0% found this document useful (0 votes)
43 views45 pages

Gift-Maundy2021 Chapter ActiveFilters

This document discusses different types of filters and provides details on active low-pass filters. It defines five basic types of filters: low-pass, high-pass, band-pass, band-stop, and all-pass. Active filters contain both passive and active components, allowing them to produce signal gain. The document focuses on the principles and circuit design of first-order low-pass filters, including non-inverting configurations with and without gain. Key aspects like cutoff frequency, gain, and frequency responses are analyzed through transfer functions.

Uploaded by

Abdallah Amro
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
43 views45 pages

Gift-Maundy2021 Chapter ActiveFilters

This document discusses different types of filters and provides details on active low-pass filters. It defines five basic types of filters: low-pass, high-pass, band-pass, band-stop, and all-pass. Active filters contain both passive and active components, allowing them to produce signal gain. The document focuses on the principles and circuit design of first-order low-pass filters, including non-inverting configurations with and without gain. Key aspects like cutoff frequency, gain, and frequency responses are analyzed through transfer functions.

Uploaded by

Abdallah Amro
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 45

Active Filters

11

A filter is an electrical network that passes signals all-pass filters. A low-pass filter passes signals
within a specified band of frequencies while with constant amplitude from DC up to a fre-
attenuating those signals that fall outside of this quency fc referred to as the cut-off frequency.
band. Passive filters utilize passive components, The signal output above fc is attenuated. This is
namely, resistors, capacitors and inductors, while shown in Fig. 11.1 where the magnitude of the
active filters contain passive as well as active output voltage of the ideal (solid line) and practi-
components such as transistors and operational cal (dashed line) responses are plotted against the
amplifiers. Passive filters have the advantage of frequency. The ideal response shows the pass-
not requiring an external power supply as do band, the range of transmitted frequencies, and
active filters. However, they often utilize the stop-band, the range of attenuated frequencies
inductors which tend to be bulky and costly, and the cut-off or break frequency fc. In the prac-
whereas active filters utilize mainly resistors and tical response, fc is the frequency at which the
capacitors. Additionally, active filters can pro- magnitude of the output voltage falls to 0.707 of
duce signal gain and have high input and low its low-frequency value which is 3 dB. Here
output impedances which allow simple cascading fc ¼ 1 Hz for the normalized response.
of systems with little or no interaction between A high-pass filter attenuates signals from DC
stages. This chapter discusses the principles of up to a frequency fc while passing all signals with
active filter operation and treats with several constant amplitude above fc. The normalized
types and configurations of active filters. At the response is shown in Fig. 11.2 where the ideal
end of the chapter, the student will be able to: and practical responses with pass-band and stop-
band are indicated. A band-pass filter passes a
• Understand the principles of operation of the band of frequencies, while frequencies outside
basic types of filters of that band are attenuated. This is shown in
• Explain the operation of a range of active Fig. 11.3 where the ideal and practical responses
filters are indicated. A band-stop filter stops a band of
• Design a range of active filters frequencies while passing all frequencies outside
that band. It performs in a manner that is exactly
opposite to the band-pass filter. The band-stop
response is shown in Fig. 11.4 with ideal and
11.1 Introduction to Filters practical responses. Finally, an all-pass filter
passes all frequencies with a constant amplitude
There are five basic types of filters: low-pass, output but with a changing phase. This is shown
high-pass, band-pass, band-stop (or notch) and in Fig. 11.5.
# Springer Nature Switzerland AG 2021 411
S. J. G. Gift, B. Maundy, Electronic Circuit Design and Application,
https://doi.org/10.1007/978-3-030-46989-4_11
412 11 Active Filters

Fig. 11.1 Low-pass filter


response

Fig. 11.2 High-pass filter


response

Fig. 11.3 Band-pass filter


response

Vo 1=sC 1
11.2 Basic First-Order LP Filter Af ¼ ¼ ¼
V i R þ 1=sC 1 þ jωRC
We first consider a low pass of the simplest kind ¼
1
ð11:1Þ
shown in Fig. 11.6(a). It is a non-inverting 1 þ jf = f c
low-pass active filter using an RC network feeding
where
into a unity-gain operational amplifier. A resistor
Ros is sometimes included in the feedback loop of f c ¼ 1=2πRC ð11:2Þ
the op-amp in order to minimize DC offset at the
output as shown in Fig. 11.6(b). For the circuit,
11.2 Basic First-Order LP Filter 413

Fig. 11.4 Band-stop filter


response
Magnitude

0.5

0
0

-45
Phase (deg)

-90

-135

-180
10-2 10-1 100 101 102
Frequency (Hz)

Fig. 11.5 Magnitude and phase responses of first-order all-pass filter

Note that for f  f c , |A f | ! 1 and for f  f c , with


|Af| ! 0. The frequency response plot for this  
circuit is shown in Fig. 11.7 which is clearly a A f  ðdBÞ ¼ 20 log 0:707 ¼ 3 dB ð11:6Þ
cf
low-pass response. The frequency fc ¼ 1/2πRC is
the cut-off frequency. The system has one pole at The signal phase is
f ¼ fc and is therefore first-order. At this fre-
quency, the filter transfer function becomes ∡A f ð f c Þ ¼ 45 ð11:7Þ

1 1 Equation (11.2) is used to design the filter.


A f ð f cÞ ¼ ¼ ð11:3Þ
1 þ jf c = f c 1 þ j
Example 11.1
This reduces to Design a low-pass non-inverting filter with unity
1 gain and a break frequency of 1 kHz.

A f ð f c Þ ¼ pffiffiffi  ¼ 0:707∠  45 ð11:4Þ
2∠45
Solution Choose C ¼ 0.01 μF. Then from
Hence Eq. (11.2), R ¼ 1/2π  103  0.01  106 ¼
  16 k. If low DC offset is required, then Ros ¼ R ¼
A f  ¼ p1ffiffiffi ¼ 0:707 ð11:5Þ 16 k.
fc
2
414 11 Active Filters

Fig. 11.6 Basic first-order


low-pass filter circuit (unity
gain)

(a) (b)

Fig. 11.7 First first-order


low-pass filter response

where fc ¼ 1/2πRC is the break frequency and


k ¼ 1 þ RR21 is the low-frequency gain. For low DC
offset, R ¼ R2//R1.
An alternative configuration for a first-order
low-pass filter is shown in Fig. 11.9(a). It is an
inverting amplifier configuration with a capaci-
tor C across the feedback resistor R2. From
Fig. 11.9(a),
Vi V o R2
¼ ¼ V o = ð11:9Þ
Fig. 11.8 First-order non-inverting low-pass filter R1 R2 ==1=sC 1 þ sCR2
with gain
which yields
11.2.1 Low-Pass Filter with Gain R2 1 k
Af ¼  : ¼ ð11:10Þ
R1 1 þ sCR2 1 þ jf = f c
The low-pass filter with gain is achieved by reduc-
ing the feedback around the op-amp as shown in where fc ¼ 1/2πR2C is the break frequency and
Fig. 11.8. The transfer function is given by k ¼ R2/R1 is the low-frequency gain. For reduced
DC offset, a resistor Ros ¼ R1//R2 can be inserted
Vo k
¼ ð11:8Þ between the non-inverting terminal and ground as
V i 1 þ jf = f c
shown in Fig. 11.9(b).
11.2 Basic First-Order LP Filter 415

(a) (b)

Fig. 11.9 First-order inverting low-pass filter with gain

higher-order filters considered later, the condition


a1 6¼ 1 will arise.
For the first-order non-inverting low-pass filter,
the transfer function is given by equation (11.12):

1 þ RR21
A f ðsÞ ¼ ð11:12Þ
1 þ RCs
Comparing coefficients in (11.11) and (11.12)
Fig. 11.10 Solution for Example 11.2 gives
R2
Example 11.2 Ao ¼ 1 þ ð11:13Þ
R1
Design a first-order non-inverting low-pass filter
with a gain of 10 and a cut-off frequency of 5 kHz a1 =2π f c ¼ RC ð11:14Þ
using Fig. 11.8.
The design steps are as follows:
1. Select C usually between 100 pF and 1 μF.
Solution Choose C ¼ 0.01 μF. Then from
2. Set the coefficient a1 ¼ 1.
fc ¼ 1/2πRC, resistor R is found to be R ¼ 1/
3. Determine R using R ¼ a1/2πfcC.
(2  π  5  103  0.01  106) ¼ 3.2 k. Since
4. Select R1 and find R2 using R2 ¼ R1(Ao  1).
1 + R2/R1 ¼ 10, for R1 ¼ 1 k, then R2 ¼ 9 k. The
solution is shown in Fig. 11.10.
For the first-order inverting low-pass filter, the
basic first-order system can be written as
Design Procedure
In preparation for the discussion on higher-order Ao
AðsÞ ¼  ð11:15Þ
filters and their realization, we wish to develop a 1 þ a1 ðs=ωc Þ
more ordered procedure for designing first-order
filters. Thus, the basic first-order system can be The transfer function for the actual system is
written as given by equation (11.16):
R2
Ao
AðsÞ ¼ ð11:11Þ A f ðsÞ ¼  R1
ð11:16Þ
1 þ a1 ðs=ωc Þ 1 þ R2 Cs
where Ao ¼ k is the low-frequency gain, ωc ¼ 2πfc Comparing coefficients in (11.15) and (11.16)
is the cut-off frequency and here a1 ¼ 1. For gives
416 11 Active Filters

R2 In order to allow optimization of a filter, the


Ao ¼ ð11:17Þ
R1 transfer function of the filter must contain com-
plex poles and hence, for a second-order filter,
a1 =2π f c ¼ R2 C ð11:18Þ
must take the following form:
The design steps are as follows: Ao
1. Select C usually between 100 pF and 1 μF. AðsÞ ¼   ð11:19Þ
1 þ a1 ðs=ωc Þ þ b1 ðs=ωc Þ2
2. Set the coefficient a1 ¼ 1.
3. Determine R2 using R2 ¼ a1/2πfcC.
where Ao is the low-frequency gain in the pass-
4. Determine R1 using R1 ¼ R2/Ao.
band and a1 and b1 are positive real filter
coefficients. These coefficients define the location
of the complex poles of the filter and thereby
11.3 Low-Pass Second-Order Filter determine the characteristics of the filter transfer
function. The angular frequency ωc ¼ 2πfc
It is possible to cascade two first-order low-pass represents the cut-off frequency of the filter.
filters with the same cut-off frequency fc each This transfer function can be normalized by
with 20 dB/dec roll-off rate to get a second- setting ωc ¼ 1 giving
order filter with a roll-off rate of 40 dB/dec.
Ao
This is shown in Fig. 11.11. Such an approach AðsÞ ¼ ð11:20Þ
has several disadvantages. The first is that the ð 1 þ a1 s þ b1 s 2 Þ
cut-off frequency of the overall filter is lower Several filter types are available:
than the cut-off frequency fc of the individual (i) Butterworth response, maximum pass-band
first-order filters. The result is that the pass-band flatness; (ii) Chebyshev response, sharpened tran-
gain begins to fall well before the cut-off fre- sition from pass-band to stop-band; (iii) Bessel
quency fc. Secondly, the transition from the response, linearized phase response within the
pass-band to the stop-band lacks sharpness as is pass-band; (iv) inverse Chebyshev response, flat
desired in a filter. Thirdly, it can be shown that the pass-band and ripples in the stop-band; and
phase response is not linear with the effect that (v) elliptic (Cauer) response, optimal approxima-
signals distortion results. A low-pass filter can be tion to the ideal low-pass filter with ripples in the
optimized to meet at least one of the following pass-band and the stop-band. These filter
three requirements: (i) maximum flatness in the responses occur for second- and higher-order
pass-band; (ii) maximum sharpness in the transi- and not for first-order where complex poles are
tion from pass-band to stop-band; and (iii) linear not possible, and therefore all filter types have
phase response. identical responses for a first-order low-pass filter
(corresponding to a1 ¼ 1 in (11.11)). Only the
Butterworth, Chebyshev and Bessel are consid-
ered in this chapter.
The amplitude responses of first-, second- and
third-order Butterworth low-pass filters using the
normalized characteristic are shown in Fig. 11.12.
The filter provides maximum flatness in the pass-
band, and as the order increases, the pass-band
flatness occurs for higher frequencies before the
break frequency. This type of filter is therefore
ideal for audio systems where high-frequency
Fig. 11.11 Two cascaded first-order filters noise needs to be removed with maximum pass-
11.3 Low-Pass Second-Order Filter 417

Fig. 11.12 Low-pass filter


response plots for
Butterworth filters

Fig. 11.13 Gain responses


of fourth-order low-pass
filters

band flatness in order to preserve signal quality. 11.3.1 Sallen-Key or


The Chebyshev low-pass filter provides increased Voltage-Controlled Voltage
roll-off rate as compared with the Butterworth Source (VCVS) Topology
filter. The price paid for this is constant ampli-
tude ripples in the pass-band. The amplitude of The circuit shown in Fig. 11.14 is one implemen-
these ripples for a filter of given order may be tation of a second-order low-pass filter referred to
varied by changing the filter coefficients; the as the Sallen-Key or VCVS topology. Using
higher the ripple amplitude, the steeper the roll- KCL, the node equations at junctions A and B are
off rate. The Bessel low-pass filters have a linear
Vi  VA VA  VB
phase response as compared with the ¼ þ ðV A  V o ÞsC 2 ð11:21Þ
R1 R2
Butterworth and the Chebyshev filters resulting
in constant group delay. The pass-band gain VA  VB
¼ V B sC 1 ð11:22Þ
however is not at flat as the Butterworth though R2
it is flatter than the Chebyshev and the roll-off
rate is less than both the Chebyshev and the where VB ¼ Vo. Eliminating VA and VB yields
Butterworth. A graphical comparison of these Vo
three filter responses is presented in Fig. 11.13. Af ¼
Vi
The second-order low-pass filter transfer func- 1
tion (11.19) can be realized around a single active ¼
1 þ sC 1 ðR1 þ R2 Þ þ s2 R1 R2 C 1 C 2
device using two topologies, the Sallen-Key and
ð11:23Þ
the multiple feedback.
418 11 Active Filters

Fig. 11.14 Sallen-Key


second-order unity-gain
low-pass filter

(a) (b)

Comparing coefficients for (11.23) and (11.19) Design Procedure


yields 1. Select C1 usually between 100 pF and 1 μF.
2. For the specific type of filter (Butterworth,
Ao ¼ 1 ð11:24Þ
Chebyshev, Bessel) and the specific filter
a1 =ωc ¼ C 1 ðR1 þ R2 Þ ð11:25Þ requirements, obtain the coefficients a1 and
b1 from the relevant table.
b1 =ωc 2 ¼ R1 R2 C 1 C 2 ð11:26Þ 3. Determine C2 using C2  4b1C1/a12.
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
a þ a1 2 4b1 C 1 =C 2
From (11.25), 4. Find R1 and R2 using R1 ¼ 1 4π f c C 1
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
a1  a1 2 4b1 C 1 =C 2
R2 ¼ a1 =ωc C1  R1 ð11:27Þ and R2 ¼ 4π f C 1 .
c

5. Determine Ros ¼ R1 + R2 for minimum offset


and hence voltage.
b1 =ωc 2 ¼ R1 C1 C2 ða1 =ωc C 1  R1 Þ ð11:28Þ
Example 11.3
Design a VCVS second-order unity-gain
After manipulation, this leads to a quadratic low-pass Butterworth filter with a cut-off fre-
equation in R1 given by quency of 10 kHz.

