Allen)  Chapter 3 Page 31
3  HIGHERORDER LADDER PASSIVE AND ACTIVE FILTERS
In Fig. 21, two approaches to designing active filters were shown. One of the
approaches uses cascaded, first and secondorder active filters and was discussed in Sec.
2. The other approach uses a ladder configuration and is presented in this section. Both
approaches are used to design successful higherorder active filters. The cascade approach
is easier to design and tune but does not have the insensitivity to component variation found
in the ladder approach.
The concepts of normalization and frequency transformation developed for cascade
active filters will also be used in the design of ladder active filters. The starting point for a
ladder design is an RLC passive circuit rather than a transfer function or the roots of a
transfer function. This means that our design manipulations will be with circuit elements
rather than with roots. These two approaches are among the more widely used approaches
for designing active filters.
Sensitivity of an Active Filter
Sensitivity is a measure of the dependence of the filter performance upon the passive
and active components of the filter design. Sensitivity is generally expressed as percentage
change of some performance aspect to the percentage change in an active or passive
component. For example, we can express the relative sensitivity of a performance aspect
designated as "p" to the variation of some component "x" as
S
p
x
=
∂p
p
∂x
x
=
x
p
∂p
∂x
. (31)
For example, the relative sensitivity of the poleQ of the TowThomas, lowpass realization
of Fig. 116 to R
1
can be found by letting p = Q and x = R
1
and applying Eq. (31) to Eq.
(159) to get
S
Q
R
1
=
R
1
R
1
C
1
R
2
R
3
C
2
.

,
`
C
1
R
2
R
3
C
2
= 1 . (32)
ECE 6414  Continuous Time Filters (P. Allen)  Chapter 3 Page 32
Furthermore, we can show that the sensitivity of the poleQ of Fig. 116 to R
2
is 0.5.
Another interpretation of sensitivity is as follows. If the component x changes by dx/x,
then the performance changes by the product of S
p
x
times dx/x. For example, the poleQ
will change by +10% if R
1
changes by +10%. If the relationship between p and x is
nonlinear, the sensitivity measure is only good for small changes. Consequently, changes
much over 10% should be calculated using nonlinear sensitivity methods or point
sensitivity calculations.
Sensitivity calculations can become complex if the expression relating the
performance to the component is complicated. Fortunately we only wish to use the concept
of sensitivity to compare the two different approaches to higherorder active filter design
shown in Fig. 21. Although we have not yet developed the ladder filter, it can be
distinguished from the cascaded filter by its dependence upon the passive and active
components of the filter realization. The performance of the ladder filter always depends
on more than one passive or active component. As a result, the performance is not strongly
dependent upon any one component but rather on all components. However, the
performance of the cascaded filter may depend upon only one component. If this
component is inaccurate or off in value, the cascade filter performance can be strongly
influenced while the same component change may not be noticed in the performance of the
ladder filter. This advantage of the ladder filter becomes a disadvantage when tuning the
filter. It is almost impossible to tune a ladder filter because no single component dominates
the performance of the filter.
Normalized, RLC, LowPass Ladder Filters
RLC, lowpass ladder filters are the result of modern network synthesis techniques
and are based on techniques well known in circuit theory
†
. The resulting realizations of
these synthesis techniques always start with a load resistor of 1 ohm and works toward the
†
M.E. Van Valkenburg, Introduction to Modern Network Synthesis, Chapter 10, John Wiley & Sons,
Inc., New York, 1960.
ECE 6414  Continuous Time Filters (P. Allen)  Chapter 3 Page 33
input of the filter. Fig. 31 shows the form for a singlyterminated RLC filter for the case
of even and odd order functions with the numbering of components going from the output
to the input of the filter.
1Ω
+

+

L
2n L
N,n
C
N1,n
C
3n
C
1n V
in
(s
n
)
(a.)
(b.)
V
out
(s
n
)
+

+

C
2n C
N1,n
L
N,n
L
3n L
1n
V
in
(s
n
)
V
out
(s
n
)
1Ω
Figure 31  Singlyterminated, RLC filters. (a.) N even. (b.) N odd.
The RLC ladder filters of Fig. 31 are normalized to a passband of 1 rps. The
denormalizations of Table 21 are applicable to the elements of Fig. 31. Fig. 32 shows
the normalized ladder filters for doublyterminated, RLC filters. It is seen that these filters
are similar to those of Fig. 31 except for a series source resistance.
(a.)
(b.)
+

+

L
2n L
N,n
C
N1,n
C
3n
C
1n V
in
(s
n
)
V
out
(s
n
)
R
1Ω
+

C
2n C
N1,n
L
N,n
L
3n L
1n
V
out
(s
n
)
+

V
in
(s
n
)
R
1Ω
Figure 32  Doublyterminated, RLC filters. (a.) N even. (b.) N odd.
The tabular information for the design of RLC filters consists of the normalized
component values of Figs. 31 and 32. Each of the many different types of filter
ECE 6414  Continuous Time Filters (P. Allen)  Chapter 3 Page 34
approximations have been tabulated for values of N up to 10 or more
†
. Tables 31 and 32
are typical of this tabularized information for the Butterworth and 1dB Chebyshev
approximation for the singlyterminated and doublyterminated, RLC filters of Figs. 31
and 32.
Use these component designations for even order of Fig. 31a and Fig. 32a.
N C
1n
L
2n
C
3n
L
4n
C
5n
L
6n
C
7n
L
8n
C
9n
L
10n
2 0.7071 1.4142
3 0.5000 1.3333 1.5000 Butterworth (1 rps passband)
4 0.3827 1.0824 1.5772 1.5307
5 0.3090 0.8944 1.3820 1.6944 1.5451
6 0.2588 0.7579 1.2016 1.5529 1.7593 1.5529
7 0.2225 0.6560 1.0550 1.3972 1.6588 1.7988 1.5576
8 0.1951 0.5576 0.9370 1.2588 1.5283 1.7287 1.8246 1.5607
9 0.1736 0.5155 0.8414 1.1408 1.4037 1.6202 1.7772 1.8424 1.5628
10 0.1564 0.4654 0.7626 1.0406 1.2921 1.5100 1.6869 1.8121 1.8552 1.5643
2 0.9110 0.9957
3 1.0118 1.3332 1.5088 1dB ripple Chebyshev (1 rps passband)
4 1.0495 1.4126 1.9093 1.2817
5 1.0674 1.4441 1.9938 1.5908 1.6652
6 1.0773 1.4601 2.0270 1.6507 2.0491 1.3457
7 1.0832 1.4694 2.0437 1.6736 2.1192 1.6489 1.7118
8 1.0872 1.4751 2.0537 1.6850 2.1453 1.7021 2.0922 1.3691
9 1.0899 1.4790 2.0601 1.6918 2.1583 1.7213 2.1574 1.6707 1.7317
10 1.0918 1.4817 2.0645 1.6961 2.1658 1.7306 2.1803 1.7215 2.1111 1.3801
L
1n
C
2n
L
3n
C
4n
L
5n
C
6n
L
7n
C
8n
L
9n
C
10n
Use these component designations for odd order of Fig. 31b.
Table 31  Normalized component values for Fig. 31 for the Butterworth and Chebyshev
singlyterminated, RLC filter approximations.
Example 31  Use of the Table 31 to Find a SinglyTerminated, RLC Lowpass Filter
Find a singlyterminated, normalized, RLC filter for a 4thorder Butterworth filter.
Solution
Using Table 31 and using the component designations at the top of the table gives
Fig. 33.
†
L. Weinburg, Network Analysis and Synthesis, McGrawHill Book Co., 1962, R.E. Krieger Publishing
Co., Huntington, NY, 1975.
A.I. Zverev, Handbook of Filter Synthesis, John Wiley & Sons, Inc., NY, 1967.
ECE 6414  Continuous Time Filters (P. Allen)  Chapter 3 Page 35
V
in
(s
n
)
+

+

V
out
(s
n
) C
1n
=
0.3827 F
C
3n
=
1.5772 F
L
2n
=1.0824 H L
4n
=1.5307 H
1Ω
Figure 33  Fourthorder, Butterworth, normalized lowpass RLC filter realization.
Use these component designations for even order of Fig. 32a, R = 1Ω.
N C
1n
L
2n
C
3n
L
4n
C
5n
L
6n
C
7n
L
8n
C
9n
L
10n
2 1.4142 1.4142
3 1.0000 2.0000 1.0000 Butterworth (1 rps passband)
4 0.7654 1.8478 1.8478 0.7654
5 0.6180 1.6180 2.0000 1.6180 0.6180
6 0.5176 1.4142 1.9319 1.9319 1.4142 0.5176
7 0.4450 1.2470 1.8019 2.0000 1.8019 1.2740 0.4450
8 0.3902 1.1111 1.6629 1.9616 1.9616 1.6629 1.1111 0.3902
9 0.3473 1.0000 1.5321 1.8794 2.0000 1.8794 1.5321 1.0000 0.3473
10 0.3129 0.9080 1.4142 1.7820 1.9754 1.9754 1.7820 1.4142 0.9080 0.3129
3 2.0236 0.9941 2.0236 1dB ripple Chebyshev (1 rps passband)
5 2.1349 1.0911 3.0009 1.0911 2.1349
7 2.1666 1.1115 3.0936 1.1735 3.0936 1.1115 2.1666
9 2.1797 1.1192 3.1214 1.1897 3.1746 1.1897 3.1214 1.1192 2.1797
L
1n
C
2n
L
3n
C
4n
L
5n
C
6n
L
7n
C
8n
L
9n
C
10n
Use these component designations for odd order of Fig. 32b, R = 1Ω.
Table 32  Normalized component values for Fig. 32 for the Butterworth and 1dB
Chebyshev doublyterminated, RLC approximations.
We note that no solution exists for the evenorder cases of the doublyterminated,
RLC Chebyshev approximations for R = 1 Ω. This is a special result for R = 1 Ω and is
not true for other values of R. We also can see that the gain in the passband will be no
more than 6 dB because of the equal source and load resistances causing an gain of 0.5 at
low frequencies where the inductors are shortcircuits and the capacitors are opencircuits.
Example 32  Use of Table 32 to Find a DoublyTerminated, RLC Lowpass Fi lter
Find a doublyterminated, RLC filter for a fifthorder Chebyshev filter approximation
having 1 dB ripple in the passband and a source resistance of 1 Ω.
Solution
ECE 6414  Continuous Time Filters (P. Allen)  Chapter 3 Page 36
Using Table 32 and using the component designations at the bottom of the table
gives Fig. 34.
V
out
(s
n
)
+

+

V
in
(s
n
)
1 Ω
C
4n
=
1.0911 F
C
2n
=
1.0911 F
L
1n
=2.1349 H L
5n
=2.1349 H L
3n
=3.0009 H
1Ω
Figure 34  Fifthorder, doublyterminated, normalized, Chebyshev lowpass RLC filter
realization.
LowPass, RLC Filter Design
In some cases, such as high frequency, the designer may choose a passive realization
over an active realization. In this case, the filter has no active elements such as op amps.
The design procedure is simply to find the normalized, lowpass, filter approximation
which satisfies the filter specifications and then denormalize the frequency and the
impedance. An example will illustrate the procedure.
Example 33  Design of a Passive, LowPass Filter
A lowpass filter using the Butterworth approximation is to have the following
specifications:
T
PB
= 3 dB, T
SB
= 40 dB, f
PB
= 500 kHz, and f
SB
= 1 MHz.
Design this filter using a passive RLC realization and denormalize the impedances by 10
3
.
Solution
First we see that Ω
n
= f
SB
/f
PB
= 2 and then try various values of N in Eq. (211)
using ε = 1 to get T
SB
= 30.1 dB for N = 5, T
SB
= 36.1 dB for N = 6, and T
SB
= 42.14
dB for N = 7. From Table 31 for N = 7 for the Butterworth approximation we get L
1n
=
0.2225 H, C
2n
= 0.6560 F, L
3n
= 1.0550 H, C
4n
= 1.3972 F, L
5n
= 1.6588 H, C
6n
=
1.7988 F, and L
7n
= 1.5576 H. Next, we use the denormalizations in the bottom row of
Table 21 to get L
1
= (10
3
)(0.2225H)/(πx10
6
) = 0.071 mH, C
2
= (0.6560F)/(πx10
9
) =
209 pF, L
3
= (10
3
)(1.0550H)/(πx10
6
) = 0.336 mH, C
4
= (1.3972F)/(πx10
9
) = 445 pF,
ECE 6414  Continuous Time Filters (P. Allen)  Chapter 3 Page 37
L
5
= (10
3
)(1.6588H)/(πx10
6
) = 0.528 mH, C
6
= (1.7988F)/(πx10
9
) = 573 pF, and L
7
=
(10
3
)(1.5576H)/(πx10
6
) = 0.496 mH. The load resistor becomes 1 kΩ. The final
realization is shown in Fig. 35.
1 kΩ
+

+

L
1
=0.071 mH L
3
=0.336 mH L
5
=0.528 mH L
7
=0.496 mH
C
2
=
209 pF
C
4
=
445 pF
C
4
=
573 pF
V
in
(s) V
out
(s)
Figure 35  Seventhorder, Butterworth filter realization of Ex. 33.
We can see by the previous example, that the design of RLC filters is very simple.
However, the actual implementation of the filter requires careful consideration and effort.
In order to get the desired results, we must have the exact values of components that were
calculated in Ex. 33. The cost of such accurate components is generally prohibitive.
Therefore, one adjusts each component before inserting it into the filter by the following
means. For each capacitor, a small variable capacitor (i.e. 50 pF) can be placed in parallel
with a larger fixed capacitor whose value is slightly less than the desired value. This
arrangement can be taken to a capacitor bridge or capacitance meter and the variable
capacitor trimmed until the desired value is achieved.
The inductors are typically made by winding lowresistance wire around a cylindrical
insulator. Inside the cylindrical insulator is a magnetic cylinder whose position is adjusted
by a screw which causes the inductance of the inductor to vary. The inductors are custom
made by winding the necessary number of turns to achieve an inductance less than desired
with the magnetic core removed. Then the magnetic core is inserted and adjusted to achieve
the desired inductance.
The above procedures work well if there are no parasitic elements that will effect the
filter performance. The important parasitics are the capacitors from each node to ground
and the resistance of the inductors. For precise filter applications, these influences must be
ECE 6414  Continuous Time Filters (P. Allen)  Chapter 3 Page 38
incorporated into the design phase. This results in a more complex filter design procedure
which is beyond the scope of our treatment.
HighPass, RLC, Ladder Filter Design
The design of highpass and bandpass, RLC ladder filters is made simple by the
frequency transformations of the previous section. The frequency transformation from the
normalized, lowpass to normalized highpass was given by Eq. (223). If we apply this
transformation to an inductor of a normalized, lowpass realization, we obtain
s
ln
L
ln
=
.

