURE 3.21 (a) Deliyannis’s band-pass filter, f0 = 200 Hz, Q = 12, K = 1; (b) frequency
response.
r = ISF × rn = 10 4 × 1 Ω = 10 kΩ
Figure 3.21a shows the designed filter for K = 1, and its frequency response is shown
in Figure 3.21b.
Figure 3.22a shows the designed filter with K as a free parameter, and its
frequency response is shown in Figure 3.22b.
114 Active Filters: Theory and Design
79.6 n
50 K
2K 79.6 n
LM741
−
+
158 K
10 K
(a)
60.00
40.00
20.00
Gain dB
0.00
–20.00
–40.00
10 100 1K 10 K
Frequency in Hz
(b)
FIGURE 3.22 (a) Deliyannis’s band-pass filter, f0 = 200 Hz, Q = 12, K: free parameter;
(b) frequency response.
R
LPF KR
−
+
vi
R vo
HPF
EXAMPLE 3.14
Design a wide-band band-reject Butterworth filter having 3 dB at 300 and 900 Hz
and greater than 40 dB of attenuation between 750 and 360 Hz, respectively, with
a gain of 1.
Solution
As the ratio of upper cutoff to lower cutoff is well in excess of an octave, a wide-
band approach can be used. First, separate the specification into individual low-pass
and high-pass requirements.
Low-pass High-pass
3 dB at 300 Hz 3 dB at 900 Hz
40 dB at 750 Hz 40 dB at 360 Hz
a = 1.932 b = 1.000
R1n = R2 n = R3n = 1 Ω ( K = 1)
2K + 1 3
C1n = = = 1.553 F
aK 1.932
a 1.932
C2 n = = = 0.644 F
(2 K + 1)b 3
Second stage
a = 1.414 b = 1.000
3
C1n = = 2.122 F
1.414
1.414
C2 n = = 0.471 F
3
116 Active Filters: Theory and Design
Third stage
a = 0.518 b = 1.000
3
C1n = = 5.792 F
0.518
0.518
C2 n = = 0.173 F
3
Denormalization
ω1
ISF = 10 4 , FSF = = 2π f1 = 2π × 300 = 6π × 10 2
ωn
R1 = R2 = R3 = ISF × R1n = 10 4 × 1 Ω = 10 kΩ
First stage
C1n 1.553
C1 = = = 82.4 nF
ISF × FSF 6π × 10 6
C2 n 0.644
C2 = = = 34.2 nF
ISF × FSF 6π × 10 6
Second stage
C1n 2.122
C1 = = = 112.6 nF
ISF × FSF 6π × 10 6
C2 n 0.471
C2 = = = 25 nF
ISF × FSF 6π × 10 6
Third stage
C1n 5.792
C1 = = = 307.3 nF
ISF × FSF 6π × 10 6
C2 n 0.173
C2 = = = 9.2 nF
ISF × FSF 6π × 10 6
(b) High-pass filter
From the Butterworth nomographs for Amax = 3 dB, Amin = 40 dB, and f2 / fs =
900 / 360 = 2.5, we have n = 5; hence, n = 6.
C1n = C2 n = C3n = 1 F ( K = 1)
First stage
1 1
R1n = = = 0.644 Ω
C1n 1.553
1 1
R2 n = = = 1.553 Ω
C2 n 0.644
Second stage
1 1
R1n = = = 0.471 Ω
C1n 2.122
1 1
R2 n = = = 2.122 Ω
C2 n 0.471
Third stage
1 1
R1n = = = 0.173 Ω
C1n 5.792
1 1
R2 n = = = 5.792 Ω
C2 n 0.173
Denormalization
ω2
ISF = 10 4 , FSF = = 2 π f2 = 2 π × 900 = 1.8 π × 10 3
ωn
Cn 1
C1 = C2 = C3 = = = 17.7 nF
ISF × FSF 1.8π × 10 7
First stage
Second stage
10 K 10 K 10 K 10 K
34.2 n 25 n 9.2 n
10 K 10 K 10 K 10 K 10 K
10 K 10 K
− − − −
+ + + +
82.4 n LM741 112.6 n LM741 307.3 n LM741 LM741
18 n 18 n 18 n
15.5 K 21.2 K 58 K
18 n 18 n 18 n 18 n 18 n 18 n 10 K
− − −
+ + +
6.4 K LM741 4.7 K LM741 1.7 K LM741
(a)
FIGURE 3.24 (a) Wide-band band-reject filter, 300 to 900 Hz, K = 1; (b) frequency response.
Active Filters: Theory and Design
MultiFeedback Filters 119
10.00
0.00
–10.00
Gain dB
–20.00
–30.00
–40.00
10 100 1K 10 K
Frequency in Hz
(b)
Third stage
Figure 3.24a shows the designed filter, and its frequency response is shown in Figure
3.24b.
1 ⎛ 1 ⎞
Vi + ⎜ + sC1 + sC2 ⎟ V1 − sC2V2 − sC1V0 = 0 (3.77)
R1 ⎝ R1 ⎠
node v2
⎛ 1 ⎞ 1
− sC2 + ⎜ + sC2 ⎟ V2 − V0 = 0 (3.78)
⎝ 2
R ⎠ R2
120 Active Filters: Theory and Design
C1
R2
R1 v C2
1 v2
−
+
R3 v3
vi
vi
R4
node v3
R4
V3 = V = bVi (3.79)
R3 + R4 i
where
R4
b= (3.80)
R3 + R4
V0 + sR2C2V1
V2 = (3.81)
1 + sR2C2
V2 = V3 = bVi (3.82)
⎡ bV (1 + sR2C2 ) V0 ⎤
(1 + sR1C1 + sR2C2 ) ⎢ i − ⎥ − sR1C2 bVi − sR1C1V0 = Vi ∴
⎣ sR2C2 sR2C2 ⎦
If we put
1
ω 20 = (3.84)
R1R2C1C2
and
1
Q= (3.85)
ω 0 R1 (C1 + C2 )
Hence
2
⎛ s ⎞ ⎛ s ⎞
b ⎜ ⎟ + ω 0 [b( R1C1 + R2C2 + R1C2 ) − R2C2 ] ⎜ ⎟ + b
⎝ ω0 ⎠ ⎝ ω0 ⎠
H (s ) = 2
(3.86)
⎛ s ⎞ 1⎛ s ⎞
⎜ ⎟ + ⎜ ⎟ +1
⎝ ω0 ⎠ Q ⎝ ω0 ⎠
1
Q= ∴
2 R1
1
R1 = (3.87)
2Q
1
R2 = = 2Q (3.88)
R1
− b + j[b( R1 + R2 + R1 ) − R2 ] + b
H ( jω 0 ) = =0 ∴
1
−1 + j + j
Q
122 Active Filters: Theory and Design
R2
b= or
2 R1 + R2
R4 R2
= ∴
R3 + R4 2 R1 + R2
R4 2Q 2Q 2
= = ∴
1 + R4 1 1 + 2Q 2
2 + 2Q
2Q
R4 = 2Q 2 (3.89)
EXAMPLE 3.15
Design a notch filter for f0 = 50 Hz and Q = 5.
Solution
Let Cn = 1 F.
R3n = 1 Ω
1 1
R1n = = = 0.05 Ω
2Q 20
R2 n = 2Q = 2 × 10 = 20 Ω
R4 n = 2Q 2 = 2 × 10 2 = 200 Ω
Denormalization
Let ISF = 10 3 .
ω0
FSF = = 2 π f0 = 100 π
ωn
Cn 1
C= = = 3.2 μF
ISF × FSF π × 10 5
R1 = ISF × R1n = 10 3 × 0.05 Ω = 50 Ω
R2 = ISF × R2 n = 10 3 × 20 Ω = 20 kΩ
R3 = ISF × R3n = 10 3 × 1 Ω = 1 kΩ
Figure 3.26a shows the designed filter, and its frequency response is shown in Figure
3.26b.
MultiFeedback Filters 123
3.3 u
20 K
47 3.3 u
−
1K
+ LM741
200 K
(a)
0.00
–1.00
–2.00
Gain dB
–3.00
–4.00
–5.00
1 10 100 1K
Frequency in Hz
(b)
⎛ Ks ⎞
V0 = − Vi ⎜ 2 + K⎟ (3.90)
⎝ s + as + ω 2
0 ⎠
where
R6
K= (3.91)
R5
124 Active Filters: Theory and Design
R2 R6
R1 C v01 R4
− −
vi + +
R3
vo
R5
and
R2
K= (3.92)
2 R1
R6
R5 = (3.93)
K
and
⎛ s2 + 1 ⎞
V0 = − K ⎜ 2 Vi ∴
⎝ s + as + ω 20 ⎟⎠
K (s 2 + 1)
H (s ) = − (3.94)
s + as + ω 20
2
EXAMPLE 3.16
Design an MFB narrow-band band-reject filter with f0 = 1 kHz, Q = 6, and K = 5.
Solution
Q 6
R1n = = = 1.2 Ω
K 5
R2 n = 2Q = 2 × 6 = 12 Ω
Q 6
R3n = = = 0.09 Ω
2Q 2 − K 72 − 5
R4 n = R6 n = 1 Ω
R6 n 2
R5n = = = 0.2 Ω
K 10
MultiFeedback Filters 125
Denormalization
Let ISF = 10 4.
ω0
FSF = = 2π f0 = 2π × 10 3
ωn
Cn 1
C= = = 15.9 nF
ISF × FSF 2π × 10 7
R1 = ISF × R1n = 10 4 × 1.2 Ω = 12 kΩ
R2 = ISF × R2 n = 10 4 × 12 Ω = 120 kΩ
R4 = ISF × R4 n = 10 4 × 1 Ω = 10 kΩ
16 n 10 K
2K
120 K −
12 K 16 n 10 K +
LM741
−
+
0.9 K LM741
(a)
16.00
12.00
8.00
Gain dB
4.00
0.00
–4.00
10 100 1K 10 K 100 K
Frequency in Hz
(b)
R6 = ISF × R6 n = 10 4 × 1 Ω = 10 kΩ
Figure 3.28a shows the designed filter, and its frequency response is shown in
Figure 3.28b.
R1R2
Req = R3 + (3.95)
R1 + R3
3. Standard resistance values of 5% tolerance normally yield acceptable
results in the lower-order cases. For orders five and six, resistances of 2%
tolerance probably should be used, and for order seven, 1% tolerance
probably should be used.
4. In the case of capacitors, percentage tolerances should parallel those given
earlier for the resistors for best results. As precision capacitors are rela-
tively expensive, it may be desirable to use capacitors of higher tolerances,
in which case trimming is generally required. In the case of lower orders
( n ≤ 4), 10% capacitors are quite often satisfactory.
5. The inverting gain of each stage of the filter is R2 / R1. Gain adjustment
can be made by using a potentiometer in lieu of R2 .
6. For minimum dc offset, a resistance equal to Req of 2 W can be placed in
the noninverting input to ground.
7. In a low-pass filter section, maximum gain peaking is very nearly equal
to Q at f1. So, as a rule of thumb:
(a) Op-amp bandwidth (BW) should be at least
BW = 100 × K × Q × f1 (3.96)
(b) For adequate full-power response, the slew rate (SR) of the op-amp must be
SR > π × V0 pp × BW f V /s (3.97)
EXAMPLE 3.17
A unity-gain 20-kHz 5-pole, 3-db ripple Chebyshev MFB filter has:
BW3 = 50 × f1 = 50 × 20 × 10 3 = 1 MHz
PROBLEMS
3.1 Design an LP Butterworth filter 3 dB at 3 kHz and attenuation 40 db at
12 kHz with a gain of 10.
3.2 Design an LP Chebyshev 1 dB filter at 800 Hz and attenuation 35 dB at
2400 Hz with a gain of 1.
3.3 Design an LP Bessel filter of order six at 1 kHz with K = 1.
3.4 Design an HP Butterworth filter at 600 Hz and attenuation 40 dB at 150
Hz with a gain of 5.
128 Active Filters: Theory and Design
aC/3
R R
−
3C +
vi a vo
3R/a
C C
−
aR +
vi 3 vo
The cutoff frequency of the low-pass and high-pass response is identical to the
center frequency of the band-pass response. In addition, the damping factor is equal
to 1/Q for a band-pass filter, and is the same for all three responses.
Figure 4.2 shows the circuit connection for the state-variable filter.
From this figure, we have:
R2 R Rq ⎛ R2 R2 ⎞
VHP = − VLP − 2 Vi + ⎜1 + + ⎟V (4.1)
R3 Rg Rq + R1 ⎜⎝ R3 Rg ⎟⎠ BP
1
VBP = − V (4.2)
sRC HP
1
VLP = − V (4.3)
sRC BP
129
130 Active Filters: Theory and Design
vi Σ ∫ ∫
R2 R Rq ⎛ R2 R2 ⎞
s 2 R 2C 2VLP = − VLP − 2 Vi − ⎜1 + + ⎟ sRCVLP ∴
R3 Rg Rq + R1 ⎜⎝ R3 Rg ⎟⎠
R3
R2 C C
Rg
− R
− R
+ −
+
vi vHP +
vLP
R1
Rq vBP
⎡ Rq ⎛ R2 R2 ⎞ R ⎤ R
⎢ s 2 R 2C 2 + ⎜⎜1 + + ⎟ sRC + 2 ⎥VLP = − 2 Vi ∴
⎢ Rq + R1 ⎝ ⎟
R3 Rg ⎠ R3 ⎥ Rg
⎣ ⎦
R2
VLP Rg
H LP = =− (4.7)
Vi Rq ⎛ R R ⎞ R
s 2 R 2C 2 + ⎜ 1 + 2 + 2 ⎟ sRC + 2
Rq + R1 ⎝ R3 Rg ⎠ R3
K
H LP = − (4.8)
s 2 + as + b
where
R3
K= (for s = 0) (4.9)
Rg
R2 1
b = ω12 = (4.10)
R3 R 2C 2
Rq ⎛ R2 R2 ⎞ 1
α= ⎜1 + + ⎟ (4.11)
Rq + R1 ⎜⎝ R3 Rg ⎟⎠ RC
R = Rg = Rq = 1 Ω (4.12)
C =1F (4.13)
R3 = K (4.14)
R2
b= ∴
K
R2 = bK (4.15)
132 Active Filters: Theory and Design
1 ⎛ bK bK ⎞ 1
a= 1+ + ∴
1 + R1 ⎜⎝ K 1 ⎟⎠ 1
1 + b(1 + K )
R1 = −1 (4.16)
α
EXAMPLE 4.1
Design an LP state-variable Butterworth filter 3 dB at 1 kHz and attenuation 40 dB
at 3.5 kHz with a gain of 1.
Solution
From Butterworth nomographs for Amax = 3 dB, Amin = 40 dB, and fs /f1 = 3.5/1 = 3.5,
we find n = 4. Hence
Rn = Rgn = Rqn = 1 Ω, Cn = 1 F
R3n = K = 1 Ω
R2 n = bK = 1 × 1 = 1 Ω
First stage
α = 1.848, b = 1.000
1 + (1 + K )b 3 3
R1n = −1= −1= − 1 = 0.623 Ω
a a 1.848
Second stage
α = 0.765, b = 1.000
3 3
R1n = −1= − 1 = 2.922 Ω
a 0.765
Denormalization
We accept ISF = 10 4.
ω1
FSF = = 2 π × 10 3
ωn
R = Rg = Rq = R2 = R3 = ISF × Rn = 10 4 × 1 Ω = 10 kΩ
Cn 1
C= = = 15.92 nF
ISF × FSF 2 π × 10 7
Filters with Three Op-Amps 133
First stage
Second stage
Figure 4.3a shows the designed filter, and its frequency response is shown in
Figure 4.3b.
EXAMPLE 4.2
Design a low-pass state-variable Chebyshev 3-dB filter at 1 kHz and attenuation 35 dB
at 2 kHz, with a gain of 5.
Solution
From Chebyshev 3-dB nomographs for Amax = 3 dB, Amin = 35 dB, and fs / f1 = 2 /
1 = 2, we find n = 4.
Rn = Rgn = Rqn = 1 Ω, Cn = 1 F
K1 = K 2 = K = 5 = 2.236
R3n = K1 = 2.236
First stage
a = 0.411, b = 0.196
1 + (1 + K )b 1.634
R1n = −1= − 1 = 2.976 Ω
a 0.411
Second stage
a = 0.170, b = 0.903
1 + (1 + K )b 3.922
R1n = −1= − 1 = 22.071 Ω
a 0.170
134
10 K 10 K
10 K 16 n 16 n 10 K 16 n 16 n
10 K 10 K 10 K 10 K 10 K 10 K
− − − − − −
+ + + + + +
LM318 LM318 LM318 LM318 LM318 LM318
6.23 K 29.2 K
10 K 10 K
(a)
FIGURE 4.3 (a) S-V LP Butterworth filter, f1 = 1 kHz, K = 1; (b) frequency response.
Active Filters: Theory and Design
Filters with Three Op-Amps 135
20.00
0.00
–20.00
Gain dB
–40.00
–60.00
–80.00
10 100 1K 10 K 100 K
Frequency in Hz
(b)
Denormalization
ω1
ISF = 10 4 , FSF = = 2 π × 10 3
ωn
R = Rg = Rq = ISF × Rn = 10 4 × 1 = 10 kΩ
Cn 1
C= = = 15.92 nF
ISF × FSF 2 π × 10 7
R3 = ISF × R3n = 10 4 × 2.236 Ω = 22.36 kΩ
First stage
Second stage
Figure 4.4a shows the designed filter, its frequency response is shown in Figure 4.4b,
and its ripple in Figure 4.4c.