ωc 2 C 1 C 2 R1 2  ωc a1 C 2 R1 þ b1 ¼ 0 ð11:29Þ Solution For this case, fc ¼ 10 kHz. Select


C1 ¼ 0.001 μF. For second-order Butterworth,
Solving for R1 we get pffiffiffi
a1 ¼ 2, b1 ¼ 1. Then C2  (4  1  0.01  106)/
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi pffiffiffi2
a1 C2 þ a1 2 C2 2  4b1 C 1 C 2 2 ¼ 0:02 μF . Select C2 ¼ 0.05 μF. Hence
R1 ¼ ð11:30Þ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
4π f c C 1 C 2 1:4142þ 2ð410:01106 =0:05106 Þ
R1 ¼ 3
0:01106
¼ 2 k
where the positive sign from the square root is pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
4π1010
1:4142 2ð410:0110 =0:01106 Þ 6

taken and and R2 ¼ 4π10103 0:01106


¼
254 Ω. Finally, for minimum offset,
C2  4b1 C 1 =a1 2 ð11:31Þ Ros ¼ 2000 + 254 ¼ 2254Ω.
Substituting (11.30) for R1 in (11.27) gives
Based on the design Eqs. (11.30) and (11.32),
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
a1 C2  a1 2 C2 2  4b1 C 1 C 2 the resistor and capacitor values can be scaled in
R2 ¼ ð11:32Þ order to get more appropriate values. Specifically,
4π f c C 1 C 2
R1, R2 ! αR1, αR2 where α may be referred to as
We can now state a design procedure for a spe- an impedance scaling factor (ISF) and C1,
cific cut-off frequency: C2 ! C1/α, C2/α will leave the filter
11.3 Low-Pass Second-Order Filter 419

Ao
AðsÞ ¼   ð11:36Þ
1 þ a1 ðs=ωc Þ þ b1 ðs=ωc Þ2

yields
Ao ¼ k ð11:37Þ

a1 =ωc ¼ ð3  kÞCR ð11:38Þ

b1 =ωc 2 ¼ R2 C 2 ð11:39Þ
Fig. 11.15 Sallen-Key second-order low-pass filter pffiffiffi
with gain For a Butterworth filter, a1 ¼ 2 and b1 ¼ 1.
Hence from (11.39),
characteristics unaffected. This can be seen by ωc ¼ 1=RC, f c ¼ 1=2πRC ð11:40Þ
simply substituting αR1, αR2 and C1/α, C2/α in
Eqs. (11.30) and (11.32) to see that the equations and from (11.38)
are unchanged. Also, the system can be designed pffiffiffi
2¼3k ð11:41Þ
for gain as shown in Fig. 11.15 where
k ¼ 1 þ Rb =Ra ð11:33Þ giving
pffiffiffi
The process of developing the associated k ¼3 2 ¼ 1:6 ð11:42Þ
design procedure follows that used for the unity-
For minimum offset
gain case.
Ra Rb R
R1 þ R2 ¼ Ra ==Rb ¼ ¼ b ð11:43Þ
Example 11.4 Ra þ Rb k
Develop a design procedure for the VCVS
Hence
Butterworth second-order low-pass filter with
gain for the special case when R1 ¼ R2 and Rb ¼ kðR1 þ R2 Þ ¼ 3:2R ð11:44Þ
C1 ¼ C2.
Also
Solution Using nodal analysis, the transfer func-
Ra Rb ðk  1Þ
tion for the filter with gain is given by R1 þ R2 ¼ Ra ==Rb ¼ ¼ Ra
Ra þ Rb k
Af ¼
Vo ð11:45Þ
Vi
k Hence
¼
1 þ sðC 1 R1 þ C 1 R2 þ C 2 R1  kC2 R1 Þ þ s2 R1 R2 C 1 C 2
kðR1 þ R2 Þ
ð11:34Þ Ra ¼ ¼ 3:2 R=0:6 ¼ 5:3 R ð11:46Þ
k1
For the case where R1 ¼ R2 ¼ R and C1 ¼ C2 ¼ C, These equations enable the determination of Ra
then (11.34) reduces to and Rb to satisfy both the gain requirement and
the minimum offset requirement. We can now
Vo k
Af ¼ ¼ ð11:35Þ state a design procedure for a specific cut-off
V i 1 þ sð3  kÞCR þ s2 R2 C2
frequency:
Comparing coefficients with (11.19)
420 11 Active Filters

Table 11.1 Butterworth, Bessel and Chebyshev (0.5 dB) coefficients


n i ai(Bu) bi(Bu) ai(Be) bi(Be) ai(.5 dB) bi(.5 dB)
1 1 1.0000 0.0000 1.000 0.0000 1.0000 0.0000
2 1 1.4142 1.0000 1.3617 0.6180 1.3614 1.3827
3 1 1.0000 0.0000 0.7560 0.0000 1.8636 0.0000
2 1.0000 1.0000 0.9996 0.4772 0.0640 1.1931
4 1 1.8478 1.0000 1.3397 0.4889 2.6282 3.4341
2 0.7654 1.0000 0.7743 0.3890 0.3648 1.1509
5 1 1.0000 0.0000 0.6656 0.0000 2.9235 0.0000
2 1.6180 1.0000 1.1402 0.4128 1.3025 2.3534
3 0.6180 1.0000 0.6216 0.3245 0.2290 1.0833
6 1 1.9319 1.0000 1.2217 0.3887 3.8645 6.9797
2 1.4142 1.0000 0.9686 0.3505 0.7528 1.8573
3 0.5176 1.0000 0.5131 0.2756 0.1589 1.0711
7 1 1.0000 0.0000 0.5937 0.000 4.0211 0.0000
2 1.8019 1.0000 1.0944 0.3395 1.8729 4.1795
3 1.2470 1.0000 0.8304 0.3011 0.4861 1.5676
4 0.4450 1.0000 0.4332 0.2381 0.1156 1.0443
8 1 1.9616 1.0000 1.1112 0.3162 5.1117 11.9607
2 1.6629 1.0000 0.9754 0.2979 1.0639 2.9365
3 1.1111 1.0000 0.7202 0.2621 0.3439 1.4206
4 0.3902 1.0000 0.3728 0.2087 0.0885 1.0407
9 1 1.0000 0.0000 0.5386 0.0000 5.1318 0.0000
2 1.8794 1.0000 1.0244 0.2834 2.4283 6.6307
3 1.5321 1.0000 0.8710 0.2636 0.6839 2.2908
4 1.0000 1.0000 0.6320 0.2311 0.2559 1.3133
5 0.3473 1.0000 0.3257 0.1854 0.0695 1.0272
10 1 1.9754 1.0000 1.0215 0.2650 6.3648 18.3695
2 1.7820 1.0000 0.9393 0.2549 1.3582 4.3453
3 1.4142 1.0000 0.7815 0.2351 0.4822 1.9440
4 0.9080 1.0000 0.5604 0.2059 0.1994 1.2520
5 0.3129 1.0000 0.2883 0.1665 0.0563 1.0263

Design Procedure 1
Aðjf Þ ¼ pffiffiffi  2 ð11:48Þ
1þj 2 f
þ j f
1. Select C1 ¼ C2 ¼ C usually between 100 pF fc fc

and 1 μF.
In the general second-order transfer function
2. Determine R ¼ 1/2πfcC.
(11.23) of Fig. 11.14, if we let R1 ¼ R2 ¼ R and
3. Set R1 ¼ R2 ¼ R.
C1 ¼ C2/2 ¼ C, then for s ¼ jω, the transfer
4. Find Ra and Rb using (11.46) and (11.44).
function (a) reduces to
Simplified Design Procedure for Second-Order Vo 1
ðjωÞ ¼ ð11:49Þ
Low-Pass Unity-Gain Butterworth Filter Vi 1 þ 2RCjω þ 2R2 C 2 ðjωÞ2
A simplified design procedure for the unity-gain
second-order Butterworth filter can be developed. Substituting ω ¼ 2πf and
Consider the transfer function (11.33) for a pffiffiffi
low-pass second-order filter: f c ¼ 1=2 2πRC ð11:50Þ

Ao yields
AðsÞ ¼   ð11:47Þ
2
1 þ a1 ðs=ωc Þ þ b1 ðs=ωc Þ Vo 1
ðjf Þ ¼ pffiffiffi  2 ð11:51Þ
Vi
1þj 2 f
þ j f
For unity gain, Ao ¼ 1 and forpaffiffiffi Butterworth fc fc

response from Table 11.1, a1 ¼ 2 and b1 ¼ 1.


Then setting s ¼ jω, (11.10) becomes
11.3 Low-Pass Second-Order Filter 421

Table 11.2 Chebyshev (1 dB, 2 dB, 3 dB) coefficients


n i ai(1 dB) bi(1 dB) ai(2 dB) bi(2 dB) ai(3 dB) bi(3 dB)
1 1 1.0000 1.0000 1.0000 0.0000 1.0000 0.0000
2 1 1.3022 1.5515 1.1813 1.7775 1.0650 1.9305
3 1 2.2156 0.0000 2.7994 0.0000 3.3496 0.0000
2 0.5442 1.2057 0.4300 1.2036 0.3559 1.1923
4 1 2.5904 4.1301 2.4025 4.9862 2.1853 5.5339
2 0.3039 1.1697 0.2374 1.1896 0.1964 1.2009
5 1 3.5711 0.0000 4.6345 0.0000 5.6334 0.0000
2 1.1280 2.4896 0.9090 2.6036 0.7620 2.6530
3 0.1872 1.0814 0.1434 1.0750 0.1172 1.0686
6 1 3.8437 8.5592 3.5880 10.4648 3.2721 11.6773
2 0.6292 1.9124 0.4925 1.9622 0.4077 1.9873
3 0.1296 1.0766 0.0995 1.0826 0.0815 1.0861
7 1 4.9520 0.0000 6.4760 0.0000 7.9064 0.0000
2 1.6338 4.4899 1.3258 4.7649 1.1159 4.8963
3 0.3987 1.5834 0.3067 1.5927 0.2515 1.5944
4 0.0937 1.0432 0.0714 1.0384 0.0582 1.0348
8 1 5.1019 14.7608 4.7743 18.1510 4.3583 20.2948
2 0.8916 3.0426 0.6991 3.1353 0.5791 3.1808
3 0.2806 1.4334 0.2153 1.4449 0.1765 1.4507
4 0.0717 1.0432 0.0547 1.0461 0.0448 1.0478
9 1 6.3415 0.0000 8.3198 0.0000 10.1759 0.0000
2 2.1252 7.1711 1.7299 7.6580 1.4585 7.8971
3 0.5624 2.3278 0.4337 2.3549 0.3561 2.3651
4 0.2076 1.3166 0.1583 1.3174 0.1294 1.3165
5 0.0562 1.0258 0.427 1.0232 0.0348 1.0210
10 1 6.3634 22.7468 5.9618 28.0376 5.4449 31.3788
2 1.1399 4.5167 0.8947 4.6644 0.7414 4.7363
3 0.3939 1.9665 0.3023 1.9858 0.2479 1.9952
4 0.1616 1.2569 0.1233 1.2614 0.1008 1.2638
5 0.0455 1.0277 0.0347 1.0294 0.0283 1.0304

Clearly, transfer function (11.51) is identical to Solution Choose C1 ¼ 0.01 μF and then
pffiffiffi
transfer function (11.48), the transfer function of C2 ¼ 0.02 μF. Using (11.36), R ¼ 1=2 2π 
a second-order unity-gain Butterworth filter. 10  103  0:01  106 ¼ 1:1 k and hence
Hence a simplified design procedure for a unity- R1 ¼ R2 ¼ 1.1 k. Ros ¼ 1.1 k  2 ¼ 2.2 k.
gain second-order Butterworth filter is as follows:
1. Choose the desired cut-off frequency fc. Example 11.6
2. Choose C1, usually an available value between Design a VCVS second-order unity-gain
about 100 pF and 1 μF. low-pass Chebyshev filter with a cut-off fre-
3. Let C2 ¼ 2C1. quency of 10 kHz and a ripple width
pffiffiffi
4. Determine R using R ¼ 1=2 2πfC 1 and then RWdB ¼ 2 dB.
set R1 ¼ R2 ¼ R.
5. For minimum offset use Ros ¼ 2R. Solution For this case, fc ¼ 10 kHz. Select
C1 ¼ 0.01 μF. From Table 11.2, the coefficients
Example 11.5 for RWdB ¼ 2 dB are a1 ¼ 1.1813, b1 ¼ 1.7775.
Using the configuration shown in Fig. 11.14, Then from (11.31), C2  (4  1.7775 
re-design the second-order Butterworth filter of 0.01  106)/(1.1813)2 ¼ 0.05 μF. Select
Example 11.5 using the simplified approach. C2 ¼ 0.1 μF. Hence from (11.30) R1 ¼
422 11 Active Filters

pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

1:1813þ ð1:1813Þ2 ð41:77750:01106 =0:1106 Þ
4π10103 0:01106
¼
1598 Ω and from (11.32) R2 ¼
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

1:1813 ð1:1813Þ2 ð41:77750:01106 =0:1106 Þ
4π10103 0:01106
¼
282 Ω. Finally, for minimum offset,
Ros ¼ 1598 + 282 ¼ 1880 Ω.