,
`
1
s
hn
L
ln
=
1
s
hn
C
hn
. (33)
Similarly, if applying the transformation to a capacitor, C
ln
, of a normalized, lowpass
realization, we obtain
1
s
ln
C
ln
=
.

,
`
s
hn
1
1
C
ln
= s
hn
L
hn
. (34)
From Eqs. (33) and (34), we see that the normalized, lowpass to normalized, highpass
frequency transformation takes an inductor, L
ln
, and replaces it by a capacitor, C
hn
, whose
value is 1/L
ln
. This transformation also takes a capacitor, C
ln
, and replaces it by an
inductor, L
hn
, whose value is 1/C
ln
. Fig. 36 illustrates these important relationships.
s
ln
→
1
s
hn
Normalized Low
Pass Network
Normalized High
Pass Network
L
ln
C
ln
L
hn
=
1
C
ln
C
hn
=
1
L
ln
Figure 36  Influence of the normalized, lowpass to normalized, highpass frequency
transformation on the inductors and capacitors.
From the above results, we see that to achieve a normalized, highpass RLC filter, we
replace each inductor, L
ln
, with a capacitor, C
hn
, whose value is 1/L
ln
and each capacitor,
C
ln
, with an inductor, L
hn
, whose value is 1/C
ln
. Once, the elements have been
ECE 6414  Continuous Time Filters (P. Allen)  Chapter 3 Page 39
transformed, then both a frequency and impedance denormalization can be performed to
result in the final filter design.
Example 34  Design of a HighPass, RLC Filter
A filter is to be designed to provide a passband with a ripple of no more than 3 dB
above 100 kHz and a stopband attenuation of at least 45 dB below 50 kHz. The output of
this filter will be connected to a 50ohm cable and will serve as the load resistor. Select a
suitable RLC filter approximation and give the values of all components.
Solution
We know from Eq. (226) that Ω
n
= 2. Using Eq. (211) gives an N = 8 (T
SB
= 
48.2 dB) for a Butterworth approximation. Eq. (219) gives N = 5 (T
SB
= 45.3 dB) for a
Chebyshev approximation. Let us choose the Chebyshev approximation since there are
almost half the components. We must choose the singlyterminated filter because the
doublyterminated filter has a maximum passband gain of 6 dB. Therefore, from Table 3
1 we have L
1ln
= 1.0674 H, C
2ln
= 1.4441 F, L
3ln
= 1.9938 H, C
4ln
= 1.5908 F, and
L
5ln
= 1.6652 H.
Next, we apply the normalized, lowpass to normalized, highpass frequency
transformation to each element to get C
1hn
= 1/L
1ln
= 1/1.0674 = 0.9369 F, L
2hn
= 1/C
2ln
= 1/1.4441 = 0.6925 H, C
3hn
= 1/L
3ln
= 1/1.9938 = 0.5016 F, L
4hn
= 1/C
4ln
= 1/1.5908
= 0.6286 H, and C
5hn
= 1/L
5ln
= 1/1.6652 = 0.6005 F. Finally, we denormalize the
frequency by 2πx100 kHz = 2πx10
5
and the impedance by 50 in order to represent the
connection of the 50ohm cable to the output of the filter. The actual values for the filter are
C
1h
= 0.9369/(50x2πx10
5
) = 59.6 nF, L
2h
= 0.6925x(50)/(2πx10
5
) = 110 µH, C
3h
=
0.5016/(50x2πx10
5
) = 31.9 nF, L
4h
= 0.6286x(50)/(2πx10
5
) = 100 µH, and C
5h
=
0.6005/(50x2πx10
5
) = 38.3 nF. Fig. 37 shows the final realization for this example.
ECE 6414  Continuous Time Filters (P. Allen)  Chapter 3 Page 310
+

+

V
in
(s)
V
out
(s)
C
1h
= 59.6 nF C
3h
= 31.9 nF C
5h
= 38.3 nF
L
2h
=
110 µH
L
4h
=
100 µH
50 ohm
cable
Figure 37  Denormalized, Chebyshev, highpass filter realization for Ex. 34.
Bandpass, RLC, Ladder Filter Design
The design of RLC bandpass ladder filter follows exactly the same approach as used
in the previous section except the transformations are applied to the passive components of
the normalized lowpass filter rather than the roots. The first step in designing a bandpass
filter is to apply the bandpass normalization of Eq. (232) to the components of the low
pass filter. Next, we apply the normalized, lowpass to normalized, bandpass
transformation of Eq. (233). Both of these steps are incorporated in Eq. (231). Lastly,
we denormalize the normalized, bandpass filter by ω
r
and by the desired impedance
denormalization.
Let us first consider the inductor, L
ln
, of a normalized, lowpass filter. Let us
simultaneously apply the bandpass normalization and the frequency transformation by
using Eq. (231). The normalized, inductance L
ln
can be expressed as
s
ln
L
ln
=
]
]
]
.

,
` ω
r
BW
.

,
`
(s
bn
+
1
s
bn
) L
ln
= s
bn.

,
` ω
r
L
ln
BW
+
1
s
bn
.

,
` ω
r
L
ln
BW
= s
bn
L
bn
+
1
s
bn
C
bn
. (35)
Thus we see that the bandpass normalization and frequency transformation takes an
inductance, L
ln
, and replaces it by an inductor, L
bn
, in series with a capacitor, C
bn
, whose
values are given as
L
bn
=
.

,
` ω
r
BW
L
ln
=
L
ln
Ω
b
(36)
and
ECE 6414  Continuous Time Filters (P. Allen)  Chapter 3 Page 311
C
bn
=
.

,
`
BW
ω
r
1
L
ln
=
Ω
b
L
ln
. (37)
Now we apply Eq. (231) to a normalized capacitance, C
ln
, to get
1
s
ln
C
ln
=
1
]
]
]
.

,
` ω
r
BW
.

,
`
(s
bn
+
1
s
bn
) C
l n
=
1
s
bn.

,
` ω
r
BW
C
l n
+
1
s
bn
.

,
` ω
r
C
ln
BW
=
1
s
bn
C
bn
+
1
s
bn
L
bn
. (38)
From Eq. (38), we see that the bandpass normalization and frequency transformation takes
a capacitor, C
ln
, in a lowpass circuit and transfors to a capacitor, C
bn
, in parallel with an
inductor, L
bn
, whose values are given as
C
bn
=
.

,
` ω
r
BW
C
ln
=
C
ln
Ω
b
(39)
and
L
bn
=
.

,
`
BW
ω
r
1
C
ln
=
Ω
b
C
ln
. (310)
Eqs. (36), (37), (39), and (310) are very important in the design of RLC bandpass
filters and are illustrated in Fig. 38.
Normalized
LowPass
Network
L
ln
C
ln
s
n
→
ω
r
BW
s
bn
+
1
s
bn
Normalized Bandpass Network
L
bn
=
ω
r
BW
L
ln
C
bn
=
BW
ω
r
1
L
ln
L
bn
=
BW
ω
r
1
C
ln
C
bn
=
ω
r
BW
C
ln
Figure 38  Illustration of the influence of the normalized, lowpass to the normalized,
bandpass transformation of Eq. (231) on an inductor and capacitor of a lowpass filter.
ECE 6414  Continuous Time Filters (P. Allen)  Chapter 3 Page 312
Once the normalized, lowpass realization has been transformed to a normalized,
bandpass realization, then all that is left is to denormalize the passive elements. An
example will serve to illustrate the design approach.
Example 35  Design of an RLC Bandpass Filter
A 1dB ripple, Chebyshev approximation is to be used to design a bandpass filter
with a passband of 10 kHz and a stopband of 50 kHz geometrically centered at 100 kHz.
The attenuation in the stopband should be at least 45 dB. Assume that the load resistor is
300 ohms. Give a final filter schematic with the actual component values.
Solution
From the specification and Eq. (241), we know that Ω
n
= 5. Substituting N = 3 into
Eq. (219) gives an attenuation of 47.85 dB. The values of the thirdorder, normalized
Chebyshev approximation are L
1ln
= 1.0118 H, C
2ln
= 1.3332 F, and L
3ln
= 1.5088 H.
The value of Ω
b
, defined in Eq. (235), is Ω
b
= 10/100 = 0.1. Applying the
normalized, lowpass to normalized, bandpass transformation gives the following results.
For L
1ln
we get L
1bn
= 1.0118/0.1 = 11.018 H and C
1bn
= 0.1/0.11018 = 0.090761 F.
For C
2ln
we get C
2bn
= 1.3332/0.1 = 13.332 F and L
2bn
= 0.1/0.13332 = 0.075008 H.
Finally, for L
3ln
we get L
3bn
= 1.5088/0.1 = 15.088 H and C
3bn
= 0.1/1.5088 = 0.066278
F.
Finally, we frequency and impedance denomalize each element by 2πx10
5
and 300,
respectively. The resulting bandpass filter is shown in Fig. 39.
300 Ω
+

+

V
in
(s)
V
out
(s)
C
3b
=3.516nF C
1b
=4.815nF
C
2b
=
70.73nF
L
2b
=
358µH
L
3b
=
7.204mH
L
1b
=
5.261mH
Figure 39  Chebyshev, bandpass realization for Ex. 35.
Design of Other Types of RLC Filters
ECE 6414  Continuous Time Filters (P. Allen)  Chapter 3 Page 313
There are several types of RLC filters that we will not cover in this section but we
should briefly mention them. One type are filters that have jω axis zeros such as the one
shown in Fig. 218. Normalized, lowpass RLC filters have been developed and tabulated
for this type of filter. They are similar in form to the ladder filters of Figs. 31 and 32
except that the series inductor is replaced by a parallel inductor and capacitor or the shunt
capacitor is replaced by a series capacitor and inductor.
There is also bandstop filters which we have not considered. These filters can be
developed by first applying the normalized, lowpass to normalized, highpass
transformation to the lowpass normalized realization. Next, the lowpass to bandpass
transformation of Eq. (231) is applied to the normalized highpass filter. The result is a
bandstop filter having a stopband of BW geometrically centered at ω
r
. The passband is
geometrically centered around ω
r
and has the width of Ω
n
BW where BW is the bandwidth
of a bandpass filter. If Ex. 35 were repeated for a bandstop filter with a stopband of 10
kHz and a passband of 50 kHz geometrically centered at 100 kHz, the result would be
similar to Fig. 39 except L
1b
and C
1b
(L
3b
and C
3b
) would in parallel and C
2b
and L
2b
would be in series. The values would be different because of the normalized, lowpass to
normalized, highpass transformation (see PA31).
State Variable Equations for RLC Ladder Filters
The next step in the design of ladder filters is to show how to use active elements and
resistors and capacitors to realize a lowpass ladder filter. Let us demonstrate the approach
by using an example. Consider the doublyterminated, fifthorder, RLC, lowpass ladder
filter of Fig. 310. Note that we have reordered the numbering of the components to start
with the source and proceed to the load. We are also dropping the "l" from the component
subscripts because we will only be dealing with lowpass structures in this discussion.
ECE 6414  Continuous Time Filters (P. Allen)  Chapter 3 Page 314
+

