Whites, EE 322
Lecture 11
Page 1 of 17
Lecture 11: Ladder Filters. Butterworth and
Chebyshev Filters. Filter Tables. ADS.
Ladder filters are circuit networks that are composed of
alternating series and shunt reactive elements, as shown below
in Figs. 1 and 2. The signal flow is assumed from the left to the
right. The source is the Thévenin equivalent representation of all
the circuitry feeding the filter (on the left) and the load is the
Thévenin equivalent representation on the circuitry that the
filtered signal is feeding.
L2
R
L4
Low Pass:
Figure 1:
C1
R
C3
C1
C5
R
C3
C5
L2
L4
High Pass:
Figure 2 :
R
Notice that the same source and load resistances are assumed.
This is called “doubly terminated” filters. All of our filters will
be doubly terminated.
Ladder filters are actually one of the oldest types of filters. They
have been around since the mid-1800’s.
© 2016 Keith W. Whites
Whites, EE 322
Lecture 11
Page 2 of 17
A circuit designer can achieve a sharper (or steeper) frequency
roll off with ladder filters than with simple RC or RL circuits.
Consequently, one can obtain more ideal low, high or band pass
filter responses and with little resistive loss.
Additionally, doubly terminated ladder filters can have a lower
sensitivity to component variation. That is a good characteristic.
There are four basic types of ladder filters:
1. Maximally flat, also called Butterworth filters,
2. Equal ripple, also called Chebyshev filters,
3. Elliptic, also called Cauer filters,
4. Linear phase filters.
We will consider the first two in this course.
The circuits in Figs. 1 and 2 can be either Butterworth or
Chebyshev filters. The topology is the same for both. Only the
values for L and C vary between the two types of filters.
We will characterize these two filter types by the response of the
loss factor L  f  magnitude versus frequency. [The loss factor is
sometimes referred to as the insertion loss = IL = 10 log10(L).]
Whites, EE 322
Lecture 11
Page 3 of 17
Maximum Available Source Power
Before further discussion of ladder filters, we must first define
maximum available source power, P+. This is the maximum time
average power that can be provided by a source, or by the
previous stage in the circuit, to a matched load.
Consider that a source or previous circuit stage has been
modeled by this Thévenin equivalent circuit:
As you determined in Homework Prob. 1, a dc source delivers
maximum power when a resistive load Rs is connected to the
output, similar to that shown above. For the ac circuit shown
here, the maximum power delivered to the load Rs is
2
 Vs 
Vs2
1 V02 1  2 
or P 
[W]
(1)
P 

8 Rs
2 Rs 2 Rs
In summary, P+ is the maximum available source power from an
ac source (or a Thévenin equivalent) with internal resistance Rs.
It is the maximum time-average power that can be delivered to a
Whites, EE 322
Lecture 11
Page 4 of 17
matched source. Very important formula. (Note that Vs is the
amplitude, not p-to-p.)
1. Maximally Flat, or Butterworth, Low Pass Filter
For this filter, the values of the inductors and capacitors are
somehow chosen so that
2n
 f 
Pi
1  
P f 
 fc 
Where LB is the loss factor as a function of f.
LB  f  
(5.1)
In this expression:
 Pi = maximum available source power (see Lecture 10),
 P = delivered power to the load,
 fc = cutoff frequency of the filter,
 n = order of the filter (number of L’s and C’s in high and
low pass filter; number of L-C pairs in bandpass filters).
For the Butterworth (maximally flat) low pass filter (Fig. 5.2a):
|P|
Flat magnitude response.
Pi
"roll off"
Pi/2
Pass band
f
fc
Whites, EE 322
Lecture 11
Page 5 of 17
2. Equal Ripple, or Chebyshev Low Pass Filter
The values of the inductors and capacitors in this type of filter
are somehow chosen so that
P
(5.3)
LC  f   i  1   Cn2  f fc 

P
argument
In this expression:
  = ripple size,
f
 Cn  f  Chebyshev polynomial of order n (see plots in Fig.
n
5.3b),
Chebyshev filters might be more susceptible to variations in
component values than Butterworth filters. This is due to the
large coefficients of the polynomials listed in Fig. 5.3.
|P|
Equal ripple.
Pi
Pi/(1+)
Sharper roll off than
Butterworth.
Pass band
f
fc
Comments

Whether to use Butterworth, Chebyshev or another filter type
depends on the specifications/requirements of the circuit
Whites, EE 322
Lecture 11
Page 6 of 17
(required rejection, roll off, phase variation, etc.), the
available components, component value variations and so on.