Example 11.7
Design a VCVS second-order unity-gain Fig. 11.16 Multiple feedback low-pass filter
low-pass Bessel filter with a cut-off frequency of
10 kHz. Noting that Vb ¼ 0 because node b is a virtual
earth and eliminating Va, we get
Solution For this case, fc ¼ 10 kHz. Select R2
Vo
C1 ¼ 0.01 μF. From Table 11.1, the Bessel ¼  R1

Vi 1 þ C 1 R2 þ R3 þ R1 s þ C1 C2 R2 R3 s2
R2 R3
coefficients are a1 ¼ 1.3617, b1 ¼ 0.6180.
Then C2  (4  0.6180  0.01  106)/
ð11:54Þ
(1.3617)2 ¼ 0.01333 μF. Select C2 ¼
0.02 μF. Hence R1 Comparing the coefficients of (11.54) with the
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1:3617þ ð1:3617Þ ð40:61800:0110 =0:0210 Þ
2 6 6
general second-order low-pass transfer function
¼ 4π10103 0:01106
¼ 1700 Ω and R2 ¼ AðsÞ ¼ 
Ao
ð11:55Þ
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1:3617 ð1:3617Þ2 ð40:61800:01106 =0:02106 Þ 1 þ a1 ðs=ωc Þ þ b1 ðs=ωc Þ2
4π10103 0:01106
¼
458 Ω. Finally, for minimum offset, (where a negative sign is introduced to account
Ros ¼ 1700 + 458 ¼ 2158 Ω. for the inversion produced by the MFB configu-
ration) yields

Ao ¼ R2 =R1 ð11:56Þ
11.3.2 Low-Pass Multiple Feedback 
Topology a1
¼ C1 R2 þ R3 þ
R2 R3
ð11:57Þ
ωc R1
Another available topology for implementation of
second-order low-pass filters is the multiple feed- b1
¼ C 1 C 2 R2 R3 ð11:58Þ
back (MFB) topology shown in Fig. 11.16. In this ωc 2
circuit there are multiple feedback paths, and the Solving for R1,R2, R3 in terms of C1 and C2, we
active element is operating in a high-gain mode. get

The amplifier introduces 180 of phase shift in pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
addition to that introduced by the network a1 þ a1 2  4b1 ð1 þ Ao ÞC 1 =C 2
R1 ¼ ð11:59Þ
elements. Assuming an ideal op-amp as the active 4π f c C 1 Ao
element, then for nodes a and b,
In order to ensure a positive value under the
Vi  Va V  Vb Va  Vo square root and hence a real solution for R1, then
¼ V a sC 2 þ a þ
R1 R3 R2
ð11:52Þ 4b1 ð1 þ Ao Þ
C2  C1 ð11:60Þ
a1 2
Va  Vb
¼ ðV b  V o ÞsC 1 ð11:53Þ Using (11.59) in (11.56) and (11.58), we get
R3
11.4 Higher-Order Low-Pass Filters 423

R2 ¼ Ao R1 ð11:61Þ 11.4 Higher-Order Low-Pass Filters


and We wish to consider those higher-order filters in
b1 which the transfer function has a denominator
R3 ¼ 2
ð11:62Þ polynomial that is of degree n corresponding to
4π 2 f c C 1 C 2 R2
the order of the filter and a numerator polynomial
with degree m < n. There are two general methods
Design Procedure for realizing these higher-order transfer functions.
The first is to synthesize the filter around a single
1. Select C1 usually between 100 pF and 1 μF. active element such as an op-amp. This works
2. For the specific type of filter (Butterworth, well up to about n ¼ 3 but problems arise for
Chebyshev, Bessel) and the specific filter filters of higher order. The second general method
requirements, obtain the coefficients a1 and is to factor the higher-order transfer function A(s)
b1 from the relevant table. into first- and second-order transfer functions
3. Determine C2 to satisfy C2  C 1 4b1 ða1þA 1
2

. A1(s) and A2(s), each of which can be synthesized
4. Find R1, R2 and R3 using and then cascaded. This approach has the advan-
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi tage of simplicity, and the resulting first- and
a þ a1 2 4b1 ð1þAo ÞC 1 =C 2
R1 ¼ 1 4π f C 1 Ao , R2 ¼ AoR1 and second-order transfer functions are those already
c

R3 ¼ 4π2 b1
2 . analysed and used in filter implementation. In this
f c C 1 C 2 R2
5. Scale values if necessary. chapter, the factorizing approach is pursued in the
development of higher-order filters.
Example 11.8 Therefore, the approach adopted here in
Design a MFB second-order low-pass realizing higher-order low-pass filters is to cas-
Butterworth filter with a gain of 10 and a cut-off cade first- and/or second-order filters in order to
frequency of 10 kHz. approximate the higher-order filter. Thus, in order
to realize a fourth-order filter, two second-order
Solution For this case, fc ¼ 10 kHz. Select filters are cascaded. In order to realize an odd-
C1 ¼ 0.001 μF. From Table 11.1, the order filter, at least one first-order must be cas-
Butterworth coefficients are a1 ¼ 1.4142, caded with one or more second-order filters. For
example, in order to synthesize a third-order filter,
b1 ¼ 1. Then C 2  C 1 4b1 ða1þA oÞ
1
2
one first-order and one second-order filters are
C2  0.001  106  4  1  (1 + 10)/ cascaded. Similarly, if a fifth-order filter is
pffiffiffi2
2 ¼ 0:022 μF . Select C2 ¼ 0.1 μF. Hence required, then one first-order and two second-
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
pffiffi 2 order filters must be cascaded.
1:4142þ ð 2Þ ð41110:001106 =0:1106 Þ
R1 ¼ The transfer function of one stage is given by
4π10103 0:001106 10
¼ 2.12 k, R2 ¼ AoR1 ¼ 2.12 k and R3 ¼ Ao
AðsÞ ¼   ð11:63Þ
1
¼ 119.5 Ω. 1 þ a1 ðs=ωc Þ þ b1 ðs=ωc Þ2
4πð104 Þ 0:001106 0:1106 21:2103
2

In a first-order transfer function, the coefficient


b1 is zero giving the transfer function
424 11 Active Filters

Ao seen earlier. Therefore, the general transfer func-


AðsÞ ¼ ð11:64Þ
1 þ a1 ðs=ωc Þ tion representing this cascade of second- and first-
order transfer functions is given by

Ao
AðsÞ ¼  
∏i 1 þ ai ðs=ωc Þ þ bi ðs=ωc Þ2
Ao
¼     
2
1 þ a1 ðs=ωc Þ þ b1 ðs=ωc Þ 1 þ a2 ðs=ωc Þ þ b2 ðs=ωc Þ2 . . . 1 þ an ðs=ωc Þ þ bn ðs=ωc Þ2
ð11:65Þ

The multiplication of the factors in the denom- first- and second-order stages is shown diagram-
inator of the transfer function yields an nth-order matically in Fig. 11.17 up to the sixth order. As
polynomial where n represents the order of the can be seen, a filter with even order comprises
filter. The filter coefficients ai and bi determine only second-order stages, while a filter with odd
the filter characteristics such as Butterworth, order has a first-order stage included.
Chebyshev and Bessel, the types being consid-
ered in this discussion. These are tabulated in
Table 11.1 up to order 10. This cascading of

Fig. 11.17 Cascading first- and second-order stages to realize higher-order filters
11.4 Higher-Order Low-Pass Filters 425

11.4.1 Third-Order Low-Pass A f ðsÞ ¼


1
ð11:69Þ
Unity-Gain Filter 1 þ R3 C3 s

Comparing coefficients in (11.68) and (11.69)


In order to develop a third-order unity-gain
gives
low-pass filter, we cascade first- and second-
order filters such that a1 =2π f c ¼ R3 C3 ð11:70Þ
Ao
AðsÞ ¼   The design steps are as follows:
ð1 þ a1 ðs=ωc ÞÞ 1 þ a2 ðs=ωc Þ þ b2 ðs=ωc Þ2 1. Select C3 usually between 100 pF and 1 μF.
ð11:66Þ 2. For the specific type of filter (Butterworth,
Chebyshev, Bessel) and the specific filter
Note that b1 ¼ 0 for the first-order part of the requirements, obtain the coefficient
transfer function. For the unity-gain system under a1(b1 ¼ 0) from the relevant table.
consideration, Ao ¼ 1. Hence (11.66) can be 3. Determine R using R3 ¼ a1/2πfcC.
written as 4. Set Ros2 ¼ R for minimum offset.
1 1
A ðsÞ ¼ :  Partial Filter 2
1 þ a1 ðs=ωc Þ 1 þ a ðs=ω Þ þ b ðs=ω Þ2
2 c 2 c The second-order system is
ð11:67Þ 1
AðsÞ ¼   ð11:71Þ
The circuit of this system is shown in 1 þ a2 ðs=ωc Þ þ b2 ðs=ωc Þ2
Fig. 11.18, comprising a unity-gain non-inverting
first-order system cascaded with a unity-gain
non-inverting second-order system. The design
process is as follows:

Partial Filter 1
The first-order system is
1
AðsÞ ¼ ð11:68Þ
1 þ a1 ðs=ωc Þ

where ωc ¼ 2πfc is the cut-off frequency.


The transfer function for the first-order non-
inverting unity-gain low-pass filter in Fig. 11.19
Fig. 11.19 First-order low-pass partial filter
is given by

Fig. 11.18 Third-order low-pass unity-gain filter


Fig. 11.20 Second-order low-pass partial filter
426 11 Active Filters

Fig. 11.21 Completed third-order low-pass Butterworth filter for Example 11.21

The transfer function for the second-order non- For this case, fc ¼ 20 kHz. Select C3 ¼ 1 nF.
inverting unity-gain low-pass filter in Fig. 11.20 From Table 11.1, a1 ¼ 1, b1 ¼ 0. Hence
is given by R3 ¼ a1/2πfcC3 ¼ 1/2π  20  103  109 ¼ 8 k.
1
Af ¼ Partial Filter 2
1 þ sC 1 ðR1 þ R2 Þ þ s2 R1 R2 C 1 C 2
For this case, fc ¼ 20 kHz. Select C1 ¼ 0.001 μF.
ð11:72Þ From Table 11.1, a2 ¼ 1, b2 ¼ 1. Then
C2  4b2C1/a22 ¼ 4  1  0.001  106 ¼
Comparing coefficients for (11.63) and (11.62)
0.004 μF. Select C2 ¼ 0.005 μF. Hence R1 ¼
yields pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1þ 1ð410:001106 =0:005106 Þ
¼ 5.8 k and R2
a2 =ωc ¼ C 1 ðR1 þ R2 Þ ð11:73Þ p
3
0:001106
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
4π2010
1 1ð410:001106 =0:005106 Þ
¼ ¼ 2.2 k.
b2 =ωc 2 ¼ R1 R2 C 1 C 2 ð11:74Þ 4π20103 0:001106
Finally, for minimum offset, Ros ¼ 5.8 k + 2.2 k
From this, the design steps are as follows: ¼ 8 k.
1. Select C1 usually between 100 pF and 1 μF.
2. For the specific type of filter (Butterworth, The completed filter is shown in Fig. 11.21.
Chebyshev, Bessel) and the specific filter
requirements, obtain the coefficients a2 and Simplified Design Procedure for Third-Order
b2 from the relevant table. Low-Pass Unity-Gain Butterworth Filter
3. Determine C2 using C2  4b2C1/a22. A simplified design procedure for the unity-gain
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
a þ a2 2 4b2 C 1 =C 2 third-order Butterworth filter can be developed.
4. Find R1 and R2 using R1 ¼ 2 4π f c C 1
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi Consider the transfer function (11.75) for a
a2  a2 2 4b2 C 1 =C 2
and R2 ¼ . low-pass third-order filter:
4π f C1 c

5. Determine Ros1 ¼ R1 + R2 for minimum offset Ao


AðsÞ ¼  
voltage. ð1 þ a1 ðs=ωc ÞÞ 1 þ a2 ðs=ωc Þ þ b2 ðs=ωc Þ2

Example 11.9 ð11:75Þ


Design a VCVS third-order unity-gain low-pass
For unity gain, Ao ¼ 1 and for a Butterworth
Butterworth filter with a cut-off frequency of
response from Table 11.1, a1 ¼ 1 and a2 ¼ 1,
20 kHz.
b2 ¼ 1. Then setting s ¼ jω, (11.10) becomes
Solution The design of each of the two partial 1 1
A¼ : ð11:76Þ
filters is considered separately. 1 þ jf = f c 1 þ jf = f c þ ðjf = f c Þ2

Partial Filter 1
11.4 Higher-Order Low-Pass Filters 427

Fig. 11.22 Third-order


low-pass unity-gain
Butterworth filter

This transfer function can be implemented using Butterworth filter with a cut-off frequency of
the configuration shown in Fig. 11.18 which com- 9 kHz.
prises cascaded first- and second-order systems.
For this system, the transfer function is given by Solution Choose C1 ¼ 0.005 μF and then
C3 ¼ 2C1 ¼ 0.01 μF and C2 ¼ 2C3 ¼ 0.02 μF.
1 1
Af ¼ : Using (11.70), R ¼ 1/2πfcC3 ¼ 1/
1 þ sR3 C 3 1 þ sC 1 ðR1 þ R2 Þ þ s2 C 1 C 2 R1 R2
2π  9  103  0.01  106 ¼ 1.8 k. Hence
ð11:77Þ R1 ¼ R2 ¼ R3 ¼ 1.8 k.
Let R1 ¼ R, R2 ¼ R, R3 ¼ R and C1 ¼ C3/2 and
C2 ¼ 2C3. Then Fig. 11.18 becomes Fig. 11.22 Example 11.11
and (11.77) becomes Design a VCVS fourth-order unity-gain low-pass
Chebyshev filter with a cut-off frequency of
1 1 20 kHz and 1 dB ripple width.
Af ¼ :
1 þ sRC 3 1 þ sC 3 R þ s2 ðC 3 RÞ2
1 1 Solution The design of each of the two partial
¼ : ð11:78Þ filters is considered separately.
1 þ jf = f c 1 þ jf = f c þ ðjf = f c Þ2

where Partial Filter 1


For this case, fc ¼ 20 kHz. Select C1 ¼ 0.001 μF.
f c ¼ 1=2πRC 3 ð11:79Þ From Table 11.2, a1 ¼ 2.5904, b1 ¼ 4.1301.
Then C2  4b1C1/a12 ¼ 4  4.1301  0.001 
which is the cut-off frequency. Clearly, transfer
106/(2.5904)2 ¼ 0.0025 μF. Select C2 ¼
function (11.78) is identical to transfer function
0.005 μF. Hence R1
(11.76). Hence a simplified design procedure for pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ffi
2:5904þ ð2:5904Þ ð44:13010:00110 =0:00510 Þ
2 6 6
a unity-gain third-order low-pass Butterworth ¼ 4π20103 0:001106
filter is as follows: ¼ 9.9 k and R2 ¼
1. Choose the desired cut-off frequency fc. pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

