Ladder Filters. Butterworth and
Chebyshev Filters. Filter Tables. ADS.
Ladder filters are networks that are composed of alternating
series and shunt elements.
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.
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 have a low
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).] Maximum Available Power
Before further discussion of ladder filters, we must first define
maximum available 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
In summary, P+ is the maximum available 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
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
Where LB is the loss factor as a function of f.
In this expression:
• Pi = maximum available power from the source (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 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
In this expression:
• ? = ripple size,
• Cnn)) Chebyshev polynomial of order n
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 Comments
• Whether to use Butterworth, Chebyshev or another filter type
depends on the specifications/requirements of the circuit
(required rejection, roll off, phase variation, etc.), the
available components, component value variations and so on.
• Once you have the specifications, then you can synthesize the
filter. The required filter specifications are:
• 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. 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
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
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
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, a 4=1.618 and a5 = 0.618.
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.
For R = 50 ? and W3=2 ? Fc=5.027 10 ×107 rad/s (at 8 MHz), then
All of these values are “in the ballpark” for the Harmonic Filter.
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. 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 comes 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 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. ADS Simulation of a Low-Pass Ladder Filter
ADS Startup Window:
To get going with ADS, you must first create a “project”:
ADS example with Rs = 50 ?:
Here is a plot of Pout/Pin in dB:
This doesn’t look like the response of a maximally flat low pass
filter. What’s wrong? Here’s a plot of |Vout| Vin| in dB:
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?
Lastly, here’s a plot of Pout/P+ in dB where P+ is the maximum
available power from the source:
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 avaliable 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. ADS example with Rs = 100 ?:
Changing the impedance “level” (from 50 ? to 100 ?) has a
dramatic effect on the performance of the filter. Can you explain
why?