Butterworth
Once you have the specifications, then you can synthesize the
filter. The required filter specifications are:
Chebyshev

Cutoff frequency, fc
Order of the filter, n (for
rejection)
Impedance level, R (for source
and load)
Ripple in passband
With these specifications, you can calculate the specific
inductor and capacitor values needed to realize the filter (i.e.,
“synthesize” it). It is a complicated procedure to derive the
formulas for these component values, but it is covered in EE
481/581 Microwave Engineering. There are entire books
devoted to this topic. (See the attachment at the end of this
lecture for a simple example.)
Instead of deriving these formulas, designers often simply use
filter tables. These are tabulated values for normalized
susceptance and reactance (collectively called immittance, a).
To un-normalize values from filter tables for low pass filters,
use
Whites, EE 322
Lecture 11
Page 7 of 17
 R   N 
L

 a [H]

R
 N  c 
 R   
C   N   N  a [F]
 R   c 
RN and N are the normalization values used in the tables
(often both = 1), while R and c are the actual circuit values.
An example will help explain this procedure.
Example N11.1: Design a fifth-order, Butterworth, low-pass
filter (see Fig. 1 above) with a cutoff frequency of 8 MHz, a
rejection of at least 23 dB at 14 MHz and an impedance level of
50 .
With a fifth order filter, n = 5. From (5.1) and f/fc = 14/8 then
  14  25 
IL 14 MHz  10log10  L   10log10 1      24.3 dB
  8  
which meets the 23 dB spec. (Note that there is also loss in the
passband. At 7 MHz, for example, IL = 10 log10[1+(7/8)10]=1.0
dB. Where does this “lost” energy go?)
Now, for this fifth-order Butterworth filter we read the
immittance coefficients from Table 5.1 to be
a1  0.618 , a2  1.618 , a3  2 , a4  1.618 , and a5  0.618 .
Whites, EE 322
Lecture 11
Page 8 of 17
For a low pass filter, these immittance coefficients are the
normalized susceptances of the shunt elements at fc and the
normalized reactances of the series elements at fc.
R=50 
C1
a
 1
R c
L2 
a2 R
c
C3
a
 3
Rc
L4 
a4 R
c
C5
a
 5
R c
R=50 
For R = 50  and  c  2 f c  5.027  107 rad/s (at 8 MHz), then
a
C1  1  2.46  1010 F = 246 pF

R c
aR
L2  2  1.61 106 H = 1.61 H

c
a3
 7.96  1010 F = 796 pF (use standard 820 pF)
R c

L4  L2 = 1.61 H

C5  C1 = 246 pF.
All of these values are “in the ballpark” for the Harmonic Filter.