2:5904 ð2:5904Þ2 ð44:13010:001106 =0:005106 Þ
2. Choose C3, usually an available value between 4π20103 0:001106
¼
about 100 pF and 1 μF. 2.6 k. Finally, for minimum offset, Ros ¼ 9.9 k +
3. Let C1 ¼ 1/2C3 and C2 ¼ 2C3. 2.6 k ¼ 12.5 k.
4. Determine R using R ¼ 1/2πfcC3. Set R1 ¼ R2
¼ R3 ¼ R. Partial Filter 2
For this case, fc ¼ 20 kHz. Select C1 ¼ 0.001
Example 11.10 μF. From Table 11.1, a2 ¼ 0.3039, b2 ¼
Using the circuit of Fig. 11.22 and the simplified 1.1697. Then C2  4b2C1/a22 ¼ 4  1.1697
design procedure, design a third-order low-pass  0.001  106/(0.3039)2 ¼ 0.05 μF. Select
C2 ¼ 0.1 μF. Hence R1
428 11 Active Filters

Fig. 11.23 Completed


third-order low-pass filter
for Example 11.13

Fig. 11.24 Completed fifth-order low-pass filter for Example 11.14

pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
0:3039þ ð0:3039Þ2 ð41:16970:001106 =0:1106 Þ For this case, fc ¼ 50 kHz. Select C1 ¼ 820
¼ 4π20103 0:001106 pF. From Table 11.1, a2 ¼ 1.6180, b2 ¼ 1. Then
¼ 2 k and R2 ¼ C2  4b2C1/a22 ¼ 4  1  820  1012/
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
0:3039 ð0:3039Þ2 ð41:16970:001106 =0:1106 Þ (1.618)2 ¼ 1.26 nF. Select C2 ¼ 1.5 nF. Hence
4π20103 0:001106
¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1:618þ ð1:618Þ2 ð410:820109 =1:5109 Þ
360 Ω. Finally, for minimum offset, Ros ¼ 2 k + R1 ¼ 4π50103 0:820109
360 Ω ¼ 2360 Ω. ¼ 4.4 k and R2 ¼
The completed filter is shown in Fig. 11.23. pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
1:618 ð1:618Þ ð410:82010 =1:510 Þ
2 9 9

4π50103 0:820109
¼ 1.9 k.
Example 11.12 Finally, for minimum offset, Ros ¼ 4.4 k + 1.9 k
Design a VCVS fifth-order unity-gain low-pass ¼ 6.3 k.
Butterworth filter with a cut-off frequency of
50 kHz. Partial Filter 3
For this case, fc ¼ 50 kHz. Select C1 ¼ 330
Solution The design of each of the three partial pF. From Table 11.1, a3 ¼ 0.6180, b3 ¼ 1. Then
filters is considered separately. C2  4b3C1/a32 ¼ 4  1  330  1012/
(0.6180)2 ¼ 3.5 nF. Select C2 ¼ 4.7 nF. Hence
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Partial Filter 1 0:618þ ð0:618Þ2 ð410:330109 =4:7109 Þ
R1 ¼
For this case, fc ¼ 50 kHz. Select C ¼ 1 nF. From 4π50103 0:330109
Table 11.1, a1 ¼ 1, b1 ¼ 0. Hence R ¼ a1/ ¼ 4.5 k and R2 ¼
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2πfcC ¼ 1/2π  50  103  109 ¼ 3.2 k. 1:618 ð1:618Þ ð410:33010 =4:710 Þ
2 9 9

4π50103 0:330109
¼ 1.5 k.

Partial Filter 2
11.5 High-Pass First-Order Filter-Butterworth Response 429

Finally, for minimum offset, Ros ¼ 4.5 k + 1.5 k included in the feedback loop in order to reduce
¼ 6 k. DC offset at the output as shown in Fig. 11.25(b).
The completed filter is shown in Fig. 11.24. For the circuit
Vo R sCR
Af ¼ ðsÞ ¼ ¼
Vi R þ 1=sC 1 þ sCR
11.5 High-Pass First-Order jf = f c
Filter-Butterworth Response ¼ ð11:80Þ
1 þ jf = f c
High-pass filters can be generated from low-pass where
filters by changing resistors to capacities and
capacities to resistors in the filter network. This f c ¼ 1=2πRC ð11:81Þ
corresponds to the transformation s ! 1/s or
Note that for f  fc, |Af| ! 1 and for f  fc, |
jð f = f c Þ ! jð f =1 f Þ . Thus, the first-order low-pass
c Af| ! 0. The frequency response plot for this
filter given by 1
1þjf = f c is transformed such that circuit is shown in Fig. 11.26, where the cut-off
jf = f c
1
1þjf = f c ! 1
1þjf =1f
¼ 1þjf = f c . A basic first-order frequency is fc ¼ 1/2πRC. At f ¼ fc, the filter
c
transfer function becomes
high-pass filter is shown in Fig. 11.25(a). It
employs a CR network to provide the high-pass jf = f c j
filtering, the output of which feeds into a unity- A f ðf cÞ ¼ ¼ ð11:82Þ
1 þ jf = f c 1 þ j
gain buffer. It is obtained from the first-order    
low-pass filter by interchanging R and C in the Hence, A f ð f c Þ ¼  j  ¼ p1ffiffi ¼ 0:707 ¼
1þj 2
filter network. A resistor ROS ¼ R may be
3 dB. Thus, at the break frequency f ¼ fc the
magnitude response falls by 3 dB from its
mid-band value as shown in Fig. 11.26.

Example 11.13
Design an RC high-pass first-order Butterworth
filter with a cut-off frequency of 1 kHz.

Solution Choose C ¼ 0.01 μF. Then from


(11.81), R ¼ 1/2πfcC ¼ 1/2π  103  0.01 
(a) (b)
106 ¼ 15.9 k. Choose Rf ¼ R ¼ 15.9 k for low
Fig. 11.25 High-pass first-order filter DC offset.

Fig. 11.26 First-order high-pass filter response


430 11 Active Filters

11.5.1 First-Order High-Pass Filter Vi V o


¼ ð11:84Þ
with Gain R1 þ 1=sC R2

The high-pass first-order filter with gain is shown Hence the transfer function is given by
in Fig. 11.27 where the active device is modified R2 sCR1 jf = f c
to produce a gain greater than unity. The transfer Af ¼  : ¼ k ð11:85Þ
R1 1 þ sCR1 1 þ jf = f c
function is given by
where k ¼ R2/R1 is the gain and
Vo kjf = f c
Af ¼ ¼ ð11:83Þ
V i 1 þ jf = f c f c ¼ 1=2πR1 C ð11:86Þ

where k ¼ 1 + R2/R1 and fc ¼ 1/2πRC. The system is the break frequency.


has a high-frequency gain of k ¼ 1 + R2/R1 and a
break frequency of fc ¼ 1/2πRC. For low DC Example 11.14
offset, R ¼ R1//R2. Design a first-order high-pass Butterworth
An alternative configuration for a first-order non-inverting filter using Fig. 11.27 with a gain
high-pass filter is shown in Fig. 11.28. It is an of 10 and a cut-off of 1 kHz.
inverting configuration with a capacitor C in
series with the input resistor R1. From Solution Choose C ¼ 0.01 μF. Then from
Fig. 11.28, we have (11.86), R ¼ 1/2πfC giving R ¼ 1/(2  π  103
 0.01  106) ¼ 15.9 k. Since 1 + R2/R1 ¼10 and
R ¼ R1//R2, therefore R2 ¼ 159 k and R1 ¼ 17.7 k.

Design Procedure
As we did for the low-pass filters, we wish to
write down an ordered procedure for designing
first-order high-pass filters. Thus, the basic first-
order system can be written as
A1
AðsÞ ¼ ð11:87Þ
1 þ a1 =ðs=ωc Þ

where A1 ¼ k is the pass-band gain, ωc ¼ 2πfc is


the cut-off frequency and a1 ¼ 1 for first-order
Fig. 11.27 First-order high-pass filter with gain filters. For higher-order filters considered later,

(a) (b)

Fig. 11.28 Inverting first-order high-pass filter


11.6 High-Pass Second-Order Filter 431

the condition a1 6¼ 1 will arise. Equation (11.87) 4. Determine R1 using R1 ¼ R2/A1.


can be written as
A1 s
AðsÞ ¼ ð11:88Þ 11.6 High-Pass Second-Order Filter
s þ a 1 ωc

For the first-order non-inverting high-pass As in the case of second-order low-pass filters, a
filter, the transfer function is given by (see earlier) high-pass filter can be optimized to meet at least
  one of the following three requirements:
1 þ RR21 (i) maximum flatness in the pass-band;
A f ðsÞ ¼ ð11:89Þ
s þ 1=RC (ii) maximum sharpness in the transition from
pass-band to stop-band; and (iii) linear phase
Comparing coefficients in (11.89) and (11.88) response. In order to allow this optimization of
gives the filter, the transfer function must contain com-
R2 plex poles and hence, for a second-order filter,
A1 ¼ 1 þ ð11:90Þ can take the following form:
R1
1 A1 s2
a 1 ωc ¼ ð11:91Þ AðsÞ ¼ ð11:96Þ
RC s2 þ a 1 ωc s þ b 1 ωc 2

The design steps are as follows: where A1 is the low-frequency gain in the pass-
1. Select C usually between 100 pF and 1 μF. band and a1 and b1 are positive real filter
2. Set the coefficient a1 ¼ 1. coefficients that define the location of the complex
3. Determine R using R ¼ 1/a12πfcC. poles of the filter and hence the characteristics of the
4. Select R1 and find R2 using R2 ¼ R1(A1  1). filter. The angular frequency ωc ¼ 2πfc represents
the cut-off frequency of the filter. This transfer
For the first-order inverting low-pass filter, the function can be normalized by setting ωc ¼ 1 giving
transfer function is given by (see earlier)
A1 s2
AðsÞ ¼ ð11:97Þ
1 þ a1 s þ b1
R2
sR1
s2
AðsÞ ¼  ð11:92Þ
s þ 1=R1 C 1
Only the Butterworth, Chebyshev and Bessel
Comparing coefficients in (11.92) with the responses are considered. These filter responses
general transfer function occur for second (and higher) order and not for
first-order where complex poles are not possible
AðsÞ ¼ 
A1 s
ð11:93Þ and therefore all three filter types have identical
s þ a1 ω c responses for a first-order high-pass filter. The
amplitude responses of first-, second- and
gives equation (11.94):
higher-order high-pass Butterworth filters using
R2 the normalized characteristic are shown in
A1 ¼ ð11:94Þ
R1 Fig. 11.29. The second-order high-pass filter
transfer function (11.97) can be realized around
1
a 1 ωc ¼ ð11:95Þ a single active device using two topologies, the
R1 C1
Sallen-Key and the multiple feedback.
The design steps are as follows:
1. Select C1 usually between 100 pF and 1 μF.
2. Set the coefficient a1 ¼ 1.
3. Determine R1 using R1 ¼ 1/a12πfcC1.
432 11 Active Filters

Fig. 11.29 High-pass filter response plots for Butterworth filters

s2 C 1 C 2 R1 R2
A ðsÞ ¼
1 þ sR2 ðC 1 þ C 2 Þ þ s2 R1 R2 C 1 C 2
ð11:100Þ

In order to simplify the development, let


C1 ¼ C2 ¼ C. Then (11.100) becomes

s2
A f ðsÞ ¼ ð11:101Þ
s2 þ 2
R1 C s þ C2 R1 R
1 2
Fig. 11.30 Second-order high-pass unity-gain filter
Comparing coefficients for (11.101) and
(11.96) yields
11.6.1 Sallen-Key or A1 ¼ 1 ð11:102Þ
Voltage-Controlled Voltage
Source (VCVS) Topology 2
a 1 ωc ¼ ð11:103Þ
R1 C
The circuit shown in Fig. 11.30 is the VCVS or 1
Sallen-Key implementation of a second-order high- b1 ω c 2 ¼ ð11:104Þ
R1 R2 C2
pass unity-gain filter. Comparing this with the
second-order low-pass filter, it can be seen that the From (11.103),
resistors and capacitors of the filter network are
1
interchanged. Using KCL at nodes A and B, we have R1 ¼ ð11:105Þ
a1 π f c C
ðV i  V A ÞsC1 ¼ ðV A  V B ÞsC2 þ ðV A  V o Þ=R2
and from (11.104),
ð11:98Þ
a1
R2 ¼ ð11:106Þ
ðV A  V B ÞsC 2 ¼ V B =R1 ð11:99Þ b1 4πfC

where VB ¼ Vo. These equations yield We can now state a design procedure for a
specific cut-off frequency:
11.6 High-Pass Second-Order Filter 433

Develop a design procedure for the VCVS


Butterworth second-order high-pass filter with
gain for the special case when R1 ¼ R2 and
C1 ¼ C2.