C
2n
L
3n L
1n
V
out
(s
n
)
+

V
in
(s
n
) C
4n
L
5n R
0n
R
6n
+

+

I
1
I
3
I
5
V
2
V
4
Figure 310  A fifthorder, lowpass, normalized RLC ladder filter.
The first step in realizing the RLC filter of Fig. 310 by active RC elements is to
assign a current I
j
to every jth series element (or combination of elements in series) of the
ladder filter and a voltage V
k
to every kth shunt element (or combination of elements in
shunt) ofthe ladder filter. These currents and voltages for the example of Fig. 310 are
shown on the figure. These variables are called state variables.
The next step is to alternatively use loop (KVL) and node (KCL) equations expressed
in terms of the state variables only. For example, we begin at the source of Fig. 310 and
write the loop equation
V
in
(s)  I
1
(s)R
0n
 sL
1n
I
1
(s)  V
2
(s) = 0 . (311)
Next, we write the nodal equation
I
1
(s)  sC
2n
V
2
(s)  I
3
(s) = 0 . (312)
We continue in this manner to get the following state equations.
V
2
(s)  sL
3n
I
3
(s)  V
4
(s) = 0 (313)
I
3
(s)  sC
4n
V
4
(s)  I
5
(s) = 0 (314)
and
V
4
(s)  sL
5n
I
5
(s)  R
6n
I
5n
(s) = 0 (315)
Eqs. (311) through (315) constitute the state equations which completely describe the
ladder filter of Fig. 310. A supplementary equation of interest is
V
out
(s) = I
5
(s)R
6n
. (316)
Once, the state equations for a ladder filter are written, then we define a voltage
analog, V
'
j
of current I
j
as
V
'
j
= R'I
j
(317)
ECE 6414  Continuous Time Filters (P. Allen)  Chapter 3 Page 315
where R' is an arbitrary resistance (normally 1 ohm). The voltage analog concept allows
us to convert from impedance and admittance functions to voltage transfer functions which
is a useful step in the activeRC implementation of the ladder filter. Now if for every
current in the state equations of Eq. (311) through Eq. (315) we replace currents I
1
, I
3
,
and I
5
by their voltage analogs, we get the following modified set of state equations.
V
in
(s) 
.

,
`
V
'
1
(s)
R'
(R
0n
+ sL
1n
)  V
2
(s) = 0 (318)
.

,
`
V
'
1
(s)
R'
 sC
2n
V
2
(s) 
.

,
`
V
'
3
( s )
R'
= 0 (319)
V
2
(s)  sL
3n
.

,
`
V
'
3
( s )
R'
 V
4
(s) = 0 (320)
.

,
`
V
'
3
( s )
R'
 sC
4n
V
4
(s) 
.

,
`
V
'
5
( s )
R'
= 0 (321)
and
V
4
(s) 
.

,
`
V
'
5
( s )
R'
(sL
5n
+ R
6n
) = 0 (322)
The next step is to use the 5 equations of Eqs. (318) through (322) to solve for each
of the state variables. The result is
V
'
1
(s) =
R'
sL
1n
]
]
V
in
(s)  V
2
(s) 
.

,
`
R
0n
R'
V
'
1
(s) (323)
V
2
(s) =
1
sR'C
2n
[V
'
1
(s)  V
'
3
(s) ] (324)
V
'
3
(s) =
R'
sL
3n
[V
2
(s)  V
4
(s)] (325)
V
4
(s) =
1
sR'C
4n
[ V
'
3
(s)  V
'
5
(s) ] (326)
and
V
'
5
(s) =
R'
sL
5n
[V
4
(s) 
R
6n
R'
V
'
5
(s) ] . (327)
ECE 6414  Continuous Time Filters (P. Allen)  Chapter 3 Page 316
However, we would prefer to have the variable V
out
(s) used in place of V
'
5
(s) . From Eq.
(315) we get
V
out
(s) =
.

,
`
R
6n
R'
V
'
5
(s) . (328)
Combining Eqs. (326) and (327) with (328) gives
V
4
(s) =
1
sR'C
4n
[ V
'
3
(s) 
.

,
`
R'
R
6n
V
out
(s)] (329)
V
out
(s) =
R
6n
sL
5n
[V
4
(s)  V
out
(s)] . (330)
ECE 6414  Continuous Time Filters (P. Allen)  Chapter 3 Page 317
Design of LowPass, Ladder Filters using ActiveRC Elements
Finally, we are ready to synthesize an activeRC realization of Fig. 310. This is
done by realizing that Eqs. (323), (324), (325), (329) and Eq. (330) are simple
integrators. In each case, the state variable is expressed as the integration of itself or other
state variables. The next step would be to go back to Sec. 4.1 and realize each of the five
equations with 5 positive integrators capable of summing more that one input. However,
as we will recall, the positive integrator did not provide the flexibility and usefulness (i.e.
summing) found in the negative or inverting integrator. We could realize the state
equations using positive integrators of the form shown in Fig. 311.
A1 A2
R
R
R R
V
j
V
k
V
k V
l
C
k
Figure 311  Positive integrator realization of the kth integrator.
We can show that the output of Fig. 311 is given as
V
k
(s) =
1
sRC
k
[V
j
(s)  V
l
(s)] (331)
where V
j
(s) is the input from the previous stage and V
l
(s) is the negative output of the next
stage. Note that the inverter at the output of the integrator allows us to get either the
positive or negative output variable.
Fig. 311 works well for Eqs. (324), (325), and (329). However, a different
approach is necessary for Eqs. (323) and (330) because the variable is a function of itself.
Solving for V
'
1
(s) and V
out
(s) in Eqs. (323) and (331) gives
V
'
1
(s) =
.

,
`
R'
L
1n
s +
R
0 n
L
1n
{ }
V
in
(s) + [V
2
(s)] (332)
and
ECE 6414  Continuous Time Filters (P. Allen)  Chapter 3 Page 318
V
out
(s) =
.

,
`
R
6n
L
5n
s +
R
0 n
L
5n
V
4
(s) , (333)
respectively. We see that Eqs. (332) and (333) are nothing more than the firstorder,
lowpass filter of Fig. 4.26c preceeded by a gain of 1 stage which also serves as a
summer for Eq. (332). The implementation of Eq. (332) is shown in Fig. 312.
A1 A2
R
R
R
R
V
1
'
(s
n
)
V
2
(s
n
)
V
in
(s
n
)
L
1n
/R
R/R
0n
R'
R'
Figure 312  Realization of Eq. (332).
A similar realization can be found for Eq. (333). This realization will only have one input
to the inverter.
Unfortunately, the straightforward application of Figs. 311 and 312 to realize the
state equations uses more op amps than is necessary. A more efficient realization is
achieved by modification of the state equations. One modification of Eqs. (323), (324),
(325), (329) and Eq. (330) is shown below.
V
'
1
(s) =
R'
sL
1n
]
]
V
in
(s)  V
2
(s) 
.

,
`
R
0n
R'
V
'
1
(s) (334)
V
2
(s) =
1
sR'C
2n
[V
'
1
(s)  V
'
3
(s) ] (335)
V
'
3
(s) =
R'
sL
3n
[V
2
(s) + V
4
(s)] (336)
V
4
(s) =
1
sR'C
4n
[ V
'
3
(s) +
.

,
`
R'
R
6n
V
out
(s)] (337)
and
V
out
(s) =
R
6n
sL
5n
[V
4
(s)  V
out
(s)] . (338)
Another modification of the state equations is given as
ECE 6414  Continuous Time Filters (P. Allen)  Chapter 3 Page 319
V
'
1
(s) =
R'
sL
1n
]
]
V
in
(s)  V
2
(s) 
.

,
`
R
0n
R'
V
'
1
(s) (339)
V
2
(s) =
1
sR'C
2n
[V
'
1
(s) + V
'
3
(s) ] (340)
V
'
3
(s) =
R'
sL
3n
[V
2
(s) + V
4
(s)] (341)
V
4
(s) =
1
sR'C
4n
[ V
'
3
(s) 
.

,
`
R'
R
6n
V
out
(s)] (342)
and
V
out
(s) =
R
6n
sL
5n
[V
4
(s)  V
out
(s)] . (343)
Fig. 313a shows the realization of the state equations of Eqs. (334) through (338).
Fig. 313b shows the realization of the state equations of Eqs. (339) through (343). The
second modification has one less op amp but only V
out
(s) is available. Fig. 313a saves
two op amps over the case where every integrator consists of two op amps and Fig. 313b
saves three op amps over the case where every integrator consists of two op amps. It is
interesting to note that Fig. 313 turns out to be the interconnection of four, secondorder,
TowThomas filters of Fig. 116. The inside filters have a Q of infinity because neither
integrator is damped. The intercoupled secondorder filters are distinctly different from the
cascade realization in that there is more than one path from the input to the output of the
filter. This type of realization was first published by Girling and Good
†
and were called
"leapfrog" filters.
†
F.E.J. Girling and E.F. Good, "Active Filters 12: The LeapFrog or ActiveLadder Synthesis," Wireless
World, vol. 76, July 1970, pp. 341345.
ECE 6414  Continuous Time Filters (P. Allen)  Chapter 3 Page 320
(
a
.
)
(
b
.
)
F
i
g
u
r
e
3

1
3

(
a
.
)
R
e
a
l
i
z
a
t
i
o
n
o
f
E
q
s
.
(
3

3
4
)
t
h
r
o
u
g
h
(
3

3
8
)
.
(
b
.
)
R
e
a
l
i
z
a
t
i
o
n
o
f
E
q
s
.
(
3

3
9
)
t
h
r
o
u
g
h
(
3

4
3
)
.
A
1
A
2
A
3
A
4
A
5
A
6
A
7
A
8
R
R
R
R
R
R
R
R
R
R
R
/
R
0
n
L
/
R
1
n
R
R
R
R
V
'
1
C
/
R
2
n

V
2
R

V
3
L
/
R
3
n
V
4
C
/
R
4
n
R
/
R
6
n
L
/
R
5
n
V
o
u
t
V
i
n
R
'
R
'
R
'
R
'
R
'
R
'
R
'
A
1
A
2
A
3
A
4
A
5
A
6
A
7
R
R
R
R
R
R
R
R
R
R
R
/
R
0
n
L
/
R
1
n
R
R
C
/
R
2
n

V
2
R
V
3
L
/
R
3
n
V
4
C
/
R
4
n
R
/
R
6
n
L
/
R
5
n
V
i
n
V
o
u
t

V
'
1

R
'
R
'
R
'
R
'
R
'
R
'
ECE 6414  Continuous Time Filters (P. Allen)  Chapter 3 Page 321
Example 36  Design of a LowPass, Butterworth Ladder Filter using the ActiveRC
Elements
Design a thirdorder, Butterworth, lowpass filter having a passband frequency of 1
kHz using the RCactive ladder realization of Fig. 313. Assume that the source resistance
of the input voltage is zero. Give a realization and the value of all components.
Solution
The normalized, lowpass ladder filter is obtain from Table 31 and is shown in Fig.
314a. Note that we have reversed the order of the subscripts to correspond with Fig. 3
10. Let us choose Fig. 313b as the realization form resulting in Fig. 314b where R
0n
=
0 and R' = R = 1 Ω. Frequency denormalizing this filter by 2πx10
3
and impedance
denormalizing by 10
4
(arbitrarily chosen) gives the realization of Fig. 314c.
+

+

V
in
(s
n
)
V
out
(s
n
)
L
1n
=1.5000 H L
3n
=0.5000 H
C
2n
=
1.3333 F
R
4n
=
1 Ω
(a.)
(b.)
(c.)
= 1.5F = 1.3333F
A1 A2 A3 A4
1Ω
L
1n
C
2n
V
2
1/R
4n
L
3n
V'
1

V
in
(s
n
) V
out
(s
n
)
=0.5 F
1Ω
1Ω
1Ω
1Ω 1Ω
1Ω
V
out
(s)
A1 A2 A3 A4
10 kΩ
V
2
V'
1