136
22.4 K 22.4 K
4.38 K 16 K 16 n 20.2 K 16 n 16 n
10 K 10 K 10 K 10 K 10 K 10 K
− − − − − −
+ + + + + +
LM318 LM318 LM138 LM318 LM318 LM318
29.8 K 220.7 K
10 K 10 K
(a)
FIGURE 4.4 (a) S-V LP Chebyshev 3 dB, f1 = 1 kHz, K = 5; (b) its frequency response; and (c) its ripple.
Active Filters: Theory and Design
Filters with Three Op-Amps 137
20.00
0.00
–20.00
Gain dB
–40.00
–60.00
–80.00
10 100 1K 10 K 100 K
Frequency in Hz
(b)
18.00
16.00
14.00
Gain DB
12.00
10.00
8.00
10 100 1K 10 K 100 K
Frequency in Hz
(c)
1 ⎛ 1 ⎞ 1
VLP = − ⎜ − VHP ⎟ = 2 2 2 VHP (4.17)
sRC ⎝ sRC ⎠ s RC
R2 1 R Rq ⎛ R2 R2 ⎞ 1
VHP = − ⋅ 2 2 2 VHP − 2 Vi − ⎜1 + + ⎟ V ∴
R3 s R C Rg Rq + R1 ⎜⎝ R3 Rg ⎟⎠ sRC HP
⎡ R 1 Rq ⎛ R2 R2 ⎞ 1 ⎤
VHP ⎢1 + 2 ⋅ 2 2 2 + ⎜⎜1 + + ⎟ ⎥ = − R2 Vi ∴
⎢ R3 s R C Rq + R1 ⎝ ⎟
R3 Rg ⎠ sRC ⎥ Rg
⎣ ⎦
⎡ Rq ⎛ R2 R2 ⎞ R ⎤ R
VHP = ⎢ s 2 R 2C 2 + ⎜⎜1 + + ⎟ sRC + 2 ⎥ = − 2 s 2 R 2C 2Vi ∴
⎢ Rq + R1 ⎝ ⎟
R3 Rg ⎠ R3 ⎥ Rg
⎣ ⎦
R2 2 2 2
s RC
V Rg
H HP = HP = − ∴
Vi Rq ⎛ R2 R2 ⎞ R
s 2 R 2C 2 + ⎜⎜1 + + ⎟ sRC + 2
Rq + R1 ⎝ ⎟
R3 Rg ⎠ R3
R2 2
s
Rg
H HP = − ∴
Rq ⎛ R2 R2 ⎞ 1 R 1
s +
2
⎜1 + + ⎟ s+ 2 2 2
Rq + R1 ⎜⎝ ⎟
R3 Rg ⎠ RC R3 R C
Ks 2
H HP = − (4.18)
s 2 + as + b
where
R2 1
b = ω 22 = (4.19)
R3 R 2C 2
Rq ⎛ R2 R2 ⎞ 1
a= ⎜1 + + ⎟ (4.20)
Rq + R1 ⎜⎝ R3 Rg ⎟⎠ RC
For s → ∞ , we have:
R2
K = H HP (∞) = (4.21)
Rg
R2 n = K (4.23)
R3n = bK (4.24)
1 + b( K + 1)
R1n = −1 (4.25)
a
EXAMPLE 4.3
Design an HP Butterworth 3-dB filter at 100 Hz and attenuation 40 dB at 28.6 Hz,
with a gain of 1.
Solution
From Butterworth nomographs for Amax = 3 dB, Amin = 40 dB, and f2/fs = 100/28.6 @
3.5, we find n = 4 .
Rqn = Rgn = Rn = 1 Ω
Cn = 1 F
R2 n = K = 1 Ω
First stage
a = 1.848, b = 1.000
1 + b( K + 1) 1+ 2
R1n = −1= − 1 = 0.623 Ω
a 1.848
R3n = bK = 1 Ω
Second stage
a = 0.765, b = 1.000
1 + b( K + 1) 3
R1n = −1= − 1 = 2.922 Ω
a 0.765
R3n = bK = 1 Ω
Denormalization
ω2
ISF = 10 4 , FSF = = 2 π f2 = 2 π × 100 = 200 π
ωn
Cn 1
C= = = 0.159 μ F
ISF × FSF 2 π × 10 6
140 Active Filters: Theory and Design
R2 = ISF × R2 n = 10 × 1 = 10 kΩ
First stage
R3 = ISF × R3n = 10 4 × 1 Ω = 10 kΩ
Second stage
R3 = ISF × R3n = 10 4 × 1 Ω = 10 kΩ
Figure 4.5a shows the designed filter, and its frequency response is shown in
Figure 4.5b.
EXAMPLE 4.4
Design an HP Chebyshev 3-dB filter at 100 Hz and attenuation 40 dB at 40 Hz,
with a gain of 5.
Solution
From Chebyshev nomographs for Amax = 3 dB, Amin = 40 dB, and f2/fs = 100/40 = 25,
we find n = 4.
K= 5
Rn = Rgn = Rqn = 1 Ω
Cn = 1 F
First stage
a = 0.411, b = 0.196
10 K 10 K
10 K 10 K 10 K 10 K 10 K 10 K
− − − − − −
+ + + + + +
LM741 LM741 LM741 LM741 LM741 LM741
6.2 K 29.2 K
10 K 10 K
(a)
FIGURE 4.5 (a) Fourth-order S-V HP Butterworth 3 dB, f2 = 100 Hz, K = 1; (b) frequency response.
141
142 Active Filters: Theory and Design
20.00
0.00
–20.00
Gain dB
–40.00
–60.00
–80.00
10 100 1K 10 K
Frequency in Hz
(b)
Second stage
a = 0.170, b = 0.903
1 + 0.903 × 3.236
R1n = − 1 = 22.071 Ω
0.170
R2 n = 2.236 Ω
Denormalization
ω2
ISF = 10 4 , FSF = = 2 π f2 = 2 π × 100
ωn
R = Rg = Rq = 10 kΩ
Cn 1
C= = = 0.16 μ F
ISF × FSF 2π × 10 6
R2 = ISF × R2 n = 10 4 × 2.236 Ω = 22.4 kΩ
First stage
Second stage
Figure 4.6a shows the designed filter, and its frequency response is shown in
Figure 4.6b.
EXAMPLE 4.5
Design a Butterworth band-pass filter 3 dB from 300 to 3000 Hz and attenuation
30 dB below 50 Hz and above 18,000 Hz, with gain 1.
Solution
(a) Low-pass filter
From Butterworth nomographs for Amax = 3 dB, Amin = 30 dB, and fs / f1 = 18,000/
3000 = 6, we find n = 2 . Hence
a = 1.414, b = 1.000
R2 n = bK = 1 × 1 = 1 Ω
1 + b + bK 3 3
R1n = −1= −1= − 1 = 1.122 Ω
a a 1.414
Denormalization
ω1
ISF = 10 4 , FSF = = 2π f1 = 2π × 3000 = 6π × 10 3
ωn
R = Rg = Rq = R2 = R3 = ISF × Rn = 10 4 × 1 Ω = 10 kΩ
Cn 1
C= = = 5.3 nF
ISF × FSF 6 π × 10 7
4.38 K 20.2 K
10 K 10 K 10 K 10 K 10 K 10 K
− − − − − −
+ + + + + +
LM741 LM741 LM741 LM741 LM741 LM741
29.8 K 220.7 K
10 K 10 K
(a)
FIGURE 4.6 (a) Fourth-order S-V HP Chebyshev 3 dB, f2 = 100 Hz, K = 5; (b) frequency response.
Active Filters: Theory and Design
Filters with Three Op-Amps 145
40.00
20.00
00.00
Gain dB
–20.00
–40.00
–60.00
10 100 1K 10 K
Frequency in Hz
(b)
Denormalization
ω 2 2 π f2
ISF = 10 4 , FSF = = = 2 π × 300 = 6 π × 10 2
ωn 1
Cn 1
C= = = 53.1 nF
ISF × FSF 6 π × 10 2
Figure 4.7a shows the designed filter, and its frequency response is shown in Figure 4.7b.
EXAMPLE 4.6
Design a wide-band band-reject Butterworth filter having 3 dB at 100 and 1000 Hz
and greater than 20 dB of attenuation between 500 and 360 Hz with a gain of 5.
Solution
Because the ratio of upper cutoff to lower cutoff is well in excess of an octave, a
wide-band approach can be used. First, separate the specification into individual
low-pass and high-pass requirements.
(a) Low-pass filter
From Butterworth nomographs for Amax = 3 dB, Amin = 20 dB, and fs /f1 = 500/100 = 5,
we find n = 2 .
a = 1.414, b = 1.000
Rn = Rgn = Rqn = 1 Ω, Cn = 1 F
146
10 K 10 K
10 K 10 K 10 K 10 K 10 K 10 K
− − − − − −
+ + + + + +
LM741 LM741 LM741 LM741 LM741 LM741
11.2 K 11.2 K
10 K 10 K
(a)
FIGURE 4.7 (a) Second-order S-V Butterworth BP 300 to 3000 Hz, K = 1; (b) frequency response.
Active Filters: Theory and Design
Filters with Three Op-Amps 147
20.00
0.00
–20.00
Gain dB
–40.00
–60.00
–80.00
10 100 1K 10 K 100 K
Frequency in Hz
(b)
For K = 1, we have:
R3n = K = 1 Ω, R2 n = bK = 1 Ω
1 + b + bK 3 3
R1n = −1= −1= − 1 = 1.122 Ω
a a 1.414
Denormalization
ω1
ISF = 10 4 , FSF = = 2 π f1 = 2 π × 100
ωn
Cn 1
C= = = 0.159 μ F
ISF × FSF 2 π × 10 6
R = Rg = Rq = R3 = R2 = ISF × Rn = 10 4 × 1 Ω = 10 kΩ
ω2
ISF = 10 4 , FSF = = 2 π f2 = 2 π × 10 3
ωn
Cn 1
C= = = 15.9 nF
ISF × FSF 2 π × 10 7
Figure 4.8a shows the designed filter, and its frequency response is shown in Figure 4.8b.
148 Active Filters: Theory and Design
10 K
160 n 160 n 50 K
10 K
10 K
10 K 10 K 10 K −
− − − +
LM741
+ + +
LM741 LM741 LM741
11.2 K
10 K
10 K
10 K 16 n 16 n
10 K 10 K 10 K
− − −
+ + +
LM741 LM741 LM741
10 K 11.2 K
10 K
(a)
20.00
15.00
10.00
Gain dB
5.00
0.00
–5.00
1 10 100 1K 10 K 100 K
Frequency in Hz
(b)
FIGURE 4.8 (a) Wide-band band-reject S-V filter, 100 to 1000 Hz, K = 5; (b) frequency
response.
Filters with Three Op-Amps 149
R2 ⎛ 1 ⎞ R Rq ⎛ R2 R2 ⎞
− sRCVBP = − ⎜ − VBP ⎟ − 2 Vi − +
⎜1+ + ⎟V ∴
R3 ⎝ sRC ⎠ Rg Rq + R1 ⎜⎝ R3 Rg ⎟⎠ BP
⎡ Rq ⎛ R2 R2 ⎞ ⎤
⎢ sRC + R2 1 + ⎜1 + + ⎟ ⎥V = 2 V
R
∴
⎢ R3 sRC Rq + R1 ⎜⎝ R3 Rg ⎟⎠ ⎥ BP Rg i
⎣ ⎦
⎡ Rq ⎛ R2 R2 ⎞ R ⎤ R
⎢ s 2 R 2C 2 + ⎜⎜1 + + ⎟ sRC + 2 ⎥ VBP = 2 sRCVi ∴
⎢ Rq + R1 ⎝ R3 Rq ⎟⎠ R3 ⎥ Rg
⎣ ⎦
R2
sRC
VBP Rg
H BP (s) = = ∴
Vi Rq ⎛ R2 R2 ⎞ R
s RC +
2 2 2
⎜⎜1 + + ⎟ sRC + 2
Rq + R1 ⎝ ⎟
R3 Rg ⎠ R3
R2
sRC
Rg
H BP = ∴
⎡ Rq ⎛ R2 R2 ⎞ 1 R 1 ⎤
R 2C 2 ⎢ s 2 + ⎜⎜1 + + ⎟ s+ 2 2 2⎥
⎢ Rq + R1 ⎝ R3 Rg ⎠⎟ RC R3 R C ⎥
⎣ ⎦
R2 1
s
Rg RC
H BP = (4.26)
Rq ⎛ R2 R2 ⎞ 1 R 1
s2 + ⎜1 + + ⎟ s+ 2 2 2
Rq + R1 ⎜⎝ R3 Rg ⎟⎠ RC R3 R C
If we put
R2 1
ω 20 = ∴
R3 R 2C 2
R2
R3
ω0 = (4.27)
RC
150 Active Filters: Theory and Design
Hence
R2 1
s
Rg RC
H BP = ∴
Rq ⎛ R R ⎞ 1
s2 + ⎜ 1+ 2 + 2 ⎟ s + ω 20
Rq + R1 ⎝ R3 Rg ⎠ RC
R2 1 ⎛ s ⎞
⎜ ⎟
Rg ω 0 RC ⎝ ω 0 ⎠
H BP =
Rq ⎛ R2 R2 ⎞ 1 ⎛ s ⎞
2
⎛ s ⎞
⎜ ⎟ + ⎜1 + + ⎟ ⎜ ⎟ +1
⎝ ω0 ⎠ Rq + R1 ⎜⎝ R3 Rg ⎟⎠ ω 0 RC ⎝ ω o ⎠
If we put
Rq ⎛ R2 R2 ⎞ 1
a= ⎜1 + + ⎟ (4.28)
Rq + R1 ⎜⎝ R3 Rg ⎟⎠ ω 0 RC
we have
R2 1 ⎛ s ⎞
⎜ ⎟
Rg ω 0 RC ⎝ ω 0 ⎠
H BP = 2
(4.29)
⎛ s ⎞ ⎛ s ⎞
⎜ ⎟ + a⎜ ⎟ +1
⎝ ω0 ⎠ ⎝ ω0 ⎠
R2 1
Rg ω 0 RC
K= ∴
Rq ⎛ R2 R2 ⎞ 1
⎜1 + + ⎟
Rq + R1 ⎜⎝ R3 Rg ⎟⎠ ω o RC
R2
Rg
K= (4.30)
Rq ⎛ R2 R2 ⎞
⎜1 + + ⎟
Rq + R1 ⎜⎝ R3 Rg ⎟⎠
R = Rg = R2 = 1 Ω (4.31)
C =1 F (4.32)
R3 = 1 Ω (4.33)
3Rq
α= ∴
Rq + R1
3Rq
Rq + R1 = = 3QRq ∴
a
R1 = (3Q − 1) Rq (4.34)
1 1 + 3Q − 1
K= = ∴
3Rq 3
Rq + (3Q − 1) Rq
K =Q (4.35)
EXAMPLE 4.7
Design a narrow-band band-pass filter with a center frequency of 1 kHz and Q = 40.
Solution
Cn = 1 F
K = Q = 40 or 32 dB
Denormalization
ω0
ISF = 10 3 , FSF = = 2 π f0 = 2 π × 10 3 ∴
ωn
152 Active Filters: Theory and Design
R = Rg = Rq = R2 = R3 = ISF × Rn = 10 3 × 1 Ω = 10 kΩ
Cn 1
C= = = 0.159 μ F
ISF × FSF 2 π × 10 6
Figure 4.9a shows the designed filter, and its frequency response is shown in Figure 4.9b.
10 K
10 K 16 n 16 n
10 K 10 K 10 K
− − −
+ LM41 + LM741 + LM741
119 K
VBP
10 K
(a)
20.00
10.00
0.00
Gain dB
–10.00
–20.00
–30.00
100 1K 10 K
Frequency in Hz
(b)
FIGURE 4.9 (a) Narrow-band S-V BPF, f0 = 1 kHz, Q = 40; (b) its frequency response.
Filters with Three Op-Amps 153
EXAMPLE 4.8
Design a 100-Hz notch filter with Q = 25.
Solution
Cn = 1 F
R1n = 3Q − 1 = 3 × 25 − 1 = 74 Ω
K = Q = 25 or 28 dB
Denormalization
ω0
ISF = 10 4 , FSF = = 2 π × 10 2
ωn
R = Rg = Rq = R2 = R3 = ISF × Rn = 10 4 × 1 Ω = 10 kΩ
Cn 1
C= = = 0.159 μ F
ISF × FSF 2 π × 10 6
To complete the design, add the two-input op-amp summing amplifier to the LP and
HP outputs, giving the final circuit of Figure 4.10a; its frequency response is shown
in Figure 4.10b.
154 Active Filters: Theory and Design
10 K
10 K 160 n 160 n
10 K 10 K 10 K
− − −
+ LM741 + LM741 + LM741
119 K
10 K 10 K
10 K
10 K
−
+ LM741
(a)
5.00
0.00
–5.00
Gain dB
–10.00
–15.00
–20.00
1 10 100 1K 10 K
Frequency in Hz
(b)
FIGURE 4.10 (a) State-variable notch filter, f0 = 100 Hz, Q = 25; (b) its frequency response.
Rf
Rq
Cf
R Cf
Rg
− R Rf
−
+ −
+ v1
vi vBP +
vLP
is needed at different center frequencies. The center frequency is easily tuned by merely
adjusting the value of R f . Also, Q may be adjusted by changing the value of Rq , and
the gain of the filter may be changed by adjusting the value of Rg . The biquad filter
is capable of attaining high values of Q, in the neighborhood of 100, and is a much
more stable network than those discussed in the previous band-pass filters.
