You are on page 1of 30

# ECE 6414 - Continuous Time Filters (P.

Allen) - Chapter 3 Page 3-1
3 - HIGHER-ORDER LADDER PASSIVE AND ACTIVE FILTERS
In Fig. 2-1, two approaches to designing active filters were shown. One of the
approaches uses cascaded, first- and second-order 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 higher-order active filters. The cascade approach
is easier to design and tune but does not have the insensitivity to component variation found
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
. (3-1)
For example, the relative sensitivity of the pole-Q of the Tow-Thomas, low-pass realization
of Fig. 1-16 to R
1
can be found by letting p = Q and x = R
1
and applying Eq. (3-1) to Eq.
(1-59) 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 . (3-2)
ECE 6414 - Continuous Time Filters (P. Allen) - Chapter 3 Page 3-2
Furthermore, we can show that the sensitivity of the pole-Q of Fig. 1-16 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 pole-Q
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 higher-order active filter design
shown in Fig. 2-1. 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
filter. It is almost impossible to tune a ladder filter because no single component dominates
the performance of the filter.
RLC, low-pass ladder filters are the result of modern network synthesis techniques
and are based on techniques well known in circuit theory

. The resulting realizations of

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 3-3
input of the filter. Fig. 3-1 shows the form for a singly-terminated 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
N-1,n
C
3n
C
1n V
in
(s
n
)
(a.)
(b.)
V
out
(s
n
)
+
-
+
-
C
2n C
N-1,n
L
N,n
L
3n L
1n
V
in
(s
n
)
V
out
(s
n
)
1Ω
Figure 3-1 - Singly-terminated, RLC filters. (a.) N even. (b.) N odd.
The RLC ladder filters of Fig. 3-1 are normalized to a passband of 1 rps. The
denormalizations of Table 2-1 are applicable to the elements of Fig. 3-1. Fig. 3-2 shows
the normalized ladder filters for doubly-terminated, RLC filters. It is seen that these filters
are similar to those of Fig. 3-1 except for a series source resistance.
(a.)
(b.)
+
-
+
-
L
2n L
N,n
C
N-1,n
C
3n
C
1n V
in
(s
n
)
V
out
(s
n
)
R
1Ω
+
-
C
2n C
N-1,n
L
N,n
L
3n L
1n
V
out
(s
n
)
+
-
V
in
(s
n
)
R
1Ω
Figure 3-2 - Doubly-terminated, 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. 3-1 and 3-2. Each of the many different types of filter
ECE 6414 - Continuous Time Filters (P. Allen) - Chapter 3 Page 3-4
approximations have been tabulated for values of N up to 10 or more

. Tables 3-1 and 3-2
are typical of this tabularized information for the Butterworth and 1-dB Chebyshev
approximation for the singly-terminated and doubly-terminated, RLC filters of Figs. 3-1
and 3-2.
Use these component designations for even order of Fig. 3-1a and Fig. 3-2a.
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 1-dB 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. 3-1b.
Table 3-1 - Normalized component values for Fig. 3-1 for the Butterworth and Chebyshev
singly-terminated, RLC filter approximations.
Example 3-1 - Use of the Table 3-1 to Find a Singly-Terminated, RLC Low-pass Filter
Find a singly-terminated, normalized, RLC filter for a 4th-order Butterworth filter.
Solution
Using Table 3-1 and using the component designations at the top of the table gives
Fig. 3-3.

L. Weinburg, Network Analysis and Synthesis, McGraw-Hill 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 3-5
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 3-3 - Fourth-order, Butterworth, normalized low-pass RLC filter realization.
Use these component designations for even order of Fig. 3-2a, 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 1-dB 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. 3-2b, R = 1Ω.
Table 3-2 - Normalized component values for Fig. 3-2 for the Butterworth and 1-dB
Chebyshev doubly-terminated, RLC approximations.
We note that no solution exists for the even-order cases of the doubly-terminated,
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 short-circuits and the capacitors are open-circuits.
Example 3-2 - Use of Table 3-2 to Find a Doubly-Terminated, RLC Low-pass Fi lter
Find a doubly-terminated, RLC filter for a fifth-order 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 3-6
Using Table 3-2 and using the component designations at the bottom of the table
gives Fig. 3-4.
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 3-4 - Fifth-order, doubly-terminated, normalized, Chebyshev low-pass RLC filter
realization.
Low-Pass, 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, low-pass, filter approximation
which satisfies the filter specifications and then denormalize the frequency and the
impedance. An example will illustrate the procedure.
Example 3-3 - Design of a Passive, Low-Pass Filter
A low-pass 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. (2-11)
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 3-1 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 2-1 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 3-7
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. 3-5.
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 3-5 - Seventh-order, Butterworth filter realization of Ex. 3-3.
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. 3-3. 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 low-resistance 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 3-8
incorporated into the design phase. This results in a more complex filter design procedure
which is beyond the scope of our treatment.
The design of high-pass and bandpass, RLC ladder filters is made simple by the
frequency transformations of the previous section. The frequency transformation from the
normalized, low-pass to normalized high-pass was given by Eq. (2-23). If we apply this
transformation to an inductor of a normalized, low-pass realization, we obtain
s
ln
L
ln
=
.
|
,
`
1
s
hn
L
ln
=
1
s
hn
C
hn
. (3-3)
Similarly, if applying the transformation to a capacitor, C
ln
, of a normalized, low-pass
realization, we obtain
1
s
ln
C
ln
=
.
|
,
`
s
hn
1

1
C
ln
= s
hn
L
hn
. (3-4)
From Eqs. (3-3) and (3-4), we see that the normalized, low-pass to normalized, high-pass
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. 3-6 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 3-6 - Influence of the normalized, low-pass to normalized, high-pass frequency
transformation on the inductors and capacitors.
From the above results, we see that to achieve a normalized, high-pass 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 3-9
transformed, then both a frequency and impedance denormalization can be performed to
result in the final filter design.
Example 3-4 - Design of a High-Pass, 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 50-ohm 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. (2-26) that Ω
n
= 2. Using Eq. (2-11) gives an N = 8 (T
SB
= -
48.2 dB) for a Butterworth approximation. Eq. (2-19) 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 singly-terminated filter because the
doubly-terminated 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, low-pass to normalized, high-pass 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 50-ohm 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. 3-7 shows the final realization for this example.
ECE 6414 - Continuous Time Filters (P. Allen) - Chapter 3 Page 3-10
+
-
+
-
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 3-7 - Denormalized, Chebyshev, high-pass filter realization for Ex. 3-4.
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 low-pass filter rather than the roots. The first step in designing a bandpass
filter is to apply the bandpass normalization of Eq. (2-32) to the components of the low-
pass filter. Next, we apply the normalized, low-pass to normalized, bandpass
transformation of Eq. (2-33). Both of these steps are incorporated in Eq. (2-31). 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, low-pass filter. Let us
simultaneously apply the bandpass normalization and the frequency transformation by
using Eq. (2-31). 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
. (3-5)
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

b
(3-6)
and
ECE 6414 - Continuous Time Filters (P. Allen) - Chapter 3 Page 3-11
C
bn
=
.
|
,
`
BW
ω
r