Solution Using nodal analysis, the transfer func-


tion for the filter with gain is given by
Vo
Af ¼
Vi
Fig. 11.31 Sallen-Key second-order high-pass filter ks2
with gain ¼ h  i
s2 þ s C11 þ C12 þ C11R2 ð1  kÞ þ R1 R21C1 C2
Design Procedure ð11:108Þ
1. Select C1 ¼ C2 ¼ C usually between 100 pF
and 1 μF. For the case where R1 ¼ R2 ¼ R and
2. For the specific type of filter (Butterworth, C1 ¼ C2 ¼ C, then (11.108) reduces to
Chebyshev, Bessel) and the specific filter
Vo ks2
requirements, obtain the coefficients a1 and Af ¼ ¼ 2 ð11:109Þ
V i s þ s RC ð3  k Þ þ 21 2
1
b1 from the relevant table. R C
3. Find R1 and R2 using R1 ¼ π f 1a1 C and R2 ¼
c Comparing coefficients with
a1
4πfb1 C .
4. Determine Ros ¼ R1 for minimum offset volt- A1 s2
AðsÞ ¼ ð11:110Þ
age if required. s2 þ a 1 ωc s þ b 1 ωc 2

yields
A1 ¼ k ð11:111Þ
Example 11.15
a1 ωc ¼ ð3  kÞCR ð11:112Þ
Design a VCVS second-order unity-gain high-
pass Butterworth filter with a cut-off frequency b 1 ωc 2 ¼ R 2 C 2 ð11:113Þ
of 10 kHz.
pffiffiffi
For a Butterworth filter, a1 ¼ 2 and b1 ¼ 1.
Solution For this case, fc ¼ 10 kHz. Select C1 ¼ Hence from (11.113),
pffiffiffi
C2 ¼ 0.01 μF. From Table 11.1, a1 ¼ 2, b1 ¼
ωc ¼ 1=RC, f c ¼ 1=2πRC ð11:114Þ
1: Then R1 ¼ π10103 p1ffiffi20:01106 ¼ 2.3 k and
pffiffi
R2 ¼ 4π10103 0:0110
2
6 ¼ 1.1 k. Finally, for and from (11.112)
minimum offset, Ros ¼ 2.3 k. pffiffiffi
2¼3k ð11:115Þ
The system can be designed for gain as shown
in Fig. 11.31. where giving
Rb pffiffiffi
k ¼1þ ð11:107Þ k ¼3 2 ¼ 1:6 ð11:116Þ
Ra
For minimum offset
The process of developing the associated
design procedure follows that used for the unity-
gain case.

Example 11.16
434 11 Active Filters

Ra Rb R s2 C 1 C 2 R1 R2
R ¼ Ra ==Rb ¼ ¼ b ð11:117Þ Af ¼
Ra þ Rb k 1 þ sR2 ðC 1 þ C 2 Þ þ s2 R1 R2 C1 C2
Hence ð11:123Þ

Rb ¼ kR ¼ 1:6 R ð11:118Þ In order to obtain the maximally flat


Butterworth response, let C1 ¼ C2 ¼ C, R1 ¼
Also R, R2 ¼ R/2. Then, (11.123) reduces to
Ra Rb ð k  1Þ s2 C2 R2 =2
R ¼ Ra ==Rb ¼ ¼ Ra ð11:119Þ Af ¼
Ra þ Rb k 1 þ sCR þ s2 R2 C 2 =2
Hence ðjf = f c Þ2
¼ pffiffiffi ð11:124Þ
kR 1 þ 2jf = f c þ ðjf = f c Þ2
Ra ¼ ¼ 1:6 R=0:6 ¼ 2:7 R ð11:120Þ
k1
where
We can now state a design procedure for a pffiffiffi
specific cut-off frequency: fc ¼ 2=2πRC ð11:125Þ

This transfer function is identical to the


Design Procedure second-order Butterworth high-pass filter in
(11.122). The design procedure is as follows:
1. Select C1 ¼ C2 ¼ C usually between 100 pF 1. Determine the desired cut-off frequency fc.
and 1 μF. 2. Choose a value of C between 100 pF and
2. Determine R ¼ 1/2πfcC. 1 μF.
3. Set R1 ¼ R2 ¼ R. 3. Set C2 ¼ C1 ¼ C.
4. Find Ra and Rb using (11.120) and (11.118). pffiffiffi
4. Determine R using R ¼ 2=2π f c C.
Simplified Design Procedure for Second-Order
Example 11.17
High-Pass Unity-Gain Butterworth Filter
Using the configuration shown in Fig. 11.30,
A simplified design procedure for the unity-gain
re-design the second-order high-pass Butterworth fil-
second-order Butterworth filter can be developed.
ter of Example 11.15 using the simplified approach.
Consider the transfer function (11.121) for a high-
pass second-order filter:
Solution For fc ¼ 10 kHz, choose
A1 s 2 C1 ¼ C2 ¼ 0.01 μF. Using (11.125),
AðsÞ ¼ ð11:121Þ pffiffiffi
s 2 þ a 1 ωc s þ b 1 ωc 2 R ¼ 2=2π  10  103  0:01  106 ¼ 2:3 k and
hence R1 ¼ 2.3 k, R2 ¼ 1.1 k and Ros ¼ 2.3 k.
For unity gain, A1 ¼ 1 and for
pffiffiaffi Butterworth
response from Table 11.1, a1 ¼ 2 and b1 ¼ 1. Example 11.18
Then setting s ¼ jω, (11.121) becomes Using C1 ¼ C2 ¼ 0.1 μF, determine the cut-off
frequency in the high-pass filter if R1 ¼ 1 k and
ðjf = f c Þ2
Aðjf Þ ¼ pffiffiffi ð11:122Þ R2 ¼ 500 Ω.
1 þ 2jf = f c þ ðjf = f c Þ2
pffiffiffi
The transfer function for a second-order high- Solution From Eq. (11.125), f c ¼ 2=2π 
pass filter using the VCVS or Sallen-Key topol- 103  0:1  106 ¼ 2:3 kHz.
ogy in Fig. 11.30 is given by
Example 11.19
11.6 High-Pass Second-Order Filter 435

Design a VCVS second-order unity-gain C1 2


C2 s
high-pass Chebyshev filter with a cut-off fre- A f ðsÞ ¼   
C 1 þC 2 þC 3
quency of 10 kHz and a ripple width s2 þ R2 C 2 C 3 s þ R1 R21C2 C3
RWdB ¼ 1 dB. ð11:128Þ

Solution For this case, fc ¼ 10 kHz. Select For simplicity of the analysis, let C1 ¼ C3 ¼ C.
C1 ¼ C2 ¼ 0.001 μF. From Table 11.2, the Then (11.128) reduces to
coefficients for RWdB ¼ 1 dB are a1 ¼ 1.3022, C 2
b1 ¼ 1.5515. Then from (11.105) and (11.106), C2 s
A f ðsÞ ¼    ð11:129Þ
R1 ¼ π10103 1:30220:00110
1
6 ¼ 24 k and R2 ¼ s2 þ 2CþC 2
R2 C 2 C s þ R1 R21C2 C
1:3022
¼ 6.7 k. Finally, for
4π10103 1:55150:001106
minimum offset, Ros ¼ 24 k. Comparing the coefficients of (11.129) with
the general second-order high-pass transfer
Example 11.20 function
Design a VCVS second-order unity-gain high- A1 s2
pass Bessel filter with a cut-off frequency of AðsÞ ¼  ð11:130Þ
s2 þ a1 ω c s þ b1 ω c 2
10 kHz.
(where a negative sign is introduced to account
Solution For this case, fc ¼ 10 kHz. Select for the inversion produced by the MFB configu-
C1 ¼ C2 ¼ 0.001 μF. From Table 11.1, the Bessel ration) yields
coefficients are a1 ¼ 1.3617, b1 ¼ 0.6180. Then
C
R1 ¼ π10103 1:36170:00110
1
6 ¼ 23.4 k and R2 ¼
A1 ¼ ð11:131Þ
C2
1:3617
¼ 17.5 k. Finally, for
4π10103 0:61800:001106
C 2 þ 2C
minimum offset, Ros ¼ 23.4 k. a 1 ωc ¼ ð11:132Þ
R2 C2 C
1
b 1 ωc 2 ¼ ð11:133Þ
R1 R2 C 2 C
11.6.2 High-Pass Second-Order
Multiple Feedback Filter Solving for R1, R2 in terms of C and C2, we get
a1
Another available topology for implementation of R1 ¼ ð11:134Þ
b1 2π f c ðC 2 þ 2C Þ
second-order high-pass filters is the multiple feed-
back (MFB) topology shown in Fig. 11.32. In
this circuit the resistors and capacitors are
interchanged compared to the corresponding
low-pass MFB circuit.
Assuming an ideal op-amp, then for nodes a
and b,
ðV i  V a ÞsC 1 ¼ ðV a  V o ÞsC 2 þ ðV a  V b Þ
Va
sC 3 þ ð11:126Þ
R1

ðV b  V o Þ
ðV a  V b ÞsC 3 ¼ ð11:127Þ
R2
Fig. 11.32 High-pass second-order multiple feedback
Noting that Vb ¼ 0 because node b is a virtual
filter
earth and eliminating Va, we get
436 11 Active Filters

ðC2 þ 2C Þ
R2 ¼ ð11:135Þ
a1 2π f c C 2 C

Design Procedure
1. Select C usually between 100 pF and 1 μF.
2. Determine C2 using C2 ¼ C/A1.
3. For the specific type of filter (Butterworth,
Chebyshev, Bessel) and the specific filter Fig. 11.33 Third-order high-pass filter
requirements, obtain the coefficients a1 and
b1 from the relevant table. A1 s
4. Find R1 and R2 using R1 ¼ b1 2π f aðC1 2 þ2CÞ and AðsÞ ¼ ð11:137Þ
c s þ a 1 ωc
R2 ¼ að1C2π2 þ2C Þ
f C2 C.
c Therefore, the general transfer function
5. Scale values if necessary. representing this cascade of second- and first-
order transfer functions is given by
Example 11.21 
Design a MFB second-order high-pass s2
A ðsÞ ¼ A 1 ∏ i 2
Butterworth filter with a gain of 10 and a cut-off s þ ai ω c s þ bi ω c 2
 
frequency of 1 kHz. s2 s2
¼ A1 2
s þ a1 ω c s þ b1 ω c 2 s2 þ a 2 ω c s þ b 2 ω c 2
Solution For this case, fc ¼ 1 kHz. Select C1 
s2
¼ C3 ¼ C ¼ 0.01 μF. Then C2 ¼ 0.01 μF/10 ... 2 ð11:138Þ
s þ a n ωc s þ b n ωc 2
¼ 0.001 μF. From Table 11.1, the Butterworth
coefficients are a1 ¼ 1.4142, b1 ¼ 1. Then R1
¼ b1 2π f aðC1 2 þ2CÞ ¼ 2π103 ð0:001þ0:02
1:4142
Þ106
¼
c
ðC 2 þ2CÞ
10.7 k and R2 ¼ a1 2π f c C 2 C ¼ 11.7.1 Third-Order High-Pass
ð0:001þ0:02Þ10 6 Unity-Gain Filter
1:41422π103 0:001106 0:01106
¼ 236 k.
In order to develop a third-order unity-gain high-
pass filter, we cascade first- and second-order
11.7 Higher-Order High-Pass Filters filters such that

s s2
We come now to higher-order filters in which the AðsÞ ¼ A1 : 2
s þ a 1 ωc ð s þ a 2 ωc s þ b 2 ωc 2 Þ
transfer function has a denominator polynomial
that is of degree n. We again use the method of ð11:139Þ
factoring the higher-order transfer function A(s) Note again that b1 ¼ 0 for the first-order part of
into first- and second-order transfer functions the transfer function. For the unity-gain system
A1(s) and A2(s), each of which is then synthesized under consideration, A1 ¼ 1. Hence (11.139) can
and cascaded. be written as
The transfer function of one stage is given by
s s2
A1 s2 AðsÞ ¼ : 2
AðsÞ ¼ 2 ð11:136Þ s þ a 1 ω c ð s þ a 2 ωc s þ b 2 ωc 2 Þ
s þ a 1 ωc s þ b 1 ωc 2
ð11:140Þ
In a first-order transfer function, the coefficient
b1 is zero giving the transfer function The circuit of this third-order system is shown
in Fig. 11.33. It comprises a unity-gain
non-inverting first-order system cascaded with a
11.7 Higher-Order High-Pass Filters 437

unity-gain non-inverting second-order system. From (11.146),


The design process is as follows:
1
R1 ¼ ð11:148Þ
a2 π f c C
Partial Filter 1
The first-order system is and from (11.147),
s a1
AðsÞ ¼ ð11:141Þ R2 ¼ ð11:149Þ
s þ a 1 ωc b2 4π f c C
where ωc ¼ 2πfc is the cut-off frequency. The We can now state a design procedure for a
transfer function for the first-order non-inverting specific cut-off frequency:
unity-gain high-pass filter in Fig. 11.33 is given by
R3 C 3 s s Design Procedure
A f ðsÞ ¼ ¼ ð11:142Þ 1. Select C usually between 100 pF and 1 μF. Set
1 þ R3 C 3 s s þ 1=R3 C 3
C1 ¼ C2 ¼ C.
Comparing coefficients in (11.142) and 2. For the specific type of filter (Butterworth,
(11.141) gives Chebyshev, Bessel) and the specific filter
requirements, obtain the coefficients a2 and
1
a 1 ωc ¼ ð11:143Þ b2 from the relevant table.
R3 C3
3. Find R1 and R2 using R1 ¼ π f 1a2 C and R2 ¼
c
The design steps are as follows: a2
4π f c b2 C .
1. Select C3 usually between 100 pF and 1 μF.
2. For the specific type of filter (Butterworth,
Example 11.22
Chebyshev, Bessel) and the specific filter
Design a VCVS third-order unity-gain high-pass
requirements, obtain the coefficient a1(b1 ¼ 0)
Butterworth filter with a cut-off frequency of
from the relevant table.
20 kHz.
3. Determine R using R3 ¼ 1/a12πfcC3.
Solution The design of each of the two partial
Partial Filter 2
filters is considered separately.
The second-order system is

s2 Partial Filter 1
AðsÞ ¼ ð11:144Þ For this case, fc ¼ 20 kHz. Select C3 ¼ 1 nF.
ðs2 þ a 2 ωc s þ b 2 ωc 2 Þ
From Table 11.1, a1 ¼ 1, b1 ¼ 0. Hence R3 ¼ 1/
The transfer function for the second-order non- 2πfca1C3 ¼ 1/2π  20  103  109 ¼ 8 k.
inverting unity-gain low-pass filter in Fig. 11.33
is given by (C1 ¼ C2 ¼ C) Partial Filter 2

s2
A f ðsÞ ¼ ð11:145Þ
s2 þ 2
R1 C s þ C2 R11 R2

Comparing coefficients for (11.145) and


(11.144) yields
2
a 2 ωc ¼ ð11:146Þ
R1 C
1
b2 ω c 2 ¼ ð11:147Þ Fig. 11.34 Solution to Example 11.24
R1 R2 C2
438 11 Active Filters