From this figure, we have:
Z Z
VBP = − Vi − V (4.36)
Rg R f LP
V1 = −VB (4.37)
1
VLP = − V (4.38)
sR f C f 1
1 1 1 + sRqC f
Y= = + sC f = (4.39)
Z Rq Rq
VLP = −
1
sR f C f
( −VBP ) =
1
V
sR f C f BP
(4.40)
Z Z 1
VBP = − V− ⋅ V ∴
Rg i R f sR f C f BP
⎛ Z ⎞ Z
VBP ⎜1 + 2 ⎟ = − Vi ∴
⎜⎝ sR f C f ⎟⎠ Rg
156 Active Filters: Theory and Design
1 + sR 2f C f Z Z
VBP = − Vi ∴
sR 2f C f Rg
VBP sR 2f C f Z sR 2f C f
H BP = =− =− ∴
Vi Rg (sR 2f C f + Z ) ⎛ sR 2C ⎞
Rg ⎜1 + f f ⎟
⎝ Z ⎠
sR 2f C f
Rg
H BP =− ∴
1 + sRqC f
1 + sR 2f C f
Rq
sR 2f C f Rq
Rg
H BP =− ∴
Rq + sR f C f (1 + sRqC f )
2
sR 2f C f Rq
Rg
H BP =− 2 2 ∴
s R f RqC f + sR 2f C f + Rq
2
1
s
RgC f
H BP =−
1
s2 + s + ω o2
RqC f
where
1
ω0 = (4.41)
Rf C f
1 ⎛ s ⎞
⎜ ⎟
ω 0 RgC f ⎝ ω 0 ⎠
H BP = − 2
(4.42)
⎛ s ⎞ 1⎛ s ⎞
⎜ ⎟ + ⎜ ⎟ +1
⎝ ω0 ⎠ Q ⎝ ω0 ⎠
where
Q = ω 0 RqC f (4.43)
Filters with Three Op-Amps 157
1
ω 0 RgC f
K = H BP (ω 0 ) = − ∴
1
ω 0 RqC f
Rq
K=− (4.44)
Rg
1
Rf = (4-45)
ω 0C f
Q
Rq = = QR f (4.46)
ω 0C f
Rq
Rg = (4.47)
K
Rq
Q= (4.48)
Rf
But
ω0
BW =
Q
ω 0 = 1 rad/s
and
Cf = 1 F (4.49)
Rf = 1 Ω (4.50)
Rq = Q (4.51)
Q
Rg = (4.52)
K
EXAMPLE 4.9
Design a narrow-band band-pass filter with fo = 1 kHz, Q = 50, and a gain of 5.
Solution
Rn = R f n = 1 Ω
Cn = 1 F
Rqn = Q = 50 Ω
Rqn Q 50
Rgn = = = = 10 Ω
K K 5
Denormalization
ω0
ISF = 10 4 , FSF = = 2 π f0 = 2 π × 10 3
ωn
R = R f = ISF × Rn = 10 4 × 1 Ω = 10 kΩ
10 K
500 K
16 n 10 K 16 n
100 K
− 10 K
− 10 K
+ −
LM741 +
LM741 +
LM741
(a)
20.00
0.00
–20.00
Gain dB
–40.00
–60.00
–80.00
10 100 1K 10 K 100 K
Frequency in Hz
(b)
FIGURE 4.12 (a) Narrow-band band-pass filter, f0 = 1 kHz, Q = 50, K = 5; (b) its frequency
response.
Cn 1
C= = = 15.92 nF
ISF × FSF 2 × 10 7
Figure 4.12a shows the designed filter, and its frequency response is shown in
Figure 4.12b.
Z Z
sR f C f VLP = − Vi − V ∴
Rg R f LP
⎛ Z⎞ Z
⎜⎜ sR f C f + ⎟⎟ VLP = − ∴
⎝ R f ⎠ Rg
⎡ Rq ⎤ Rq
⎢ sR f C f + ⎥ VLP = − V ∴
⎢⎣ R f (1 + sRqC f ) ⎥⎦ Rg (1 + sRqC f ) i
⎛ Rq ⎞ Rq
⎜⎜ sR f C f (1 + sRqC f ) + ⎟⎟ VLP = − Vi ∴
⎝ R f ⎠ Rg
⎛ Rq ⎞ Rq
⎜⎜ s 2 R f RqC 2f + sR f C f + ⎟⎟ VLP = − Vi ∴
⎝ Rf ⎠ Rg
Rq
VLP Rg
H LP = =− ∴
Vi Rq
s 2 R f RqC 2f + sR f C f +
Rf
1
R f RqC 2f
H LP =−
1 1
s2 + s+ 2 2
RqC f Rf C f
1
R f RgC 2f
H LP =− 2 (4.54)
s + as + b
where
1
b = ω12 = (4.55)
R C 2f
2
f
and
1
α= (4.56)
RqC f
Filters with Three Op-Amps 161
Rf
H LP = K = − (4.57)
Rg
1
Rf = (4.58)
Cf b
1
Rq = (4.59)
aC f
Rf
Rg = (4.60)
K
Cf n = 1 F (4.61)
we have:
1
R fn = (4.62)
b
1
Rqn = (4.63)
a
R fn 1
Rqn = = (4.64)
K K b
EXAMPLE 4.10
Design a second-order low-pass Butterworth filter with a cutoff frequency of 1 kHz
and a gain of 10.
162 Active Filters: Theory and Design
Solution
a = 1.414, b = 1.009
1 1
Rqn = = = 0.708 Ω
a 1.414
1 1
Rgn = = = 0.1 Ω
K b 10 × 1
1 1
R fn = = =1 Ω
b 1
Denormalization
ω1
ISF = 10 4 , FSF = = 2 π f1 = 2 π × 10 3 ∴
ωn
C fn 1
Cf = = = 15.92 nF
ISF × FSF 2 π × 10 7
R f = ISF × R fn = 10 4 × 1 Ω = 10 kΩ
R = ISF × Rn = 10 4 × 1 Ω = 10 kΩ
Figure 4.13a shows the designed filter, and its frequency response is shown in
Figure 4.13b.
EXAMPLE 4.10
Design an LP Chebyshev 3-dB filter at 1 kHz and attenuation 40 dB at 2.5 kHz,
with a gain of 10.
Solution
From Chebyshev nomographs for Amax = 3 dB, Amin = 40 dB, and fs /f1 = 2.5/1 = 2.5,
we find n = 4 .
K = 10 = 3.162
Rn = 1 Ω, C fn = 1 F
Filters with Three Op-Amps 163
10 K
7.1 K
16 n 10 K 16 n
1K 10 K 10 K
− − −
+ LM741 + LM741 + LM741
(a)
30.00
20.00
10.00
Gain dB
0.00
–10.00
–20.00
10 100 1K 10 K 100 K
Frequency in Hz
(b)
FIGURE 4.13 (a) Low-pass Butterworth filter, f1 = 1 kHz, K = 10; (b) frequency response.
First stage
a = 0.411, b = 0.196
1 1
Rqn = = = 2.433 Ω
a 0.411
1 1
Rgn = = = 0.714 Ω
K b 3.162 × 0.196
1 1
R fn = = = 2.259 Ω
b 0.196
164 Active Filters: Theory and Design
Second stage
a = 0.170, b = 0.903
1 1
Rqn = = = 5.882 Ω
a 0.170
1 1
Rgn = = = 0.333 Ω
K b 3.162 × 0.903
1 1
R fn = = = 1.052 Ω
b 0.903
Denormalization
ω1
ISF = 10 4 , FSF = = 2 π f1 = 2 π × 10 3
ωn
R = ISF × Rn = 10 4 × 1 Ω = 10 kΩ
C fn 1
Cf = = = 15.9 nF
ISF × FSF 2 π × 10 7
First stage
Second stage
Figure 4.14a shows the designed filter, and its frequency response is shown in
Figure 4.14b.
22.6 K 10.5 K
24.3 K 16 n
16 n 10 K 16 n 58.8 K 10 K 16 n
(a)
Filters with Three Op-Amps
25.00
20.00
15.00
Gain dB
10.00
5.00
0.00
1 10 100 1K 10 K 100 K
Frequency in Hz
(b)
165
FIGURE 4.14 (a) LP Chebyshev 3 dB filter, f1 = 1 kHz, K = 10; (b) its frequency response.
166 Active Filters: Theory and Design
PROBLEMS
4.1 Design an LP state-variable Butterworth 3 dB at 1 kHz and attenuation
45 dB at 2.5 kHz, with a gain of 10.
4.2 Design an LP state-variable Chebyshev 2 dB filter at 350 Hz and attenu-
ation 35 dB at 700 Hz, and with a gain of 1.
4.3 Design a BP state-variable filter with a center frequency 100 Hz, Q = 20,
K = 10, and n = 4.
4.4 Figure P4.1 shows the state-variable filter with four op-amps. With this
circuit the gain and Q are independent of each other.
R3
R2 C C
Rg
K
− R
− R
+ −
+
vi vHP +
R1 vLP
R4
R5
−
+
vBP
4.12 Prove that the following filter can be used to realize the general biquadratic
function (Figure P4.2):
Rf
Rq
Cf R Cf R
Rg
− R Rf
− R
+ vBP −
+ −
vi + vLP
R +
vo
R
s 2 + b1s + b0
H (s ) = − K
s 2 + a1s + a0
5 Sensitivity
5.1 INTRODUCTION
A desired function is realized by interconnecting electrical components of carefully
chosen values in a filter network. These components are subject to change due to
variations in temperature, humidity, aging, and tolerances in manufacturing. In order
to measure the change in filter performance due to drift or change in component
values, we use the sensitivity concept.
The basis for all modern-day sensitivity analysis methods is rooted in the work
of W.H. Bode. S.J. Mason further expanded Bode’s definition, and it is Mason’s
definition that is most often cited, and hence used here.
Sensitivity is a measure of deviation in some performance characteristic of the
circuit due to some change in the nominal value of one or more of the elements of
the filter. Low-sensitivity circuits are naturally preferred over high-sensitivity cir-
cuits. The relative sensitivity is defined as
x ∂ H ∂ ln H
S xH = = (5.1)
H ∂x ∂ ln x
where H is the system transfer function and x is the parameter or network element
that is causing H to change.
In general, the network function will be a ratio of polynomials in s, such that
N (s , x )
H (s , x ) = (5.2)
D(s, x )
and
where
∂ N (s , x )
N ′(s , x ) =
∂x
(5.4)
∂ D( s , x )
D′(s, x ) =
∂x
169
170 Active Filters: Theory and Design
Hence,
⎡ N ′(s, x ) D′(s, x ) ⎤
S xH = x ⎢ − ⎥ (5.6)
⎣ N (s, x ) D(s, x ) ⎦
EXAMPLE 5.1
Determine S xH for the network function:
K
H (s , x ) =
s + xs + 9
2
where K is some constant, and x will be assumed to have a nominal value of unity.
Then,
⎡0 s ⎤
S xH = x ⎢ − 2 ⎥ ∴
⎣ K s + xs + 9 ⎦
xs
S xH = −
s 2 + xs + 9
−j −j
S xH = = ∴
−1 + j + 9 8 + j
− j(8 − j) −1 − j8
S xH = = = 0.015 − j 0.123
82 + 1 65
The interpretation of the answer is that the real part of the sensitivity specifies a
normalized change in the magnitude of the given network function, whereas the
imaginary part indicates a change in the argument (phase) of the given network
function.
Consequently, the normalized magnitude change of S xH with respect to normal-
ized change in x is 0.015, and the phase change with respect to a normalized change
in x is 0.123 radians.
Sensitivity 171
ln H = a ln h1 + b ln h2 + c ln h3 (5.8)
∂ ln H
ShH1 = =a (5.9)
∂ ln h1
Similarly, we have:
ShH2 = b (5.10)
ShH3 = c (5.11)
H = H1H 2 (5.12)
∂ ln H ∂ ln( H1 ⋅ H 2 ) ∂ ln H1 ∂ ln H 2
S xH = = = + (5.13)
∂ ln x ∂ ln x ∂ ln x ∂ ln x
showing that
S xH = − S 1/x H (5.16)
S xH n = n S xH (5.19)
1 H
S xHn = S (5.20)
n x
EXAMPLE 5.2
Perform sensitivity analysis for the low-pass Sallen–Key filter of Figure 2.9.
G1G2
K
C1C2
H (s ) = (5.21)
⎡ G + G2 (1 − K )G2 ⎤ G1G2
s +⎢ 1
2 + ⎥s +
⎣ C2 C1 ⎦ C1C2
G1G2
ω12 = b = = R1−1/ 2 R2−1/ 2C1−1/ 2C2−1/ 2 (5.22)
C1C2
ω1 G1 + G2 (1 − K )G2
B = aω1 = = + ∴
Q C2 C1
ω1 1 1 1− K
B= = + + (5.23)
Q R1C2 R2C2 R2C1
k = K R1−1R2−1C1−1C2−1 (5.24)
where K = 1 + Rb / Ra
For convenience, we write
SKk = 1 (5.27)
1
S Rω,1R2 ,C1 ,C2 = − (5.28)
2
Sensitivity 173
S xQ = S xω1 − S xB (5.30)
where
x ∂B Q ∂B
S xB = ⋅ =x ⋅ (5.31)
B ∂x ω1 ∂ x
∂B 1
=− 2
∂ R1 R1 C1
⎛ 1 ⎞
S RB1 = R1Q( R1 R2C1C2 )−1/ 2 ⎜ − 2 ⎟ = − QR1−1/22 R2−1/ 2C1−1/ 2C2−1/ 2 (5.32)
⎝ R1 C1 ⎠
1 RC
S RQ1 = − +Q 2 2 (5.33)
2 R1C1
Similarly, we obtain
1 ⎡ RC RC ⎤
S RQ2 = − + Q ⎢ 1 2 + (1 − K ) 1 1 ⎥ (5.34)
2 ⎢⎣ R2C1 R2C2 ⎥⎦
1 ⎡ RC R2C2 ⎤
SCQ1 = − + Q⎢ 1 2 + ⎥ (5.35)
2 ⎢⎣ R2C1 R1C1 ⎥⎦
1 RC
SCQ2 = − + (1 − K ) Q 1 1 (5.36)
2 R2C2
R1C1
SKQ = KQ (5.37)
R2C2
174 Active Filters: Theory and Design
R1C1
S RQa = − S RQb = (1 − K )Q (5.38)
R2C2
1− K
S Rka = − S Rkb = (5.39)
K
All of these sensitivity functions have been derived for general values of the filter
parameters. Once a specific circuit design is chosen, these parameters are known,
and the numerical values sensitivities can be ascertained. As a specific example,
consider the normalized Butterworth transfer function of
2
H (s ) = (5.40)
s + 2s +1
2
R1 = R2 = 1 Ω, C1 = C2 = 1 F, Ra = 10 Ω, Rb = 5.86 Ω (5.42)
SKQ = 1.121
S RQa = − S RQb = −0.414
S Rka = − S Rkb = −0.369
H ( jω ) = H ( jω ) e jφ(ω ) (5.43)
x ∂
S xH ( jω ) = ⋅ ⎡ H ( jω ) e jφ(ω ) ⎤ (5.44)
H ( jω ) ∂ x ⎣ ⎦
Sensitivity 175
which can be expanded by making use of the product rule for differentiation of a
product to give
x ∂ H ( jω ) ∂φ(ω )
S xH ( jω ) = ⋅ + jx
H ( jω ) ∂x ∂x
or
S xH ( jω ) = Re S xH ( jω ) (5.46)
1
S xφ(ω ) = Im S xH ( jω ) (5.47)
φ (ω )
These equations state that the magnitude and phase sensitivity of a transfer function
with respect to an element are simply related to the real and imaginary parts of the
transfer function sensitivity with respect to the same element.
EXAMPLE 5.3
Find the sensitivity of the low-pass transfer function
K ω12
H (s ) = (5.48)
ω
s 2 + 1 s + ω12
Q
The sensitivity function is found from Equation (.5.1) to be
ω1
s
Q ∂ H (s ) Q
SQH ( s ) = ⋅ = (5.49)
H (s) ∂Q ω
s 2 + 1 s + ω12
Q
To compute the sensitivity function of H ( jω ) with respect to Q , we apply Equation
(5.46) by first substituting s by jw in Equation (5.49) and then taking the real part.