1
L
ln
=

b
L
ln
. (3-7)
Now we apply Eq. (2-31) 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
. (3-8)
From Eq. (3-8), we see that the bandpass normalization and frequency transformation takes
a capacitor, C
ln
, in a low-pass 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

b
(3-9)
and
L
bn
=
.
|
,
`
BW
ω
r

1
C
ln
=

b
C
ln
. (3-10)
Eqs. (3-6), (3-7), (3-9), and (3-10) are very important in the design of RLC bandpass
filters and are illustrated in Fig. 3-8.
Normalized
Low-Pass
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 3-8 - Illustration of the influence of the normalized, low-pass to the normalized,
bandpass transformation of Eq. (2-31) on an inductor and capacitor of a low-pass filter.
ECE 6414 - Continuous Time Filters (P. Allen) - Chapter 3 Page 3-12
Once the normalized, low-pass 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 3-5 - Design of an RLC Bandpass Filter
A 1-dB 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. (2-41), we know that Ω
n
= 5. Substituting N = 3 into
Eq. (2-19) gives an attenuation of 47.85 dB. The values of the third-order, 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. (2-35), is Ω
b
= 10/100 = 0.1. Applying the
normalized, low-pass 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. 3-9.
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 3-9 - Chebyshev, bandpass realization for Ex. 3-5.
Design of Other Types of RLC Filters
ECE 6414 - Continuous Time Filters (P. Allen) - Chapter 3 Page 3-13
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. 2-18. Normalized, low-pass RLC filters have been developed and tabulated
for this type of filter. They are similar in form to the ladder filters of Figs. 3-1 and 3-2
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, low-pass to normalized, high-pass
transformation to the low-pass normalized realization. Next, the low-pass to bandpass
transformation of Eq. (2-31) is applied to the normalized high-pass 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. 3-5 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. 3-9 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, low-pass to
normalized, high-pass transformation (see PA3-1).
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 low-pass ladder filter. Let us demonstrate the approach
by using an example. Consider the doubly-terminated, fifth-order, RLC, low-pass ladder
filter of Fig. 3-10. 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 low-pass structures in this discussion.
ECE 6414 - Continuous Time Filters (P. Allen) - Chapter 3 Page 3-14
+
-
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 3-10 - A fifth-order, low-pass, normalized RLC ladder filter.
The first step in realizing the RLC filter of Fig. 3-10 by active RC elements is to
assign a current I
j
to every j-th series element (or combination of elements in series) of the
ladder filter and a voltage V
k
to every k-th shunt element (or combination of elements in
shunt) ofthe ladder filter. These currents and voltages for the example of Fig. 3-10 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. 3-10 and
write the loop equation
V
in
(s) - I
1
(s)R
0n
- sL
1n
I
1
(s) - V
2
(s) = 0 . (3-11)
Next, we write the nodal equation
I
1
(s) - sC
2n
V
2
(s) - I
3
(s) = 0 . (3-12)
We continue in this manner to get the following state equations.
V
2
(s) - sL
3n
I
3
(s) - V
4
(s) = 0 (3-13)
I
3
(s) - sC
4n
V
4
(s) - I
5
(s) = 0 (3-14)
and
V
4
(s) - sL
5n
I
5
(s) - R
6n
I
5n
(s) = 0 (3-15)
Eqs. (3-11) through (3-15) constitute the state equations which completely describe the
ladder filter of Fig. 3-10. A supplementary equation of interest is
V
out
(s) = I
5
(s)R
6n
. (3-16)
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
(3-17)
ECE 6414 - Continuous Time Filters (P. Allen) - Chapter 3 Page 3-15
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 active-RC implementation of the ladder filter. Now if for every
current in the state equations of Eq. (3-11) through Eq. (3-15) 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 (3-18)
.

|
,

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

|
,

`
V
'
3
( s )
R'
= 0 (3-19)
V
2
(s) - sL
3n
.

|
,

`
V
'
3
( s )
R'

- V
4
(s) = 0 (3-20)
.

|
,

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

|
,

`
V
'
5
( s )
R'
= 0 (3-21)
and
V
4
(s) -
.

|
,

`
V
'

5
( s )
R'
(sL
5n
+ R
6n
) = 0 (3-22)
The next step is to use the 5 equations of Eqs. (3-18) through (3-22) 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) (3-23)
V
2
(s) =
1
sR'C
2n
[V
'
1
(s) - V
'
3
(s) ] (3-24)
V
'
3
(s) =
R'
sL
3n
[V
2
(s) - V
4
(s)] (3-25)
V
4
(s) =
1
sR'C
4n
[ V
'
3
(s) - V
'
5
(s) ] (3-26)
and
V
'
5
(s) =
R'
sL
5n
[V
4
(s) -
R
6n
R'
V
'
5
(s) ] . (3-27)
ECE 6414 - Continuous Time Filters (P. Allen) - Chapter 3 Page 3-16
However, we would prefer to have the variable V
out
(s) used in place of V
'
5
(s) . From Eq.
(3-15) we get
V
out
(s) =
.
|
,
`
R
6n
R'
V
'
5
(s) . (3-28)
Combining Eqs. (3-26) and (3-27) with (3-28) gives
V
4
(s) =
1
sR'C
4n
[ V
'

3
(s) -
.
|
,
`
R'
R
6n
V
out
(s)] (3-29)
V
out
(s) =
R
6n
sL
5n
[V
4
(s) - V
out
(s)] . (3-30)
ECE 6414 - Continuous Time Filters (P. Allen) - Chapter 3 Page 3-17
Design of Low-Pass, Ladder Filters using Active-RC Elements
Finally, we are ready to synthesize an active-RC realization of Fig. 3-10. This is
done by realizing that Eqs. (3-23), (3-24), (3-25), (3-29) and Eq. (3-30) 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. 3-11.
A1 A2
R
R
R R
V
j
V
k
-V
k -V
l
C
k
Figure 3-11 - Positive integrator realization of the k-th integrator.
We can show that the output of Fig. 3-11 is given as
V
k
(s) =
1
sRC
k
[V
j
(s) - V
l
(s)] (3-31)
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. 3-11 works well for Eqs. (3-24), (3-25), and (3-29). However, a different
approach is necessary for Eqs. (3-23) and (3-30) because the variable is a function of itself.
Solving for V
'
1
(s) and V
out
(s) in Eqs. (3-23) and (3-31) gives
V
'
1
(s) =
.