23.87 nF
21.22 nF 7.96 nF
V
in
(s)
10 kΩ
10 kΩ 10 kΩ
10 kΩ
10 kΩ 10 kΩ
Figure 314  Results of Ex. 36. (a.) Normalized, lowpass RLC filter. (b.) Normalized,
lowpass, activeRC filter. (c.) Denormalized, lowpass, activeRC filter.
Example 37  A FifthOrder, 1 dB Chebyshev, LowPass, ActiveRC Filter Design
ECE 6414  Continuous Time Filters (P. Allen)  Chapter 3 Page 322
A fifthorder, lowpass, Chebyshev filter with a 1 dB ripple in the passband is to be
designed for a cutoff frequency of 1 kHz. Design an activeRC ladder realization of this
filter. Give a schematic and the value of all components. The input voltage source is
assumed to have a source resistance of 1 kΩ.
Solution
From Table 32, we get the normalized, lowpass, RLC ladder filter elements of R
0n
= 1 Ω, L
1n
= 2.1349 H, C
2n
= 1.09011 F, L
3n
= 3.0009 H, C
4n
= 1.09011, L
5n
= 2.1349
H, and R
6n
= 1 Ω. Note that we have reversed the order of the subscripts in Table 32 to
correspond to that of Fig. 310 (although in this particular case it makes no difference). Let
us again select Fig. 313b as the realization because it has one less op amp. The
component values above are directly substituted into Fig. 313b to achieve a normalized,
lowpass, activeRC, ladder realization. To get the denormalized filter, we select R = 1000
Ω and frequency denormalize by 2πx10
3
. The values of the capacitor of each integrator,
from input to output are C
1
= L
1n
/2πx10
6
= 2.1349/2πx10
6
= 0.340 µF, C
2
= C
2n
/2πx10
6
= 1.09011/2πx10
6
= 0.174µF, C
3
= L
3n
/2πx10
6
= 3.0009/2πx10
6
= 0.478 µF, C
4
=
C
4n
/2πx10
6
= 1.09011/2πx10
6
= 0.174µF, and C
5
= L
5n
/2πx10
6
= 2.1349/2πx10
6
=
0.340 µF.
Other Types of Ladder Filters using ActiveRC Elements
Highpass, bandpass, and bandstop ladder filters can also be designed using active
RC elements. The design methods follow a simular procedure as for the lowpass filter.
Unfortunately, in the highpass ladder filters, the state variables are realized by
differentiators rather than integrators if the state variables are the currents in the series
elements and the voltages across the shunt elements. Integrator realization of the state
variables is possible if the state variables are the voltages across the series elements and the
currents through the shunt elements. However, the equations are not as straightforward to
ECE 6414  Continuous Time Filters (P. Allen)  Chapter 3 Page 323
write as for the lowpass case. Examples of highpass ladder filter design can be found in
other references
†
.
We will illustrate the design of bandpass ladder filters using activeRC elements
because it is a straightforward extension of the lowpass ladder filter design. Once a
normalized, lowpass ladder filter has been found which meets the bandpass filter
specifications, we apply the element transformations illustrated in Fig. 38. Thus, is the j
th series element consists of a series resistor, R
jn
, and a series inductor, L
jln
, become the
series RLC circuit shown in Fig. 315a. If the kth shunt element consists of a shunt
resistor, R
kn
, and a shunt capacitor, C
kln
, the bandpass equivalent is the parallel RLC
circuit shown in Fig. 315b. Normally, R
jn
(R
kn
) is zero (infinity) except for the series or
shunt elements which include the terminating resistors of the ladder filter. The values of
the series normalized bandpass components are
L
jbn
=
.

,
` ω
r
BW
L
jln
, C
jbn
=
.

,
`
BW
ω
r
1
L
jln
. (344)
The values of the shunt normalized components are
C
kbn
=
.

,
` ω
r
BW
C
jln
, L
jbn
=
.

,
`
BW
ω
r
1
C
jln
. (345)
(b.) (a.)
L
jbn
C
jbn
Series Element
R
jn
+

+

V
j1
(s
nb
) V
j+1
(s
nb
)
I
j
(s
bn
)
Y
jn
(s
bn
)
C
kbn
L
kbn
R
kn
Shunt Element
+

Z
kn
(s
bn
)
I
k1
(s
bn
) I
k+1
(s
bn
)
V
k
(s
bn
)
+

V
k
(s
bn
)
Figure 315  (a.) Series ladder element and (b.) shunt ladder element after the normalized,
lowpass to normalized, bandpass transformation of Fig. 38.
†
P.E. Allen and E.Sa' nchezSinencio, Switched Capacitor Circuits, Chapt. 4, Van Nostrand Reinhold
Company, Inc., New York, 1984.
ECE 6414  Continuous Time Filters (P. Allen)  Chapter 3 Page 324
Each element of the bandpass filter can be realized by considering the admittance,
Y
jn
, of Fig. 315a and the impedance, Z
kn
, of Fig. 315b. These drivingpoint functions
can be written as
Y
jn
(s
bn
) =
I
j
(s
bn
)
V
j
(s
bn
)
=
I
j
(s
bn
)
V
j1
(s
bn
)V
j+1
(s
bn
)
=
.

,
`
1
L
jbn
s
bn
s
2
bn
+
.

,
`
R
jn
L
jbn
s
bn
+
1
L
jbn
C
jbn
(346)
and
Z
kn
(s
bn
) =
V
k
(s
bn
)
I
k
(s
bn
)
=
V
k
(s
bn
)
I
k1
(s
bn
)I
k+1
(s
bn
)
=
.

,
`
1
C
kbn
s
bn
s
2
bn
+
.

,
`
1
R
kn
C
kbn
s
bn
+
1
L
jbn
C
jbn
. (347)
These drivingpoint functions can be turned into voltage transfer functions using the
concept of voltage analogs for the currents I
j
and I
k
. Multiplying Eq. (346) by R' gives
T
Yj
(s
bn
) =
R'I
j
(s
bn
)
V
j
(s
bn
)
=
V
'
j
(s
bn
)
V
j1
(s
bn
)V
j+1
(s
bn
)
=
.

,
`
R'
L
jbn
s
bn
s
2
bn
+
.

,
`
R
jn
L
jbn
s
bn
+
1
L
jbn
C
jbn
(348)
or
V
'
j
(s
bn
) =
.

,
`
.

,
`
R'
L
jbn
s
bn
s
2
bn
+
.

,
`
R
jn
L
jbn
s
bn
+
1
L
jbn
C
jbn
[V
j1
(s
bn
)V
j+1
(s
bn
)] . (349)
If R
jn
= 0, then Eq. (349) reduces to
V
'
j
(s
bn
) =
.

,
`
.

,
`
R'
L
jbn
s
bn
s
2
bn
+
1
L
jbn
C
jbn
[V
j1
(s
bn
)V
j+1
(s
bn
)] . (350)
Multiplying Eq. (347) by 1/R' gives
T
Zk
(s
bn
) =
V
k
(s
bn
)
R'I
k
(s
bn
)
=
V
k
(s
bn
)
V
'
k1
(s
bn
)V
'
k+1
(s
bn
)
=
.

,
`
1
R'C
kbn
s
bn
s
2
bn
+
.

,
`
1
R
kn
C
kbn
s
bn
+
1
L
jbn
C
jbn
(3
51)
or
ECE 6414  Continuous Time Filters (P. Allen)  Chapter 3 Page 325
V
k
(s
bn
) =
.

,
`
.

,
`
1
R'C
kbn
s
bn
s
2
bn
+
.

,
`
1
R
kn
C
kbn
s
bn
+
1
L
jbn
C
jbn
[V
'
k1
(s
bn
) V
'
k+1
(s
bn
) ] . (352)
If R
kn
= ∞, then Eq. (352) reduces to
V
k
(s
bn
) =
.

,
`
.

,
`
1
R'C
kbn
s
bn
s
2
bn
+
1
L
jbn
C
jbn
[V
'
k1
(s
bn
) V
'
k+1
(s
bn
) ] . (353)
The secondorder, bandpass, TowThomas filter of Fig. 123 is a perfect realization
of Eq. (349) or Eq. (352). The modification necessary is to add another input which is
simple to do because the op amps have their positive input terminals grounded. Fig. 316
shows the general secondorder realization of either Fig. 315a for the jth series elements.
The realization of the kth shunt element is exactly the same except for the subscript j. For
the cases of Eq. (350) and Eq. (353), R
j4
= ∞. The availability of both positive and
negative outputs simplifies the bandpass realization over the previous lowpass realizations.
A1 A3
R
R
j th element
V
j
'
(s
nb
)
+V
j
'
(s
nb
)
V
j1
(s
nb
)
V
j+1
(s
nb
)
R
j1
R
j1
R
j2
R
j3
R
j4
C
j1
C
j2
A2
Figure 316  ActiveRC realization of either Fig. 315a or Fig. 315b (replace subscript j
by k and interchange primes).
Table 33 shows the design relationships for the implementation of Fig. 315 by the
TowThomas circuit of Fig. 316. An example will illustrate the approach to designing
activeRC bandpass, ladder filters.
ECE 6414  Continuous Time Filters (P. Allen)  Chapter 3 Page 326
Parameters of Fig. 316 Design of Fig. 315a Design of Fig. 315b
R
j1
or R
k1
R
j1
=
R
R'
L
jbn
C
jbn
R
k1
= R'R
C
kbn
L
kbn
R
j4
or R
k4
R
j4
=
R
R
jn
L
jbn
C
jbn
R
k4
= RR
kn
C
kbn
L
kbn
Choose R
j2
= R
j3
= R
or R
k2
= R
k3
= R, a convenient value:
C
j1
= C
j2
or C
k1
= C
k2
C
j1
= C
j2
=
L
jbn
C
jbn
R
C
k1
= C
k2
=
L
kbn
C
kbn
R
Choose C
j1
= C
j2
= C or C
k1
= C
k2
= C, a convenient value:
R
j2
= R
j3
or R
k2
= R
k3
R
j2
= R
j3
=
L
jbn
C
jbn
C
R
k2
= R
k3
=
L
kbn
C
kbn
C
R' is the scaling resistance of the voltage analog concept and is normally 1 Ω
Table 33  Design relationships for the implementation of Fig. 315a or Fig. 315b by the
activeRC filter of Fig. 316.
Example 38  Design of a Butterworth, Bandpass, Ladder ActiveRC Filter
Design a sixthorder, Butterworth, bandpass, doublyterminated, ladder filter using
activeRC elements. Assume the passband is an octave, centered geometrically about 1
kHz. Impedance denormalize by a factor of 10
4
and give a schematic, or equivalent, with
all component values.
Solution
The first thing we must do is find the bandwidth of the bandpass filter. We know
from the specification that f
PB2
= 2f
PB1
. We also know that 1 kHz = f
BP2
f
BP1
.
Combining these 2 relationships gives f
BP1
= 1/ 2 kHz and f
PB2
= 2 kHz. Thus the
bandwidth is 1/ 2 kHz.
ECE 6414  Continuous Time Filters (P. Allen)  Chapter 3 Page 327
From Table 32, for a thirdorder, doublyterminated Butterworth approximation we
find that L
1n
= 1 H, C
2n
= 2 F, and L
3n
= 1 H. Since the filter is symmetrical, we don't
have to reorder the subscripts but can use them as they are. The resulting, normalized,
lowpass filter is shown in Fig. 317a.
If we apply the normalized, lowpass to normalized, bandpass transformation of Eqs.
(344) and (345) to Fig. 317a we get the following normalized bandpass elements of Fig.
317b. L
1bn
= (ω
r
/BW)L
1ln
= (R(2))(1) = 2 H, C
1bn
= (BW/ω
r
/)(1/L
1ln
) = 1/ 2 F,
C
2bn
= (ω
r
/BW)C
2ln
= ( 2 )(2) = 2 2 F, L
2bn
= (BW,ω
r
)(1/C
2ln
) = 1/2 2 H, L
3bn
=
(ω
r
/BW)L
3ln
= ( 2 )(1) = 2 H, and C
3bn
= (BW/ω
r
)(1/L
3ln
) = 1/ 2 F. Therefore the
values of Y
1n
are R
1n
= 1 Ω, L
1bn
= 2 H, and C
1bn
= 1/ 2 F. The values for Z
2n
are
R
2n
= ∞, C
2bn
= 2 2 F, and L
2bn
= 1/2 2 H. The values for Y
3n
are R
3n
= 1 Ω, L
3bn
=
2 H, and C
3bn
= 1/ 2 F.
Finally, we use the design relationships of Table 33 to design three, secondorder
active filters of the form given in Fig. 316. We will select R = R' = 1 Ω. For the first
element, Y
1n
, which is series we choose C
11
= C
12
= 1 F to get R
12
= R
13
= 1 Ω, and R
11
= R
14
= 2 Ω. For the second element, Z
2n
, which is shunt we choose C
21
= C
22
= 1F to
get R
22
= R
23
= 1 Ω, R
21
= 2 2 Ω, and R
24
= ∞. For the third element, Y
3n
, the values
are identical to those for Y
n1
. If we frequency denormalize by 2πx10
3
and impedance
denormalize by 10
4
we get the realization of Fig. 317c.
The above example illustrates the general method by which bandpass ladder filters are
realized using activeRC elements. This method is applicable to any bandpass filter whose
passband and stopband are geometrically centered about the frequency, ω
r
. Other examples
of activeRC bandpass ladder filter design can be found in the problems.
ECE 6414  Continuous Time Filters (P. Allen)  Chapter 3 Page 328
A11 A13
10 kΩ
A12
A21 A23
10kΩ
A22
A31 A33
10 k Ω
A32
C11 =
15.91nF
R11 =
R
11 =
14.14kΩ
R13 = 10kΩ
R
12 =
10kΩ
10 kΩ
C12 = 15.91nF C22 = 15.91nF
C21 = 15.91nF
10kΩ
R
21
=
28.28kΩ
R21 =
28.28kΩ
R23 = 10kΩ
R22 =
10kΩ
R31 =
R34 =
14.14kΩ
R33 = 10kΩ
R32 =
10kΩ
C32 = 15.91nF
C31 =
15.91nF
10 kΩ Vin(s) V
out(s)
Y
1
Z
2 Y3
14.14kΩ
14.14kΩ
R14 =
14.14kΩ
V'
1
V
2
V2
V
out
+

