CONCZBNING THE MIIti" NUMBER OF RESONATORS

THE UNUXJIED Q NEEDED IN A FIL3'ER

Bdilton Dishal
Federal Telecommunication Laboratories,
Nutley, N e w Jersey

Given a s e l e c t i v e c i r c u i t o r f i l t e r s p e c i f i c a t i o n , w
t o important questions
which a r i s e are: what is the m i n i m u m number of elements which can be used t o
s a t i s f y t h e specification, andwhat is t h e minimu;n unloaded Q which these ele-
ments must have?
This paper answers these questions by means of three pairs of design equa-
tions, Equations 6 and Equations 7 12, and Equations 8 and 13 f o r f i l t e r U-,
networks supplying attenuation zeros (i.e. f i l t e r s e q u i v a l e n t t o c o n s t a n t K
t y p e f i l t e r s ) &?d the corresponding three pairs of equations, Equations 22 and 32
Equations 23 and 33, andEquations 2b and 34, f o r f i l t e r s s u p p l y i n g b o t h a t t e n u -
ation zeros and a t t e n u a t i o n p o l e s ( i . e . f i l t e r s e q u i v a l e n t t o m-derivedtype
filters

I. Introduction

Becausesimple c l a s s i c a l image parameter f i l t e r t h e o r y does not supply

accurate information, t h i s paper is w r i t t e n t o supply the engineer with some
of the detailed mathemtics which w i l l enable him t o deal more accurately with
the below s i t u a t i o n ; he is t o l d t o b u i l d t h e s e l e c t i v e c i r c u i t s (of a receiver
f o r example)which w i l l produce 6 db (for exanple) "accept ba&widthtl
no l e s s t h a n some numberof cycles which we w i l l c a l l Bypa, and 100 db down
( f o r example) "reject bandwidth" no greater than some number of cycles which
we c a l l Bwr. Inside the 6 db down points, i.e. inside the accept band he
i s allowed t o have, if he desires, ripples with a peak-to-valley ratio .as large
as some number (V$Jv); and outside the 100 db down points, i.e. in the reject.
band, he is allowed t o have any type of response he desires long as t h e at-
tenuation is never l e s s t h a n 100 db f o r a band whose width is a t l e a s t t e n .
reject bandwidths. Eg. 1 i l k s t r a t e s the above described problem.
And, as is usually the case, there are size, weight, econonic, etc. limita-
t i o n s which make it mandatory t h a t the engineer satisfy t h e s e l e c t i v i t y s p e c i -
f i c a t i o n with as few resonators as possible, and furthermore he must use resona-
t o r s w i t h unloaded which most certainly are not infinite. This paper will
deal i n d e t a i l w i t h j u s t one part of the overall problem, i.e. the exact rrumber
of elements required to meet the above described f i l t e r requirement, and the
lowest unloaded which the resonators may have.

bplanation o f symbols and reference k k s will be found a t t h e end

nf t h e a r t i c l e .
35

Authorized licensd use limted to: IE Xplore. Downlade on May 10,2 at 19:0354 UTC from IE Xplore. Restricon aply.
 Fbr t h e sake of brevity, only bandpass filters producbg symmetrical
response shapes w i l l be r e f e r r e d t o i n t h i s article, but the reader should
realize that the discussion applies equally well to low pass c i r c u i t s when
t h e analogous frequency variables and circuit constants are used.
o n l y "node" t y p e c i r c u i t s be shown i n the figures; the reader should
r e a l i z e that the discussion applies equally to the dual nresh c b c u i t s which
can be shply obtained from the node c i r c u i t s by the p r i n c i p l e o f duality.