C3 
Of course, one generally needs to use standard values of
components for the filter, unless you build your own inductors
and/or capacitors. Consequently, the circuit may need to be
“tweaked” after completing this synthesis step.
Whites, EE 322
Lecture 11
Page 9 of 17
Advanced Design System (ADS)
This tweaking process can be performed using analysis software
such as SPICE, Puff, or Advanced Design System (ADS).
Your text uses the passive microwave circuit simulator called
Puff, which used to come with your text. It is DOS-based and
requires the use of “scattering parameters” to characterize the
behavior of circuits, including filters. (S parameters are
discussed extensively in EE 481/581 Microwave Engineering.)
For these, and other, reasons we will NOT be using Puff in this
course. Instead, we will be using Advanced Design System
(ADS) from Agilent Technologies. Consequently, all of the text
problems that refer to Puff have been rewritten to use ADS.
These can be found on the course web site.
The manual “Getting Started with ADS” has been written to help
you get going with ADS. It can also be found on the course web
site. ADS has just a couple of nuances. Other than that, it is very
straightforward to use.
To illustrate the use of ADS, we will verify the proper operation
of the low-pass filter designed previously.
Whites, EE 322
Lecture 11
Page 10 of 17
ADS Simulation of a Low-Pass Ladder Filter
ADS Startup Window:
To get going with ADS, you must first create a “project”:
Whites, EE 322
Lecture 11
Page 11 of 17
ADS example with Rs = 50 :
AC
Var
Eqn
AC
AC1
Start=1.0 MHz
Stop=20.0 MHz
Step=0.1 MHz
VAR
VAR1
Vs=1.0
Rs=50
Vsource
Vin
V_AC
Vsource
Vac=polar(Vs,0) V
Freq=freq
R
R1
R=Rs Ohm
I_Probe
Iin
C
C1
C=246 pF
Meas
Eqn
L
L2
L=1.61 uH
R=
C
C3
C=796 pF
L
L4
L=1.61 uH
R=
I_Probe
C
Iout
C5
C=246 pF
Vout
R
R2
R=Rs Ohm
MeasEqn
Meas1
Pavail=abs((Vs)**2/(8*Rs))
Pout=abs(Vout*conj(Iout.i))/2.
Pin=abs(Vin*conj(Iin.i))/2.
Ps=abs(Vs*conj(Iin.i))/2.
Here is a plot of Pout/Pin in dB:
m1
freq= 7.000MHz
10*log10(Pout/Pin)=-1.813
m1
10*log10(Pout/Pin)
0
-10
-20
-30
-40
-50
0
2
4
6
8
10
12
14
16
18
20
freq, MHz
This doesn’t look like the response of a maximally flat low pass
filter. What’s wrong?
Whites, EE 322
Lecture 11
Page 12 of 17
Here’s a plot of Vout Vin in dB:
20*log10(abs(Vout/Vsource))
m2
freq=7.000MHz
20*log10(abs(Vout/Vsource))=-7.037
0
m2
-10
-20
-30
-40
-50
0
2
4
6
8
10
12
14
16
18
20
freq, MHz
This plot has the general shape of a maximally flat filter, but
there is an extra 6 dB of attenuation at the design frequency of 7
MHz. What’s going on here?
Here’s a plot of Pout/P+ in dB where P+ is the maximum
available power from the source:
Whites, EE 322
Lecture 11
m3
freq=7.000MHz
10*log10(abs(Pout/Pavail))=-1.017
m3
10*log10(abs(Pout/Pavail))
Page 13 of 17
0
-10
-20
-30
-40
-50
0
2
4
6
8
10
12
14
16
18
20
freq, MHz
Alas, this is the plot we’ve been looking for. Why? Because by
definition, insertion loss is the ratio of the output power to the
maximum available source power. See (5.1) as an example.
From this last plot, we can see that ADS predicts an insertion
loss of –1.017 dB at 7.000 MHz. This is very close to our design
prediction of –1.0 dB at 7 MHz.
Whites, EE 322
Lecture 11
Page 14 of 17
ADS example with Rs = 100 :
AC
Var
Eqn
AC
AC1
Start=1.0 MHz
Stop=20.0 MHz
Step=0.1 MHz
VAR
VAR1
Vs=1.0
Rs=100
Vsource
Vin
V_AC
Vsource
Vac=polar(Vs,0) V
Freq=freq
R
R1
R=Rs Ohm
I_Probe
Iin
C
C1
C=246 pF
Meas
Eqn
L
L2
L=1.61 uH
R=
C
C3
C=796 pF
I_Probe
C
Iout
C5
C=246 pF
L
L4
L=1.61 uH
R=
Vout
R
R2
R=Rs Ohm
MeasEqn
Meas1
Pavail=abs((Vs)**2/(8*Rs))
Pout=abs(Vout*conj(Iout.i))/2.
Pin=abs(Vin*conj(Iin.i))/2.
Ps=abs(Vs*conj(Iin.i))/2.
m1
freq= 7.000MHz
10*log10(Pout/Pavail)=-1.034
m1
10*log10(Pout/Pavail)
0
-10
-20
-30
-40
-50
0
2
4
6
8
10
12
14
16
18
20
freq, MHz
Changing the impedance “level” (from 50  to 100 ) has a
dramatic effect on the performance of the filter. Can you explain
why?
Whites, EE 322
Lecture 11
Page 15 of 17
From David M. Pozar, Microwave Engineering. New York: John Wiley & Sons, second ed.,
1998:
Whites, EE 322
Lecture 11
Page 16 of 17
Whites, EE 322
Lecture 11
Page 17 of 17