+

R
4n
=
1 Ω
Vin(sln) V
out
(s
ln
)
R
0n
=1 Ω
L
1ln
=1 H L
3ln
=1 H
C
2ln
= 2 F
+

+

L1bn 2 H
C
1bn
= 2F
C
2bn
L
2bn
=
L
3bn
2 H
C
3bn
R
4n
=1Ω
Vin (sbn)
V
out
(s
bn
)
Y
1n
Y
3n
Z2n
R
0n
=1Ω
1/ =
2F =
2F 1/ =
2 1/ 2 2 H
=
(a.)
(b.)
(c.)
Figure 317  Bandpass, activeRC ladder filter of Ex. 38. (a.) Normalized, lowpass,
ladder filter. (b.) Normalized, bandpass ladder filter. (c.) Denormalized, activeRC
bandpass ladder filter.
Summary
The block diagram in Fig. 21 illustrated the two general design approaches normally
used to design active filters. The first approach was the cascade of first or secondorder
stages and was covered in the previous section. The second approach was the design of
ladder filters which has been the subject of this section. Basically, the cascade approach is
easier to tune but is more sensitive to component tolerance than the ladder filters.
The starting point of the ladder active filter design is with the normalized, lowpass
RLC ladder filter. These filter structures are the result of network synthesis methods and
are widely tabulated for the designer. As in Sec. 2, we restricted ourselves to filters whose
zeros were at infinity or the origin of the complex frequency plane. The design of high
ECE 6414  Continuous Time Filters (P. Allen)  Chapter 3 Page 329
pass and bandpass filters was achieved by direct application of the normalized frequency
transformations to the elements of the normalized, lowpass RLC filter.
To achieve an activeRC realization of the RLC ladder filters, we introduced the
concept of state variable. By correctly selecting the state variables and using the concept of
voltage analog for current, we were able to synthesize the state variables using summing
integrators. The resulting circuits were very similar to interconnected, secondorder, low
pass TowThomas filters. We saw that highpass, activeRC ladder filters required a little
cleverness to be able to express all of the state variables so that they could be realized by
integrators. Each element of the normalized, lowpass RLC resulted in a secondorder
bandpass circuit when the normalized lowpass to normalized bandpass frequency
transformation was applied. These secondorder, bandpass circuits were easily realizable
as secondorder, bandpass, TowThomas filters.
One of the themes that hopefully has become evident to the reader is how filter design
all builds from the normalized, lowpass filters. Using simple normalizations and
frequency transformations permits the design of complex filters. One important constraint
to remember is that the bandpass and bandstop filters designed by these methods must have
passbands and stopbands geometrically centered about a common frequency. Other design
methods, not covered here, will allow the design of filters not subject to this constraint.
Some the important points of this section are summarized below for the convenience
of the reader.
• Sensitivity is a measure of some aspect of the filter performance on the components
of the filter.
• The normalized lowpass to normalized highpass frequency transformation turns
inductors (capacitors) into capacitors (inductors) whose value is the reciprocal of the
lowpass component value.
• The normalized lowpass to normalized bandpass frequency transformation turns an
inductor into an inductor in series with a capacitor. The value of the inductor is
ECE 6414  Continuous Time Filters (P. Allen)  Chapter 3 Page 330
(ω
r
/BW)L
ln
and the value of the capacitor is (BW/ω
r
L
ln
). whose value is (inductors)
whose value is the reciprocal of the lowpass component value.
• The normalized lowpass to normalized bandpass frequency transformation turns a
capacitor into a capacitor in parallel with an inductor. The value of the capacitor is
(ω
r
/BW)C
ln
and the value of the inductor is (BW/ω
r
C
ln
).
• In order to realize the state variables by integrators, the current through an inductor
and the voltage across a capacitor should be chosen as state variables.
• The TowThomas secondorder realization becomes a very useful activeRC filter for
realizing RLC ladder filters.
ECE 6414  Continuous Time Filters (P. Allen)  Chapter 3
Page 32
Furthermore, we can show that the sensitivity of the poleQ of Fig. 116 to R2 is 0.5. Another interpretation of sensitivity is as follows. If the component x changes by dx/x, p then the performance changes by the product of Sx times dx/x. For example, the poleQ will change by +10% if R1 changes by +10%. If the relationship between p and x is nonlinear, the sensitivity measure is only good for small changes. Consequently, changes much over 10% should be calculated using nonlinear sensitivity methods or point sensitivity calculations. Sensitivity calculations can become complex if the expression relating the performance to the component is complicated. Fortunately we only wish to use the concept of sensitivity to compare the two different approaches to higherorder active filter design shown in Fig. 21. Although we have not yet developed the ladder filter, it can be
distinguished from the cascaded filter by its dependence upon the passive and active components of the filter realization. The performance of the ladder filter always depends on more than one passive or active component. As a result, the performance is not strongly dependent upon any one component but rather on all components. However, the If this
performance of the cascaded filter may depend upon only one component.
component is inaccurate or off in value, the cascade filter performance can be strongly influenced while the same component change may not be noticed in the performance of the ladder filter. This advantage of the ladder filter becomes a disadvantage when tuning the filter. It is almost impossible to tune a ladder filter because no single component dominates the performance of the filter. Normalized, RLC, LowPass Ladder Filters RLC, lowpass ladder filters are the result of modern network synthesis techniques and are based on techniques well known in circuit theory†. The resulting realizations of these synthesis techniques always start with a load resistor of 1 ohm and works toward the
† M.E. Van Valkenburg, Introduction to Modern Network Synthesis, Chapter 10, John Wiley & Sons, Inc., New York, 1960.
ECE 6414  Continuous Time Filters (P. Allen)  Chapter 3
Page 33
input of the filter. Fig. 31 shows the form for a singlyterminated RLC filter for the case of even and odd order functions with the numbering of components going from the output to the input of the filter. + Vin (sn ) (a.) + Vin (sn ) (b.) Figure 31  Singlyterminated, RLC filters. (a.) N even. (b.) N odd. The RLC ladder filters of Fig. 31 are normalized to a passband of 1 rps. The LN,n CN1,n L3n C2n L1n 1Ω + Vout (sn ) LN,n CN1,n L2n C3n C1n 1Ω + Vout (sn ) 
denormalizations of Table 21 are applicable to the elements of Fig. 31. Fig. 32 shows the normalized ladder filters for doublyterminated, RLC filters. It is seen that these filters are similar to those of Fig. 31 except for a series source resistance. + Vin (sn ) (a.) + Vin (sn ) (b.) Figure 32  Doublyterminated, RLC filters. (a.) N even. (b.) N odd. The tabular information for the design of RLC filters consists of the normalized component values of Figs. 31 and 32. Each of the many different types of filter R LN,n CN1,n L3n C2n L1n 1Ω + Vout (sn ) R LN,n CN1,n L2n C3n C1n 1Ω + Vout (sn ) 
5155 0.4037 1.5908 1.7287 1. 1967.1736 0.7215 2.5529 1. 31 for the Butterworth and Chebyshev singlyterminated.6961 2.5088 1dB ripple Chebyshev (1 rps passband) 1.2817 1.Normalized component values for Fig.5451 0. RLC filters of Figs. 1975.1111 1. 31b.4126 1. Table 31 . Inc.0437 1. A.7213 2.6652 1.7988 1. NY.1583 1.7118 1.8121 1.6736 2. John Wiley & Sons.8424 1..6489 1..3820 1.0601 1.5772 1.0270 1.3333 1. NY.5529 0.0773 1.3801 L1n C2n L3n C4n L5n C6n L7n C8n L9n C10n Use these component designations for odd order of Fig.5576 0.6588 1.9110 0.0918 1.5643 0.0674 1. † L. .5607 0. McGrawHill Book Co.5307 0. 33. 32a.0550 1.1192 1.7021 2. 31 and 32.0118 1.7626 1. Allen) . RLC filter approximations.7071 1.0824 1.9093 1.5000 Butterworth (1 rps passband) 0.5576 0.6507 2. 31a and Fig.1564 0.0491 1.9938 1.1574 1. R..2225 0.4694 2.6202 1.9370 1.8552 1. Handbook of Filter Synthesis.1803 1.0495 1.8246 1.6869 1.7579 1.9957 1. normalized. Huntington.Use of the Table 31 to Find a SinglyTerminated.1408 1. Tables 31 and 32 are typical of this tabularized information for the Butterworth and 1dB Chebyshev approximation for the singlyterminated and doublyterminated.5283 1.2588 0.6707 1.4654 0.3090 0. Network Analysis and Synthesis.ECE 6414 .0537 1. Krieger Publishing Co.Chapter 3 Page 34 approximations have been tabulated for values of N up to 10 or more† . RLC filter for a 4thorder Butterworth filter.2921 1. Solution Using Table 31 and using the component designations at the top of the table gives Fig.3972 1.1658 1.4817 2.6850 2.7306 2.7593 1.0645 1.8944 1.I. Example 31 .0922 1.5000 1.2016 1.4751 2.4601 2.3691 1.0899 1.6944 1.0832 1.0872 1.5100 1. 1962.3332 1.1951 0.4142 0. Zverev. N 2 3 4 5 6 7 8 9 10 2 3 4 5 6 7 8 9 10 Use these component designations for even order of Fig.4441 1.5628 0.4790 2.6918 2.3827 1.0406 1. RLC Lowpass Filter Find a singlyterminated.7772 1.7317 1.2588 1.1453 1.3457 1.8414 1.Continuous Time Filters (P. C1n L2n C3n L4n C5n L6n C7n L8n C9n L10n 0. Weinburg.6560 1.E.
6180 0.0236 1dB ripple Chebyshev (1 rps passband) 1.3129 2. RLC Chebyshev approximations for R = 1 Ω.9319 1.9754 1.6629 1. 32b.1214 1.1192 2.5321 1. RLC approximations.9319 1.1111 0.1897 3. R = 1Ω.6180 1.9616 1. RLC filter for a fifthorder Chebyshev filter approximation having 1 dB ripple in the passband and a source resistance of 1 Ω.1666 2.4142 1.8019 1.5321 1.3129 0.Fourthorder. 32 for the Butterworth and 1dB Chebyshev doublyterminated. Allen) . RLC Lowpass Filter Find a doublyterminated.9616 1.8478 0.ECE 6414 .8794 2.0000 1.4450 0.1797 C2n L3n C4n L5n C6n L7n C8n L9n C10n Use these component designations for odd order of Fig.0936 1.7654 1.Continuous Time Filters (P.0911 2.6629 1. This is a special result for R = 1 Ω and is not true for other values of R.Use of Table 32 to Find a DoublyTerminated.Normalized component values for Fig.7820 1.9754 1. R = 1Ω.6180 0. normalized lowpass RLC filter realization.0009 1.0000 0.3902 0.3902 1.4142 2. Use these component designations for even order of Fig.1797 L1n Table 32 .0911 3.0824 H + Vin (sn ) C3n = 1.8478 1.1349 2.8794 1.1897 3.7820 1.9080 0. Solution .2470 1.4142 0. Example 32 .1349 1.1192 3. We note that no solution exists for the evenorder cases of the doublyterminated.0236 2.4142 1.0936 1.3473 0.1111 1.0000 1.9080 1.5 at low frequencies where the inductors are shortcircuits and the capacitors are opencircuits. N 2 3 4 5 6 7 8 9 10 3 5 7 9 C1n 1.0000 1.2740 0.5772 F C1n = 0.0000 1.1746 1.1666 1.0000 Butterworth (1 rps passband) 1.8019 2.3827 F 1Ω + Vout (sn )  Figure 33 .7654 0.4142 0.3473 0. Butterworth.0000 0.1735 3.0000 1.5307 H L2n =1.5176 0. 32a. L2n C3n L4n C5n L6n C7n L8n C9n L10n 1.Chapter 3 Page 35 L4n =1.4450 1.1214 1.1115 2.5176 1.9941 2.6180 2. We also can see that the gain in the passband will be no more than 6 dB because of the equal source and load resistances causing an gain of 0.4142 1.1115 3.