|
,

`
R'
L
1n
s +
R
0 n
L
1n
{ }
V
in
(s) + [-V
2
(s)] (3-32)
and
ECE 6414 - Continuous Time Filters (P. Allen) - Chapter 3 Page 3-18
V
out
(s) =
.

|
,

`
R
6n
L
5n
s +
R
0 n
L
5n
V
4
(s) , (3-33)
respectively. We see that Eqs. (3-32) and (3-33) are nothing more than the first-order,
low-pass filter of Fig. 4.2-6c preceeded by a gain of -1 stage which also serves as a
summer for Eq. (3-32). The implementation of Eq. (3-32) is shown in Fig. 3-12.
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 3-12 - Realization of Eq. (3-32).
A similar realization can be found for Eq. (3-33). This realization will only have one input
to the inverter.
Unfortunately, the straight-forward application of Figs. 3-11 and 3-12 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. (3-23), (3-24),
(3-25), (3-29) and Eq. (3-30) is shown below.
V
'
1
(s) =
R'
sL
1n

]
]
V
in
(s) - V
2
(s) -
.
|
,
`
R
0n
R'
V
'
1
(s) (3-34)
-V
2
(s) =
-1
sR'C
2n
[V
'
1
(s) - V
'
3
(s) ] (3-35)
-V
'
3
(s) =
R'
sL
3n
[-V
2
(s) + V
4
(s)] (3-36)
V
4
(s) =
-1
sR'C
4n
[- V
'

3
(s) +
.
|
,
`
R'
R
6n
V
out
(s)] (3-37)
and
V
out
(s) =
R
6n
sL
5n
[V
4
(s) - V
out
(s)] . (3-38)
Another modification of the state equations is given as
ECE 6414 - Continuous Time Filters (P. Allen) - Chapter 3 Page 3-19
-V
'
1
(s) =
-R'
sL
1n

]
]
V
in
(s) - V
2
(s) -
.
|
,
`
R
0n
R'
V
'
1
(s) (3-39)
-V
2
(s) =
1
sR'C
2n
[-V
'
1
(s) + V
'
3
(s) ] (3-40)
V
'
3
(s) =
-R'
sL
3n
[-V
2
(s) + V
4
(s)] (3-41)
V
4
(s) =
1
sR'C
4n
[ V
'

3
(s) -
.
|
,
`
R'
R
6n
V
out
(s)] (3-42)
and
-V
out
(s) =
-R
6n
sL
5n
[V
4
(s) - V
out
(s)] . (3-43)
Fig. 3-13a shows the realization of the state equations of Eqs. (3-34) through (3-38).
Fig. 3-13b shows the realization of the state equations of Eqs. (3-39) through (3-43). The
second modification has one less op amp but only -V
out
(s) is available. Fig. 3-13a saves
two op amps over the case where every integrator consists of two op amps and Fig. 3-13b
saves three op amps over the case where every integrator consists of two op amps. It is
interesting to note that Fig. 3-13 turns out to be the interconnection of four, second-order,
Tow-Thomas filters of Fig. 1-16. The inside filters have a Q of infinity because neither
integrator is damped. The intercoupled second-order 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 Leap-Frog or Active-Ladder Synthesis," Wireless
World, vol. 76, July 1970, pp. 341-345.
ECE 6414 - Continuous Time Filters (P. Allen) - Chapter 3 Page 3-20
(
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 3-21
Example 3-6 - Design of a Low-Pass, Butterworth Ladder Filter using the Active-RC
Elements
Design a third-order, Butterworth, low-pass filter having a passband frequency of 1
kHz using the RC-active ladder realization of Fig. 3-13. Assume that the source resistance
of the input voltage is zero. Give a realization and the value of all components.
Solution
The normalized, low-pass ladder filter is obtain from Table 3-1 and is shown in Fig.
3-14a. Note that we have reversed the order of the subscripts to correspond with Fig. 3-
10. Let us choose Fig. 3-13b as the realization form resulting in Fig. 3-14b 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. 3-14c.
+
-
+
-
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 3-14 - Results of Ex. 3-6. (a.) Normalized, low-pass RLC filter. (b.) Normalized,
low-pass, active-RC filter. (c.) Denormalized, low-pass, active-RC filter.
Example 3-7 - A Fifth-Order, 1 dB Chebyshev, Low-Pass, Active-RC Filter Design
ECE 6414 - Continuous Time Filters (P. Allen) - Chapter 3 Page 3-22
A fifth-order, low-pass, Chebyshev filter with a 1 dB ripple in the passband is to be
designed for a cutoff frequency of 1 kHz. Design an active-RC 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 3-2, we get the normalized, low-pass, 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 3-2 to
correspond to that of Fig. 3-10 (although in this particular case it makes no difference). Let
us again select Fig. 3-13b as the realization because it has one less op amp. The
component values above are directly substituted into Fig. 3-13b to achieve a normalized,
low-pass, active-RC, 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 Active-RC Elements
High-pass, bandpass, and bandstop ladder filters can also be designed using active-
RC elements. The design methods follow a simular procedure as for the low-pass filter.
Unfortunately, in the high-pass 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 straight-forward to
ECE 6414 - Continuous Time Filters (P. Allen) - Chapter 3 Page 3-23
write as for the low-pass case. Examples of high-pass ladder filter design can be found in
other references

.
We will illustrate the design of bandpass ladder filters using active-RC elements
because it is a straight-forward extension of the low-pass ladder filter design. Once a
normalized, low-pass ladder filter has been found which meets the bandpass filter
specifications, we apply the element transformations illustrated in Fig. 3-8. 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. 3-15a. If the k-th 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. 3-15b. 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
. (3-44)
The values of the shunt normalized components are
C
kbn
=
.
|
,
` ω
r
BW
C
jln
, L
jbn
=
.
|
,
`
BW
ω
r

1
C
jln
. (3-45)
(b.) (a.)
L
jbn
C
jbn
Series Element
R
jn
+
-
+
-
V
j-1
(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
k-1
(s
bn
) I
k+1
(s
bn
)
V
k
(s
bn
)
+
-
V
k
(s
bn
)
Figure 3-15 - (a.) Series ladder element and (b.) shunt ladder element after the normalized,
low-pass to normalized, bandpass transformation of Fig. 3-8.