A. The C i r c u i t s 1IRhich Are Used

To meet a given selectivity requirement the first s t e p which the engineer
must take i s usually that of deciding w h a t network configuration he is going
t o use. In most p r a c t i c a l s i t u a t i o n s the desire will be t o use a ladder type
network, and f o r t h e last t h r e e decades it has been r e a l i z e d that the basic
ladder configurations are those described by the mrds %onstant K", L e .
inverse a m configurations, and "m-derivedI1 configurations, i.e. configurations
i n which e i t h e r one or the other, but not both, of the above mentioned inverse
arm9 is mdif'ied t h a t it produces infinite attenuation a t f i n i t e f r e q u e n c i e s
near the edges of t h e r e j e c t band rather t h a n a t f r e q u e n c i e s infinitely f a r
from t h e of the r e j e c t band.
It is important for the reader to r e a l i z e t h a t when the percentage band-
width required is small, then by the of simple straightforward equivalences
the configuration of the basic inverse arm (constant K ) bandpass ladder (Figt2A)
can be transformed t o a chain of coupled resonant circuits shown i n Fig.
wherein any or all of the well known methods of coupling resonators can be
used, i.e. mutualinductance,highsidecapacity, larr side capacity, o lw side
inductance, etc.; and the number of resonators which must be used is exactly
the as the mubr of resonators appearing in the constant K configuration.
When small percentage bandwidths are required it is almost mandatory t o use the
coupled resonant circuit configuration, because, for t h e small percentage band-
width case the element values required for the constant K configuration become
physically impractical.

With r e s p e c t t o t h e r e l a t i o n s h i p between t h e number of arm ( n ) in the

basic iwerse ladder of F i g . 2A and t h e number of physical resonators which
t h e engine- d t supply,simplenr4mination of Fig. s h m t h a t the Tnrmber of
resonators required ie i d e n t i c a l l y e q u a l t o n f o r t h e networks of both Fig. 2A
and 2B.
Similarly by the use of straightforward equivalances, t h e configuration of
the basic d e r i v e d bandpass ladder, Fig. SA, can be transformed t o t h e t h r e e
configurations shown in Flg. SB, SC, and SD which require more easily obtainable
elementvalues. It should be r e a l i z e d t h a t a l l t h e ' r e s o n a n t c i r c u i t s shown in
Fig. are true resonators, tuned t o e i t h e r t h e mid-band frequency or t o one of
the freqpencies of i n f i n i t e r e j e c t i o n i 1 Thus when the engineer counts t h e number
of resonators that he must physically construct, he must count every one of the
resonators sham i n that c i r c u i t of 5 which he decides t o use Fig.

Authorized licensd use limted to: IE Xplore. Downlade on May 10,2 at 19:0354 UTC from IE Xplore. Restricon aply.
 With r e s p e c t t o t h e r e l a t i o n s h i p between t h e number of arms ( n ) in the
basic m-derived configuration ladder of FLg.sA, and t h e m b e r of physical
resonators which the engineer must suyply, simple exanination of F'ig.5 shows
t h a t f o r the c i r c u i t s of Fig. SA, B and C,
n+ n-1 n odd
1)umber of resonators 2
n n even
n+
2

and f o r t h e c i r c a i t of Fig .by

2n 1 n odd
Number of resonators
2n n even

5. The Three Types of iiesonators(Electrical, Mechanical, aad Quartz Crystal)

iPhich a r e Used

I n recent years there has been an increasing use of mechanical resonators

(made of Ni-Span C a l l o y ) i n f i l t e r s

Where small percentage bandwidth (i.e, 10% o r l e s s ) work is concerned, it

is important f o r the reader t o r e a l i z e t h a t i n s o f a r as f i l t e r i n g is concerned,
t h e r e i s no f u n d m e n t a l d i f f e r e n c e whatsoever between mechanical resonator and
an e l e c t r i c a l r e s o n a t o r , i.a. i n a f i l t e r t h e unloaded Q of each resonator, the
loaded Q of the end resonators, and the c o e f f i c i e n t of coupling between adjacent
resonators has exactly the same meaning f o r both mechanical and e l e c t r i c a l f i l -
ter
O f course, the high unloaded Q which is available to the designer when
mechanical resonators are used i s advantageous f o r t h e following reason: it en-
ables the designer to erect a required selectivity (Le. required value of
3Wr@'?a, and an allowable (Vp/Vv) i n the passband, a t higher midfrequency;
i.e. w i l l be seen l a t e r , as soon as a r e q u i r e d s e l e c t i v i t y i s specified w
to
t h i n g s are inmediately fixed: first t h e minimum number of resonators which can
be used, and second the highest r a t i o of f r a c t i o n a l midfrequency fo/BWa t o un-
loaded resonator Q ( Q ) which can be used.
Conversely when t h e d e s i m e r has the freedom t o choose h i s midfrequency,
highunloaded Q ~ do s not, as is often implied, enable better selectivities
t o be obtained; with both h i g h or low unloaded exactly the samenumber of
resonators w i l l be required, and i n o r d e r t o s a t i s f y t h e abovementioned maximum
allowable (fo/BGPa)/Qo r a t i o it i s simply necessary to use a lower ddfrequency
with the lower resonators. And t h e r e a r e many practical situations wherein
the designer does some freedom t o choose h i s midfrcquency, e.g. the I t I F '
frequencies11 in our a h s t universally used single and multiple superheterodyne
receivers or, in general, in the many cases where it is permissible t o e r e c t

8?