The result is given by
H ( jω ) (ω / ω1 )2
SQ = Re SQH ( jω ) = (5.50)
Q [1 − (ω / ω1 )2 ] 2 + (ω / ω1 )2
2
H ( jω )
SQ ( jω 1 ) = 1 (5.51)
176 Active Filters: Theory and Design
The above equation is the sensitivity function of H ( jω ) due to the variations in the
cutoff frequency ω1. Applying Equation (5.1) yields
⎛ω ⎞
2s 2 + ⎜ 1 ⎟ s
⎝Q⎠
SωH1( s ) = (5.52)
⎛ω ⎞
s 2 + ⎜ 1 ⎟ s + ω12
⎝Q⎠
giving
⎛ ω⎞ ⎡ ⎛ ω⎞ ⎤
2 2
1
⎢ 2
⎜⎝ ω ⎟⎠ ⎢ ⎜⎝ ω ⎟⎠ + − 2 ⎥
Q2 ⎥
H ( jω )
Sω1 = Re SωH1( jω ) =
1
⎣ 1
⎦ (5.53)
2
⎡ ⎛ ω⎞ ⎤ ⎛ ω ⎞
2 2
⎢1 − ⎜ ⎟ ⎥ + ⎜
⎢ ⎝ ω1 ⎠ ⎥ ⎝ Qω1 ⎟⎠
⎣ ⎦
H ( jω )
Sω1 ( jω 1 ) = 1 (5.54)
∂s j ∂s j
S x− s j = =x (5.55)
∂x / x ∂x
K
H (s ) = k
(5.56)
∏ (s − p )
j =1
j
kj
Sensitivity 177
ln H (s) = ln K − ∑ k ln (s − p )
j =1
j j (5.57)
∂ pj
k kj
∂ ln H (s) 1 ∂ K
∂x
= ⋅
K ∂x
+ ∑ s −∂px
j =1 j
(5.58)
∂ pj
k kjx
x ∂ H (s ) x ∂ K ∂x
⋅
H (s ) ∂ x
= ⋅
K ∂x
+ ∑ j =1
s − pj
(5.59)
This gives a relationship between the function sensitivity and pole sensitivity:
k
k j S xp j
S xH = S xK + ∑s− p
j =1 j
(5.60)
EXAMPLE 5.4
Consider the transfer function of the second-order active filter
K
H (s ) = (5.61)
ω
s 2 + 1 s + ω12
Q
In the only case of real interest, the poles are complex (Q > 0.5), so that p1 and p2
are conjugate, i.e., p2 = p1* , with
⎛ 1 1 ⎞
p1 = −ω1 ⎜ − j 1− ⎟ (5.62)
⎝ 2Q 4Q 2 ⎠
From the preceding equation and Equation (5.55), assuming that Q and ω1 are func-
tions of x, we calculate
⎛ S xQ ⎞
S xp1 = p1 ⎜ S xω1 − j ⎟ (5.63)
⎝ 4Q 2 − 1 ⎠
178 Active Filters: Theory and Design
and Qxp2 = (Qxp1 )*. From the preceding equation we observe that the location of a
pole is 4Q 2 − 1 ≅ 2Q times more sensitive to variations in ω1 than it is to variation
in Q.
Having established an expression for the pole sensitivity S xp1 in Equation (5.63),
we now investigate its effect on the passband of the biquadratic transfer function
[Equation (5.61)]. To this end we evaluate Equation (5.58) for H(s) under the
assumption that the parameter x does not affect the gain constant:
S xp1 (S p1 )*
S xH ( s ) = + x * (5.64)
s − p1 s − p1
⎛ 2 ω1 ⎞ ω
⎜⎝ 2ω1 + Q s ⎟ S xω1 − 1 sS xQ
⎠ Q
S xH ( s ) = − (5.65)
ω
s 2 + 1 s + ω12
Q
2 2
⎛ω ⎞ ⎛ ωn ⎞
2 (1 − ω n2 ) + ⎜ n ⎟ ⎜Q⎟
⎝Q⎠ ⎝ ⎠
S xH ( jω ) = − 2
S xω1 + 2
S xQ
⎛ω ⎞ ⎛ω ⎞
(1 − ω 2n )2 + ⎜ n ⎟ (1 − ω 2n )2 + ⎜ n ⎟
⎝Q⎠ ⎝Q⎠
or
H H
S xH ( jω ) = Sω1 S xω1 + SQ S xQ (5.66)
2
⎛ω ⎞
2 (1 − ω ) + ⎜ n ⎟
2
⎝ Q⎠
n
H
Sω1 = − 2
(5.67)
⎛ω ⎞
(1 − ω 2n )2 + ⎜ n ⎟
⎝ Q⎠
2
⎛ ωn ⎞
⎜⎝ Q ⎟⎠
H
SQ = 2
(5.68)
⎛ω ⎞
(1 − ω 2n )2 + ⎜ n ⎟
⎝ Q⎠
Sensitivity 179
A better appreciation of the meaning of Equations (5.67) and (5.68) is obtained from
the plot of SωH1 and SQH . As shown in Figure 5.1, both SωH1 and SQH are strong functions
of frequency.
From the aforementioned equations:
At ω = ω1 , we have
( )=1
max SQ
H
(5.69)
( )
max Sω1 ≅
H
1+
Q
1
at
⎛
ω ≅ ω1 ⎜ 1 +
⎝ 2Q ⎠
1 ⎞
⎟ (5.70)
( ) H
min Sω1 ≅ −
1+
1
Q
at
⎛
ω ≅ ω1 ⎜ 1 +
⎝ 2Q ⎠
1 ⎞
⎟ (5.71)
12
Q = 10
8 Q=5
4 Q=2
|H| 0.5
Swt
0
1.0 1.5 2.0 f
–4
–8
–12
1.0
Q=2
0.8
Q=5
|H|
SQ 0.6
Q = 10
0.4
0.2
0.0
0 0.5 1.0 1.5 2.0 f
FIGURE 5.1 Transfer function magnitude sensitivities of pole frequency w1 and quality
factor Q.
180 Active Filters: Theory and Design
H
Note that the extreme values of Sω1 occur approximately at the 3 dB frequencies of
high-Q second-order functions.
For good practical second-order sections with high values of Q , it is more
important to pay attention to low values of S xω1 than to small values of S xQ .
PROBLEMS
5.1. For the circuit shown in Figure P5.1 (assume an ideal op-amp):
(a) Find the transfer function.
(b) To synthesize a normalized Butterworth two-pole transfer function,
namely,
10
H (s ) =
s2 + 2 s + 1
C2
R1 R2
−
+
vi C1
Rb
vo
Ra
5.2 For the circuit shown in Figure P5.2, assume an ideal op-amp:
(a) Find the transfer function.
(b) Determine the various sensitivity functions S xω 0 and S xQ , where x
denotes the R’s and the C’s, respectively.
R2
R1 C
−
+
vi vo
5.3 For the circuit shown in Figure P5.3 (assume an ideal op-amp):
(a) Find the transfer function.
(b) Determine the various sensitivity functions S xH , where x denotes the
R’s and the C’s, respectively.
Sensitivity 181
R2
C2
C1 R1
−
+
vi vo
vi L C vo
5.5 For the circuit shown in Figure P5.5, assume an ideal op-amp:
(a) Find the transfer function.
(b) Determine the sensitivities S xa , S xω1 , S xK , where x denotes the R’s and
the C’s, respectively.
R4
C3
R1 R3
−
+
vi C2 vo
5.6 For the circuit shown in Figure P5.6 (assume an ideal op-amp):
(a) Find the transfer function.
(b) Determine the sensitivities S xω 2 , S xa , SCK1 , SCK4, where x denotes the R’s
and the C’s, respectively.
C4
R3
C1 C3
−
+
vi R2
vo
182 Active Filters: Theory and Design
5.7 For the circuit shown in Figure P5.7 (assume an ideal op-amp):
(a) Find the transfer function.
(b) Determine the sensitivities S xω 0 , S xQ , where x denotes the R’s and the
C’s, respectively.
C4
R3
R1 C3
−
+
vi R2
vo
5.8 For the circuit shown in Figure P5.8 (assume an ideal op-amp):
(a) Find the transfer function.
(b) Determine the sensitivities S xω 2 , S xa, where x denotes the R’s and the
C’s, respectively.
R1
C1 C2
−
+
vi R2
Rb
vo
Ra
5.9 For the circuit shown in Figure P5.9 (assume an ideal op-amp):
(a) Find the transfer function.
(b) Determine the sensitivities S xω 0 , S xQ , where x denotes the R’s and the
C’s, respectively.
C2
R2
C1 R1
−
+
vi vo
6 Filters with GIC
6.1 INTRODUCTION
Impedance converters are active RC circuits designed to simulate frequency-dependent
elements such as impedances for use in active filter synthesis. Among the various
configurations is the generalized impedance converter (GIC), which can be used to
simulate inductances.
This topology allows one to easily realize active filters beginning from a passive
filter design. In addition, the GIC filter provides extremely low distortion and noise,
at a reasonable cost. Compared with more familiar feedback techniques, such as
Sallen–Key filter topologies, the GIC filter has superior noise gain characteristics,
making it particularly suitable for audio and DSP type applications.
Figure 6.2 shows the way to implement a GIC using op-amps. The equivalent Z of
this circuit is V/I. Because each op-amp has v + = v − , the voltage at the input nodes
of both op-amps is v. By Ohm’s law, we have:
node v1
node v2
183
184 Active Filters: Theory and Design
GIC
Z1(s) K(s) Z2(s)
Eliminating V01 and V02, and solving for the ratio V/I, we have:
Y14 + Y15
V01 = V (6.5)
Y14
Y12Y14 − Y13Y15
V02 = V (6.6)
Y12Y14
⎡ Y Y −Y Y ⎤
I = ⎢V − 12 14 13 15 V ⎥ Y11 ∴
⎢⎣ Y12Y14 ⎥⎦
vi Z11 vo
+ vo1
Z12
− −
v1 v
vo2 Z13
+
Z14
v2
v
Z15
V Y Y
Z= = 12 14 ∴
I Y11Y13Y15
Z11Z13Z15
Z= (6.7)
Z12 Z14
1. If Z11 , Z13 , Z14 , and Z15 are resistances and Z12 = 1/ sC , Equation (6.7) gives
R11R13 R15
Z= sC = sL (6.8a)
R14
where
R11R13 R15
L= C (6.8b)
R14
indicating that the circuit simulates a grounded inductance (Figure 6.3). If desired,
this inductance can be adjusted by varying one of the resistances.
2. If Z12 , Z13 , and Z14 are resistances and Z15 are capacitances, Equation (6.7)
gives
R13 1
Z= =− 2 (6.9a)
2
s R13 R14C11C15 ω D
vi R11 vo
+
C12
R11R13R15
− − L= C12
R14
+ R13
R14
R15
C11 vo
R12R14C11C15
+ D=
R13
R12
− − D
+ R13
R14
C15
where
R12 R14C11C15
D= (6.9b)
R13
TABLE 6.1
The 1/s Impedance Transformation
Transformed
Element Impedance Element Impedance
R R C R/s
L sL R L
C 1/sC D 1/s 2C
54767_C006.fm Page 187 Tuesday, September 18, 2007 4:20 PM
EXAMPLE 6.1
A third-order low-pass Butterworth filter must be designed with cutoff frequency 1 kHz.
Solution
From Appendix E we find the normalized RLC filter.
Rs L1 L3
1Ω 1.000 1.000
Vi RL
2.000
1Ω
(a)
1.000 1.000
vi 2.000 D 1.000
1 1.000 1.000
vi 1 vo
1 R11
+
−
1 R12
−
1 C13
+
2 R14 D R11
R14 = =2
R12 • C13 • C15
1 C15
(b)
We cannot use GIC simulation because the RLC filter contains floating induc-
tances. This obstacle is overcome by applying the 1/s transformation, after which
resistances are changed to capacitances, the inductances to resistances, and the
capacitances to D elements.
The D elements are realized using the GIC as shown in the following figure:
Frequency and impedance scaling (FSF = 2π × 10 3 and for C = 10 nF \ ISF =
15.92 ¥ 103) gives the designed active GIC filter (Figure 6.5a) and its frequency
response (Figure 6.5b).
500 K −
15.92 K 15.92 K
+ LM741
10 n
10 n 500 K
15.92 K
15.92 K + LM 741
− −
10 n
LM741 +
31.83 K
10 n
(c)
20.00
0.00
–20.00
Gain dB
–40.00
–60.00
–80.00
100 1K 10 K
Frequency in Hz
(d)
We use 500 KW resistance at the input to provide a dc path for the op-amps. To
ensure a dc gain of 0.5, this resistance must be counterbalanced by a 500 KW resistance
at the output. To avoid loading problems, an output buffer is used. The FDNR can
be tuned by adjusting one of its resistances.
1 L1 L3 L5
C2 0.977 C4 0.794
(a)
1 1.977 1.719
0.188 0.520
Vi 1 Vo
0.977 0.794
(b)
0.188 0.520 1F
vi vo
1 1
+
+
1 − 1
−
1 1
− −
+ +
0.977 0.794
1 1
(c)
FIGURE 6.6 LP elliptic filter: (a) normalized low-pass RLC filter; (b) circuit after 1/s trans-
formation; (c) normalized configuration using GICs for D elements; (d) denormalized filter;
(e) frequency response.
190 Active Filters: Theory and Design
6.37 n 6.37 n
+ LM741
+ LM741
10 K − 10 K
−
10 K 10 K
− −
LM741 + LM741 +
9.8 K 7.9 K
6.37 n 6.37 n
(d)
0.00
−20.00
–40.00
Gain dB
–60.00
–80.00
–100.00
100 1K 100 K 100 K
Frequency in Hz
(e)
EXAMPLE 6.2
A fifth-order low-pass elliptic filter must be designed with f1 = 2.5 kHz, 1 dB pass-
band ripple, and ω s = 1.50.
Solution
From Appendix E we find the normalized RLC filter (Figure 6.6).
Filters with GIC 191
EXAMPLE 6.3
A fifth-order high-pass Butterworth filter must be designed with cutoff frequency
5 kHz.
Solution
From Appendix E we find the normalized RLC filter, and from this we design the
GIC filter (Figure 6.7).
ISF = 10 4; also, we calculate FSF = 2π f2 = 2π × 10 4 and C = Cn /( ISF × FSF ).
(a)
C1 C3 C5
R
1 1 1 1
0.618 2.000 0.618
1 L2 1 L4 R 1
0.618 1.618
(b)
FIGURE 6.7 Fifth-order Butterworth HP filter: (a) prototype low-pass filter; (b) transformed
high-pass filter; (c) high-pass filter using GIC; (d) frequency- and impedance-scaled filter;
(e) its frequency response.
192 Active Filters: Theory and Design
vi 1
vo
1 + 1 +
− −
1 1
1 1
− −
+ +
1F 1F
0.618 0.618
(c)
500 K
10 K 1.59 n 5.15 n
5.15 n
10 K
10 K 10 K
+ LM318 + LM318
3.18 n 3.18 n
− −
− −
10 K 10 K
LM318 + LM318 +
10 K 10 K
6.18 K 6.18 K
(d)
EXAMPLE 6.4
A third-order high-pass elliptic filter 0.1 dB and ω s = 1.5 must be designed with
cutoff frequency 10 kHz.
Filters with GIC 193
0.00
−20.00
–40.00
Gain dB
–60.00
–80.00
–100.00
100 1K 10 K 100 K 1MEG
Frequency in Hz
(e)
Solution
From Appendix E we find the prototype RLC filter from which we design the GIC
filter (Figure 6.8).
R L1 L3
1 0.770 0.770
L2 0.478
vi R 1 vo
C2 0.746
(a)
1 1.299 1.299
2.092
vi 1 vo
1.340
(b)
FIGURE 6.8 Elliptic high-pass filter: (a) RLC normalized low-pass filter; (b) transformed
high-pass filter; (c) high-pass filter using GIC; (d) frequency- and impedance-scaled filter;
(e) frequency response.
194 Active Filters: Theory and Design
1 1.299 1.299
vi 2.092 vo
1
1
+
−
1
−
1
+
1.340
(c)
10 K 2.07 n 2.07 n
3.33 n 10 K
10 K
+ LM741
1.59 n
−
−
10 K
LM741 +
10 K
13.4 K
(d)
0.00
−10.00
–20.00
Gain dB
–30.00
–40.00
–50.00
100 1K 10 K 100 K 1MEG
Frequency in Hz
(e)
Vo G sGL
H (s ) = = = 2 ∴
Vi 1 s LC + sLG + 1
sC + + G
sL
1⎛ s ⎞
Q ⎜⎝ ω o ⎟⎠
H (s ) = 2
(6.10)
⎛ s ⎞ 1⎛ s ⎞
⎜⎝ ω ⎟⎠ + Q ⎜⎝ ω ⎟⎠ + 1
o o
where
1
ω 2o = (6.11)
LC
and
Q = ω o RC (6.12)
vi C L vo
EXAMPLE 6.5
Design a narrow-band band-pass filter with GIC circuit with fo = 1 kHz and Q = 10.
Solution
(a) Design the RLC normalized filter.
For ω o = 1 rad/s, choose C = 1 F ∴ L = 1 H and Q = 10, R = 10 W.
(b) We specify the components for the GIC circuit.
We accept equal resistors and capacitors:
∴
R = ISF × Rn = 10 4 × 10 Ω = 100 kΩ
Cn 1
C = C12 = = = 15.9 nF
k 2π × 10 7
Figure 6.10 depicts the designed narrow-band band-pass filter with GIC.
10
Vi 1 Vo
1
+
1 −
−
1
+
(a)
FIGURE 6.10 (b) Narrow-band band-pass active filter with GIC circuit; (c) its frequency response.
Filters with GIC 197
100 K
Vi 15.9n Vo
10 K
+ LM 741
−
15.9n
LM 741 −
10 K
+
10 K
10 K
(b)
20.00
0.00
–20.00
Gain dB
–40.00
–60.00
–80.00
10 100 1K 10 K 100 K
Frequency in Hz
(c)
1
V Z2 sL +
H (s ) = o = = sC ∴
Vi Z1 + Z 2 1
R + sL +
sC
198 Active Filters: Theory and Design
C
vi vo
2
⎛ s ⎞
⎜ ⎟ +1
⎝ ωo ⎠
H (s ) = 2
(6.13)
⎛ s ⎞ 1⎛ s ⎞
⎜ ⎟ + ⎜ ⎟ +1
⎝ ωo ⎠ Q ⎝ ωo ⎠
where
1
ω 2o = (6.14)
LC
and
ω oL
Q= (6.15)
R
EXAMPLE 6.6
Design a narrow-band band-reject filter with GIC circuit with f0 = 1 kHz and Q = 10.
Solution
For ω o = 1 rad/s, choose C = 1 F and from Equations (6.14) and (6.15), we have L =
1 F and R = 1/Q = 0.1 Ω .
We use equal resistors and capacitors for the GIC circuit:
0.1
Vi 1 Vo
1
+
1 −
−
1
+
(a)
100
159n
Vi Vo
15.9n
1K
+
−
159n
−
1K
+
1K
1K
(b)
FIGURE 6.12 (b) Narrow-band band-reject active filter with GIC f0 = 1 kHz, Q = 10;
(c) frequency response.