P.E. Allen and E.Sa' nchez-Sinencio, Switched Capacitor Circuits, Chapt. 4, Van Nostrand Reinhold
Company, Inc., New York, 1984.
ECE 6414 - Continuous Time Filters (P. Allen) - Chapter 3 Page 3-24
Each element of the bandpass filter can be realized by considering the admittance,
Y
jn
, of Fig. 3-15a and the impedance, Z
kn
, of Fig. 3-15b. These driving-point functions
can be written as
Y
jn
(s
bn
) =
I
j
(s
bn
)
V
j
(s
bn
)
=
I
j
(s
bn
)
V
j-1
(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
(3-46)
and
Z
kn
(s
bn
) =
V
k
(s
bn
)
I
k
(s
bn
)
=
V
k
(s
bn
)
I
k-1
(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
. (3-47)
These driving-point functions can be turned into voltage transfer functions using the
concept of voltage analogs for the currents I
j
and I
k
. Multiplying Eq. (3-46) by R' gives
T
Yj
(s
bn
) =
R'I
j
(s
bn
)
V
j
(s
bn
)
=
V
'
j
(s
bn
)
V
j-1
(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
(3-48)
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
j-1
(s
bn
)-V
j+1
(s
bn
)] . (3-49)
If R
jn
= 0, then Eq. (3-49) reduces to
V
'
j
(s
bn
) =
.

|
,

`
.
|
,
`
R'
L
jbn
s
bn
s
2
bn
+
1
L
jbn
C
jbn
[V
j-1
(s
bn
)-V
j+1
(s
bn
)] . (3-50)
Multiplying Eq. (3-47) by 1/R' gives
T
Zk
(s
bn
) =
V
k
(s
bn
)
R'I
k
(s
bn
)
=
V
k
(s
bn
)
V
'
k-1
(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 3-25
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
'
k-1
(s
bn
) -V
'
k+1
(s
bn
) ] . (3-52)
If R
kn
= ∞, then Eq. (3-52) reduces to
V
k
(s
bn
) =
.

|
,

`
.
|
,
`
1
R'C
kbn
s
bn
s
2
bn
+
1
L
jbn
C
jbn
[V
'
k-1
(s
bn
) -V
'
k+1
(s
bn
) ] . (3-53)
The second-order, bandpass, Tow-Thomas filter of Fig. 1-23 is a perfect realization
of Eq. (3-49) or Eq. (3-52). The modification necessary is to add another input which is
simple to do because the op amps have their positive input terminals grounded. Fig. 3-16
shows the general second-order realization of either Fig. 3-15a for the j-th series elements.
The realization of the k-th shunt element is exactly the same except for the subscript j. For
the cases of Eq. (3-50) and Eq. (3-53), R
j4
= ∞. The availability of both positive and
negative outputs simplifies the bandpass realization over the previous low-pass realizations.
A1 A3
R
R
j- th element
-V
j
'
(s
nb
)
+V
j
'
(s
nb
)
V
j-1
(s
nb
)
V
j+1
(s
nb
)
R
j1
R
j1
R
j2
R
j3
R
j4
C
j1
C
j2
A2
Figure 3-16 - Active-RC realization of either Fig. 3-15a or Fig. 3-15b (replace subscript j
by k and interchange primes).
Table 3-3 shows the design relationships for the implementation of Fig. 3-15 by the
Tow-Thomas circuit of Fig. 3-16. An example will illustrate the approach to designing
ECE 6414 - Continuous Time Filters (P. Allen) - Chapter 3 Page 3-26
Parameters of Fig. 3-16 Design of Fig. 3-15a Design of Fig. 3-15b
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 3-3 - Design relationships for the implementation of Fig. 3-15a or Fig. 3-15b by the
active-RC filter of Fig. 3-16.
Example 3-8 - Design of a Butterworth, Bandpass, Ladder Active-RC Filter
Design a sixth-order, Butterworth, bandpass, doubly-terminated, ladder filter using
active-RC 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 3-27
From Table 3-2, for a third-order, doubly-terminated 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,
low-pass filter is shown in Fig. 3-17a.
If we apply the normalized, low-pass to normalized, bandpass transformation of Eqs.
(3-44) and (3-45) to Fig. 3-17a we get the following normalized bandpass elements of Fig.
3-17b. 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
=

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 3-3 to design three, second-order
active filters of the form given in Fig. 3-16. 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. 3-17c.
The above example illustrates the general method by which bandpass ladder filters are
realized using active-RC elements. This method is applicable to any bandpass filter whose
passband and stopband are geometrically centered about the frequency, ω
r
. Other examples
of active-RC bandpass ladder filter design can be found in the problems.
ECE 6414 - Continuous Time Filters (P. Allen) - Chapter 3 Page 3-28
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 3-17 - Bandpass, active-RC ladder filter of Ex. 3-8. (a.) Normalized, low-pass,
Summary
The block diagram in Fig. 2-1 illustrated the two general design approaches normally
used to design active filters. The first approach was the cascade of first- or second-order
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, low-pass
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 3-29
pass and bandpass filters was achieved by direct application of the normalized frequency
transformations to the elements of the normalized, low-pass RLC filter.
To achieve an active-RC 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, second-order, low-
pass Tow-Thomas filters. We saw that high-pass, active-RC 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, low-pass RLC resulted in a second-order
bandpass circuit when the normalized low-pass to normalized bandpass frequency
transformation was applied. These second-order, bandpass circuits were easily realizable
as second-order, bandpass, Tow-Thomas filters.
One of the themes that hopefully has become evident to the reader is how filter design
all builds from the normalized, low-pass 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
• Sensitivity is a measure of some aspect of the filter performance on the components
of the filter.
• The normalized low-pass to normalized high-pass frequency transformation turns
inductors (capacitors) into capacitors (inductors) whose value is the reciprocal of the
low-pass component value.
• The normalized low-pass 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 3-30

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 low-pass component value.
• The normalized low-pass to normalized bandpass frequency transformation turns a
capacitor into a capacitor in parallel with an inductor. The value of the capacitor is

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 Tow-Thomas second-order realization becomes a very useful active-RC filter for

ECE 6414 - Continuous Time Filters (P. Allen) - Chapter 3

Page 3-2

Furthermore, we can show that the sensitivity of the pole-Q of Fig. 1-16 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 pole-Q 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 higher-order active filter design shown in Fig. 2-1. 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.