For this case, fc ¼ 20 kHz. Select C ¼ 0.001 μF f c ¼ 1=2πCR3 ð11:154Þ


and select C1 ¼ C2 ¼ C. From Table 11.1, a2 ¼ 1,
b2 ¼ 1. Hence R1 ¼ π f 1a2 C ¼ π20103 0:00110
1
6 ¼
which is the cut-off frequency. This transfer func-
c
tion (11.153) is the same as (11.151). We can there-
15.9 k and R2 ¼ a2
4π f c b2 C ¼ 1
4π20103 0:001106
¼
fore write down a simplified design procedure:
4 k. The completed filter is shown in Fig. 11.34.
1. Choose the desired cut-off frequency fc.
Simplified Design Procedure for Third-Order 2. Choose C, usually an available value between
High-Pass Unity-Gain Butterworth Filter about 100 pF and 1 μF. Let C1 ¼ C2 ¼ C3 ¼ C
A simplified design procedure for the unity-gain 3. Determine R3 using R3 ¼ 1/2πfcC. Set
high-pass third-order Butterworth filter can be R1 ¼ 2R3 and R2 ¼ R3/2.
developed. Consider the transfer function for a
high-pass second-order filter: Example 11.23
Using the simplified design procedure, design a
s s2
AðsÞ ¼ A1 : 2 third-order high-pass Butterworth filter with a
s þ a 1 ωc ð s þ a 2 ωc s þ b 2 ωc 2 Þ
cut-off frequency of 9 kHz.
ð11:150Þ

For unity gain, A1 ¼ 1 and for a Butterworth Solution Choose C ¼ 0.01 μF and then C1 ¼
response from Table 11.1, a1 ¼ 1 and a2 ¼ 1, C2 ¼ C3 ¼ 0.01 μF. Using (11.154), R3 ¼ 1/
b2 ¼ 1. Then setting s ¼ jω, (11.150) becomes 2πfcC ¼ 1/2π  9  103  0.01  106 ¼
1.8 k. Hence R1 ¼ 2R3 ¼ 3.6 k and R2 ¼ R3/
ðjf = f c Þ2 jf = f c 2 ¼ 1.8 k/2 ¼ 900 Ω.
A¼ : ð11:151Þ
1 þ jf = f c þ ðjf = f c Þ2 1 þ jf = f c
Example 11.24
where fc is the cut-off frequency. This transfer Design a VCVS third-order unity-gain Bessel
function can be implemented using the configura- high-pass filter using Fig. 11.33 with a cut-off
tion shown in Fig. 11.33 which comprises cas- frequency of 1 kHz.
caded first- and second-order systems. For this
system, the transfer function is given by Solution The design of each of the two partial
filters is considered separately.
s2 C 1 C 2 R 1 R 2 sC 3 R3
Af ¼ :
1 þ sR2 ðC 1 þ C 2 Þ þ s2 C 1 C 2 R1 R2 1 þ sR3 C 3
Partial Filter 1
ð11:152Þ For this case, fc ¼ 1 kHz. Select C3 ¼ 0.1 μF.
From Table 11.1, a1 ¼ 0.756, b1 ¼ 0. Hence
Let C1 ¼ C2 ¼ C3 ¼ C and R1 ¼ 2R3 and
R3 ¼ 1/2πfca1C¼ 1/2π  103  0.756  0.1 
R2 ¼ R3/2. Then (11.152) becomes

s2 ðCR3 Þ2 sCR3
Af ¼ :
1 þ sR3 C þ s2 ðCR3 Þ2 1 þ sCR3
ðjf = f c Þ2 jf = f c
¼ :
2 1 þ jf = f
1 þ jf = f c þ ðjf = f c Þ c

ð11:153Þ

where

Fig. 11.35 Solution to Example 11.24

106 ¼ 2.1 k.
11.7 Higher-Order High-Pass Filters 439

Fig. 11.36 Band-pass filter response

Fig. 11.37 Normalized


band-pass filter response
for varying values of Q

Partial Filter 2 lower cut-off frequency and upper cut-off fre-


For this case, fc ¼ 1 kHz. Select C ¼ 0.1 μF and quency, respectively. The bandwidth B is the
select C1 ¼ C2 ¼ C. From Table 11.1, frequency difference given by
a2 ¼ 0.9996, b2 ¼ 0.4772. Hence R1 ¼ π f 1a2 C ¼
c B ¼ fU  fL ð11:155Þ
3
1
π10 0:99960:1106
¼ 3.2 k and R2 ¼ a2
4π f c b2 C ¼
1
¼ 1.7 k. The completed filter The relationship betweenpfUffiffiffiffiffiffiffiffiffiffiffiffi
, fL and the reso-
4π103 0:47720:1106
is shown in Fig. 11.35. nant frequency fR is f R ¼ f L f U . Band-pass
filters are generally classified as being either
narrow-band or wide-band. A narrow-band filter
11.7.2 Band-Pass Filter is one having a bandwidth less than one-tenth of
the resonant frequency, otherwise the filter is
A band-pass filter passes frequencies within a classified as wide-band. The quality factor Q of
particular frequency band and attenuates all the circuit is the ratio of resonant frequency to
frequencies outside that band. This is shown in bandwidth given by
Fig. 11.36. This filter has a maximum gain at a
frequency fR referred to as the resonant or centre Q ¼ f R =B ð11:156Þ
frequency. The frequencies fL and fU where the
output falls to 0.707 of its maximum value are the
440 11 Active Filters

Fig. 11.38 Band-pass filter using cascaded low-pass and


high-pass filters
Fig. 11.40 Sallen-Key band-pass filter

filters can be combined in an inverting amplifier


C2 as shown in Fig. 11.39.
In narrow-band filters, B < 0.1fR. This type of
filter has a general second-order transfer function
C1 R
Vi
1 R2 given by
Vo
Vo sωo G Q
¼ 2 ð11:157Þ
V i s þ s ωQo þ ωo 2

where ωo is the centre frequency, G is the centre


Fig. 11.39 Band-pass filter using combined first-order frequency gain and Q is the quality factor. There
filters are several topologies that can be used to imple-
ment this transfer function including the Sallen-
Key, multiple feedback and Wien.
It is a measure of the selectivity of the band-
pass filter, i.e. its ability to select a specific band
of frequencies while rejecting signals outside the
band. The higher the Q, the more selective the 11.7.3 Sallen-Key Band-Pass Filter
band-pass filter as shown in Fig. 11.37. From
(11.156), B < 0.1fR is equivalent to Q > 10 The circuit shown in Fig. 11.40 is a second-order
which corresponds to a narrow-band filter, and Sallen-Key (VCVS) band-pass filter. Node analy-
B > 0.1fR is equivalent to Q < 10 which sis at A and B yields the equations
corresponds to a wide-band filter.

In general, a band-pass filter can be realized by 1 1 Vi V
cascading a low-pass filter and a high-pass filter VA þ þ sðC 1 þ C 2 Þ ¼ þ sC 2 V B þ o
R1 R2 R1 R2
of the same order and gain with the respective ð11:158Þ
cut-off frequencies suitably separated with appro-

priate overlap. Such a circuit is shown in 1
Fig. 11.38. VB þ sC 2 ¼ sC 2 V A ð11:159Þ
R3
The lower cut-off frequency is set by the high-
pass filter, and the upper cut-off frequency is set Vo
VB ¼ ð11:160Þ
by the low-pass filter. The pass-band gain will be k
the gain of each filter and will be constant for a where
band of frequencies. The roll-off rate on each side
of the frequency response characteristic would be Rb
k ¼1þ ð11:161Þ
determined by the order of the filters. In the case Ra
of first-order filters, the low-pass and high-pass This gives
11.7 Higher-Order High-Pass Filters 441

Vo have k ¼ 3  1/Q ¼ 3  1/5 ¼ 2.8. Since


ðsÞ ¼
Vi k ¼ 1 + Rb/Ra, for Ra ¼ 1 k then Rb ¼ 1.8 k.
k
s Note that the resulting value for G using (11.169)
 C 1 R1
 is G ¼ 2.8/(3  2.8) ¼ 14.
s2 þ s C11 1
þ 1
þ 1
ð1  k Þ þ CC2 R1 3 þ C1 CR21RþR 2
R1 R3 R2 1 R2 R3 Thus, this design process allows designing
ð11:162Þ for a particular fo and either Q or G. If greater
design freedom is desirable for specified values
Let C1 ¼ C2 ¼ C and R1 ¼ R2 ¼ R and of fo, Q and G, then the conditions C1 ¼ C2 ¼ C
R3 ¼ 2R. Then (11.162) reduces to and R1 ¼ R2 ¼ R, R3 ¼ 2R would need to be
k relaxed.
Vo CR s
ðsÞ ¼ ð11:163Þ
Vi s2 þ s ð3kÞ þ RC
1
C 2 R2

Comparing coefficients in Eqs. (11.163) and 11.7.4 Multiple Feedback Band-Pass


(11.157) yields Filter
1
ωo 2 ¼ ð11:164Þ A simple circuit to implement the band-pass
R2 C2 transfer function is shown in Fig. 11.41. It is a
ωo ð3  k Þ multiple feedback configuration referred to as the
¼ ð11:165Þ Delyiannis-Friend circuit. Setting the capacitor
Q RC
values equal such that C1 ¼ C2 ¼ C simplifies
Gωo
¼
k
ð11:166Þ the analysis without compromising the filter. The
Q RC node equations are given by
From this, Vi  VA VA
¼ þ ðV A  V o ÞsC þ V A sC
R1 R2
1
fo ¼ ð11:167Þ ð11:170Þ
2πRC
1 Vo
Q¼ ð11:168Þ V A sC ¼  ð11:171Þ
3k R3

k
G¼ ð11:169Þ
3k
Eliminating VA yields
In the design process, a value of C is selected, s
and then using (11.167), R is determined for a Vo CR1
¼ 2 ð11:172Þ
given fo. Following this, either a particular Vi s þ 2s
CR3 þ CR2 R1 þR 2
1 R2 R3

Q using (11.168) or G using (11.169) can be


designed for by selecting k. Note that as
Q increases, the value of (3  k) becomes smaller
indicating that k approaches 3(k < 3) and hence
the circuit gain G in (11.169) changes.

Example 11.25
Design a Sallen-Key second-order band-pass fil-
ter with a centre frequency of 5 kHz and a quality
factor of 5.

Solution We select C ¼ 0.01 μF. Then using


(11.167), R ¼ 1/2π  5  103  0.01  106
¼ 3.2 k. Using (11.168) (11.143), for Q ¼ 5 we Fig. 11.41 Multiple feedback band-pass filter
442 11 Active Filters

Comparing terms in Eqs. (11.172) and Q


R1 ¼
2π f o GC
Vo sωo G
¼ 7=2π  103  10  0:01  106
Q
¼ 2 ð11:173Þ
Vi s þ s ωQo þ ωo 2
¼ 11:1 k;
yields
2Q
R3 ¼
1 1 ωo C
ωo ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi , f o ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi  
C R1 ==R2 :R3 2πC R1 ==R2 :R3 ¼ 2  7= 2  π  103  0:01  106
ð11:174Þ ¼ 222 k;
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 
1 R3 G
Q¼ ð11:175Þ R2 ¼ 1=4π f o CQ 1  2
2 R1 ==R2 2Q
¼ 1=4π  103  0:01  106  7  ð1  0:1Þ
1 R3
G¼ ð11:176Þ ¼ 1:1 k
2 R1

Solving for R1, R2, R3 yields (In the case where the centre frequency gain G is
not specified, then G can be selected to ensure that
Q
R1 ¼ ð11:177Þ the realizability condition is satisfied with equality.
2π f o GC
Then resistor R2 can be removed from the circuit.)
2Q If the realizability condition is not met, then the
R3 ¼ ð11:178Þ
ωo C requirement C1 ¼ C2 would have to be relaxed.

1 G
G2 ¼ ¼ 4π f o CQ 1  2 ð11:179Þ
R2 2Q
11.7.5 Wien Band-Pass Filter
where G2 is conductance. This must be positive
and therefore the following realizability condition A third band-pass filter is shown in Fig. 11.42. It
must be satisfied: utilizes a Wien network which has a band-pass
 response. At fo ¼ 1/2πRC, the output of the net-
G G
1 2 0) 1 ð11:180Þ work is at a maximum of 1/3 of the input voltage.
2Q 2Q2 By placing this network in a positive feedback
loop, the band-pass response is sharpened. The
If the realizability condition is satisfied such
transfer function is given by
that G/2Q2 ¼ 1, then G2 ¼ 0 which means
R2 ! 1(open circuit). In these circumstances,
resistor R2 is removed from the circuit.

Example 11.26
Design a second-order MFB band-pass filter with
f0 ¼ 1 kHz, Q ¼ 7 and G ¼ 10.

Solution Checking the realizability condition,


we have G/2Q2 ¼ 10/2  72 ¼ 0.1 < 1. Hence
the condition is satisfied. Select C1 ¼ C2 ¼ C ¼ 0.01
μF. Then Fig. 11.42 Wien band-pass filter
11.8 Band-Stop Filter 443

Vo α RR21 ωo s 11.8 Band-Stop Filter


¼   ð11:181Þ
V i s2 þ 3  α R 2 ω s þ ω 2
R1 o o
The general transfer function of a second-order
band-stop or notch filter is given by
where
Vo s2 þ ω o 2
1 1 ¼ 2 ð11:185Þ
ωo ¼ , f ¼ ð11:182Þ V i s þ s ωQo þ ωo 2
RC o 2πRC


1
ð11:183Þ where ωo ¼ 2πfo is the notch frequency and Q is
3  α RR23 the quality factor. The normalized response for
this filter is shown in Fig. 11.43 for varying
R2 values of Q.
G¼α Q ð11:184Þ
R1

In designing this filter, C is first chosen and


then R is selected to achieve the desired centre 11.8.1 Twin-T Notch Filter
frequency using (11.182). Following this, R2, R3
and α are determined to give the desired A very popular notch filter is the twin-T circuit
Q using (11.183) and then R1 is adjusted to shown in Fig. 11.44. It utilizes a twin-T network
give the desired centre frequency gain using at the input to create the notch at frequency fo and
(11.184). This procedure is called orthogonal sharpens the notch with positive feedback. The
tuning. feedback here is applied to both legs of the T
network, and the level of this feedback is
Example 11.27
Design a second-order Wien band-pass filter with
f0 ¼ 1 kHz, Q ¼ 7 and G ¼ 10.