Cn 1
C = C12 = = = 15.9 nF
k 2π × 10 7
Figure 6.12a depicts the designed filter, and its frequency response is shown in
Figure 6.12b.
200 Active Filters: Theory and Design
0.50
–0.50
–1.50
Gain dB
–2.50
–3.50
–4.50
10 100 1K 10 K 100 K
Frequency in Hz
(c)
PROBLEMS
6.1 Provided R = L/2C , the circuit of Figure P6.1 yields a third-order low-
pass Butterworth response with 3 dB frequency ω1 = 1/ 2 LC .
(a) Specify suitable components for w1 =1 rad/s.
(b) Convert the circuit to a GIC realization for f1 = 1 kHz
R L
vi C C R vo
6.2 Provided R = L/2C , the circuit of Figure P6.2 yields a third-order high-
pass Butterworth response with 3 dB frequency ω 2 = 1/ 2 LC .
(a) Specify suitable components for ω 2 = 1 rad/s.
(b) Convert the circuit to a GIC realization for f2 = 3 kHz.
R C C
Vi L R Vo
Filters with GIC 201
R L L
1Ω 1.864 H 1.864 H
vi C R vo
1.280 F 1Ω
6.6 (a) Find the transfer function of the filter of Figure P6.4:
(b) If the normalized transfer function of the preceding filter is
1
H n (s ) =
s2 + 2 s +1
Vi Vo
C R
R = 1Ω
6.7 Find an FDNR realization of the second-order high-pass filter using the
results of Problem 6.6 for f2 = 300 Hz.
6.8 Using GICs and information of Appendix E, design a fourth-order But-
terworth LP filter with f1 = 500 Hz.
6.9 Using GICs and information of Appendix E, design a seventh-order low-
pass elliptic 1-dB passband ripple with ω s = 2.00 and f1 = 10 kHz.
6.10 Using GICs and information of Appendix E, design a third-order high-
pass 0.1-dB passband ripple with ω s = 2.00 and f2 = 1 kHz.
7 OTA Filters
7.1 INTRODUCTION
An ideal operational transconductance amplifier (OTA) is a voltage-controlled cur-
rent source. What is important and useful about the OTA’s transconductance param-
eter is that it is controlled by an external current, the amplifier current I b , so that
one obtains
20
gm = I (7.1)
Vi B
io = gm vd (7.2)
or
io = gm (v2 − v1 ) (7.3)
203
204 Active Filters: Theory and Design
node v1
IB
v2 + Io
v1 –
(a)
io
v2 +
gmvd
v1 –
(b)
io
v2
vd Ri
gmvd Ro
v1
(c)
FIGURE 7.1 (a) OTA: symbol, (b) ideal model, (c) equivalent circuit.
OTA Filters 205
vi +
vo vi +
–
gm(vi – v1)
Z2 v1 Z2
v1 – vo
Z1 Z3 Z1 Z3
(a) (b)
FIGURE 7.2 (a) Single OTA filter with three impedance, (b) equivalent circuit.
(Y2 − gm ) Y2
(Y2 + Y3 ) Vo − Vo = gmVi ∴
Y1 + Y2
Vo gm (Y1 + Y2 )
H1 (s) = = (7.6)
Vi Y1Y2 + Y2Y3 + Y1Y3 + gmY2
and
V1 gmY2
H 2 (s ) = = (7.7)
Vi Y1Y2 + Y2Y3 + Y1Y3 + gmY2
From these expressions we can derive different first- and second-order filters.
vi + io
vo vi +
–
gm(vi – vo)
va va – vo
R C R C
(a) (b)
I o = gm (Vi − Vo ) (7.8)
Vo = I o Z L (7.9)
YL = sC + G ∴
1 1
ZL = = (7.10)
YL sC + G
gm (Vi − Vo )
Vo = ∴
sC + G
Vo gm
H (s ) = = ∴
Vi sC + gm + G
K
H (s ) = (7.11)
s
1+
ω1
where
1 + gm R
ω1 = (7.12)
RC
gm R
K= (7.13)
1 + gm R
node vo
vi +
vo vi +
–
gm(vi – vo)
C va C
– vo
R1 R2 R1 R2
(a) (b)
FIGURE 7.4 (a) First-order high-pass OTA filter, (b) its equivalent circuit.
(sC + G1 )(sC + G2 )
Va + ( gm − sC ) Va = gmVi ∴
sC
Va 3gmC
H (s ) = = ∴
Vi sC ( gm + G1 + G2 ) + G1G2
⎛ s ⎞
K⎜ ⎟
⎝ ω2 ⎠
H (s ) = (7.16)
s
1+
ω2
where
gm gm
K= = (7.17)
gm + G1 + G2 1 + gm R
R = R1 / / R2 (7.18)
and
G1G2 1
ω2 = = (7.19)
C ( gm + G1 + G2 ) C (gm R1R2 + R1 + R2 )
(sC1 + G ) Vo − G Va = 0 (7.20)
208 Active Filters: Theory and Design
vi +
vi +
–
gm(vi – vo)
R vo R
vo – va
C1 C2 C1 C2
(a) (b)
node va
(sC1 + G )(sC2 + G )
Vo + ( gm − G ) Vo = gmVi ∴
G
Vo gmG
H (s ) = = 2 ∴
Vi s C1C2 + s(C1 + C2 ) G + gmG
gmG
C1C2
H (s ) = ∴
(C + C 2)G g G
s2 + 1 s+ m
C1C2 C1C2
1
H (s ) = 2
(7.22)
⎛ s ⎞ (C1 + C2 ) G ⎛ s ⎞
⎜⎝ ω ⎟⎠ + ω C C ⎜⎝ ω ⎟⎠ + 1
1 1 1 2 1
where
gm G
ω12 = b = (7.23)
ω1C1C2
and
(C1 + C2 ) G
a= (7.24)
ω1C1C2
OTA Filters 209
Design Equations
For C1n = C2 n = Cn = 1 F and ω1 = 1 rad/s ∴
From Equation (7.24) we have:
2
Rn = (7.25)
a
2b
gmn = (7.26)
a
EXAMPLE 7.1
Design a second-order Butterworth low-pass filter with a cutoff frequency of 100 kHz.
Solution
From Butterworth coefficients we find a = 1.414 and b = 1.000, hence
2 2
Rn = = = 1.414 Ω
a 1.414
2b 2
gmn = = = 1.414 S
a 1.414
ω1
ISF = 10 4 ∴ FSF = = 2 π f1 = 2 π × 10 5 ∴
ωn
k = ISF × FSF = 2 π × 10 9 ∴
Cn 1
C= = = 159.2 pF
k 2 π × 10 9
gmn 1.414
gm = = = 141.4 μS
ISF 10 4
1
Sg1/ma = − S11//Ra = (7.27)
2
SC1/ a = 0 (7.28)
210 Active Filters: Theory and Design
gm = 1414 m S
vi +
R
vo
14.14 KΩ
C1 159.2 pF C2 159.2 pF
These results indicate extremely low-sensitivity performance of this OTA filter. This
is generally true for other OTA filters.
node va
vi +
vi +
–
gm(vi – vo)
R1 vo R1 va
vo –
C1 C2 R2 C1 C2 R2
(a) (b)
FIGURE 7.7 (a) Second-order low-pass filter with four passive components, (b) equivalent
circuit.
OTA Filters 211
Vo gmG1
H (s ) = = ∴
Vi s 2C1C2 + s[C1 (G1 + G2 ) + C2G1 ] + G1 ( gm + G2 )
K
H (s ) = 2
(7.31)
⎛ s ⎞ ⎛ s ⎞
⎜ ⎟ + a⎜ ⎟ +1
⎝ ω1 ⎠ ⎝ ω1 ⎠
where
gm R2
K= (7.32)
1 + gm R2
( gm + G2 ) G1
ω12 = b = (7.33)
C1C2
and
C1 (G1 + G2 ) + C2G1
a= (7.34)
C1C2ω1
If C1 = C2 = C and R1 = R2 = R, we have:
gm R
K= (7.35)
1 + gm R
( gm + G ) G
ω12 = b = (7.36)
C2
3G
a= (7.37)
C ω1
Design Equations
For normalized filter, ω1 = 1 rad/s and Cn = 1 F, we have:
3
Rn = (7.38)
a
212 Active Filters: Theory and Design
1
gmn +
Rn
=b ∴
Rn
9 b − a2
gmn = (7.39)
3a
EXAMPLE 7.2
Design a second-order Butterworth low-pass filter with f1 = 100 kHz.
Solution
From Butterworth coefficients we find a = 1.414, b = 1.000.
3 3
Rn = = = 2.122 Ω
a 1.414
9 b − a 2 9 × 1 − 1.414 2
gmn = = = 1.650 S
3a 3 × 1.414
ω1
ISF = 10 4 ∴ FSF = = 2 π f1 = 2 π × 10 5 ∴
ωn
k = ISF × FSF = 2 π × 10 9 ∴
Cn 1
C= = = 159.2 pF,
k 2π × 10 9
gmn 1.650
gm = = = 165 μS
ISF 10 4
gm R 165 × 10 −6 × 21.2 × 10 3
K= = = 0.779
1 + gm R 1 + 165 × 10 −6 × 21.2 × 10 3
1
SCω1 = − S1ω/ R1 = − (7.40)
2
1 ⎡⎢ ⎛ a ⎞ ⎤
2
Sgωm1 = 1− ⎥ (7.41)
2 ⎢ ⎜⎝ 3 ⎟⎠ ⎥
⎣ ⎦
OTA Filters 213
gm = 165 mS
vi +
R
vo
21 KΩ
C 159 pF 159 pF C R 21 KΩ
SCK = 0 (7.42)
2
⎛ a⎞
S K =⎜ ⎟ (7.43)
gm
⎝ 3⎠
node va
vi +
vi +
–
gm(vi – vo)
C vo C
vo va – va
R C R R C R
(a) (b)
⎡ (sC + G )(2sC + G ) ⎤
⎢ − sC + gm ⎥ Vo = gmVi ∴
⎣ sC ⎦
Vo sC gm
H (s ) = = ∴
Vi s 2C 2 + sC (3G + gm ) + G 2
⎛ s ⎞
Ko ⎜ ⎟
⎝ ωo ⎠
H (s ) = 2
(7.46)
⎛ s ⎞ 1⎛ s ⎞
⎜ ⎟ + ⎜ ⎟ +1
⎝ ωo ⎠ Q ⎝ ωo ⎠
where
G
ωo = (7.47)
C
1 3G + gm
= (7.48)
Q ω oC
gm
Ko = (7.49)
ω oC
For s = j ω o ∴
Kj
H( j ωo) = K = = K oQ ∴
1
j
Q
K
Ko = (7.50)
Q
Design Equations
For ω o = 1 rad/s, Cn = 1 F ∴
Rn = 1 Ω (7.51)
1
= 3 + gm ∴
Q
3Q − 1
gmn = (7.52)
Q
OTA Filters 215
K o = gmn (7.53)
K = Qgmn (7.54)
EXAMPLE 7.3
Design a narrow-band band-pass filter with fo = 100 kHz and Q = 10.
Solution
3Q − 1 3 × 10 − 1
gmn = = = 2.9 S
Q 10
Cn = 1 F and Rn = 1 Ω
ISF = 10 4 ∴
ωo
FSF = = 2 π fo = 2 π × 10 5 ∴
ωn
Cn 1
C= = = 159 pF
ISF × FSF 2 π × 10 9
R = ISF × Rn = 10 4 × 1 Ω = 10 kΩ
gmn 2.9
gm = = = 290 μS
ISF 10 4
gm = 290 mS
vi +
C
vo
159 pF
R R
10 K C 159 pF 10 K
FIGURE 7.10 Narrow-band band-pass filter with fo = 100 kHz and Q = 10.
216 Active Filters: Theory and Design
Va = V1+ (7.56)
For OTA2:
1 V − Vo
V2+ = I o1 = gm1 a (7.58)
sC1 sC1
gm1gm 2 (Va − Vo )
I o 2 = gm 2 (V2+ − V2− ) = (7.59)
sC1
I o2
V3+ = Vb and V3− = Vo = + Vc (7.60)
sC2
v1 – v2
gm1 +
va v3
+ gm2 –
C1 – gm3 vo
C2 +
vc
vb
At the output of both OTA2 and OTA3, Kirchhoff’s current law gives:
gm1gm 2 (Va − Vo )
+ gm3 (Vb − Vo ) = (Vo − Vc ) sC2 ∴
sC1
For Vc = 0, Va = 0, and Vb = Vi ∴
Vo sgmC1
H BP (s) = = 2 ∴
Vi s C1C2 + sgm3C1 + gm1gm 2
1⎛ s ⎞
⎜ ⎟
Q ⎝ ωo ⎠
H BP (s) = 2
(7.64)
⎛ s ⎞ 1⎛ s ⎞
⎜ ⎟ + ⎜ ⎟ +1
⎝ ωo ⎠ Q ⎝ ωo ⎠
where
gm1gm 2
ω 2o = (7.65)
C1C2
ω oC2
Q= (7.66)
gm3
By setting gm1 = gm 2 = gm , it is clear from Equation (7.65) that the center frequency
can be literally dependent on gm .
gm
ωo = or
C1C2
gm
fo = (7.67)
2 π C1C2
218 Active Filters: Theory and Design
At the same time, gm3 can be separately adjusted to yield a controllable value for
the filter Q:
1 gm
Q= C2 ∴
gm3 C1C2
gm C2
Q= (7.68)
gm3 C1
Therefore, the band-pass filter realized with this circuit has independently controlled
fo and Q.
If Vb = Vc = 0 and Va = Vi ∴
Vo gm1gm 2
H LP (s) = = ∴
Vi s 2C1C2 + gm3C1 s + gm1gm 2
1
H LP (s) = 2
(7.69)
⎛ s ⎞ gm3 ⎛ s ⎞
⎜ ⎟ + ⎜ ⎟ +1
⎝ ω1 ⎠ ω1C2 ⎝ ω1 ⎠
where
gm1gm 2
ω12 = b = (7.70)
C1C2
gm3
a= (7.71)
C2
For gm1 = gm 2 = gm ∴
gm3
C2 = (7.72)
a
and
a gm2
C1 = (7.73)
b gm3
If Vb = Va = 0 and Vc = Vi ∴
OTA Filters 219
s 2C1C2
H HP (s) = ∴
s C1C2 + gm3C1 s + gm1gm 2
2
2
⎛ s ⎞
⎜ ⎟
⎝ ω2 ⎠
H HP (s) = 2
(7.74)
⎛ s ⎞ gm3 ⎛ s ⎞
⎜ ⎟ + ⎜ ⎟ +1
⎝ ω2 ⎠ ω 2C2 ⎝ ω 2 ⎠
where
gm3
a= (7.75)
ω 2C2
and
gm1gm 2
ω 22 = (7.76)
C1C2
a gm2
C1n = (7.77)
b gm3
gm3
C2 n = (7.78)
a
io
v2 +
gm(s)(v2 – v1)
Ri Ci
Ro Co
v1 –
output capacitances but also those contributed by wiring. The input resistor Ri is
from >10 MW up to >100 MW (in CMOS circuits), so that its effect can be neglected.
The output Ro is usually small, of the order of 100 KW or less, so that its effect must
be included.
The transconductance frequency dependence can be modeled via a dominant
pole or excess phase shift:
gmo
gm (s) = (7.79)
s
1+
ωb
where ω b is the bandwidth of the OTA and gmo is the dc transconductance. The phase
shift model is also often used, which is
where φ is the phase delay, τ = 1/ ω b is the time delay, and φ = ωτ, when ω << ω b .
Some typical values of OTA (CMOS) parameters are:
PROBLEMS
7.1 For the circuit of Figure P7.1, find the input impedance.
ii
vi – io
Zi gm
+
OTA Filters 221
v1 +
gm1
–
–
gm3 vo
+
v2 +
gm2
–
gm1 g
Vo = V1 + m 2 V2
gm3 gm3
v1 +
gm vo
v2 –
C
∫ (v − v ) dt
gm t
vo = 2 1
C 0
vi +
gm vo
–
Ans.
gm (sRC + 1)
H (s ) =
sC + gm
222 Active Filters: Theory and Design
7.5 Prove that the transfer function of the filter of Figure P7.5 is
vi +
gm
–
C
vo
R C R
⎛ s ⎞
Ko ⎜ ⎟
Vo ⎝ ωo ⎠
H (s ) = =
Vi ⎛ s ⎞ 2 1 ⎛ s ⎞
⎜ ⎟ + ⎜ ⎟ =1
⎝ ωo ⎠ Q ⎝ ωo ⎠
where
1 ω RC 1
ωo = , Q= o and K o =
RC 3 + gm R Q
vi +
gm vo
–
Ans.
gm
H (s ) =
sC + gm
vi +
gm vo
–
R C
OTA Filters 223
Ans.
gm
H (s ) =
sC + G
7.8 Find the transfer functions of Figure P7.8.
Vo V
(a) H (s) = , (b) H (s) = a
Vi Vi
vi +
gm vo
–
R1
va
C R2
Ans.
sgmC + gmG1
(a) H (s) =
sC (G1 + G2 ) + G1 ( gm + G2 )
gmG1
(b) H (s) =
sC (G1 + G2 ) + G1 ( gm + G2 )
v1 –
gm1 –
v2 + gm2 vo
C +
Ans.
gm1
Vo = (V − V )
sC + gm 2 2 1
– R R
gm vo
vi +
C C R
224 Active Filters: Theory and Design
Ans.
