You are on page 1of 7

ON THE

DESIGN OF
CAUER
FILTERS
SOURCE:-RESEARCH PAPER
BY KLAUS HUBER
Mohammad Imad Nizami
Roll No: - 141300045
EEE-2nd Year
Subject: - Circuit Theory 2- Assignment

PAGE 0

Table of Contents
Summary

Introduction

The Designing Equations

A few examples of Cauer Filters


Conclusion

Result4
References
..4

PAGE 0

1. Summary
1.i. Introduction
An Cauer filter (also known as an Elliptical filter, named after Wilhelm Cauer,
or as a Zolotarev filter, after Yegor Zolotarev) is a signal processing filter with
equalized ripple behavior in both the passband and the stopband. The amount of
ripple in each band is independently adjustable, and no other filter of equal order
can have a faster transition in gain between the passband and the stopband, for
the given values of ripple. Alternatively, one may give up the ability to adjust
independently the passband and stopband ripple, and instead design a filter
which is maximally insensitive to component variations.
For the design of low-pass filters usually the following de-sign parameters are
given. The maximal attenuation in the pass band max, the minimum attenuation
in the stop band min, and the normalized (circle) frequency s, at which the stop
band starts. We assume that the pass band extends from = 0 to = 1.
For given degree n, Elliptic- or Cauer filters achieve the best approximation in the
frequency domain.
For elliptic filters the following equations hold, if the attenuation is measured in
decibels.

max =

10 log10 (1+

min = 10 log10 [1 + ( g2)]


s =

1
k

The design triple (max, min, s), is easily transferred to the triple (, g, k). Given
such a triple one usually first determines the degree n which is able to fulfill the
specification. The degree is found by rounding the following expression to the
next integer:
2

K ( 1g
K (g)
(1)

1k

K
K (k )

The function K(k) is the complete elliptic integral of the first kind

K (k) =

d
1k 2 sin2 ( )

The value of K(k) is most easily found using the arithmetic geometric mean M by
PAGE 1

K (k) =

1, 1k
2. M

Where the arithmetic geometric mean M (a, b) of the two numbers a and b is
given by the limit of the recursion ai+1 = (ai+bi)/2, bi+1 =ai-bi with starting
values a0=a, b0=b. The values of k and g lie in the interval [0, 1].
As n is obtained from equation (1) by rounding, the designer is usually given
some freedom to improve the original triple (max, min, s). To this end, in this
letter we give simple design equations g= fn (k) and k = fn1 (g), which relate k
and g in an easily computable way such that
2

K ( 1f n(x )2 )

1 x

K
K (x)

K (f n (x ))
(2)

Holds

1.ii. The Designing Equations


We assume that n is fixed. Then, given two values from the triple ( max, min, s)
or from the equivalent triple (, g, k) we want to be able to compute the
remaining third value. The only non-trivial dependence is the relation between k
and g. In this section we set g= fn (k) and
k= fn1 (g) and determine fn(x) and
its inverse.
First we note that for a composite number n=n1n2 the following holds

fn(x) = fn1( fn2(x)).

(3)

This relation follows immediately from equation (2) setting


2

K ( 1f n2 (x )2)
K (f n 2 (x ))
(x)
f n2

1f n 1 ( 2)
K (f n1 ( f n 2 ( x ) ) )
K

1 x

K
K (x)

= n2

1f n 2( 2)
K
K (f n 2 ( x ) )

= n1

PAGE 2

And multiplying n1 with n2


Second we find that the inverse function of fn(x) is given by
1

f n (x)

1f ( 1x )
2

(4)

1x2

This also follows from equation (2). Simply set u =

and v=

1f (x)

to get
2

K ( 1u
K (u)

1v

K
K (v )

From which we conclude u=fn (v), hence v=


get

f n (u)

2 2

1f n (1u )

=n

f n (u)

1f ( x )

for which we

which was to be shown. We now explicitly give

the function fn(x). We therefore consider two cases. The case n=2, and then the
case of odd n. The case n=2 is given by the so-called Landen-Gauss
transformation. We get

g = f2 (k) =
(5)

1 1k

1+ 1k

and

2g
1
k = f n (k) = 1+g

For odd n the solution is given by a formula of Jacobi,

( n1 ) K
2K
4k
g = fn(k) = kn { cd n .cd n ..cd
n

( ) ( )

)}

(6)

In the formula K = K(k), The function cd(x) = cd(x, k) is the quotient of the
elliptic function cn(x, k) and the elliptic function dn(x, k) which is easy to
compute. In fact most modern scientific libraries and computer algebra packages
have these functions available. Equation (6) can be put into another form and
can be extended for even n. We eventually get

K
3k
LK
g = fn(k) = kn { sn n . sn n .. sn n

( ) ( )

Where L =.

( )}

n1 for n even
n2 for n odd

(7)
PAGE 3

Equation (7) for odd n follows immediately from equation (6) using cd(x) = sn
(x+K). The formula for even n is obtained from the odd case using well-known
properties of elliptic functions, namely

sn

K
( ,k) =
2

1k

where

'

k = 1k
'

And the Landen equation

sn

2
(( 1+k ' ) x , k 2 2 ) = ( 1+k ' ) . sn(x ,k)cd(x ,k) ,
(1+k )

And by taking into account that


'

1k
K 1+k ' )

1+k '
2 .K (k)

1.iii. A few Examples of Cauer Filters

Fig1. A Cauer Low Pass Filter

Fig2. A Basic Cauer Filter

2. Conclusion
PAGE 4

Although Cauer filters are a well-established part in filter design, there are useful
mathematical relationships which are not widely-known in the filter community
.In this report, simple formulas have been given which relate the parameters for
the design of elliptic filters. Apart from theoretical considerations the results
given facilitate the design of filters in hard- or software and in particular make it
easier to construct universal filters whose parameters can be set easily.

3. Result
The designing of a Cauer filter, as stated in this paper, is derived from a low pas
filter. Also for designing a Cauer filter, one should know simple formulas which
relate to the parameters of Cauer. Foreknowledge on the working of filters and
some engineering mathematical formulas is required.
The gain of a Cauer filter is given by

G=

0
,/

2 2
1+ Rn

Where Rn is the nth-order elliptic rational function

is the cutoff frequency

is the ripple factor


is the selectivity factor

4. References
1. Research paper published by Klaus Huber (from Sciencedirect.com)
2. Google for images
3. Wikipedia for additional information
4. Engineering mathematics for the mathematical formula

PAGE 5