† 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 3-3

input of the filter. Fig. 3-1 shows the form for a singly-terminated 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 3-1 - Singly-terminated, RLC filters. (a.) N even. (b.) N odd. The RLC ladder filters of Fig. 3-1 are normalized to a passband of 1 rps. The LN,n CN-1,n L3n C2n L1n 1Ω + Vout (sn ) LN,n CN-1,n L2n C3n C1n 1Ω + Vout (sn ) -

denormalizations of Table 2-1 are applicable to the elements of Fig. 3-1. Fig. 3-2 shows the normalized ladder filters for doubly-terminated, RLC filters. It is seen that these filters are similar to those of Fig. 3-1 except for a series source resistance. + Vin (sn ) (a.) + Vin (sn ) (b.) Figure 3-2 - Doubly-terminated, 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. 3-1 and 3-2. Each of the many different types of filter R LN,n CN-1,n L3n C2n L1n 1Ω + Vout (sn ) R LN,n CN-1,n L2n C3n C1n 1Ω + Vout (sn ) -

5155 0.4037 1.5908 1.7287 1. 1967.1736 0.7215 2.5529 1. 3-1 for the Butterworth and Chebyshev singly-terminated.6961 2.5088 1-dB ripple Chebyshev (1 rps passband) 1.2817 1.Normalized component values for Fig.5451 0. RLC filters of Figs. 1975.1111 1. 3-1b.4126 1. Table 3-1 . 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. McGraw-Hill Book Co.5307 0. 3-3. 3-2a.0550 1.1192 1.7021 2. 3-1 and 3-2.0118 1.7626 1. Allen) . RLC filter approximations.7071 1.0824 1.9093 1.5000 Butterworth (1 rps passband) 0.5576 0.6507 2. 3-1a 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 3-1 to Find a Singly-Terminated.1408 1. Tables 3-1 and 3-2 are typical of this tabularized information for the Butterworth and 1-dB Chebyshev approximation for the singly-terminated and doubly-terminated.5283 1.2588 0.6707 1.4654 0.3090 0. Network Analysis and Synthesis.ECE 6414 .0537 1. Krieger Publishing Co.Chapter 3 Page 3-4 approximations have been tabulated for values of N up to 10 or more† . RLC filter for a 4th-order Butterworth filter.2921 1. Solution Using Table 3-1 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 3-1 .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 Low-pass Filter Find a singly-terminated.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 1-dB ripple Chebyshev (1 rps passband) 1.3129 2. RLC Chebyshev approximations for R = 1 Ω.9319 1.9754 1.6629 1. 3-2b.1214 1.1192 2.5321 1. RLC approximations.9319 1.1111 0.1897 3. R = 1Ω.6180 1.9616 1. RLC filter for a fifth-order 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.Fourth-order. 3-2 for the Butterworth and 1-dB Chebyshev doubly-terminated. Allen) . RLC Low-pass Filter Find a doubly-terminated.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 3-2 to Find a Doubly-Terminated.Normalized component values for Fig.7820 1.9754 1. R = 1Ω.6180 0. normalized low-pass 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 3-2 .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 3-2 .1349 1.1192 3. We note that no solution exists for the even-order cases of the doubly-terminated.0236 2.4142 1.0936 1.3473 0.1111 1.0000 1.9080 1.5 at low frequencies where the inductors are short-circuits and the capacitors are open-circuits. 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 3-3 .7654 0.4142 0.3473 0. Butterworth.0000 0.1735 3.0000 1.5307 H L2n =1.5176 0. 3-2a. L2n C3n L4n C5n L6n C7n L8n C9n L10n 1.Chapter 3 Page 3-5 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 3-6 Using Table 3-2 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 3-3 . C4n = 1.14 dB for N = 7. . Low-Pass Filter A low-pass filter using the Butterworth approximation is to have the following specifications: TPB = -3 dB. From Table 3-1 for N = 7 for the Butterworth approximation we get L1n = 0. Chebyshev low-pass RLC filter realization.071 mH.6560F)/(πx109) = 209 pF. L5n = 1. fPB = 500 kHz.3972F)/(πx109) = 445 pF. and L7n = 1. In this case.Fifth-order. such as high frequency. (2-11) using ε = 1 to get TSB = -30.2225H)/(πx106) = 0. C2n = 0. Low-Pass.2225 H. Allen) . we use the denormalizations in the bottom row of Table 2-1 to get L1 = (103)(0. TSB = -40 dB. and TSB = -42. TSB = -36. low-pass.0550H)/(πx106) = 0. doubly-terminated.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 3-4 . 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. 3-4.

071 mH + Vin (s) C4 = 573 pF C4 = 445 pF C2 = 209 pF 1 kΩ + Vout (s) Figure 3-5 . 3-3.6588H)/(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 low-resistance wire around a cylindrical insulator. This The load resistor becomes 1 kΩ.7988F)/(πx109) = 573 pF. 3-5. The important parasitics are the capacitors from each node to ground and the resistance of the inductors. 3-3.528 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 3-7 L5 = (103)(1.Seventh-order. 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. high-pass RLC filter.Influence of the normalized. Lln Cln sln → s1 hn Chn = 1 Lln Lhn = 1 Cln Normalized HighPass Network Normalized LowPass Network Figure 3-6 . If we apply this transformation to an inductor of a normalized. of a normalized. with a capacitor. and replaces it by an inductor. (3-3) Similarly. the elements have been . From the above results. with an inductor. (3-3) and (3-4). low-pass 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. low-pass realization. whose value is 1/Lln and each capacitor. we replace each inductor. Once. Allen) .Continuous Time Filters (P. (3-4) From Eqs. if applying the transformation to a capacitor. 3-6 illustrates these important relationships. High-Pass. Lhn. high-pass frequency transformation takes an inductor. C ln. Lhn.Chapter 3 Page 3-8 incorporated into the design phase. we see that the normalized. RLC ladder filters is made simple by the frequency transformations of the previous section. low-pass to normalized high-pass was given by Eq. This transformation also takes a capacitor. whose value is 1/Cln. (2-23). Chn. RLC. Lln. Ladder Filter Design The design of high-pass 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 . high-pass frequency transformation on the inductors and capacitors. low-pass to normalized. low-pass 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 50-ohm 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 singly-terminated filter because the doubly-terminated filter has a maximum passband gain of -6 dB.ECE 6414 . Fig. Using Eq. Solution We know from Eq. . (2-26) that Ωn = 2.6 nF. low-pass 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 50-ohm cable and will serve as the load resistor. Therefore. (2-11) gives an N = 8 (TSB = 48.6652 = 0. Finally.5016 F. high-pass 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. (2-19) gives N = 5 (TSB = -45. L2h = 0. 3-7 shows the final realization for this example.Design of a High-Pass. L3ln = 1.6286 H.Chapter 3 Page 3-9 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 3-4 .