K gm G ( gm + G )
H (s ) = , K= , ω12 =
2
⎛ s ⎞ 1⎛ s ⎞ gm + G 2C 2
⎜ ⎟ + ⎜ ⎟ +1
⎝ ω1 ⎠ Q ⎝ ω1 ⎠
2ω1RC
Q=
5
vi +
gm1 +
C1 – gm2 vo
C2 –
Switched Capacitor
8 Filters
8.1 INTRODUCTION
Many active filters with resistors and capacitors have been replaced with a filter
called a switched capacitor filter (SC filter). The switched capacitor filter allows for
very sophisticated, accurate, and tunable analog circuits to be manufactured without
using resistors. The main reason is that resistors are hard to build on integrated
circuits because they take up a lot of space, and the circuits can be made to depend
on ratios of capacitor values, which can be set accurately, and not absolute values,
which vary between manufacturing runs.
Δq = C (V2 − V ) (8.1)
n
Δ q = C (V2 − V ) (8.2)
Δt
The left-hand side of this equation represents current, and the number of cycles per
unit time is the switching frequency or clock frequency, fCLK , hence:
i = C (V2 − V ) fCLK
Rearranging, we get
V2 − V1 1
Req = = (8.3)
i C fCLK
225
226 Active Filters: Theory and Design
Req
V1 V2 V1 V2
C
(a) (b)
f1 f2 f1
V1 V2
f2
C
(c)
FIGURE 8.1 (a) Switched capacitor; (b) approximate resistor; (c) MOS implementation;
clock signals φ1 and φ2 .
This equation indicates that the switched capacitor behaves approximately like a
resistor. The value of the resistor decreases with increasing switching frequency or
increasing capacitance.
q = C1Vi
R Clock C2
−
+
vi vo Vi −
+ Vo
C1
(a) (b)
Because the clock frequency fCLK is much higher than the frequency being filtered, Vi
does not change, while C1 is being changed.
During the second half of the clock period, the charge C1Vi is transferred to the
feedback capacitor because v − = v + = 0. The total transfer of charge in one clock
cycle is
q = C1Vi (8.4)
q C1Vi
ii = = = C1Vi fCLK (8.5)
T T
Vi 1
R= = (8.6)
ii C1 fCLK
C2
RC = (8.7)
C1 fCLK
1
f1 =
2π RC
C1
f1 = f (8.8)
2π C2 CLK
fCLK fCLK
fc = or (8.10)
50 100
N/AP/HPA
+
VA S1A BPA LPA
7 3 5 2 1
4
INVA − –
+
–
∫ ∫
+
15
AGND
10
CLKA Level Non Overl
Shift Clock
12 6
50/100 Control SAB
9
L.Sh.
FIGURE 8.3 Half-block diagram of the MF100 Universal monolithic dual SC filter.
Switched Capacitor Filters 229
The pins are described in the data sheet, and we will describe a few of them:
Each section can be configured for a variety of different modes. The mode of Figure
8.4 is referred to as the state variable mode (mode 3) because it provides the high-
pass, band-pass, and low-pass responses by direct consecutive integrations; other
modes can be found in the data sheets and application notes.
node va
1
VBP = VHP (8.12)
s
R4
R3
R2
R1
vi − –
va + +
– ∫ ∫ VLP
+
VHP VBP
1
VLP = VBP (8.13)
s
VLP G1
H LP (s) = =− 2 ∴
Vi s G2 + s G3 + G4
R2
R1
H LP (s) = −
R R
s2 + 2 s + 2
R3 R4
For s → 0 ∴
R4
H LP (0) = K = (8.16)
R1
R2
R1
H LP (s) = − 2 (8.17)
s + as + b
where
R2
a= (8.18)
R3
R2
ω12 = b = (8.19)
R4
Switched Capacitor Filters 231
R4 n = K R1n = K Ω (8.21)
R2 n = bK Ω (8.22)
R2 n b K
R3n = = Ω (8.23)
a a
EXAMPLE 8.1
Design a low-pass Butterworth filter with the following specifications: f1 = 1 kHz,
fs = 2.5 kHz, Amax = 3 dB, and K = 1.
Solution
From the Butterworth nomographs, we find: n = 5, and from Butterworth coefficients,
we have:
For K = 1 we have:
R1n = R2 n = R4 n = 1 Ω
First stage
R2 n 1
R3n = = = 0.617 Ω
a 1.618
Second stage
R2 n 1
R3n = = = 1.618 Ω
a 0.618
Third stage
Rn = 1 Ω
1
Cn = =1F
b
232 Active Filters: Theory and Design
ISF = 2 × 10 4; we calculate
ω1
FSF = = 2π f1 = 2π × 10 3
ωn
R1 = R2 = R4 = ISF × R1n = 2 × 10 4 × 1 Ω = 20 kΩ
First stage
Second stage
Third stage
R = ISF × Rn = 2 × 10 4 × 1 Ω = 20 kΩ
Cn 1
C= = ≅ 8 nF
ISF × FSF 2 × 10 4 × 2π × 10 3
fCLK = 50 f1 = 50 × 1 kHz = 50 kHz
Figure 8.5 shows the designed filter using MF5 (half of MF100).
⎛1 ⎞ ⎛1 ⎞
G4 ⎜ VBP ⎟ + G2VHP + G3 ⎜ VHP ⎟ = −G1Vi ∴
⎝s ⎠ ⎝s ⎠
⎛11 ⎞ G
G4 ⎜ VHP ⎟ + G2VHP + 3 VHP = −G1Vi ∴
⎝s s ⎠ s
VHP s 2G1
H HP (s) = =− 2 ∴
Vi s G2 + sG3 + G4
R2 2
s
R1
H HP (s) = − (8.24)
R R
s2 + 2 s + 2
R3 R4
1 14 8 nF R3 12 K 1 14
R3 33 K
BP O/P LP O/P C BP O/P LP O/P
R2 20 K 2 13 R R2 20 K 2 13
N/AP/HP VO2 N/AP/HP VO2
Switched Capacitor Filters
3 12 20 K 3 12
vi INV1 INV2 INV1 INV2
R1 20 K 4 11 R1 20 K 4 11
S1 MF5 AGND S1 MF5 AGND
5 10 −5 V −5 V 5 10
−5 V SA −V SA −V −5 V
6 9 +5V 6 9
+5 V +V 50/100 100 nF +V 50/100 100 nF
7 8 7 8
100 nF L Sh CLK L Sh CLK
100 nF
5V
0 FCLK = 50 kHz
For s → ∞ ∴
R2
H HP (∞) = K = (8.25)
R1
K
H HP (s) = − (8.26)
s + as + b
2
where
R2
ω 22 = b = (8.27)
R4
R2
a= (8.28)
R3
R2 n = K (8.29)
1
R3n = (8.30)
a
K
R4 n = (8.31)
b
EXAMPLE 8.2
Design an SC high-pass Butterworth filter with the following specifications:
Solution
From the Butterworth nomographs, we find n = 4.
First stage
a = 1.848, b = 1.000
1 1
R1n = 1 Ω, R2 n = K = 1 Ω, R3n = = = 0.541 Ω
a 1.848
K 1
R4 n = = = 1Ω
b 1
Switched Capacitor Filters 235
Second stage
a = 0.766, b = 1.000
R1n = 1 Ω, R2 n = K = 1 Ω
1 1
R3n = = = 1.305 Ω
a 0.766
K
R4 n = = 1Ω
b
ISF = 2 × 10 4 ∴
ω1
FSF = = 2π f1 = 2π × 430 = 8.6π × 10 2
ωn
R1 = R2 = R4 = ISF × R1n = 2 × 10 4 Ω = 20 kΩ
First stage
Second stage
⎛1 ⎞
G4 ⎜ VBP ⎟ + G2 (s VBP ) + G3VBP = − G1Vi ∴
⎝s ⎠
R2
V R1
H BP (s) = BP = − ∴
Vi R R
s2 + 2 s + 2
R3 R4
236 Active Filters: Theory and Design
R1B 20 K
R4A 20 K 1 20 R4B 20 K
LPA LPB
R3A 10.8 K 2 19 R3B 26.1 K
BPA BPB
R2A 20 K 3 18 R2B 20 K
N/AP/HPA N/AP/HPB
4 17
vi INVA INVB
R1A 20 K 5 16 vo
SIA SIB
6 LMF100 15
−5 V SA/B AGND
7 14
VA+ VA
8 13
+5 V
+
VD
VD −5 V
9 12 100 nF
50/100/CL
100 nF L Sh.
11
10 CLKB
CLKA
5V
0 fCLK = 43 kHz
R2 ⎛ s ⎞
ω o R1 ⎜⎝ ω o ⎟⎠
H BP (s) = − 2
(8.32)
⎛ s ⎞ 1⎛ s ⎞
⎜⎝ ω ⎟⎠ + Q ⎜⎝ ω ⎠⎟ + 1
o
where
R2
ω 2o = (8.33)
R4
1 R2
= (8.34)
Q ω o R3
R3
H BP ( j ω o ) = K = − (8.35)
R1
Switched Capacitor Filters 237
R1n = 1 Ω (8.36)
From Equations (8.35) and (8.33), we have:
R3n = K (8.37)
K
R2 n = R4 n = (8.38)
Q
EXAMPLE 8.3
Design a narrow-band band-pass filter with fo = 750 Hz, Q = 10, and K = 5.
Solution
K 5
R1n = 1 Ω, R2 n = = = 0.5 Ω, R3n = K = 5 Ω
Q 10
R4 n = R2 n = 0.5 Ω
ISF = 2 × 10 4 ∴
R1 = ISF × R1n = 2 × 10 4 × 1 Ω = 20 kΩ
R2 = R4 = ISF × R2 n = 2 × 10 4 × 0.5 Ω = 10 kΩ
EXAMPLE 8.4
Design a Butterworth band-pass filter with the following specifications: BW = 2 to 8 kHz ,
fs1 = 16 kHz, fs 2 = 1 kHz, Amax = 3 dB, Amin = 12 dB, and K = 9.
Solution
From the Butterworth nomographs, we find n = 2.
(a) Low-pass filter
a = 1.414, b = 1.000
K1 = K 2 = K = 9 =3
R1n = 1 Ω, R2 n = K R1n = 3 × 1 Ω = 3 Ω, R4 n = 3 Ω
3 3
R3n = = = 2.121 Ω
a 1.414
238
R4 5 K
vo
R3 100 K 1 14
BP O/P LP O/P
R1 5 K 2 13
N/AP/HP VO2
R1 20 K 3 12
vi INV1 INV2
4 11
S1 MF5 AGND
5 10
−5 V SA −V −5 V
6 + 9
+5 V V 50/100
100 nF 7 8 5V
L Sh CLK 0
fCLK = 75 kHz
100 nF
ISF = 2 × 10 4 ∴
R1 = ISF × R1n = 2 × 10 4 × 1 Ω = 20 kΩ
R2 = R4 = ISF × R2 n = 2 × 10 4 × 3 Ω = 60 kΩ
R1n = 1 Ω, R2 n = 3R1n = 3 Ω, R4 n = 3 Ω
3 3
R3n = = = 2.121 Ω
a 1.414
ISF = 2 × 10 4 ∴
R1 = ISF × R1n = 2 × 10 4 × 1 Ω = 20 kΩ
R2 = R4 = ISF × R2 n = 2 × 10 4 × 3 Ω = 60 kΩ
R3A 42.4 K 1 60 K
LPA LPB
2 R3B 42.4 K
BPA BPB
R2A 60 K 3 R2B 60 K
N/AP/HPA N/AP/HPB
R1A 20 K 4 vo
vi INVA INVB
5 16
SIA SIB
6 LMF100 15
−5 V SA/B AGND
14
7
VA+ VA−
8 −
+5 V + VD −5 V
VD
100 nF 9 12 100 nF
50/100/CL
L Sh. 11
10 CLKB
CLKA
5V
0
fCLK = 400 kHz
R R
SC Filter
C C
vi vo
FIGURE 8.9 RC passive antialias filter, f1a, and to remove sampling steps from the output of
SC filters, f1b.
PROBLEMS
8.1 The circuit of Figure P8.1 provides the notch, band-pass, and low-pass
responses (node 1), with the notch frequency fz and the response frequency fo
independent prove that
R2 R R
K N = K LP = , K BP = 3 , Q = 3
R1 R1 R2
Switched Capacitor Filters 241
R4
R2 N S1 BP
LP
R1
vi − –
+
– ∫ ∫
+
AGND
R3
fCLK f
fz = fo = f1 , f1 = or f1 = CLK
100 50
8.2 The circuit of Figure P8.2 provides the all-pass and low-pass responses
(mode 4). Prove that
R2 AP BP LP
S1
R1
vi − –
+
– ∫ ∫
+
AGND
R3
R3 R R
Qp = , Qz = 3 , K = 2
R2 R1 R1
⎛ R ⎞ R ⎛ R2 ⎞
K LP = ⎜ 1 + 2 ⎟ , K BP = 3 ⎜⎝ 1 + R ⎟⎠
⎝ R1 ⎠ R2 1
8.3 The circuit of Figure P8.3 provides the notch, band-pass, and low-pass
responses (mode 1). Prove that
R2 R R
K LP = K N = , K BP = 3 , Q = 3 ( f N = fo )
R1 R1 R2
242 Active Filters: Theory and Design
R2 N S1 BP LP
− –
+
– ∫ ∫
+
SA/B AGND
R3
V+
8.4 The circuit of Figure P8.4 provides band-pass and low-pass responses
(mode 2). Prove that
R2 N S1 BP LP
− –
+
– ∫ ∫
+
SA/B AGND
R3
V+
fCLK R fCLK R2
fo = 1+ 2 or 1+ , fo : center frequency
100 R4 50 R4
R2 R2
1+
R4 R1 R
Q= , K LP = , K BP = 3
R2 R2 R1
1+
R4 R4
R2
R1 R ⎛ f ⎞
KN1 = ( f → 0), K N 2 = 2 ⎜ f → CLK ⎟
R2 R1 ⎝ 2 ⎠
1+
R4
Switched Capacitor Filters 243
8.5 The circuit of Figure 8.5 provides numerator complex zeros, BP, and LP
responses (mode 5). Prove that
R4
R2 C.Z BP
LP
R1
vi − –
+
– ∫ ∫
+
SA/B AGND
R3
V+
fCLK R fCLK R2
fo = 1+ 2 or 1+
100 R4 50 R4
fCLK R fCLK R1
fo = 1+ 1 or 1+
100 R4 50 R4
R3 R2 R3 R1
Qp = 1+ , Qz = 1+
R2 R4 R1 R4
R2 ( R4 − R1 ) R ⎛ f ⎞
K z1 = ( f → 0), K z 2 = 2 ⎜ f → CLK ⎟
R1 ( R2 + R4 ) R1 ⎝ 2 ⎠
R3 ⎛ R2 ⎞ R4 ⎛ R1 + R2 ⎞
K BP = ⎜⎝ 1 + R ⎟⎠ , K LP = R ⎜⎝ R + R ⎟⎠
R2 1 1 2 4
8.6 The circuit of Figure P8.6 provides single-pole HP and LP filters (mode 6a).
Prove that
R2 HP S1 LP
R1
vi − –
+
– ∫ ∫
+
SA/B AGND
R3
−V
244 Active Filters: Theory and Design
fCLK R2 fCLK R2
fc = or fc = , fc : f1 or f2
100 R3 50 R3
R3 R
K LP = − , K HP = − 2
R1 R1
8.7 The circuit of Figure P8.7 provides single-pole LP (inverting and nonin-
verting) (mode 6b). Prove that
R2 HP S1 LP
R1
vi − –
+
– ∫ ∫
+
SA/B AGND
R3
−V
fCLK R2 fCLK R2
fc = or fc =
100 R3 50 R3
R3
K LP1 = 1 (noninverting), K LP2 = (inverting)
R2
The ratio of the clock frequency to the low-pass cutoff frequency is internally set
to 100:1.
Design a Butterworth filter to satisfy the following specifications: Amax = 3 dB,
f1 = 3 kHz, 3 kHz, Amin = 35 dB, and fs = 6 kHz.
Vout INV1 V01
3 13 4
7 6th order +
Vin Op-Amp 1
Switched Capacitor Filters
butterworth LPF
f1 f2
Vos 8 5
AGND
adj
V+
2
Non-overlapping V02
clock generator V−
14
INV2
CLK 9 −
+
in +
− 1
Op-Amp2 n. INV2
11 12 6 10
CLK R L. Sh V+ V−
245
Appendix A: Node Voltage
Network Analysis
The node-voltage method of circuit analysis is a method in which the Kirchhoff
voltage law (KVL) equations are written implicitly on the circuit diagram so that
only the Kirchhoff current law (KCL) equations need to be solved. The method also
permits a minimum number of voltage variables to be assigned. The method will
be developed through a study of the circuit of Figure A.1. In this circuit, two unknown
voltages, v1 and v2, are chosen. The voltage v1 is chosen as a voltage rise from node 3
to node 1; v2 is similarly chosen as a voltage rise from node 3 to node 2. As node 3
is the point from which the unknown voltages are measured, it is called the reference
node.