Solution Select C ¼ 0.01 μF and then R ¼ 1/


2π  103  0.01  106 ¼ 15.9 k. To get the
specified Q, choosing α ¼ 1 in (11.183) gives
7 ¼ 1R2 . Selecting R3 ¼ 10 k we get R2 ¼
3R
3

8.6 k. Finally, using (11.184), in order to achieve


the desired gain, we get 10 ¼ 8:6 R1  7 which
k

yields R1 ¼ 6 k. Fig. 11.44 Twin-T notch filter

Fig. 11.43 Normalized band-stop filter response for varying values of Q


444 11 Active Filters

adjustable with the potential divider R1, R2. Both


op-amps provide buffering. The transfer function
is given by

Vo s2 þ ω o 2
¼ 2 ð11:186Þ
V i s þ s ωQo þ ωo 2

where
1 1
ωo ¼ , f ¼ ð11:187Þ
RC o 2πRC
1
Q¼ ð11:188Þ
4ð 1  ρÞ

R1 Fig. 11.45 Wien notch filter


ρ¼ ð11:189Þ
R1 þ R2

With this notch filter, Qs up to about 50 are R2 ==R3


k¼ ð11:193Þ
R1
achievable.
where k is the low-frequency gain. Thus, R and
Example 11.29 C are selected to give a particular notch fre-
Using the circuit in Fig. 11.44, design a twin-T quency, and R2 and R3 are selected to give the
notch filter to have a notch frequency of 60 Hz desired Q after which R1 is chosen to achieve a
and a Q of 20. particular gain.

Solution Select C ¼ 1 μF. Then using (11.187), Example 11.30


R ¼ 1/2π  60  106 ¼ 2.7 k. Using (11.188) Using the circuit in Fig. 11.45, design a Wien
20 ¼ 1/4  (1  ρ) from which ρ ¼ 0.9875. From notch filter to have a notch frequency of 60 Hz,
(11.189) with R1 ¼ 10 k, then R2 ¼ 127 Ω. a Q of 20 and a gain of 10.

Solution Select C ¼ 1 μF. Then using (11.191),


11.8.2 Wien Notch Filter R ¼ 1/2π  60  106 ¼ 2.7 k. Using (11.192),
 
20 ¼ 13 1 þ RR23 , and therefore for R3 ¼ 10 k,
The notch filter shown in Fig. 11.45 is easier to
tune than the twin-T filter, and higher Qs can be R2 ¼ 590 k. From (11.193) and noting that R2//
achieved. It utilizes the Wien network in a feed- R3 ¼ 9.8 k, we have 10 ¼ R2 R==R
1
3
¼ 9:8 k
R1 . Hence
back loop to increase Q. The transfer function is R1 ¼ 980 Ω.
given by

Vo s 2 þ ωo 2
¼k 2 ð11:190Þ 11.8.3 All-Pass Filters
Vi s þ s ωQo þ ωo 2

where An all-pass filter is one in which the output signal


amplitude is constant for all frequencies but
1 1 whose phase varies with frequency. Because of
ωo ¼ ,f ¼ ð11:191Þ
RC o 2πRC these properties, all-pass filters are used to correct
 spurious signal phase shifts and to introduce sig-
1 R
Q¼ 1þ 2 ð11:192Þ nal delay. The all-pass filter needs to have a
3 R3
constant group delay over the relevant frequency
11.8 Band-Stop Filter 445

band, in order that transmitted signal suffer mini- Vo


ðDC Þ ¼ 1 ð11:197Þ
mum distortion. The group delay refers to the Vi
time by which signal frequencies within the
The phase response is given by
band are delayed. It is defined as

V
t gr ¼ 

ð11:194Þ ∡ o ¼∡ð1  s=ωo Þ  ∡ð1 þ s=ωo Þ
dω Vi
¼  tan 1 ω=ωo  tan 1 ω=ωo
where ϕ is the phase difference between output
and input. ¼ 2 tan 1 ω=ωo ð11:198Þ

This system produces zero phase at low


11.8.4 First-Order All-Pass Filter frequencies with no inversion and 180 at
Realization high frequencies. The magnitude and phase
responses are shown in Fig. 11.46. At f ¼ fo,
The transfer function of a first-order all-pass filter ∡(Vo/Vi) ¼ 90 . Using (11.194), the group
is given by delay produced by this filter is given by

1  ωso dϕ d  
Vo t gr ¼  ¼ 2 tan 1 ω=ωo
ðsÞ ¼ ð11:195Þ dω dω
Vi 1 þ ωso
1=π f o
¼ ð11:199Þ
The amplitude response is given by 1 þ ð f = f o Þ2
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
   
V o  1  s=ωo  1 þ ðω=ωo Þ2
 ¼  ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼1 This has a value
 V i  1 þ s=ωo 
1 þ ðω=ωo Þ2
t gr ¼ 1=π f o , f  f o ð11:200Þ
ð11:196Þ
which is the maximum group delay occurring at
which is constant for all frequencies. Note that at low frequencies. A plot of tgr against f is shown in
low frequencies, Fig. 11.47.
Magnitude

0.5

0
0

-45
Phase (deg)

-90

-135

-180
0.01fo 0.1fo fo 10fo 100fo

Frequency (Hz)

Fig. 11.46 Magnitude and phase response of first-order all-pass filter


446 11 Active Filters

Fig. 11.49 First-order all-pass filter (lead)

Fig. 11.47 Group delay for first-order all-pass filter (lag)

This system produces zero phase at low


frequencies with no inversion and 180 at
high frequencies. At f ¼ fo ¼ 1/2πRC, ∡(Vo/Vi)
¼ 90 . Using (11.194), the group delay pro-
duced by this filter is given by
dϕ d  
t gr ¼  ¼ 2 tan 1 ωRC
dω dω
RC
¼2 ð11:205Þ
1 þ ω2 R2 C2
Fig. 11.48 First-order all-pass filter (lag)
This has a value tgr ¼ 2RC, ω  1/RC which is
the maximum group delay occurring at low
frequencies.
First-Order Al- Pass Filter Realization
A first-order all-pass filter which produces this
response is shown in Fig. 11.48. It is based on a A first-order all-pass filter which produces a
lead in the phase is shown in Fig. 11.49. In this
single op-amp with negative feedback to the
inverting terminal but where the non-inverting circuit, R and C are interchanged as compared to
terminal is fed through an RC network. For this the first-order all-pass lag circuit. For this circuit
system at the non-inverting terminal, R sCR
VA ¼ ¼ ð11:206Þ
R þ 1=sC 1 þ sCR
1=sC 1
VA ¼ ¼ ð11:201Þ
R þ 1=sC 1 þ sCR Noting that at the two input terminals, VA ¼ VB,
we have
Noting that at the two input terminals, VA ¼ VB,
we have Vi  VB VB  Vo
¼ ð11:207Þ
RA RA
Vi  VB VB  Vo
¼ ð11:202Þ
RA RA Therefore
Therefore V o ð1  sCRÞ
¼ ð11:208Þ
Vi 1 þ sCR
V o 1  sCR
¼ ð11:203Þ
V i 1 þ sCR This is the transfer function of an all-pass lead
This is the transfer function of an all-pass lag network. At low frequencies,
network with Vo
ðDC Þ ¼ 1 ð11:209Þ
Vi
ωo ¼ 1=RC, f o ¼ 1=2πRC ð11:204Þ
The amplitude response is given by
11.9 State Variable Filter 447

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
   
V o  ð1  sCRÞ 1 þ ðωCRÞ2
 ¼  ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ 1
 V i   1 þ sCR 
1 þ ðωCRÞ2
ð11:210Þ

which is constant for all frequencies. The phase


response is given by
 Fig. 11.50 Block diagram of state variable filter system
V
∡ o ¼ ∡ð1Þ þ ∡ð1  sCRÞ  ∡ð1 þ sCRÞ
Vi
 X ðsÞ
¼ 180  tan 1 ωRC  tan 1 ωRC Vo ¼ ð11:214Þ

s2
1
¼ 180  2 tan ωRC ð11:211Þ

This system produces 180 phase shift at low


frequencies which is inversion and zero and
phase shift at high frequencies. At f ¼ fo ¼ 1/
2πRC, ∡(Vo/Vi) ¼ +90 . X ðsÞ X ðsÞ
V i ¼ X ðsÞ þ a1 þ ao 2 ð11:215Þ
s s
Expressing Eqs. (11.214) and (11.215) in the
11.9 State Variable Filter time domain (by taking the inverse Laplace trans-
form) yields
The state variable filter also called the KHN Z Z Z
(Kerwin, Huelsman, Newcomb) filter is another xðtÞ ¼ V i ðt Þ  a1 xðtÞdt  a0 xðt Þdt dt
versatile filter realization. Because of its great flex-
ibility, its good performance and low sensitivity to ð11:216Þ
component variation, it is one of the most widely Z Z
used active filters. The filter name arises from the V o ðt Þ ¼ xðt Þdt dt ð11:217Þ
state variable method used to solve differential
R R R
equations that is used here to develop the filter where the quantities x(t), x(t)dt and ( x(t)dt)dt
topology. In outlining the method, consider the are referred to as state variables and hence the
second-order low-pass transfer function name of the filter. A block diagram of this system
Vo 1 is shown in Fig. 11.50, and an implementation of
¼ ð11:212Þ this using op-amps is shown in Fig. 11.51. Here
V i s 2 þ a1 s þ ao
op-amp 1 provides the summing action, while
In this equation, an arbitrary function X(s)/s2 is op-amp 2 and op-amp3 are used as integrators.
introduced such that (11.212) becomes For this system, the output Vo3 of op-amp
X ðsÞ
3 represents the low-pass output Vo as given in
Vo
¼ s2 (11.212), i.e.Vo3 ¼ Vo ¼ VLP . Also, the output Vo2
V i X ð2sÞ ðs2 þ a1 s þ ao Þ of op-amp 2 is Vo2 ¼  sVo3 ¼  VBP, and the
s
X ðsÞ output Vo1 of op-amp 1 is Vo1 ¼  sVo2 ¼ VHP.
¼ s2
ð11:213Þ The actual transfer functions are given by setting
X ðsÞ
X ðsÞ þ a1 s þ ao Xsð2sÞ R4 ¼ R5 ¼ R, C1 ¼ C2 ¼ C and R2 ¼ R3 ¼ Rx, then

Equating numerator and denominator, we get


448 11 Active Filters

Fig. 11.51 Circuit implementation of state variable filter system

V HP  RR1x s2
¼ R7 ð2R1 þRx Þ
ð11:218Þ
Vi s2 þ CRR s þ C21R2
1 ðR6 þR7 Þ

of general form

V HP ks2
¼ 2 ð11:219Þ
Vi s þ s ωQo þ ωo 2

Rx 1 Fig. 11.52 Block diagram of modified state variable filter


V BP R1 CR s system
¼ R7 ð2R1 þRx Þ
ð11:220Þ
Vi s2 þ CRR s þ C21R2
1 ðR6 þR7 Þ

Rx
of general form G¼ ¼ kQ ð11:227Þ
βð2R1 þ Rx Þ
V BP sωo G Q
¼ 2 ð11:221Þ
Vi s þ s ωQo þ ωo 2 Example 11.31
Using the state variable filter, design a band-pass
V LP  RR21 C21R2 filter with a centre frequency fo ¼ 10 kHz, a
¼ R7 ð2R1 þRx Þ
ð11:222Þ
Vi s2 þ CRR s þ C21R2 quality factor Q ¼ 50 and a centre frequency
1 ðR6 þR7 Þ
gain G ¼ 10.
of general form
Solution Choosing C ¼ 0.001 μF, then R ¼ 1/
V LP kωo 2
¼ 2 ð11:223Þ 2π  10  103  0.001  106 ¼ 15.9 kΩ.
Vi s þ s ωQo þ ωo 2
G ¼ kQ giving 10 ¼ 50 k from which k ¼ 1/5.
This yields Rx ¼ 10 kΩ and R1 ¼ 50 kΩ. Finally,
From these for β ¼ R7/(R6 + R7),
Q ¼ 1/β(2 + k) which becomes 50 ¼ 1/β(2 + 0.2)
f 0 ¼ 1=2πRC ð11:224Þ which yields β ¼ 1/110. For R7 ¼ 1 kΩ then
R6 ¼ 109 kΩ.
k ¼ Rx =R1 ð11:225Þ

R1 1
Q¼ ¼ ð11:226Þ 11.9.1 Modified State Variable Filter
βð2R1 þ Rx Þ βð2 þ kÞ
We wish to exploit the excellent summing
properties of the virtual earth of an op-amp in the
inverting configuration. Thus, in the state variable
11.9 State Variable Filter 449

Fig. 11.53 Circuit


implementation of modified
state variable filter system

Fig. 11.54 Universal filter

filter configuration
R in Fig. 11.50, instead of adding high-pass, band-pass and low-pass outputs.
the term a1 x(t)dt derived from the band-pass Moreover, the availability of quad op-amp
output at theR non-inverting terminal of the op-amp, packages such as the LM148 series or the superior
the term  x(t)dt is inverted (sign changed) using OPA4134 facilitates the implementation of this
R
an inverting amplifier and the term a1 x(t)dt added system.
at the inverting terminal using the normal summing For this system, the transfer functions are as
process via the virtual earth. The resulting modified follows:
block diagram is shown in Fig. 11.52, and the  
modified circuit is shown in Fig. 11.53. VN  R1
VR1 s 2
þ 1
2
VR3 C 2

The inverting amplifier is op-amp 4 with the ¼ 2 ð11:228Þ


Vi s þ VR2RVR2
3C
s þ VR 12 C2
3
output signal being applied to the virtual earth of
the input op-amp 1 via R6. This modified topol- R1 R2 2
V HP VR1 VR2 s
ogy achieves the same results as the original ¼ 2 ð11:229Þ
Vi s þ VR2RVR
2
3C
s þ VR 12 C2
circuit. However, the presence of op-amp 4 allows 3

the input to be moved from op-amp 1 to op-amp


V BP  VRR11 RVR2 2 CVR
1
s
4 such that the notch filter response becomes ¼ 2 3
ð11:230Þ
available at the output of op-amp 4. This is so Vi s þ VR2 VR3 C s þ VR 12 C2
R 2
3

since in general, Vi  VBP ¼ VN with the summing


action resulting in VN. The fully redrawn
circuit is shown in Fig. 11.54 and may be consid-
ered a universal filter as it provides band-stop,
450 11 Active Filters

R1 R2 1
V LP VR1 VR2 C 2 VR3 2
11.10 Applications
¼ 2 ð11:231Þ
Vi s þ VR2RVR
2
3C
s þ VR 12 C2
3
Several filter applications are presented in this
For this system, section.