Denormalized. high-pass filter realization for Ex.ECE 6414 . RLC. bandpass transformation of Eq. (2-31). Let us first consider the inductor.Chapter 3 Page 3-10 C5h = 38. The first step in designing a bandpass filter is to apply the bandpass normalization of Eq. 3-4.Continuous Time Filters (P. and replaces it by an inductor. of a normalized. Allen) . low-pass to normalized. Let us simultaneously apply the bandpass normalization and the frequency transformation by using Eq. low-pass filter.6 nF + Vin (s) L2h = 110 µH L4h = 100 µH 50 ohm cable + Vout (s) - Figure 3-7 . Next. Lbn. (2-31). Lln. (3-5) 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 low-pass 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 (3-6) . 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. (2-33). Bandpass. (2-32) to the components of the lowpass filter.

(3-9). we see that the bandpass normalization and frequency transformation takes a capacitor.Continuous Time Filters (P. 3-8. (3-7). Cbn. Cln. (2-31) on an inductor and capacitor of a low-pass filter. . in parallel with an inductor. (3-8).Illustration of the influence of the normalized.  ωr  (3-9) (3-10) Eqs.ECE 6414 . and (3-10) are very important in the design of RLC bandpass filters and are illustrated in Fig. (3-6). (2-31) to a normalized capacitance. in a low-pass circuit and transfors to a capacitor. 1 s bn C bn + s L bn bn 1 (3-7) (3-8) 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 Low-Pass Network Figure 3-8 . bandpass transformation of Eq. Lbn.Chapter 3 Page 3-11 Ωb BW 1 Cbn =  Lln = Lln . whose values are given as Cln  ωr  Cbn = BW Cln = Ωb and Ωb BW 1 Lbn =  Cln = Cln . low-pass to the normalized. to get 1 1 slnCln =  ω  r 1    BW  (s b n + s )  C l n  bn    = 1  ωr  1  ωrCln sbnBW 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 . (2-41). Example 3-5 . and L3ln = 1.Chebyshev.Chapter 3 Page 3-12 Once the normalized.018 H and C1bn = 0. For L1ln we get L1bn = 1. The resulting bandpass filter is shown in Fig. low-pass to normalized. The values of the third-order.Design of an RLC Bandpass Filter A 1-dB ripple. is Ωb = 10/100 = 0.5088 = 0. (2-19) 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. 3-9.Continuous Time Filters (P.3332/0. An example will serve to illustrate the design approach.1/0.5088/0.066278 F. 3-5.088 H and C3bn = 0. respectively. normalized Chebyshev approximation are L1ln = 1.261mH L2b = 300 Ω 358µH + Vout (s) - Figure 3-9 . 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 . low-pass realization has been transformed to a normalized. Allen) .075008 H .815nF L1b = 5. bandpass realization for Ex. (2-35).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.

low-pass 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. fifth-order. the low-pass to bandpass transformation of Eq. 3-10. low-pass RLC filters have been developed and tabulated for this type of filter. low-pass ladder filter of Fig.Chapter 3 Page 3-13 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. low-pass to normalized. (2-31) is applied to the normalized high-pass filter. high-pass transformation (see PA3-1).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 doubly-terminated. These filters can be developed by first applying the normalized. 3-1 and 3-2 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 low-pass ladder filter. They are similar in form to the ladder filters of Figs. 2-18. high-pass transformation to the low-pass normalized realization. If Ex. RLC. We are also dropping the "l" from the component subscripts because we will only be dealing with low-pass structures in this discussion.ECE 6414 . One type are filters that have jω axis zeros such as the one shown in Fig. 3-9 except L1b and C1b (L3b and C3b) would in parallel and C2b and L2b would be in series. 3-5 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 (3-15) (3-13) (3-14) (3-12) (3-11) Eqs. 3-10 by active RC elements is to assign a current Ij to every j-th series element (or combination of elements in series) of the ladder filter and a voltage Vk to every k-th 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. 3-10 are shown on the figure.sC4nV4(s) . 3-10.I5(s) = 0 and V4(s) . These variables are called state variables.Chapter 3 Page 3-14 I1 + Vin (sn ) R0n L1n C2n + - I3 L3n V2 C4n + - I5 L5n V4 R6n + Vout (sn ) - Figure 3-10 . 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) . (3-11) through (3-15) constitute the state equations which completely describe the ladder filter of Fig. we write the nodal equation I1(s) . (3-16) Once. low-pass. then we define a voltage analog.V2(s) = 0 . V' of current I as j j Vj' = R'Ij (3-17) . A supplementary equation of interest is Vout(s) = I5(s)R6n . normalized RLC ladder filter. Allen) . Next.A fifth-order. 3-10 and write the loop equation Vin(s) .sL3nI3(s) .

V4(s) = 0    V '( s )  V '( s )  3  .Chapter 3 Page 3-15 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) . (3-11) through Eq. and I5 by their voltage analogs.V2(s) = 0    V '(s)  V '( s )  1  . 5 5 (3-27) (3-23) (3-24) (3-25) (3-26) . we get the following modified set of state equations.sC4nV4(s) . (3-15) 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 active-RC implementation of the ladder filter.ECE 6414 . I3 . Allen) . R '  (sL5n + R6n) = 0   (3-22) (3-18) (3-19) (3-20) (3-21) 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 '  . (3-18) through (3-22) 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 3-16 However. (3-26) and (3-27) with (3-28) gives 1  R'  V4(s) = sR'C 4n [ V '(s) . (3-28) Combining Eqs.Continuous Time Filters (P. (3-29) (3-30) .Vout(s)] . we would prefer to have the variable Vout(s) used in place of V '(s) . Allen) . 5 (3-15) we get  R6n Vout(s) =  R '  V '(s) 5 .ECE 6414 .