The voltage rise from node 2 to node 1 is the third unknown voltage in the
circuit; it is found from the KVL equation to be
V21 = V1 − V2
There are three nodes in the circuit, and as a consequence, two independent KCL
equations can be written, assuming that all branch currents are leaving the node.
node v1
V1 − Va V1 V1 − V2
+ + =0 (A.1)
Z1 Z2 Z3
V2 − V1 V2 V2 − Vb
+ + =0 (A.2)
Z3 Z4 Z5
⎛ 1 1 1⎞ 1 1
⎜⎝ Z + Z + Z ⎟⎠ V1 − Z V2 = Z Va (A.3)
1 2 3 3 1
1 ⎛ 1 1 1⎞ 1
− V1 + ⎜ + + ⎟ V2 = V (A.4)
Z3 ⎝ Z3 Z 4 Z5 ⎠ Z5 b
247
248 Active Filters: Theory and Design
Z1 v1 Z3 v2 Z5
va Z2 Z4 vb
v3
or
Examination of Equations (A.5) and (A.6) shows a pattern that will permit
equations of this type to be written readily by inspection. In Equation (A.5), written
at node 1, the coefficient of v1 is the positive sum of the admittance connected to
node 1. The coefficient of v2 is the negative sum of the admittances connected
between nodes v1 and v2. The right-hand side of the equation is the sum of the current
sources feeding into node 1.
Now consider Equation (A.6), written for node 2. An analogous situation exists:
The coefficient of v2 is the positive sum of the admittances connected to node 2; the
coefficient of v1 is the negative sum of the admittances between nodes 2 and 1; and
the right-hand side of the equation is the sum of the current sources feeding into 2.
The fact that these two equations are similar in structure is not a coincidence. It
follows from KCL equations and the manner in which the voltage variables are
selected.
The formal procedure for writing equations of the type represented by Equations
(A.5) and (A.6) is called the node-voltage method.
EXAMPLE A.1
Use the node-voltage to determine the transfer function in the circuit of Figure A.2.
v2 = v0
R v1 R v2
vi C C vo
Solution
node v1
node v2
sC + G
V1 = V0 (A.9)
G
(sC + G )(sc + 2G )
V0 − GV0 = GVi ∴
G
V0 G2
H (s ) = = 2 2 ∴
Vi s C + 3sCG + G 2
1
H (s ) = (A.10)
s R C + 3sRC + 1
2 2 2
Appendix B: Filter Design
Nomograph
B.1 BUTTERWORTH LP FILTER DESIGN NOMOGRAPH
f1 fs f
Amax
Passband Stopband
Amin
Transition Region
12
23
21
19
? 11 17
15
13
? 12
11
10
? 10
?
? 9
?
9
?
8 8
?
? ?
? 7 7
? ?
? 6
? ?
? 6 5
? ?
?
? ? 5 4
?
?
? 4
? ? 3
?
? ?
? 3
2
?
?
2
1
1
0
1 2 3 4 5 6 7 8 9 10
fs/fl
251
252 Active Filters: Theory and Design
f1 fs f
Amax
Passband Stopband
Amin
Transition Region
12
? 11 13
10
? 11
9 8
? 10
7
?
? 9
? 6
? 5
8
? 4
? ?
? 7
? ?
?
? ?
? 6
? 3
?
?
? ? 5
?
?
? 4
? ? 2
?
? ?
? 3
?
?
2
1
0
1 2 3 4 5 6 7 8 910
fs/fl
Appendix C: First- and Second-
Order Factors of Denominator
Polynomial
n
D = (s + ci ) ∏ (s
i =1
2 + ái s + bi )
Butterworth
n
1 (s + 1)
2 (s 2 + 1.414 s + 1)
3 (s + 1)(s 2 + s + 1)
4 (s 2 + 1.848s + 1)(s 2 + 0.765s + 1)
5 (s + 1)(s 2 + 1.618s + 1)(s 2 + 0.765s + 1)
6 (s 2 + 1.932s + 1)(s 2 + 1.414 s + 1)(s 2 + 0.518s + 1)
7 (s + 1)(s 2 + 1.802s + 1)(s 2 + 1.247s + 1)(s 2 + 0.445s + 1)
8 (s 2 + 1.962s + 1)(s 2 + 1.663s + 1)(s 2 + 1.111s + 1)(s 2 + 0.390 s + 1)
Chebyshev 0.1 dB
n
1 (s + 6.552)
2 (s 2 + 2.372s + 3.314)
3 (s + 0.969)(s 2 + 0.969s + 1.690)
4 (s 2 + 1.275s + 0.623)(s 2 + 0.528s + 1.330)
5 (s + 0.539)(s 2 + 0.872s + 0.636)(s 2 + 0.333s + 1.195)
6 (s 2 + 0.856s + 0.263)(s 2 + 0.626s + 0.696)(s 2 + 0.229s + 1.129)
7 (s + 0.337)(s 2 + 0.679s + 0.330)(s 2 + 0.470 s + 0.753) (s 2 + 0.168s + 1.069)
8 (s 2 + 0.643s + 0.146)(s 2 + 0.545s + 0.416)(s 2 + 0.364 s + 0.779)(s 2 + 0.128s + 1.069)
253
254 Active Filters: Theory and Design
Chebyshev 0.5 dB
n
1 (s + 2.863)
2 (s 2 + 1.426s + 1.516)
3 (s + 0.626)(s 2 + 0.626s + 1.142)
4 (s 2 + 0.847s + 0.356)(s 2 + 0.351s + 1.064)
5 (s + 0.362)(s 2 + 0.586s + 0.477)(s 2 + 0.224 s + 1.036)
6 (s 2 + 0.580 s + 0.157)(s 2 + 0.424 s + 0.590)(s 2 + 0.155s + 1.023)
7 (s + 0.256)(s 2 + 0.462s + 0.254)(s 2 + 0.319s + 0.677) (s 2 + 0.114 s + 1.016)
8 (s 2 + 0.439s + 0.088)(s 2 + 0.372s + 0.359)(s 2 + 0.248s + 0.741)(s 2 + 0.087s + 1.012)
Chebyshev 1 dB
n
1 (s + 1.965)
2 (s 2 + 1.098s + 1.103)
3 (s + 0.494)(s 2 + 0.494 s + 0.994)
4 (s 2 + 0.674 s + 0.279)(s 2 + 0.279s + 0.987)
5 (s + 0.289)(s 2 + 0.468s + 0.429)(s 2 + 0.179s + 0.988)
6 (s 2 + 0.464 s + 0.125)(s 2 + 0.340 s + 0.558)(s 2 + 0.124 s + 0.991)
7 (s + 0.205)(s 2 + 0.370 s + 0.230)(s 2 + 0.256s + 0.653) (s 2 + 0.091s + 0.993)
8 (s 2 + 0.352s + 0.070)(s 2 + 0.298s + 0.341)(s 2 + 0.199s + 0.724)(s 2 + 0.070 s + 0.994)
Chebyshev 2 dB
n
1 (s + 1.308)
2 (s 2 + 0.804 s + 0.823)
3 (s + 0.369)(s 2 + 0.369s + 0.886)
4 (s 2 + 0.506s + 0.222)(s 2 + 0.210s + 0.929)
5 (s + 0.218)(s 2 + 0.353s + 0.393)(s 2 + 0.135s + 0.952)
6 (s 2 + 0.351s + 0.100)(s 2 + 0.257s + 0.533)(s 2 + 0.094 s + 0.966)
7 (s + 0.155)(s 2 + 0.280 s + 0.212)(s 2 + 0.194 s + 0.635) (s 2 + 0.069s + 0.975)
8 (s 2 + 0.266s + 0.057)(s 2 + 0.226s + 0.327)(s 2 + 0.151s + 0.710)(s 2 + 0.053s + 0.980)
Appendix C: First- and Second-Order Factors of Denominator Polynomial 255
Chebyshev 3 dB
n
1 (s + 1.002)
2 (s 2 + 0.645s + 0.708)
3 (s + 0.299)(s 2 + 0.299s + 0.839)
4 (s 2 + 0.411s + 0.196)(s 2 + 0.170 s + 0.903)
5 (s + 0.178)(s 2 + 0.287s + 0.377)(s 2 + 0.110 s + 0.936)
6 (s 2 + 0.285s + 0.089)(s 2 + 0.209s + 0.522)(s 2 + 0.07s + 0.955)
7 (s + 0.126)(s 2 + 0.228s + 0.204)(s 2 + 0.158s + 0.627) (s 2 + 0.056s + 0.966)
8 (s 2 + 0.217s + 0.050)(s 2 + 0.184 s + 0.321)(s 2 + 0.123s + 0.704)(s 2 + 0.043s + 0.974)
Bessel
n
1 (s + 1.000)
2 (s 2 + 3.000 s + 3.000)
3 (s + 2.322)(s 2 + 3.678s + 6.459)
4 (s 2 + 4.208s + 11.488)(s 2 + 5.792s + 9.140)
5 (s + 3.647)(s 2 + 4.649s + 18.156)(s 2 + 6.704 s + 14.272)
6 (s 2 + 5.032s + 26.514)(s 2 + 7.471s + 20.853)(s 2 + 8.497s + 18.801)
7 (s + 4.972)(s 2 + 5.371s + 36.597)(s 2 + 8.140 s + 28.937)(s 2 + 9.517s + 25.666)
8 (s 2 + 5.678s + 48.432)(s 2 + 8.737s + 38.569)(s 2 + 10.410 s + 33.935)(s 2 + 11.176s + 31.977)
Appendix D: Formulas
of Normalized Filters
ω c = ω1 = ω 2 = ω 0 = 1 rad/s
R
+
R(Ω) C(F)
−
vi C
Rb 1
vo 1
b
Ra
(a)
(b) Second-order
C2
R = 1 Ω, Ra = 1 Ω, Rb = K − 1 Ω
R K C 1( f ) C2( f )
R
+
a+ a2 + 8b (K − 1) 1
− >1
C1 4b bC1
vi Rb
vo a 2
1
2b a
Ra a 1 1
3−
b b b
(b)
257
258 Active Filters: Theory and Design
R2
C2
R 1 = R3 = R = 1 Ω
R1 R3
− R2 (Ω) C1(F ) C2 (F )
C1 +
vi vo 2K + 1 a
1 aK (2K + 1)b
(a)
(b) Band-pass
(i) BPF with one op-amp
C2
R2 C=1F Q ≤ 10
K R1(Ω) R2(Ω) R3(Ω)
R1 C1
− R2 Q Q
2R1 K 2Q 2Q2 − K
+
vi R3 vo
2Q2 Q ∞
2Q
K
kQ Q(1 + M) k
k = 22, 32, ... M = K=
2Q − k M
C Q
R Q
− R3 Q2 − (Q + 1)
−
+
vi R1 +
vo
(b)
Appendix D: Formulas of Normalized Filters 259
(c) Notch
R2
R1 C C=1F R3 = 1 Ω Q ≤ 10
−
+ R1(Ω) R2(Ω) R4(Ω)
R3 1
2Q 2Q2
vo 2Q
vi
R4
(c)
R3
R2 C C
Rg
− R R
−
+ −
vi +
vHP +
vLP
R1
Rq vBP
C=1F Rg = Rq = R = 1 Ω Q ≤ 100
(a)
260 Active Filters: Theory and Design
Rf
Rq
Cf Cf
R
Rg
− R Rf
−
+ −
vi +
vHP +
vLP
R=1Ω Cf = 1 F Q ≤ 100
Rg(Ω) Rq(Ω) Rf (Ω)
1 1 1
LPF
K b a b
Rq
BPF Q 1
K
BPF 1 Q 1 K = Q Free Parameter
(b)
Appendix E
TABLE E1
Element Values for Low-Pass LC Butterworth Filters
w1 = 1 rad/s
1 L1 L3 L2 K + 1
C2 C4 Cn 1 1
n L1 C2 L3 C4 L5 C6 L7 C8 L9 C10
2 1.414 1.414
3 1.000 2.000 2.000
4 0.765 1.848 1.848 0.765
5 0.618 1.618 2.000 1.618 0.618
6 0.518 1.414 1.932 1.932 1.414 0.518
7 0.445 1.247 1.802 2.000 1.802 1.247 0.445
8 0.390 1.111 1.663 1.962 1.962 1.663 1.111 0.390
9 0.347 1.000 1.532 1.879 2.000 1.879 1.532 1.000 0.347
10 0.313 0.908 1.414 1.975 1.975 1.975 1.782 1.414 0.908 0.313
n C1 L2 C3 L4 C5 L6 C7 L8 C9 L10
1 L2 Ln
C1 C3 1 C2n + 1 1
TABLE E2
Element Values for Low-Pass Chebyshev Filters
Chebyshev 0.01 dB
n L1 C2 L3 C4 L5 C6 L7 C8 L9
3 1.181 1.821 1.181
5 0.977 1.685 2.037 1.685 0.977
7 0.913 1.595 2.002 1.870 2.002 1.595 0.913
9 0.885 1.551 1.961 1.862 2.072 1.862 1.961 1.551 0.885
n C1 L2 C3 L4 C5 L6 C7 L8 C9
261
262 Active Filters: Theory and Design
Chebyshev 0.1 dB
n L1 C2 L3 C4 L5 C6 L7 C8 L9
3 1.433 1.594 1.433
5 1.301 1.556 2.241 1.556 1.301
7 1.262 1.520 2.239 1.680 2.239 1.520 1.262
9 1.245 1.502 2.222 1.683 2.296 1.683 2.222 1.502 1.245
n C1 L2 C3 L4 C5 L6 C7 L8 C9
Chebyshev 0.25 dB
n L1 C2 L3 C4 L5 C6 L7 C8 L9
3 1.633 1.436 1.633
5 1.540 1.435 2.440 1.435 1.540
7 1.512 1.417 2.453 1.535 2.453 1.417 1.512
9 1.500 1.408 2.445 1.541 2.508 1.541 2.445 1.408 1.500
n C1 L2 C3 L4 C5 L6 C7 L8 C9
Chebyshev 0.5 dB
n L1 C2 L3 C4 L5 C6 L7 C8 L9
3 1.864 1.280 1.864
5 1.807 1.302 2.691 1.302 1.807
7 1.790 1.296 2.718 1.385 2.718 1.296 1.790
9 1.782 1.292 2.716 1.392 2.773 1.392 2.716 1.292 1.782
n C1 L2 C3 L4 C5 L6 C7 L8 C9
Chebyshev 1 dB
n L1 C2 L3 C4 L5 C6 L7 C8 L9
3 2.216 1.088 2.216
5 2.207 1.128 3.102 1.128 2.207
7 2.204 1.131 3.147 1.194 3.147 1.131 2.204
9 2.202 1.131 3.154 1.202 3.208 1.202 3.154 1.131 2.202
n C1 L2 C3 L4 C5 L6 C7 L8 C9
Chebyshev 2 dB
n L1 C2 L3 C4 L5 C6 L7 C8 L9
3 2.800 0.860 2.800
5 2.864 0.909 3.827 0.909 2.864
7 2.882 0.917 3.901 0.959 3.901 0.917 2.882
9 2.890 0.920 3.920 0.968 3.974 0.968 3.920 0.920 2.890
n C1 L2 C3 L4 C5 L6 C7 L8 C9
Chebyshev 3 dB
n L1 C2 L3 C4 L5 C6 L7 C8 L9
3 3.350 0.712 3.350
5 3.482 0.762 4.538 0.762 3.482
7 3.519 0.772 4.639 0.804 4.639 0.772 3.519
9 3.534 0.776 4.669 0.812 4.727 0.812 4.669 0.776 3.534
n C1 L2 C3 L4 C5 L6 C7 L8 C9
Appendix E 263
TABLE E3
Element Values for Elliptic Low-Pass Filters
1 L1 L3 L2 k + 1
L2 L4 L2 k + 2
C2 C4 C2 k + 1
7 1.10 0.988 1.167 0.244 1.277 0.597 1.357 1.040 0.679 0.967 0.583
1.50 1.116 1.336 0.097 1.757 1.152 0.372 1.638 1.125 0.268 0.956
2.00 1.149 1.380 0.038 1.920 1.352 0.177 1.857 1.270 0.127 1.067
8 1.10 0.779 1.133 0.338 1.089 0.707 1.151 0.866 0.927 0.745 1.096 1.061
1.50 0.978 1.384 0.109 1.561 1.343 0.321 1.417 1.504 0.218 1.358 1.081
2.00 1.033 1.456 0.053 1.725 1.575 0.153 1.614 1.697 0.104 1.440 1.085
n ωs L1 C2 L2 L3 C4 L4 L5 C6 L6 L7 C8
264 Active Filters: Theory and Design
n ws L1 C2 L2 L3 C4 L4 L5 C6 L6 L7 C8
7 1.10 1.910 0.927 0.307 1.936 0.480 1.688 1.553 0.593 1.107 1.420
1.50 2.079 1.048 0.100 2.614 0.874 0.490 2.440 0.905 0.333 1.877
2.00 2.124 1.080 0.049 2.844 1.016 0.235 2.753 1.006 0.160 2.019
8 1.10 1.466 1.101 0.348 1.288 0.759 1.072 0.900 1.113 0.620 1.078 1.760
1.50 1.719 1.283 0.118 1.837 1.369 0.315 1.438 1.772 0.185 1.261 1.830
2.00 1.791 1.334 0.058 2.028 1.589 0.152 1.626 1.996 0.089 1.321 1.848
n ws L1 C2 L2 L3 C4 L4 L5 C6 L6 L7 C8
Appendix F: Coefficients
of Denominator Polynomial
s n + bn−1s n−1 + + b2s 2 + b1s + bo
TABLE F1
Butterworth Filters
n b0 b1 b2 b3 b4 b5 b6 b7
1 1.000
2 1.000 1.414
3 1.000 2.000 2.000
4 1.000 2.613 3.414 2.613
5 1.000 3.236 5.236 5.236 3.236
6 1.000 3.864 7.464 9.141 7.464 3.864
7 1.000 4.494 10.098 14.592 14.592 10.098 4.494
8 1.000 5.126 13.137 21.846 25.688 21.846 13.137 5.126
TABLE F2
0.1-dB Chebyshev Filter
n b0 b1 b2 b3 b4 b5 b6 b7
1 6.552
2 3.313 2.372
3 1.638 2.630 1.939
4 0.829 2.026 2.627 1.804
5 0.410 1.436 2.397 2.771 1.744
6 0.207 0.902 2.048 2.779 2.966 1.712
7 0.102 0.562 1.483 2.705 3.169 3.184 1.693
8 0.052 0.326 1.067 2.159 3.419 3.565 3.413 1.681
TABLE F3
0.5-dB Chebyshev Filter
n b0 b1 b2 b3 b4 b5 b6 b7
1 2.863
2 1.516 1.426
3 0.716 1.535 1.253
265
266 Active Filters: Theory and Design
TABLE F4
1-dB Chebyshev Filter
n b0 b1 b2 b3 b4 b5 b6 b7
1 1.965
2 1.103 1.098
3 0.491 1.238 0.988
4 0.276 0.743 1.454 0.953
5 0.123 0.581 0.974 1.689 0.937
6 0.069 0.307 0.939 1.202 1.931 0.928
7 0.031 0.214 0.549 1.358 1.429 2.176 0.923
8 0.017 0.107 0.448 0.847 1.837 1.655 2.423 0.920
TABLE F5
2-dB Chebyshev Filter
n b0 b1 b2 b3 b4 b5 b6 b7
1 1.308
2 0.823 0.804
3 0.327 1.022 0.738
4 0.206 0.517 1.256 0.716
5 0.082 0.459 0.693 1.450 0.706
6 0.051 0.210 0.771 0.867 1.746 0.701
7 0.020 0.166 0.383 1.144 1.039 1.994 0.698
8 0.013 0.070 0.360 0.598 1.580 1.212 2.242 0.696
TABLE F6
3-dB Chebyshev Filter
n b0 b1 b2 b3 b4 b5 b6 b7
1 1.002
2 0.708 0.645
3 0.251 0.928 0.597
4 0.177 0.405 1.169 0.582
5 0.063 0.408 0.549 1.415 0.574
6 0.044 0.163 0.699 0.691 1.663 0.571
7 0.016 0.146 0.300 1.052 0.831 1.912 0.568
8 0.011 0.056 0.321 0.472 1.467 0.972 2.161 0.567
Appendix F: Coefficients of Denominator Polynomial 267
TABLE F7
Bessel Filter
n b0 b1 b2 b3 b4 b5 b6
1 1
2 3 3
3 15 15 6
4 105 105 45 10
5 945 945 420 105 15
6 10395 10395 4725 1260 210 21
7 135135 135135 62370 17325 3150 378 28
Bibliography
1. Acar, C., Anday, F., and Kutman, H., On the realization of OTA-C filters, Int. J.
Circuit Theory Applications, 21(4), 331–341, 1993.