the filter has no active elements such as op amps. The design procedure is simply to find the normalized. filter approximation which satisfies the filter specifications and then denormalize the frequency and the impedance. L3 = (103)(1. L3n = 1.Continuous Time Filters (P. C2 = (0.6588 H.Chapter 3 Page 36 Using Table 32 and using the component designations at the bottom of the table gives Fig.1 dB for N = 6. Next. An example will illustrate the procedure. Design this filter using a passive RLC realization and denormalize the impedances by 103.3972 F. normalized. the designer may choose a passive realization over an active realization.336 mH. Example 33 . C4n = 1.14 dB for N = 7. . LowPass Filter A lowpass filter using the Butterworth approximation is to have the following specifications: TPB = 3 dB. From Table 31 for N = 7 for the Butterworth approximation we get L1n = 0. Chebyshev lowpass RLC filter realization.071 mH.6560F)/(πx109) = 209 pF. L5n = 1. fPB = 500 kHz.3972F)/(πx109) = 445 pF. and L7n = 1. In this case.Fifthorder. such as high frequency. (211) using ε = 1 to get TSB = 30.2225H)/(πx106) = 0. C2n = 0. LowPass.2225 H. Allen) . we use the denormalizations in the bottom row of Table 21 to get L1 = (103)(0. TSB = 40 dB. and TSB = 42. TSB = 36. lowpass.0550H)/(πx106) = 0. doublyterminated.1349 H L3n =3.0550 H. Solution First we see that Ωn = fSB/fPB = 2 and then try various values of N in Eq.1 dB for N = 5.5576 H.0009 H L1n =2. C4 = (1.0911 F 1Ω Vout (sn )  Figure 34 . L5n =2. C6n = 1.7988 F.6560 F. and fSB = 1 MHz.1349 H + Vin (sn ) 1Ω + C4n = 1.ECE 6414 .0911 F C2n = 1. RLC Filter Design In some cases.Design of a Passive. 34.
071 mH + Vin (s) C4 = 573 pF C4 = 445 pF C2 = 209 pF 1 kΩ + Vout (s) Figure 35 . 33.6588H)/(πx106) = 0. For each capacitor.Continuous Time Filters (P.e. these influences must be . one adjusts each component before inserting it into the filter by the following means. Inside the cylindrical insulator is a magnetic cylinder whose position is adjusted by a screw which causes the inductance of the inductor to vary. For precise filter applications. In order to get the desired results. The above procedures work well if there are no parasitic elements that will effect the filter performance.496 mH L5 =0.336 mH L1 =0.ECE 6414 . Then the magnetic core is inserted and adjusted to achieve the desired inductance. realization is shown in Fig. that the design of RLC filters is very simple.528 mH.496 mH. We can see by the previous example. L7 =0. The final arrangement can be taken to a capacitor bridge or capacitance meter and the variable capacitor trimmed until the desired value is achieved. Therefore. we must have the exact values of components that were calculated in Ex. C6 = (1. The inductors are typically made by winding lowresistance wire around a cylindrical insulator. This The load resistor becomes 1 kΩ.7988F)/(πx109) = 573 pF. 35. The important parasitics are the capacitors from each node to ground and the resistance of the inductors. 33.528 mH L3 =0.5576H)/(πx106) = 0. the actual implementation of the filter requires careful consideration and effort. a small variable capacitor (i. and L7 = (103)(1.Chapter 3 Page 37 L5 = (103)(1.Seventhorder. The cost of such accurate components is generally prohibitive. However. 50 pF) can be placed in parallel with a larger fixed capacitor whose value is slightly less than the desired value. Allen) . The inductors are custom made by winding the necessary number of turns to achieve an inductance less than desired with the magnetic core removed. Butterworth filter realization of Ex.
Lln. whose value is 1/Lln. highpass RLC filter.Influence of the normalized. Lln Cln sln → s1 hn Chn = 1 Lln Lhn = 1 Cln Normalized HighPass Network Normalized LowPass Network Figure 36 . If we apply this transformation to an inductor of a normalized. of a normalized. with a capacitor. and replaces it by an inductor. (33) Similarly. the elements have been . From the above results. with an inductor. (33) and (34). lowpass realization. Chn. we obtain 1 1 slnLln = shn Lln = shnChn . and replaces it by a capacitor. C ln. we see that to achieve a normalized. The frequency transformation from the normalized. lowpass realization. whose value is 1/Lln and each capacitor. we replace each inductor. Once. Allen) .Continuous Time Filters (P. (34) From Eqs. if applying the transformation to a capacitor. 36 illustrates these important relationships. HighPass. Lhn. highpass frequency transformation takes an inductor. C ln. Lhn.Chapter 3 Page 38 incorporated into the design phase. we see that the normalized. RLC ladder filters is made simple by the frequency transformations of the previous section. lowpass to normalized highpass was given by Eq. This transformation also takes a capacitor. whose value is 1/Cln. (223). Chn. RLC. Lln. Ladder Filter Design The design of highpass and bandpass. we obtain 1 shn 1 slnCln = 1 Cln = shnLhn . Fig. This results in a more complex filter design procedure which is beyond the scope of our treatment.ECE 6414 . highpass frequency transformation on the inductors and capacitors. lowpass to normalized. lowpass to normalized. whose value is 1/Cln. C ln.
6652 H. RLC Filter A filter is to be designed to provide a passband with a ripple of no more than 3 dB above 100 kHz and a stopband attenuation of at least 45 dB below 50 kHz. we denormalize the frequency by 2πx100 kHz = 2πx105 and the impedance by 50 in order to represent the connection of the 50ohm cable to the output of the filter. C3hn = 1/L3ln = 1/1. Eq. L2hn = 1/C2ln = 1/1.6286x(50)/(2πx105) = 100 µH. We must choose the singlyterminated filter because the doublyterminated filter has a maximum passband gain of 6 dB.ECE 6414 . Fig. Using Eq. Solution We know from Eq. . (226) that Ωn = 2.6 nF. lowpass to normalized.5908 = 0. and C5h = 0. Select a suitable RLC filter approximation and give the values of all components.0674 H.9369/(50x2πx105) = 59.4441 = 0. The output of this filter will be connected to a 50ohm cable and will serve as the load resistor. Therefore. (211) gives an N = 8 (TSB = 48.6652 = 0. Finally.5016 F. highpass frequency transformation to each element to get C1hn = 1/L1ln = 1/1.6005 F. L4h = 0.6925x(50)/(2πx105) = 110 µH. Let us choose the Chebyshev approximation since there are almost half the components. (219) gives N = 5 (TSB = 45. L2h = 0. 37 shows the final realization for this example.Design of a HighPass. L3ln = 1.6286 H.Chapter 3 Page 39 transformed. C2ln = 1.0674 = 0.5908 F.3 nF.9369 F. and C5hn = 1/L5ln = 1/1. L4hn = 1/C4ln = 1/1.4441 F. Next.9938 H.2 dB) for a Butterworth approximation.6925 H.Continuous Time Filters (P. Allen) . we apply the normalized. then both a frequency and impedance denormalization can be performed to result in the final filter design. The actual values for the filter are C1h = 0. from Table 31 we have L1ln = 1.9 nF. C3h = 0.3 dB) for a Chebyshev approximation. and L5ln = 1.9938 = 0. C4ln = 1.6005/(50x2πx105) = 38.5016/(50x2πx105) = 31. Example 34 .
Denormalized. highpass filter realization for Ex.ECE 6414 . RLC. bandpass transformation of Eq. (231). Let us first consider the inductor.Chapter 3 Page 310 C5h = 38. The first step in designing a bandpass filter is to apply the bandpass normalization of Eq. 34.Continuous Time Filters (P. and replaces it by an inductor. of a normalized. Allen) . lowpass to normalized. Let us simultaneously apply the bandpass normalization and the frequency transformation by using Eq. lowpass filter.6 nF + Vin (s) L2h = 110 µH L4h = 100 µH 50 ohm cable + Vout (s)  Figure 37 . Next. Lbn. (231). Lln. (35) Thus we see that the bandpass normalization and frequency transformation takes an inductance. Lastly. Ladder Filter Design The design of RLC bandpass ladder filter follows exactly the same approach as used in the previous section except the transformations are applied to the passive components of the normalized lowpass filter rather than the roots. bandpass filter by ωr and by the desired impedance denormalization.9 nF C1h = 59. in series with a capacitor. Lln. Both of these steps are incorporated in Eq. The normalized.3 nF C3h = 31. Cbn. we apply the normalized. Chebyshev. whose values are given as Lln ωr Lbn = BW Lln = Ωb and (36) . inductance Lln can be expressed as ωr 1 slnLln = BW (s b n + s ) Lln bn ω r L ln 1 ω r L ln 1 = sbn BW + sbn BW = sbnLbn + sbnCbn . we denormalize the normalized. (233). Bandpass. (232) to the components of the lowpass filter.
(39). we see that the bandpass normalization and frequency transformation takes a capacitor.Continuous Time Filters (P. 38. (37). Cbn. Cln. (231) on an inductor and capacitor of a lowpass filter. . in parallel with an inductor. (38).Illustration of the influence of the normalized. ωr (39) (310) Eqs.ECE 6414 . and (310) are very important in the design of RLC bandpass filters and are illustrated in Fig. (36). (231) to a normalized capacitance. in a lowpass circuit and transfors to a capacitor. 1 s bn C bn + s L bn bn 1 (37) (38) From Eq. ωr Now we apply Eq. Lbn = ωr Lln Cbn = BW 1 ωr Lln BW sn → ωr sbn + s1 BW bn Cbn = ωr Cln BW Lbn = BW 1 ωr Cln Normalized Bandpass Network Lln Cln Normalized LowPass Network Figure 38 . bandpass transformation of Eq. Lbn.Chapter 3 Page 311 Ωb BW 1 Cbn = Lln = Lln . whose values are given as Cln ωr Cbn = BW Cln = Ωb and Ωb BW 1 Lbn = Cln = Cln . lowpass to the normalized. to get 1 1 slnCln = ω r 1 BW (s b n + s ) C l n bn = 1 ωr 1 ωrCln sbnBW C l n + s BW bn = . Allen) . Cln.
for L3ln we get L3bn = 1. Chebyshev approximation is to be used to design a bandpass filter with a passband of 10 kHz and a stopband of 50 kHz geometrically centered at 100 kHz.1 = 13.332 F and L2bn = 0.ECE 6414 . (241). Example 35 . and L3ln = 1.Chebyshev.Chapter 3 Page 312 Once the normalized.018 H and C1bn = 0. For L1ln we get L1bn = 1. The resulting bandpass filter is shown in Fig. lowpass to normalized. The values of the thirdorder.Design of an RLC Bandpass Filter A 1dB ripple. is Ωb = 10/100 = 0.5088 = 0. (219) gives an attenuation of 47.85 dB.0118 H. bandpass transformation gives the following results.1 = 11. Solution From the specification and Eq.5088 H. C2ln = 1. defined in Eq. 39.Continuous Time Filters (P.3332/0. An example will serve to illustrate the design approach.1/0.5088/0.066278 F. 35.088 H and C3bn = 0. respectively. normalized Chebyshev approximation are L1ln = 1.261mH L2b = 300 Ω 358µH + Vout (s)  Figure 39 . Design of Other Types of RLC Filters . Applying the normalized. Finally.0118/0.204mH C2b = 70. bandpass realization.11018 = 0. For C2ln we get C2bn = 1. then all that is left is to denormalize the passive elements.1 = 15. C3b =3. The value of Ωb. Assume that the load resistor is 300 ohms.090761 F . lowpass realization has been transformed to a normalized. Allen) .075008 H .815nF L1b = 5. bandpass realization for Ex. (235).1. The attenuation in the stopband should be at least 45 dB. we frequency and impedance denomalize each element by 2πx105 and 300.13332 = 0. we know that Ωn = 5.516nF + Vin (s) L3b = 7.1/1. Substituting N = 3 into Eq.73nF C1b =4. Give a final filter schematic with the actual component values.1/0. Finally.3332 F.
lowpass to normalized. . Normalized. Let us demonstrate the approach by using an example. Next. Note that we have reordered the numbering of the components to start with the source and proceed to the load. fifthorder. the lowpass to bandpass transformation of Eq. 310. lowpass RLC filters have been developed and tabulated for this type of filter. lowpass ladder filter of Fig.Chapter 3 Page 313 There are several types of RLC filters that we will not cover in this section but we should briefly mention them. The passband is geometrically centered around ωr and has the width of ΩnBW where BW is the bandwidth of a bandpass filter. lowpass to normalized. (231) is applied to the normalized highpass filter. highpass transformation (see PA31).Continuous Time Filters (P. The result is a bandstop filter having a stopband of BW geometrically centered at ωr. the result would be similar to Fig. Consider the doublyterminated. These filters can be developed by first applying the normalized. 31 and 32 except that the series inductor is replaced by a parallel inductor and capacitor or the shunt capacitor is replaced by a series capacitor and inductor. There is also bandstop filters which we have not considered. Allen) . The values would be different because of the normalized. State Variable Equations for RLC Ladder Filters The next step in the design of ladder filters is to show how to use active elements and resistors and capacitors to realize a lowpass ladder filter. They are similar in form to the ladder filters of Figs. 218. highpass transformation to the lowpass normalized realization. If Ex. RLC. We are also dropping the "l" from the component subscripts because we will only be dealing with lowpass structures in this discussion.ECE 6414 . One type are filters that have jω axis zeros such as the one shown in Fig. 39 except L1b and C1b (L3b and C3b) would in parallel and C2b and L2b would be in series. 35 were repeated for a bandstop filter with a stopband of 10 kHz and a passband of 50 kHz geometrically centered at 100 kHz.