1 Second-Order Speech Filter


fo ¼ ð11:232Þ
2πVR3 C This system is a straightforward application of a
VR2 wide-band filter comprising a second-order
Q¼ ð11:233Þ low-pass filter with a cut-off frequency of
R2
300 Hz and a second-order high-pass filter with

R1
ð11:234Þ cut-off frequency 3.4 kHz. Both filters are Sallen
VR1 and Key types which employ a simplified
G design procedure such that the cut-off frequency
k¼ ð11:235Þ is given by fc ¼ 1/2πRC and the circuit gain is set
Q
at 1.6 by resistors Ra ¼ 39 k and Rb ¼ 22 k
From these equations it can be seen that fo is (Fig. 11.55).
continuously variable using a variable resistor for
VR3, Q is continuously variable using a variable Ideas for Exploration (i) Compare the perfor-
resistor for VR2, and G is variable using a variable mance of the 741 op-amp with the OPA134
resistor for VR1. op-amp in this application.

All-Pass Notch
Example 11.32 Figure 11.56 shows a notch filter made using
Design a universal filter with a cut-off frequency two first-order all-pass filters. The output of the
 
of 20 kHz, a quality factor of 100 and a centre second all-pass filter is V A ¼ 1Ts1þTs 1þTs V i ¼
: 1Ts

frequency gain of 10.

Solution Using Eq. (11.232) and choosing


C ¼ 0.001 μF, we get VR3 ¼ 7957 Ω. For
Q ¼ 100 from Eq. (11.233), we get VR2 ¼ 100 k
and R2 ¼ 1 k. Finally for G ¼ 10 from
Eq. (11.234), we have VR1 ¼ 10 k and
R1 ¼ 100 k.

Fig. 11.56 All-pass notch filter

Fig. 11.55 Second-order speech filter


11.10 Applications 451

 
12TsþT 2 s2
V i where T ¼ RC. This output is Rx1 ¼ Rx2 ¼ 10 k. In the circuit the potentiome-
1þ2TsþT 2 s2
ter VR ¼ 1 k has been added to enable an
applied to one end of the potential divider adjustment for the deepest possible notch.
comprising two series connected equal value Finally, for a notch frequency fo ¼ 1 kHz,
resistors Rx1 ¼ Rx2 (Assume initially VR ¼ 0). choose C ¼ 0.01 μF and therefore R ¼ 1/2π 
The input signal is applied at the other end 103  0.01  106 ¼ 15.9 k.
of these two resistors and op-amp 3 connected
as a buffer takes its input from the junction Ideas for Exploration (i) Replace fixed resistors
of these two resistors. The voltage across R by ganged variable resistors in series with fixed
these resistors is therefore V A  V i ¼ resistors such that the notch frequency can be
    continuously varied.
12TsþT 2 s2
2 2  1 Vi ¼ 
4Ts
2 2 V i . Half of
1þ2TsþT s 1þ2TsþT s
this is dropped across each resistor Rx1 and is Variable Wide-Band Filter
 
given by V Rx1 ¼  1þ2TsþT
2Ts
2 2 Vi
s
. Therefore, The circuit in Fig. 11.57 is a variable wide-band
filter whose cut-off frequencies are made variable
the voltage Vo at the output Vo of op-amp 3 is
 2 2
 by including variable resistors in series with fixed
given by V o ¼ V i þ V Rx1 ¼ 1þ2Ts2TsþT
2 2
1þ2TsþT s
s
Vi ¼ resistors. Each filter uses equal value resistors and
 capacitors and unity gain for simplicity. In order
1 þ T 2 s2
V i:
1 þ 2Ts þ T s 2 2 to realize a Butterworth response, the gain of each
op-amp must be set to 1.6. Such a response is
This goes to zero at fo ¼ 1/2πRC and is especially useful in audio systems as a scratch and
non-zero at all other frequencies. This represents rumble filter to remove unwanted high and low
a notch filter with notch frequency fo ¼ 1/2πRC frequencies. The Butterworth response ensures
and Q ¼ 1/2. We can choose RA ¼ 10 k and minimum amplitude distortion in the pass-band.

Fig. 11.57 Variable wide-band filter

Fig. 11.58 Universal filter system


452 11 Active Filters

The dual-ganged potentiometers are 100 k linear 1 to 10. The circuit is assembled around a quad
type. op-amp such as the OPA2134.

Ideas for Exploration (i) Adjust the gains to 1.6 Ideas for Exploration (i) Use a bank of
for Butterworth response. (ii) Compare the per- switched capacitors to increase the frequency
formance of the 741 op-amp with the OPA134 range.
op-amp in this application.
Research Project 1
Universal Filter This research project involves the analysis and
The circuit in Fig. 11.58 is a universal filter based optimization of the variable frequency band-
on the modified sate variable filter. With the values pass filter in Fig. 11.59. It uses cascaded
selected (C ¼ 0.01 μF and the ganged potentiom- all-pass filters in a feedback loop such that at the

eter VR3 ¼ 100 k), the resonant frequency ( fo ¼ 1/ frequency where the phase shift is 180 , the feed-
2πCVR3) is variable from about 160 Hz to about back around the input op-amp goes to zero such
16 kHz, the quality factor (Q ¼ (VR2 + 1 k)/R2) is that the gain is at its maximum. At all other
variable from 0.1 to 100, and the centre frequency frequencies, the gain is reduced by the feedback.
gain (G ¼ R1/(VR1 + 1 k)) is variable from about The loop gain is kept less than unity to ensure
there is no oscillation. Potentiometer VR1 ¼ 20 k

Fig. 11.59 Variable frequency band-pass filter

Fig. 11.60 Active cross-over system


Problems 453

adjusts the quality factor, while the ganged


potentiometers VR2 vary the centre frequency of
the band-pass filter.

Ideas for Exploration (i) Introduce a bank of


switched capacitors for C in order to realize a
wide frequency range.

Research Project 2
Fig. 11.61 Circuit for Question 7
This project involves research into and the design
of an active cross-over system used in audio
speaker systems. A diagrammatic representation
of such a system is shown in Fig. 11.60. This
system separates the high frequencies and the
low frequencies of a line level audio signal com-
ing from a preamplifier using a high-pass filter
and a low-pass filter, respectively. The output of
the high-pass filter drives a power amplifier which
delivers power to the tweeter in the speaker enclo- Fig. 11.62 Circuit for Question 12
sure, while the output of the low-pass filter drives
a separate power amplifier which drives the 4. Show that in a first-order low-pass filter, the
woofer in the speaker system. Such an approach amplitude response is down 3 dB at the
allows the signal separation to take place at low cut-off frequency and determine the phase
signal levels, thereby avoiding the use of the angle of the output relative to the input at
expensive components found in passive cross- that frequency.
over filters used in speaker systems. 5. Using the Sallen-Key configuration, design a
second-order low-pass Butterworth filter
Ideas for Exploration (i) Examine the introduc- with unity gain and a cut-off frequency of
tion of an active band-pass filter that separates the 50 kHz. Ensure that offset voltage is
mid-band frequencies and used to drive a third minimized.
power amplifier that delivers power to a 6. Show that in the Sallen-Key second-order
mid-range speaker. low-pass Butterworth filter, for the case
where the network resistors are equal and
the network capacitors are equal, the gain
Problems must be 1.6. Hence repeat the design of
Question 5 using this modified approach.
7. Show that in the Sallen-Key second-order
1. Design a first-order low-pass non-inverting unity-gain low-pass filter configuration
filter with unity gain and a break frequency shown in Fig. 11.61, if R1 ¼ R2 and
of 5 kHz. C2 ¼ 2C1, then the transfer function reduces
2. Design a first-order non-inverting low-pass to a second-order Butterworth response.
filter with gain of 8 and a cut-off frequency of 8. Design a VCVS second-order unity-gain
17 kHz. low-pass Chebyshev filter with a cut-off fre-
3. Design a first-order low-pass filter with gain quency of 22 kHz and a ripple width
of 12 and a cut-off frequency of 25 kHz using RWdB ¼ 1dB.
the inverting configuration.
454 11 Active Filters

9. Design a VCVS second-order unity-gain 19. Using the Sallen-Key configuration, design a
low-pass Bessel filter with a cut-off fre- second-order high-pass Butterworth filter
quency of 550Hz. with unity gain and a cut-off frequency of
10. Design a MFB second-order low-pass 5 kHz.
Butterworth filter with a gain of 8 and a 20. Show that in the Sallen-Key second-order
cut-off frequency of 3 kHz. high-pass Butterworth filter, for the case
11. Design a VCVS third-order unity-gain where the network resistors are equal and
low-pass Butterworth filter with a cut-off fre- the network capacitors are equal, the gain
quency of 33 kHz. What is the ultimate roll- must be 1.6. Hence repeat the design of
off rate of such a filter? Question 19 using this modified approach.
12. Show that in the Sallen-Key third-order 21. Show that in the Sallen-Key second-order
unity-gain low-pass filter configuration unity-gain high-pass filter configuration
shown in Fig. 11.62, if R1 ¼ R2 ¼ R3 and shown in Fig. 11.63, if R1 ¼ 2R2 and
C2 ¼ 2C3 and C1 ¼ C3/2, then the transfer C1 ¼ C2, then the transfer function reduces
function reduces to a third-order Butterworth to a second-order high-pass Butterworth
response. Hence repeat Question 11 using response.
this simplified design procedure. 22. Design a VCVS second-order unity-gain
13. Design a VCVS fourth-order unity-gain high-pass Chebyshev filter with a cut-off fre-
low-pass Chebyshev filter with a cut-off quency of 14 kHz and a ripple width
frequency of 12 kHz and 3 dB ripple width. RWdB ¼ 1dB.
14. Design a VCVS fifth-order unity-gain 23. Design a VCVS second-order unity-gain
low-pass Butterworth filter with a cut-off fre- high-pass Bessel filter with a cut-off fre-
quency of 1200 Hz. quency of 5 kHz.
15. Design a first-order high-pass non-inverting 24. Design a MFB second-order high-pass
filter with unity gain and a break frequency of Butterworth filter with a gain of 6 and a
5 kHz. cut-off frequency of 2 kHz.
16. Design a first-order non-inverting high-pass
filter with gain of 8 and a cut-off frequency of
17 kHz.
17. Design a first-order high-pass filter with gain
of 15 and a cut-off frequency of 10 kHz using
the inverting configuration.
18. Show that in a first-order high-pass filter, the
amplitude response is down 3 dB at the
cut-off frequency and determine the phase
angle of the output relative to the input at
that frequency. Fig. 11.64 Circuit for Question 26

Fig. 11.63 Circuit for Question 21 Fig. 11.65 Circuit for Question 29
Bibliography 455

Fig. 11.66 MFB band-pass filter for Question 31

25. Design a VCVS third-order unity-gain high-


pass Butterworth filter with a cut-off fre-
quency of 900 Hz. What is the ultimate roll-
off rate of such a filter? Fig. 11.67 Circuit for Question 35
26. Show that in the Sallen-Key third-order
unity-gain low-pass filter configuration C1 ¼ C2. Hence, design a second-order
shown in Fig. 11.64, if C1 ¼ C2 ¼ C3 and MFB band-pass filter having fo ¼ 1 kHz,
R1 ¼ 2R3 and R2 ¼ R3/2, then the transfer Q ¼ 5 and G ¼ 10.
function reduces to a third-order Butterworth 33. Design a second-order Wien band-pass filter
response. Hence repeat Question 25 using with f0 ¼ 60 Hz, Q ¼ 10 and G ¼ 2.
this simplified design procedure. 34. Design a twin-T notch filter having a notch
27. Design a VCVS fourth-order unity-gain high- frequency of 2.5 kHz and Q ¼ 25.
pass Chebyshev filter with a cut-off fre- 35. Using the circuit shown in Fig. 11.67, design
quency of 1200 Hz and 2 dB ripple width. a Wien notch filter to have a notch frequency
28. Design a VCVS fifth-order unity-gain high- of 200 Hz, a Q of 25 and a gain of 2.
pass Bessel filter with a cut-off frequency of 36. By using a ganged potentiometer for R and
6 kHz. two banks of switched capacitors for C,
29. Using the VCVS band-pass filter shown in convert the Wien notch filter in Fig. 11.67
Fig. 11.65, design a Sallen-Key second-order to a variable frequency circuit and specify the
band-pass filter having C1 ¼ C2, R1 ¼ R2 frequency range.
with a centre frequency of 13 kHz and a 37. Develop the transfer function for a
quality factor of 8. Tow-Thomas biquadratic filter.
30. For the VCVS band-pass filter shown in
Fig. 11.65, develop design equations for R1,
R2 and R3 if C1 ¼ C2 ¼ C and 1 + Rb/Ra ¼ 2.
Hence, design a VCVS band-pass filter hav- Bibliography
ing fo ¼ 20 kHz, Q ¼ 6 and G ¼ 10.
31. Design a second-order MFB band-pass filter R. Schaumann, M. Van Valkenburg, Design of Analog
with f0 ¼ 2 kHz, Q ¼ 9 and G ¼ 5. Use the Filters, 1st edn. (Oxford University Press, 2001)
circuit shown in Fig. 11.66. T. Kugelstadt, in Active filter design techniques in op amps
for everyone, ed. by R. Mancini, (Texas Instruments,
32. For the second-order MFB band-pass filter 2002)
shown in Fig. 11.66, develop design A. Waters, Active filter design (Mc Graw Hill, 1991)
equations without imposing the condition

You might also like