2. Allen, P. E. and Sanchez-Sinencio, E., Switched Capacitor Circuits, Van Nostrand
Reinhold, New York, 1984.
3. Berlin, H. M., Design of Active Filters, H. W. Sams, New York, 1981.
4. Bruton, L. T., RC-Active Circuits, Prentice-Hall, Englewood Cliffs, New Jersey, 1981.
5. Budak, A., Passive and Active Network Analysis and Synthesis, Hougston, Mifflin,
1974.
6. Chen, C., Active Filter Design, Hayden, 1982.
7. Christian, E., LC Filters, John Wiley and Sons, New York, 1983.
8. Clayton, G. B., Linear Applications Handbook, TAB Books, 1975.
9. Daniels, R. W., Approximation Method of Electronic Filter Design, McGraw-Hill,
New York, 1974.
10. Davis, TH. W. and Palmer, R. W., Computer-Aided Analysis of Electrical Networks,
C. E. Merrill Publ. Co., Columbus, OH, 1973, chap. 7.
11. Daryanani, G., Principle of Active Network Synthesis and Design, John Wiley & Sons,
1976.
12. Deliyannis, T., High-Q Factor Circuit with Reduced Sensitivity, Electronics Letters,
Vol. 4, No. 26, 1968.
13. Deliyannis, T., Sun, Y., and Fidler, J. K., Continuous-Time Active Filter Design, CRC
Press, Boca Raton, FL, 1999.
14. DePian, L., Linear Active Network Theory, Prentice-Hall, Englewood Cliffs, NJ, 1962.
15. Floyd, T., Electronic Devices: Conventional-Flow Version, 4th ed., Prentice-Hall,
Englewood Cliffs, NJ, 1995.
16. Geiger, R. L. and Sanchez-Sinencio, E., Active Filters Design Using Operational
Transconductance Ampifiers: A Tutorial, IEEE, Circuit and Devices, Vol. 1, pp. 20–32,
1985.
17. Geffe, P. R., Simplified Modern Filter Design, J. Rider, 1963.
18. Ghausi, M. S. and Laker, K. R., Modern Filter Design, Prentice Hall, 1981.
19. Graeme, J. D., Tobey, G. E., and Huelsman, L. P. (Eds.), Operational Amplifiers:
Design and Applications, McGraw-Hill, New York, 1971 (Burr-Brown).
20. Grebene, A. B., Bipolar and MOS Analog Integrated Circuit Design, John Wiley &
Sons, New York, 1984.
21. Gregorian, R. and Tems, G. C., Analog MOS Integrated Signal Processing, John Wiley
& Sons, New York, 1986.
22. Guillemin, E. A., Introductory Circuit Theory, John Wiley & Sons, New York, 1953.
23. Hilburn, J. L. and Johnson, D. E., Rapid Practical Design of Active Filters, John Wiley
& Sons, New York, 1975.
24. Huelsman, L. P., Theory and Design of RC Active Circuits, TMH edition, McGraw-
Hill, New York, 1968.
25. Huelsman, L. P., Active Filters: Lumped, Distributed, Integrated, Digital & Parametric,
McGraw-Hill, New York, 1970.
26. Huelsman, L. P. and Allen, P. E., Introduction to the Theory and Design of Active
Filters, McGraw-Hill, New York, 1980.
269
270 Active Filters: Theory and Design
27. Kincaid, R. and Shirley, F., Active Bandpass Filter Design Is Made Easy with Com-
puter Program, Electronics, May 16, 1974.
28. Kuo, F. F., Network Analysis and Synthesis, John Wiley & Sons, New York, 1966.
29. Lacanette, K. (Ed.), Switched-Capacitor Filter Handbook, Natonal Semiconductor
Corp., Santa Clara, 1985.
30. Lancaster, D., Active-Filter Cookbook, Sams, 1975.
31. Mitra, S. K., Analysis and Synthesis of Linear Active Network, John Wiley & Sons,
New York, 1969.
32. Moberg, G. O., Multiple-Feedback Filter Has Low Q and High Gain, Electronics,
December 9, 1976, pp. 97–99.
33. Moshytz, G. S. and Horn, P., Active Filter Design Handbook, John Wiley & Sons,
New York, 1981.
34. Moshytz, G. S., Inductorless Filters: A Survey: Part III: Linear, Active & Digital
Filters, IEEE Spectrum, September 1970, pp. 63–76.
35. Pactitis, S. A. and Kossidas, A. T., Design of Sallen-Key Active Filters, IASTED,
Intern. Symp., August 1983.
36. Pactitis, S. A. and Kossidas, A. T., Computer Implementation of VCVS Active Filters,
Advance in System Modeling and Simulation, North Holland, Amsterdam, 1989.
37. Peyton, A. J. and Walsh, V., Analog Electronics with Op Amps, Cambridge University
Press, New York, 1993.
38. Russel, H. T., Design Active Filters with Less Effort, Electronic Design, January 7,
1971, pp. 82–85.
39. Sallen, R. P. and Key, E. L., A practical Method of Design RC Active Filters, Trans.
IRE, CT-2, pp. 74–85, 1955.
40. Schaumann, R. and Van Valkenburg, M. E., Design of Analog Filters, Oxford Uni-
versity Press, New York, 2000.
41. Schaumann, R., Continous-time integrated filters, in The Circuit and Filters Hand-
book, Chen, W. K., Ed., CRC Press, Boca Raton, FL, 1995.
42. Schaumann, R., Ghausi, M. S., and Laker, K. R., Design of Analog Filters, Passive,
Active RC and Switching Capacitors, Prentice-Hall, Englewood Cliffs, NJ, 1990.
43. Sedra, A. S. and Brakett, P., Filter Theory and Design: Active and Passive, Matrix
Publishers, London, 1978.
44. Sun, Y. and Fidler, J. K., Novel OTA-C realization of biquadratic transfer functions,
Int. J. Electronics, 75, 333–348, 1993.
45. Temes, G. C. and LaPatra, S. K., Circuit Synthesis and Design, McGraw-Hill,
New York, 1977.
46. Tobey, Graeme, and Huelsman, Operational Amplifier Design and Applications,
McGraw-Hill, New York, 1971.
47. Tomlison, G. H., Electrical Networks and Filters, Prentice Hall International, London,
1991.
48. Toumazou, C., Lidgey, F. J., and Haigh, D. G. (Eds.), Analogue IC Design: The Current
Mode Approach, Peter Peregrinous, 1990.
49. Van Valkenburg, M. E., Analog Filter Design, Holt, Rinehart & Winston, New York,
1982.
50. Van Valkenburg, M. E., Modern Network Synthesis, John Wiley & Sons, New York, 1962.
51. Volpe, G. T. and Premisler, L., Universal Building Blocks Simplify Active Filter
Design, EDN, September 5, 1976, pp. 91–95.
52. Chen, W. -K., Passive and Active Filters, Theory and Implementations, John Wiley
& Sons, New York, 1986.
53. Williams, A. B., Electronic Filter Design Handbook, McGraw-Hill, New York, 1981.
Index
A band-pass, 90–92
band-pass state-variable, 143–145
ACF7092 circuit, 153 band-reject, 64–68
Active filter denominator polynomial factors, 253
advantages of, 17–18 design, 24–25, 27–28
applications, 18–19 designing, 51–52
definition, 1, 17 design using SC filter, 235–239
from passive RLC filters, 186 equal-component, 42
programmable high-frequency, 203 fifth-order high-pass, 191–192
replaced by switched capacitor (SC) filter, 225 high-pass, 45–46, 234–235
AF100 circuit, 219 high-pass multifeedback, 79–80
Amax, 3 high-pass state-variable, 139–140
Amin, 3 low-pass, 53–55, 231–232, 251
Amplitude response, 2 low-pass cascading, 83–87
Audio system applications, 19 low-pass LC element values, 261
low-pass multifeedback, 75–76
B low-pass second order, 161–162
low-pass state-variable, 132–133
Band-pass filter mathematics of, 7ff
biquad, 155ff nomograph, 251
definition, 3 perfect flatness from, 15
Deliyanis’s, 107–114 pole locations, 16
formulas, 258 response curves, 10
narrow-band, 92ff, 102–105, 155ff from SC filter, 228
narrow-band multifeedback, 92ff second-order low-pass, 34–37, 209–210,
second-order using OTA, 213–215 212–213
wide-band, 90ff third-order low-pass, 187–190
Band-reject filter, 114ff wide-band, 60–64
Butterworth wide-band, 143–145 wide-band band-reject, 143–145
definition, 5
narrow-band, 119–123 C
narrow-band multifeedback, 123–126
wide-band, 114–119 CA3280A circuit, 219
Bessel filter, 14ff Cauer filter, 17, 228
denominator polynomial factors, 255 Chebyshev filter
design, 40–41 denominator polynomial factors, 253–254
low-pass, 58–59 designing, 52, 52–53
pole locations, 16 equal-component, 42–44, 47–49
from SC filter, 228 high-pass, 47, 55–58
step response, 16 high-pass cascading, 87–90
Bessel-Thomson filter, see Bessel filter high-pass multifeedback, 80–83
Biofeedback applications, 19 high-pass state-variable, 140–143
Biquad filter, 18 low-pass, 37–39, 162–165, 252
definition, 154 low-pass element values, 261–262
design, 154ff low-pass multifeedback, 76–78
formulas, 260 low-pass state-variable, 133–137
Burr-Brown UAF41 circuit, 153 mathematics of, 9ff
Butterworth filter nomograph, 252
271
272 Active Filters: Theory and Design
G
E
General Instrument Corporation ACF7092 circuit,
Elliptic filter, 17
153
fifth-order low-pass, 190–191
Generalized impedance converter (GIC), 183ff
low-pass element values, 263
gm—C filter, 216
third-order high-pass using GIC, 192–194
Grounded frequency-dependent negative
Equal ripple response filter, 9ff
resistance (FDNR) element, 186
F H
f 1, 3 Harris OTA circuit, 219
FDNR (Grounded frequency-dependent negative Higher-order filter, 50ff, 83ff
resistance element), 186 High-pass filter, 44ff, 79ff
Filter; see also specific filter types definition, 3
active, defined, 1 design using generalized impedance circuit,
band-pass, defined, 3 191ff
band-reject, defined, 5 design using OTA, 206–207
cascading, defined, 83ff first-order, 25–28
characteristic curves, 1 state-variable, 137ff
definition, 1 using SC filter, 232–234
equal ripple response, 9ff
frequency-domain effects, 1
high-pass, defined, 3 I
ideal, defined, 5
Laplace transform analysis, 1 Ideal filter, 5
low-pass, defined, 3 Impedance-scaling factor (ISF), 21–22
maximally flat response, 7ff Instrumentation applications, 19
narrow, defined, 4 Integrators, 129
for noise suppression, 1 ISF, see Impedance-scaling factor (ISF)
order, defined, 3
passive, defined, 1 K
types, 2–5
types compared, 18 KCL, see Kirchhoff current law (KCL) equations
Filter design nomographs, 251–252 Kinetic Technology Corporation FS-60, 153
Index 273
Kirchhoff current law (KCL) equations, 247–248 Narrow-band band-reject filter, 119–123
Kirchhoff voltage law (KVL) equations, 247–248 using GIC, 197ff
KVL, see Kirchhoff voltage law (KVL) equations Narrow filter, 4
National Semiconductor
AF100, 153
L LM3600, 219
Lab signal sources, 19 MF10, 228
Laplace transform filter analysis, 1 MF100, 228
Linear-phase properties, 15 MF6, 244
Linear Technology OTA, 219
LTC 1060 circuit, 228 SC filter, 228
SC filter, 228 NE5517 circuit, 219
LM3600 circuit, 219 Network function
LMF100 circuit, 228–229 filter frequency-domain behavior, 1
Low-pass filter mathematics of, 5ff
biquad, 159ff Node-voltage method, 248, 248–249
definition, 2–3 Noise and signal, 1
design using OTA, 205–206, 207–209 Nomographs for filter design, 53, 251ff
first-order, 22–25 Normalized filter formulas, 257–260
formulas, 257–258 Notch filter, 120ff
generalized impedance, 187–190 design, 153–154
general transfer function, 50 formulas, 259
multifeedback, 73–74 Nyquist limit, 240
second-order using OTA, 209ff
state-variable, 130–132
O
transfer function poles, 31ff
using SC filter, 230–232 OP-07 op-amp, 19
LTC 1060 circuit, 228 Operational transconductance amplifier (OTA),
203ff
commercially available, 219–220
M
definition, 203
Magnitude function, 174–176 OTA-C filter, 216–219
Manufacturing tolerances, 169 programmable high-frequency, 203
Maximally flat response filter, 7ff OTA filter, see Operational transconductance
MF10, 228 amplifier (OTA)
MF100 circuit, 228–229 Overshoot, 14
MF10 circuit, 228
MF5 circuit, 232
P
MF6 circuit, 244
MFB filter, see Multifeedback (MFB) filter Passband gain, 3
Multifeedback circuit, 18 Passband limit ( f1), 3
Multifeedback (MFB) filter, 73ff Passband ripple, 3
band-pass, 127 in Chebyshev filter, 9
design tolerance, 126, 127 Pass-band ripple filter, element values, 264
formulas, 257–259 Passive filter, defined, 1
high-pass, 127 Passive RC network, 73ff
low-pass, 126 transfer function poles, 28
narrow-band band reject, 123–126 Perfect flatness, 15
Phase function sensitivity analysis, 174–176
N Phase response, 2
Phase shift
Narrow-band band-pass filter, 155ff and phase response, 2
design using SC filter, 235–237 varying with frequency, 14
design with OTA filter, 215–216 Philips NE5517 circuit, 219
using GIC, 194ff Pole locations, 16
274 Active Filters: Theory and Design
T
R
Telephony applications, 154–155
RC network, passive, 73ff
Tolerance factors, 126–127, 169
Reference node, 247
Tone signaling applications, 19
Relative sensitivity, 169
Transfer function
Ringing, 14
filter frequency-domain behavior, 1
Ripple height, 11
filter response described by, 1–2
RLC filter, 17
general for low-pass filter, 50
Rolloff rate, 3
inverting, 73
Root sensitivity, 176–180
mathematics of, 5ff
S
U
Sallen-Key filter, 18, 21ff, 79
formulas, 257 UAF41 circuit, 153
sensitivity analysis of, 172–174 Universal filter, 129ff
SC filter, 225ff Universal SC filter, 228ff
limitations, 240 structure, 228
Second order filters, 28ff, 34ff
Selectivity factor Q, 5 V
Sensitivity, 169ff
Sensitivity analysis, 172–174 Voltage-controlled voltage source (VCVS) filter,
Signal and noise, 1 19–20
Signal dispersion, 15 high-pass, 68–70
Signal distortion, 14 low-pass filter, 31ff, 68–69
Spectrum analyzers, 154 Voltage gain, 51
State-variable circuit, 18
commercially available, 153
W
formulas, 259
narrow-band band-pass, 149ff Wide-band filter
State-variable (SV) filter, 129ff band-reject, 64ff, 114–119
Summing amplifier, 129 definition, 60
SV filter, 129ff low-pass, 60ff