sL5nI5(s) .R6nI5n(s) = 0 (315) (313) (314) (312) (311) Eqs. 310 by active RC elements is to assign a current Ij to every jth series element (or combination of elements in series) of the ladder filter and a voltage Vk to every kth shunt element (or combination of elements in shunt) ofthe ladder filter. The first step in realizing the RLC filter of Fig.ECE 6414 .V4(s) = 0 I3(s) . These currents and voltages for the example of Fig. We continue in this manner to get the following state equations.I3(s) = 0 . The next step is to alternatively use loop (KVL) and node (KCL) equations expressed in terms of the state variables only. 310 are shown on the figure.sC4nV4(s) . 310.I5(s) = 0 and V4(s) . These variables are called state variables.Chapter 3 Page 314 I1 + Vin (sn ) R0n L1n C2n +  I3 L3n V2 C4n +  I5 L5n V4 R6n + Vout (sn )  Figure 310 . we begin at the source of Fig. For example.Continuous Time Filters (P.sL1nI1(s) .sC2nV2(s) . the state equations for a ladder filter are written.I1(s)R0n . V2(s) . (311) through (315) constitute the state equations which completely describe the ladder filter of Fig. we write the nodal equation I1(s) . (316) Once. lowpass. then we define a voltage analog.V2(s) = 0 . V' of current I as j j Vj' = R'Ij (317) . A supplementary equation of interest is Vout(s) = I5(s)R6n . normalized RLC ladder filter. Allen) . Next.A fifthorder. 310 and write the loop equation Vin(s) .sL3nI3(s) .
V4(s) = 0 V '( s ) V '( s ) 3 .Chapter 3 Page 315 where R' is an arbitrary resistance (normally 1 ohm). The result is R' R0n V '(s) = sL V in (s) .R ' V '(s) ] . R ' V '(s) 1 1 1n 1 V2(s) = sR'C 2n [V'(s) . (311) through Eq. and I5 by their voltage analogs.V2(s) = 0 V '(s) V '( s ) 1 . 5 5 (327) (323) (324) (325) (326) . we get the following modified set of state equations.sC4nV4(s) . (315) we replace currents I1.sC2nV2(s) . R ' (R0n + sL1n) .V 2 (s) .V '(s) ] 1 3 R' V '(s) = sL3n [V2(s) .Continuous Time Filters (P. 5 = 0 R' R' and V5' ( s ) V4(s) .V '(s) ] 3 5 and R6n R' V '(s) = sL5n [V4(s) . The voltage analog concept allows us to convert from impedance and admittance functions to voltage transfer functions which is a useful step in the activeRC implementation of the ladder filter.ECE 6414 . I3 . Allen) . R ' (sL5n + R6n) = 0 (322) (318) (319) (320) (321) The next step is to use the 5 equations of Eqs. V '(s) 1 Vin(s) . 3 = 0 R' R' V '( s ) 3 V2(s) . Now if for every current in the state equations of Eq.sL3n R ' . (318) through (322) to solve for each of the state variables.V4(s)] 3 1 V4(s) = sR'C 4n [ V '(s) .
From Eq. R6n Vout(s)] 3 R6n Vout(s) = sL5n [V4(s) .Chapter 3 Page 316 However. (326) and (327) with (328) gives 1 R' V4(s) = sR'C 4n [ V '(s) . (328) Combining Eqs.Continuous Time Filters (P. (329) (330) .Vout(s)] . we would prefer to have the variable Vout(s) used in place of V '(s) . Allen) . 5 (315) we get R6n Vout(s) = R ' V '(s) 5 .ECE 6414 .
(325). 4. summing) found in the negative or inverting integrator. 311. We could realize the state equations using positive integrators of the form shown in Fig.1 and realize each of the five equations with 5 positive integrators capable of summing more that one input.ECE 6414 .Positive integrator realization of the kth integrator. Note that the inverter at the output of the integrator allows us to get either the positive or negative output variable. and (329). as we will recall. (330) are simple integrators.Continuous Time Filters (P. (329) and Eq. we are ready to synthesize an activeRC realization of Fig. (324). a different approach is necessary for Eqs. However. the positive integrator did not provide the flexibility and usefulness (i. (324). 310. (323) and (330) because the variable is a function of itself. (325). The next step would be to go back to Sec.e. Ladder Filters using ActiveRC Elements Finally. 311 is given as 1 Vk(s) = sRCk [Vj(s) .Chapter 3 Page 317 Design of LowPass. Allen) . In each case. Solving for V'(s) and V (s) in Eqs. This is done by realizing that Eqs. the state variable is expressed as the integration of itself or other state variables. However. R R A1 Ck Vk R R Vk Vl Vj A2 Figure 311 . (323) and (331) gives 1 out V '(s) 1 and L1n = R 0 n{ V in(s) + [V2 (s)]} s + L1n R' (332) . We can show that the output of Fig. 311 works well for Eqs.Vl(s)] (331) where Vj(s) is the input from the previous stage and Vl(s) is the negative output of the next stage. Fig. (323).
Realization of Eq. A similar realization can be found for Eq. (332) is shown in Fig. (330) is shown below.26c preceeded by a gain of 1 stage which also serves as a summer for Eq.ECE 6414 . (332). One modification of Eqs. R'R/R0n V2 (sn ) Vin (sn ) R R A1 R R A2 L1n /R R' V'1 (sn ) Figure 312 . (325).Vout(s)] .V '(s) + R Vout(s)] 3 4n 6n and R6n Vout(s) = sL [V4(s) . (333) respectively. R' R0n V '(s) = sL1n V in (s) .V 2 (s) . the straightforward application of Figs. (324). 5n Another modification of the state equations is given as (338) (334) (335) (336) (337) . (332). We see that Eqs. This realization will only have one input to the inverter. 312. (329) and Eq. A more efficient realization is achieved by modification of the state equations. 4. (332) and (333) are nothing more than the firstorder. (333). 311 and 312 to realize the state equations uses more op amps than is necessary. (323).V '(s) ] 1 3 R' V '(s) = sL3n [V2(s) + V4(s)] 3 1 R' V4(s) = sR'C [.Continuous Time Filters (P. R ' V '(s) 1 1 1 V2(s) = sR'C 2n [V'(s) . The implementation of Eq.Chapter 3 Page 318 6n R5n L Vout(s) = R 0 n s + L 5n V4(s) . lowpass filter of Fig. Allen) . Unfortunately.
ECE 6414 . pp. Girling and E. Good.J.V 2 (s) . The intercoupled secondorder filters are distinctly different from the cascade realization in that there is more than one path from the input to the output of the filter. This type of realization was first published by Girling and Good† and were called "leapfrog" filters. secondorder. The inside filters have a Q of infinity because neither integrator is damped. 313b saves three op amps over the case where every integrator consists of two op amps.Continuous Time Filters (P.Vout(s)] .E. 341345. It is interesting to note that Fig. vol. R ' V '(s) 1 1 1 V2(s) = sR'C 2n [V'(s) + V '(s) ] 1 3 R' V '(s) = sL3n [V2(s) + V4(s)] 3 1 R' V4(s) = sR'C [ V '(s) ." Wireless World. The second modification has one less op amp but only Vout(s) is available. TowThomas filters of Fig. R Vout(s)] 3 4n 6n and R6n Vout(s) = sL [V4(s) . (334) through (338). Fig. † F.F.Chapter 3 Page 319 R' R0n V '(s) = sL1n V in (s) . 5n (339) (340) (341) (342) (343) Fig. Fig. July 1970. 313a saves two op amps over the case where every integrator consists of two op amps and Fig. "Active Filters 12: The LeapFrog or ActiveLadder Synthesis. 76. 313 turns out to be the interconnection of four. (339) through (343). 313b shows the realization of the state equations of Eqs. 313a shows the realization of the state equations of Eqs. . Allen) . 116.
(a.) Realization of Eqs. (339) through (343).) R R R R V' 1 A2 A3 V 2 R R A4 V3 R'C2n/R L3n/R R' R A5 R R A6 V4 R R R' C4n /R R A7 ECE 6414 .) Realization of Eqs. (b.) Figure 313 .R R R R' R/R0n R R'C2n/R R R R A4 V2 V 3 V4 A5 A6 R R A7 R' R/R6n L3n/R R' R' C 4n /R R R A8 R R A2 V' 1 A3 L /RR' 1n L5n /R R' Vout V in R R A1 (a.Continuous Time Filters (P. (334) through (338). Page 320 .Chapter 3 R' R/R0n R/R 6n L5n /R R' Vout L1n/R R' Vin R A1 (b. Allen) .
Note that we have reversed the order of the subscripts to correspond with Fig. lowpass RLC filter. Butterworth.) Denormalized. 314b where R0n = 0 and R' = R = 1 Ω.) 1Ω L1n= 1. activeRC filter.) Normalized. lowpass filter having a passband frequency of 1 kHz using the RCactive ladder realization of Fig. Allen) . 313b as the realization form resulting in Fig.5000 H R4n = C2n = Vin (sn ) 1.Design of a LowPass. activeRC filter.Chapter 3 Page 321 Example 36 .5 F Vin (sn ) 1Ω A1 V' 1 1Ω Vout (sn ) (b. LowPass. 1 dB Chebyshev. (a.87 nF 10 kΩ 10 kΩ A1 V' 1 A2 10 kΩ A3 10 kΩ 21. Solution The normalized.) Figure 314 . Assume that the source resistance of the input voltage is zero.Continuous Time Filters (P. Example 37 .) 10 kΩ 23. + L1n =1.3333 F 1Ω (a. lowpass ladder filter is obtain from Table 31 and is shown in Fig. lowpass. lowpass. 36.96 nF Vin (s) 10 kΩ Vout (s) V2 (c. Butterworth Ladder Filter using the ActiveRC Elements Design a thirdorder. ActiveRC Filter Design . 310. Let us choose Fig. (b.) Normalized. 313.5000 H L3n =0.3333F 1Ω A2 A3 V2 1Ω A4 + Vout (sn ) 1/R 4n L3n=0. (c.A FifthOrder. 314a.22 nF 10 kΩ A4 7. 314c. Frequency denormalizing this filter by 2πx103 and impedance denormalizing by 104 (arbitrarily chosen) gives the realization of Fig.ECE 6414 .Results of Ex. Give a realization and the value of all components.5F 1Ω 1Ω C2n= 1.
Give a schematic and the value of all components.09011/2πx106 = 0. 313b as the realization because it has one less op amp. the state variables are realized by differentiators rather than integrators if the state variables are the currents in the series elements and the voltages across the shunt elements. L1n = 2.1349 H. and R6n = 1 Ω.Continuous Time Filters (P. C2 = C2n/2πx106 = 1. and bandstop ladder filters can also be designed using activeRC elements. lowpass.09011/2πx106 = 0.340 µF.174µF.Chapter 3 Page 322 A fifthorder. Allen) . Chebyshev filter with a 1 dB ripple in the passband is to be designed for a cutoff frequency of 1 kHz. C4n = 1. in the highpass ladder filters.478 µF. The component values above are directly substituted into Fig. RLC ladder filter elements of R0n = 1 Ω. The input voltage source is assumed to have a source resistance of 1 kΩ. ladder realization. The values of the capacitor of each integrator. C3 = L3n/2πx106 = 3. Solution From Table 32. 313b to achieve a normalized. Unfortunately. However. The design methods follow a simular procedure as for the lowpass filter.09011.09011 F. Note that we have reversed the order of the subscripts in Table 32 to correspond to that of Fig.ECE 6414 .0009 H.1349/2πx106 = 0.340 µF. the equations are not as straightforward to . To get the denormalized filter. Integrator realization of the state variables is possible if the state variables are the voltages across the series elements and the currents through the shunt elements. Other Types of Ladder Filters using ActiveRC Elements Highpass. Let us again select Fig. we select R = 1000 Ω and frequency denormalize by 2πx103.1349 H. activeRC. from input to output are C1 = L1n/2πx106 = 2.1349/2πx106 = 0. we get the normalized. lowpass. L5n = 2. C2n = 1.174µF. L3n = 3. bandpass. Design an activeRC ladder realization of this filter.0009/2πx106 = 0. and C5 = L5n/2πx106 = 2. lowpass. C4 = C4n/2πx106 = 1. 310 (although in this particular case it makes no difference).