(3-25). 4. summing) found in the negative or inverting integrator. 3-11. 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 k-th integrator. Note that the inverter at the output of the integrator allows us to get either the positive or negative output variable. and (3-29). as we will recall. (3-30) are simple integrators.Continuous Time Filters (P. (3-29) and Eq. we are ready to synthesize an active-RC realization of Fig. (3-24). a different approach is necessary for Eqs. However. the positive integrator did not provide the flexibility and usefulness (i. (3-24). 3-10. (3-23) and (3-30) because the variable is a function of itself. (3-25). The next step would be to go back to Sec.e. Ladder Filters using Active-RC Elements Finally. 3-11 is given as 1 Vk(s) = sRCk [Vj(s) .Chapter 3 Page 3-17 Design of Low-Pass. 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 3-11 . (3-23) and (3-31) gives 1 out V '(s) 1 and  L1n  = R 0 n{ V in(s) + [-V2 (s)]} s + L1n  R' (3-32) . We can show that the output of Fig. 3-11 works well for Eqs.Vl(s)] (3-31) where Vj(s) is the input from the previous stage and -Vl(s) is the negative output of the next stage. Fig. (3-23).

Realization of Eq. A similar realization can be found for Eq. (3-32) is shown in Fig. (3-30) is shown below.2-6c preceeded by a gain of -1 stage which also serves as a summer for Eq.ECE 6414 . (3-32). One modification of Eqs. R'R/R0n -V2 (sn ) Vin (sn ) R R A1 R R A2 L1n /R R' V'1 (sn ) Figure 3-12 . (3-25).Vout(s)] .V '(s) +  R  Vout(s)] 3 4n  6n and R6n Vout(s) = sL [V4(s) . (3-33) respectively. R'   R0n  V '(s) = sL1n V in (s) .V 2 (s) . the straight-forward application of Figs. (3-24). 5n Another modification of the state equations is given as (3-38) (3-34) (3-35) (3-36) (3-37) . (3-32). We see that Eqs. This realization will only have one input to the inverter. 3-12. (3-29) and Eq. A more efficient realization is achieved by modification of the state equations. 4. (3-32) and (3-33) are nothing more than the first-order. (3-33). 3-11 and 3-12 to realize the state equations uses more op amps than is necessary. (3-23).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 3-18 6n  R5n  L   Vout(s) = R 0 n s + L   5n V4(s) . low-pass filter of Fig. Allen) . Unfortunately.

ECE 6414 . pp. Girling and E. Good.J.V 2 (s) . The intercoupled second-order 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. second-order. The inside filters have a Q of infinity because neither integrator is damped. 3-13b saves three op amps over the case where every integrator consists of two op amps.Continuous Time Filters (P.Vout(s)] .E. 341-345. 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. Tow-Thomas filters of Fig. R  Vout(s)] 3 4n  6n and -R6n -Vout(s) = sL [V4(s) . (3-34) through (3-38). Fig. † F.F.Chapter 3 Page 3-19 -R'   R0n  -V '(s) = sL1n V in (s) . 5n (3-39) (3-40) (3-41) (3-42) (3-43) Fig. Fig. July 1970. 3-13a saves two op amps over the case where every integrator consists of two op amps and Fig. "Active Filters 12: The Leap-Frog or Active-Ladder Synthesis. 76. 3-13 turns out to be the interconnection of four. (3-39) through (3-43). 3-13b shows the realization of the state equations of Eqs. 3-13a shows the realization of the state equations of Eqs. . Allen) . 1-16.

(a.) Realization of Eqs. (3-39) through (3-43).) 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 3-13 .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. (3-34) through (3-38). Page 3-20 .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. low-pass RLC filter. Butterworth.) Denormalized. 3-14b where R0n = 0 and R' = R = 1 Ω.) 1Ω L1n= 1. active-RC filter.) Normalized. low-pass filter having a passband frequency of 1 kHz using the RC-active ladder realization of Fig. Allen) . 3-13b as the realization form resulting in Fig.5000 H R4n = C2n = Vin (sn ) 1.Design of a Low-Pass. active-RC filter.Chapter 3 Page 3-21 Example 3-6 .5 F Vin (sn ) 1Ω A1 -V' 1 1Ω Vout (sn ) (b. Low-Pass. 1 dB Chebyshev. (a.87 nF 10 kΩ 10 kΩ A1 -V' 1 A2 10 kΩ A3 10 kΩ 21. Solution The normalized.) Figure 3-14 . Assume that the source resistance of the input voltage is zero.Continuous Time Filters (P. Example 3-7 .) 10 kΩ 23. + L1n =1.3333 F 1Ω (a. low-pass ladder filter is obtain from Table 3-1 and is shown in Fig. low-pass. low-pass. 3-6.96 nF Vin (s) 10 kΩ Vout (s) -V2 (c. Butterworth Ladder Filter using the Active-RC Elements Design a third-order. Active-RC Filter Design . 310. Let us choose Fig. (b.) Normalized. 3-13.5000 H L3n =0.3333F 1Ω A2 A3 -V2 1Ω A4 + Vout (sn ) 1/R 4n L3n=0. (c.A Fifth-Order. 3-14a.22 nF 10 kΩ A4 7. 3-14c. 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. 3-13b 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. low-pass.09011/2πx106 = 0.340 µF.174µF.Chapter 3 Page 3-22 A fifth-order. 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 high-pass 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 3-2. 3-13b to achieve a normalized. Unfortunately. However. The design methods follow a simular procedure as for the low-pass filter.09011.09011 F. Note that we have reversed the order of the subscripts in Table 3-2 to correspond to that of Fig.ECE 6414 .0009 H.1349/2πx106 = 0.340 µF. the equations are not as straight-forward 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 Active-RC Elements High-pass. Let us again select Fig. we select R = 1000 Ω and frequency denormalize by 2πx103.1349 H. active-RC. from input to output are C1 = L1n/2πx106 = 2.1349/2πx106 = 0. we get the normalized. low-pass. L5n = 2. C2n = 1.174µF. L3n = 3. bandpass. Design an active-RC ladder realization of this filter.0009/2πx106 = 0. and C5 = L5n/2πx106 = 2. low-pass. C4 = C4n/2πx106 = 1. 3-10 (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. low-pass 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 k-th shunt element consists of a shunt resistor. 1984.  ωr  Cjln Ik-1 (sbn ) + Vj+1 (snb ) + Zkn (sbn ) Vk (sbn ) Rkn Lkbn Ckbn Ik+1 (sbn ) + Vk (sbn ) (3-45) Ij (sbn ) Rjn + Vj-1 (snb ) - Ljbn Cjbn Series Element Yjn (sbn ) (a. Once a normalized. is the jth series element consists of a series resistor.Sa nchez-Sinencio. 3-8. We will illustrate the design of bandpass ladder filters using active-RC elements because it is a straight-forward extension of the low-pass ladder filter design.ECE 6414 . low-pass ladder filter has been found which meets the bandpass filter specifications.Continuous Time Filters (P.) Shunt Element (b.) Figure 3-15 . R jn.Chapter 3 Page 3-23 write as for the low-pass case. Van Nostrand Reinhold ' Company. 3-15a.  ωr  Ljln (3-44) 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 high-pass ladder filter design can be found in other references†. Normally. Inc. 3-8. 4. † P. Chapt. Thus.E. BW 1 Ljbn =  .. . 3-15b.