BW 1 Cjbn = . Allen) . New York. the bandpass equivalent is the parallel RLC circuit shown in Fig. C kln. and a series inductor. lowpass to normalized. The values of the series normalized bandpass components are ωr Ljbn = BW Ljln . Ljln. and a shunt capacitor. Rkn. Allen and E. Switched Capacitor Circuits. bandpass transformation of Fig.) shunt ladder element after the normalized.(a.) Series ladder element and (b. Rjn (Rkn) is zero (infinity) except for the series or shunt elements which include the terminating resistors of the ladder filter. If the kth shunt element consists of a shunt resistor. 1984. ωr Cjln Ik1 (sbn ) + Vj+1 (snb ) + Zkn (sbn ) Vk (sbn ) Rkn Lkbn Ckbn Ik+1 (sbn ) + Vk (sbn ) (345) Ij (sbn ) Rjn + Vj1 (snb )  Ljbn Cjbn Series Element Yjn (sbn ) (a. Once a normalized. is the jth series element consists of a series resistor.Sa nchezSinencio. 38. We will illustrate the design of bandpass ladder filters using activeRC elements because it is a straightforward extension of the lowpass ladder filter design.ECE 6414 . lowpass ladder filter has been found which meets the bandpass filter specifications.Continuous Time Filters (P.) Shunt Element (b.) Figure 315 . R jn.Chapter 3 Page 323 write as for the lowpass case. Van Nostrand Reinhold ' Company. 315a. ωr Ljln (344) The values of the shunt normalized components are ωr Ckbn = BW Cjln . become the series RLC circuit shown in Fig. we apply the element transformations illustrated in Fig. Examples of highpass ladder filter design can be found in other references†. Normally. Inc. 38. 4. † P. Chapt. Thus.E. BW 1 Ljbn = .. . 315b.
315a and the impedance. Zkn. Allen) . (349) reduces to Vj (sbn) = ' LR ' sbn jbn sb2n + L 1C jbn jbn [V (s )V (s )] j1 bn j+1 bn . (347) 1 2 1 s sb n + RknCkbn b n + LjbnCjbn (346) These drivingpoint functions can be turned into voltage transfer functions using the concept of voltage analogs for the currents Ij and Ik.Continuous Time Filters (P. (346) by R' gives ' Vj (sbn) R'Ij(sbn) TYj(sbn) = V (s ) = V (s )V (s ) = j bn j1 bn j+1 bn or ' Vj (sbn) R ' s Ljbn bn = Rjn 1 [Vj1(sbn)Vj+1(sbn)] 2 sb n + Ljbns b n + LjbnCjbn R ' s Ljbn bn 1 2 Rjn sb n + Ljbn s b n + LjbnCjbn (348) . These drivingpoint functions can be written as 1 s Ij(sbn) Ij(sbn) Ljbn bn Yjn(sbn) = V (s ) = V (s )V (s ) = j bn j1 bn j+1 bn 1 2 Rjn sb n + Ljbn s b n + LjbnCjbn and Vk(sbn) Vk(sbn) Zkn(sbn) = I (s ) = I (s )I (s ) = k bn k1 bn k+1 bn 1 s Ckbn bn . (347) by 1/R' gives Vk(sbn) TZk(sbn) = R'Ik(sbn) = 51) or 1 s Vk(sbn) R'Ckbn bn = (3' ' 1 2 1 s Vk1(sbn)Vk+1(sbn) sb n + RknCkbn b n + LjbnCjbn . (349) If Rjn = 0.Chapter 3 Page 324 Each element of the bandpass filter can be realized by considering the admittance. (350) Multiplying Eq.ECE 6414 . then Eq. Yjn. of Fig. 315b. of Fig. Multiplying Eq.
315a for the jth series elements. (349) or Eq. then Eq.th element Figure 316 .ActiveRC realization of either Fig. (352) If Rkn = ∞.Continuous Time Filters (P. bandpass. 315 by the TowThomas circuit of Fig.ECE 6414 . ladder filters. (353) The secondorder. (352) reduces to 1 R'Ckbnsbn = 2 sb n + L 1C Vk(sbn) jbn jbn [V ' (s ) V ' (s ) ] k1 bn k+1 bn . 315b (replace subscript j by k and interchange primes). For the cases of Eq. (352). 316 shows the general secondorder realization of either Fig. Fig. Allen) . 123 is a perfect realization of Eq. Rj3 Rj4 Rj1 Vj1 (snb ) Vj+1 (snb ) Rj1 Cj1 A1 j. Rj4 = ∞. The modification necessary is to add another input which is simple to do because the op amps have their positive input terminals grounded. 316. An example will illustrate the approach to designing activeRC bandpass. TowThomas filter of Fig. R R A3 Cj2 A2 V'j (snb ) +V'j (snb ) Rj2 . Table 33 shows the design relationships for the implementation of Fig. (353). 315a or Fig. (350) and Eq.Chapter 3 Page 325 Vk(sbn) 1 s R'Ckbn bn [V ' (sbn) V ' (sbn) ] = k+1 sb2n + 1 s b n + 1 k1 LjbnCjbn RknCkbn . The realization of the kth shunt element is exactly the same except for the subscript j. The availability of both positive and negative outputs simplifies the bandpass realization over the previous lowpass realizations.
Continuous Time Filters (P.Chapter 3 Page 326 Parameters of Fig. a convenient value: Cj1 = Cj2 or Ck1 = Ck2 Choose Cj1 = Cj2 = C or Ck1 = Ck2 = C. with all component values. doublyterminated. centered geometrically about 1 kHz. 315b by the activeRC filter of Fig. 316 Rj1 or Rk1 Design of Fig. 316. Solution The first thing we must do is find the bandwidth of the bandpass filter.Design relationships for the implementation of Fig. ladder filter using activeRC elements. bandpass. Example 38 . or equivalent. Ladder ActiveRC Filter Design a sixthorder. 2 kHz. Combining these 2 relationships gives fBP1 = 1/ 2 kHz and fPB2 = bandwidth is 1/ 2 kHz. Butterworth. Assume the passband is an octave. 315a or Fig. 315b R Rj1 = R ' Ljbn Cjbn Ljbn Cjbn R k1 = R'R Ckbn Lkbn Ckbn Lkbn Rj4 or Rk4 R Rj4 = Rjn Rk4 = RRkn Choose Rj2 = Rj3 = R or Rk2 = Rk3 = R. We know from the specification that fPB2 = 2fPB1.ECE 6414 . Thus the . Allen) . a convenient value: Rj2 = Rj3 or Rk2 = Rk3 Rj2 = Rj3 = Cj1 = Cj2 = LjbnCjbn R Ck1 = Ck2 = LkbnCkbn R LjbnCjbn C Rk2 = Rk3 = LkbnCkbn C R' is the scaling resistance of the voltage analog concept and is normally 1 Ω Table 33 .Design of a Butterworth. 315a Design of Fig. We also know that 1 kHz = fBP2fBP1 . Bandpass. Impedance denormalize by a factor of 104 and give a schematic.
(344) and (345) to Fig. For the third element. and C1bn = 1/ 2 F. C2n = 2 F.ECE 6414 . and L2bn = 1/2 2 H. which is shunt we choose C21 = C22 = 1F to get R22 = R23 = 1 Ω. L2bn = (BW. For the second element. Z2n. 316. Other examples of activeRC bandpass ladder filter design can be found in the problems. and R24 = ∞. which is series we choose C11 = C12 = 1 F to get R12 = R13 = 1 Ω. If we frequency denormalize by 2πx103 and impedance denormalize by 104 we get the realization of Fig. 317b. Y3n. we don't have to reorder the subscripts but can use them as they are. Therefore the values of Y1n are R1n = 1 Ω. The above example illustrates the general method by which bandpass ladder filters are realized using activeRC elements. Y1n. 317a. lowpass filter is shown in Fig. ωr. L1bn = 2 H. If we apply the normalized. secondorder active filters of the form given in Fig. C1bn = (BW/ωr/)(1/L1ln) = 1/ 2 F . R21 = 2 2 Ω. The values for Y3n are R3n = 1 Ω. The values for Z2n are R2n = ∞. The resulting.Continuous Time Filters (P. L3bn = (ωr/BW)L3ln = ( 2 )(1) = 2 H. Allen) . lowpass to normalized. 317a we get the following normalized bandpass elements of Fig. bandpass transformation of Eqs. doublyterminated Butterworth approximation we find that L1n = 1 H. Since the filter is symmetrical. C2bn = 2 2 F. and C3bn = 1/ 2 F. 317c. For the first element. . We will select R = R' = 1 Ω. L3bn = 2 H. and L3n = 1 H.Chapter 3 Page 327 From Table 32. we use the design relationships of Table 33 to design three. and R11 = R14 = 2 Ω. the values are identical to those for Yn1. normalized. and C3bn = (BW/ωr)(1/L3ln) = 1/ 2 F.ωr)(1/C2ln) = 1/2 2 H. Finally. C2bn = (ωr/BW)C2ln = ( 2 )(2) = 2 2 F. This method is applicable to any bandpass filter whose passband and stopband are geometrically centered about the frequency. for a thirdorder. L1bn = (ωr/BW)L1ln = (R(2))(1) = 2 H.
14k Ω Vin(s) V' 1 10 kΩ Vout 10 k Ω 10 kΩ A33 Vout(s) V 2 A31 A11 Y1 Z2 (c. The second approach was the design of ladder filters which has been the subject of this section.91nF R11 = R14 = 14.) bandpass ladder filter.) Y3 Figure 317 . The design of highDenormalized. These filter structures are the result of network synthesis methods and are widely tabulated for the designer.91nF 28. bandpass ladder filter. (a. ladder filter. The first approach was the cascade of first.28kΩ V2 10 kΩ A13 R21 = 28. activeRC ladder filter of Ex. lowpass RLC ladder filter.91nF R32 = 10kΩ A12 A22 A32 R11 = 14. 2. As in Sec.91nF R34 = 14.91nF R12 = 10kΩ R21 = C21 = 15. we restricted ourselves to filters whose zeros were at infinity or the origin of the complex frequency plane.) Y1n C 1bn=1/ 2F L3bn = 2 H R0n =1Ω + Vin (sbn ) L1bn= 2 H C2bn = 2 2F Z2n (b. Summary The block diagram in Fig.) Normalized. lowpass. The starting point of the ladder active filter design is with the normalized.14kΩ 15.14kΩ C31 = R31 = 14.or secondorder stages and was covered in the previous section. Allen) .Continuous Time Filters (P. 21 illustrated the two general design approaches normally used to design active filters.) Normalized.Bandpass. (c. Basically. 38.28kΩ A21 10kΩ 10kΩ A23 R23 = 10kΩ C22 = 15. (b.ECE 6414 .91nF R22 = 10kΩ C 3bn=1/ 2 F L2bn = 1/ 2 2 H R 4n =1Ω R33 = 10kΩ C32 = 15. activeRC . the cascade approach is easier to tune but is more sensitive to component tolerance than the ladder filters.14k Ω C11 = 15.) R13 = 10kΩ C12 = 15.14k Ω 14.Chapter 3 Page 328 + R 0n =1 Ω L1ln =1 H C 2ln Vin(sln) =2F  L3ln =1 H R 4n = 1Ω + Vout (sln ) Y3n + Vout (sbn )  (a.
These secondorder. By correctly selecting the state variables and using the concept of voltage analog for current. The value of the inductor is . • The normalized lowpass to normalized highpass frequency transformation turns inductors (capacitors) into capacitors (inductors) whose value is the reciprocal of the lowpass component value. lowpass TowThomas filters. We saw that highpass. Some the important points of this section are summarized below for the convenience of the reader. secondorder. we were able to synthesize the state variables using summing integrators. bandpass. TowThomas filters. To achieve an activeRC realization of the RLC ladder filters. not covered here. activeRC ladder filters required a little cleverness to be able to express all of the state variables so that they could be realized by integrators. One important constraint to remember is that the bandpass and bandstop filters designed by these methods must have passbands and stopbands geometrically centered about a common frequency. Using simple normalizations and frequency transformations permits the design of complex filters. Each element of the normalized. lowpass filters. we introduced the concept of state variable.ECE 6414 . Other design methods.Continuous Time Filters (P. • The normalized lowpass to normalized bandpass frequency transformation turns an inductor into an inductor in series with a capacitor. lowpass RLC filter. Allen) .Chapter 3 Page 329 pass and bandpass filters was achieved by direct application of the normalized frequency transformations to the elements of the normalized. bandpass circuits were easily realizable as secondorder. lowpass RLC resulted in a secondorder bandpass circuit when the normalized lowpass to normalized bandpass frequency transformation was applied. One of the themes that hopefully has become evident to the reader is how filter design all builds from the normalized. will allow the design of filters not subject to this constraint. The resulting circuits were very similar to interconnected. • Sensitivity is a measure of some aspect of the filter performance on the components of the filter.
whose value is (inductors) whose value is the reciprocal of the lowpass component value. .ECE 6414 . • The TowThomas secondorder realization becomes a very useful activeRC filter for realizing RLC ladder filters. Allen) .Chapter 3 Page 330 (ωr/BW)Lln and the value of the capacitor is (BW/ωrLln). • In order to realize the state variables by integrators. the current through an inductor and the voltage across a capacitor should be chosen as state variables. • The normalized lowpass to normalized bandpass frequency transformation turns a capacitor into a capacitor in parallel with an inductor. The value of the capacitor is (ωr/BW)Cln and the value of the inductor is (BW/ωrCln).Continuous Time Filters (P.