3-15a and the impedance. Zkn. Allen) . (3-49) reduces to Vj (sbn) = '  LR ' sbn   jbn sb2n + L 1C  jbn jbn   [V (s )-V (s )]  j-1 bn j+1 bn  . (3-47) 1 2  1 s sb n +  RknCkbn b n + LjbnCjbn (3-46) These driving-point functions can be turned into voltage transfer functions using the concept of voltage analogs for the currents Ij and Ik.Continuous Time Filters (P. (3-46) by R' gives ' Vj (sbn) R'Ij(sbn) TYj(sbn) = V (s ) = V (s )-V (s ) = j bn j-1 bn j+1 bn or ' Vj (sbn)  R ' s    Ljbn bn   = Rjn  1  [Vj-1(sbn)-Vj+1(sbn)] 2  sb n +  Ljbns b n + LjbnCjbn  R ' s  Ljbn bn 1 2  Rjn sb n +  Ljbn s b n + LjbnCjbn  (3-48) . These driving-point functions can be written as  1 s Ij(sbn) Ij(sbn)  Ljbn bn Yjn(sbn) = V (s ) = V (s )-V (s ) = j bn j-1 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 k-1 bn k+1 bn  1 s  Ckbn bn . (3-47) 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 Vk-1(sbn)-Vk+1(sbn) sb n +  RknCkbn b n + LjbnCjbn . (3-49) If Rjn = 0.Chapter 3 Page 3-24 Each element of the bandpass filter can be realized by considering the admittance. (3-50) Multiplying Eq.ECE 6414 . then Eq. Yjn. of Fig. 3-15b. of Fig. Multiplying Eq.

3-15a for the j-th series elements. (3-49) or Eq. then Eq.th element Figure 3-16 .Active-RC realization of either Fig. (3-52) If Rkn = ∞.Continuous Time Filters (P. bandpass. 3-15 by the Tow-Thomas circuit of Fig.ECE 6414 . ladder filters. (3-53) The second-order. (3-52) reduces to 1  R'Ckbnsbn   = 2 sb n + L 1C  Vk(sbn) jbn jbn   [V ' (s ) -V ' (s ) ]  k-1 bn k+1 bn  . 3-15b (replace subscript j by k and interchange primes). For the cases of Eq. (3-52). 3-16 shows the general second-order realization of either Fig. Fig. Allen) . 1-23 is a perfect realization of Eq. Rj3 Rj4 Rj1 Vj-1 (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. 3-16. An example will illustrate the approach to designing active-RC bandpass. Tow-Thomas filter of Fig. R R A3 Cj2 A2 -V'j (snb ) +V'j (snb ) Rj2 . Table 3-3 shows the design relationships for the implementation of Fig. (3-53). 3-15a or Fig. (3-50) and Eq.Chapter 3 Page 3-25 Vk(sbn)  1 s    R'Ckbn bn   [V ' (sbn) -V ' (sbn) ] = k+1 sb2n +  1 s b n + 1  k-1 LjbnCjbn  RknCkbn  . The realization of the k-th 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 low-pass realizations.

Continuous Time Filters (P.Chapter 3 Page 3-26 Parameters of Fig. a convenient value: Cj1 = Cj2 or Ck1 = Ck2 Choose Cj1 = Cj2 = C or Ck1 = Ck2 = C. with all component values. doubly-terminated. centered geometrically about 1 kHz. 3-15b by the active-RC filter of Fig. 3-16 Rj1 or Rk1 Design of Fig. 3-16. 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 active-RC elements. bandpass. Example 3-8 . or equivalent. Ladder Active-RC Filter Design a sixth-order. 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. 3-15a or Fig. 3-15b 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 3-3 .Design of a Butterworth. 3-15a Design of Fig. We also know that 1 kHz = fBP2fBP1 . Bandpass. Impedance denormalize by a factor of 104 and give a schematic.

(3-44) and (3-45) 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. 3-16. Other examples of active-RC 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. 3-17b. 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 active-RC elements. Y1n. 3-17a. low-pass filter is shown in Fig. ωr. L1bn = 2 H. If we apply the normalized. second-order 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) . low-pass to normalized. 3-17a we get the following normalized bandpass elements of Fig. bandpass transformation of Eqs. doubly-terminated Butterworth approximation we find that L1n = 1 H. Since the filter is symmetrical. C2bn = 2 2 F. and C3bn = 1/ 2 F. 3-17c. For the first element. . We will select R = R' = 1 Ω. L3bn = 2 H. and L3n = 1 H.Chapter 3 Page 3-27 From Table 3-2. we use the design relationships of Table 3-3 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 third-order. 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 3-17 . 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. active-RC ladder filter of Ex. low-pass 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. low-pass. The starting point of the ladder active filter design is with the normalized.14kΩ 15.14kΩ C31 = R31 = 14.or second-order stages and was covered in the previous section. Allen) .Continuous Time Filters (P. 2-1 illustrated the two general design approaches normally used to design active filters.) Normalized.Bandpass. (c. Basically. 3-8.28kΩ 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. active-RC . 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 3-28 + 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 second-order. By correctly selecting the state variables and using the concept of voltage analog for current. The value of the inductor is . • The normalized low-pass to normalized high-pass frequency transformation turns inductors (capacitors) into capacitors (inductors) whose value is the reciprocal of the low-pass component value. lowpass Tow-Thomas filters. We saw that high-pass. Some the important points of this section are summarized below for the convenience of the reader. second-order. we were able to synthesize the state variables using summing integrators. bandpass. Tow-Thomas filters. To achieve an active-RC realization of the RLC ladder filters. not covered here. active-RC 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. low-pass filters. we introduced the concept of state variable.ECE 6414 . Other design methods.Continuous Time Filters (P. • The normalized low-pass to normalized bandpass frequency transformation turns an inductor into an inductor in series with a capacitor. low-pass RLC filter. Allen) .Chapter 3 Page 3-29 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 second-order. low-pass RLC resulted in a second-order bandpass circuit when the normalized low-pass 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 low-pass component value. .ECE 6414 . • The Tow-Thomas second-order realization becomes a very useful active-RC filter for realizing RLC ladder filters. Allen) .Chapter 3 Page 3-30 (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 low-pass 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.