Authorized licensd use limted to: IE Xplore. Downlade on May 10,2 at 19:0354 UTC from IE Xplore. Restricon aply.
 r e q u i r e d s e l e c t i v i t y a t a lower midfrequency and then make t h i s s e l e c t i v i t y
appear a t a higher midfrequency by the use of a l o c a l o s c i l l a t o r and
tube. It i s of courseunderstood t h a t whenever frequencyconversion i s used
the accompanying problems o f spurious responses, cross- and inter-modulation,
and desensitization, must always be considered and s a t i s f a c t o r i l y s o l v e d .
To stress t h e s h i l a r i t y between mechanical resonator f i l t e r s and elec-
t r i c a l r e s o n a t o r f i l t e r s , F i g . 3 shows t h e e q u i v a l e n t e l e c t r i c a l c i r c u i t of a
t r i p l e t u n e d ( f o r example) mechanical f i l t e r of the h a l f wave resonator-quarter
wave coupling type. The e q u i v a l e n t e l e c t r i c a l c i r c u i t i s t h e node network of
Fig. if we say voltage i s equivalent to velocity and current is equivalent
t o f o r c e . If say current i s equivalent t o v e l o c i t y and voltage is equivalent
t o force,then obtain t h e dual mesh network shown on Fig.3. Patently when
t h e designer knows how t o design any one of the networks of Fig.2and F’ig.3,
he then possesses the design information for a l l the other networks.

With reference t o %he use o f q d a r t z c r y s t a l r c s o r a t o r s in ladder network

filters, the configuration of Fig.sa is the only one considered i n t.his paper.
Here the c r y s t a l s u p p l i e s t h e s e r i e s r e s o n a n t c i r c u i t s of the s e r i e s arms, and
the crystal holder capacity is absorbed in the parallel resonant c b c u i t which
is across the series r e s o n a n t c i r c u i t .

C. The SpecificQuestionConsidered i n t h i s Paper

Having decided upon t h e c i r c u i t t o be used, t h e question which arises

i s the one with which t h i s paper deals; it has the following three parts: e x a c t l y
how many physical resonators must be used t o satisfy a s e l e c t i v i t y requirement
of the type given in the Introduction? what r i p p l e in t h e passband,
passband i.3. what
r;quFya;t
ratio be accepted w i t h t h i s number of resonators? and f o r a given
rmdfrequemy what is t h e minimum unloaded Q which these rasonators must possess?

Both p r a c t i c a l l y and t h e o r e t i c a l l y t h e above three questions are intim-

ately related that they should always go together as t h e t h r e e p a r t s o f one
question, for,as rsill be seen,themathematics that gives the right answw t o
any one of t h e above parts also gives the answer t o t h e o t h e r two parts.
The material which follows i s divided into three parts: P a r t I1 uses the
r e s u l t s of ‘the derivation of P a r t s III and I V t o give the reader a numerical
example of t h e use of the derived design equations; P a r t I11 gives the derivation
of the design equations f o r the constant-K configuration or coupled-resonant-
c i r c u i t f i l t e r s ; and Part I V gives the equivalent derivation for the m-derived
configuration filters.

e8

Authorized licensd use limted to: IE Xplore. Downlade on May 10,2 at 19:0354 UTC from IE Xplore. Restricon aply.
 11. Numerical Ecamples

A-1. Numerical E a n p l e Br the Constant K Configuration Filters of Fig. 2

1-r t h e s e f i l t e r s , which supply n attenuation zeros and no attenuation
poles, the three pairs of design equations are Equations 6 and 11, Equations
7 and 12, andEquations 8 and 13.
As a numerical example of the useof these design equations, we con-
s i d e r a p r a c t i c a l c a s e which a r i s e s i n t h e v e h i c u l a r communication f i e l d . In
t h i s f i e l d t h e number of users and p o t e n t i a l u s e r s is t o f a r outrun
t h e number of availablechannels. There is therefore desire t o greatly
increase the numberof available channels. An obviousprocedure is t o use, i n
the receivers involved, the smallest possible acceptance bandwidth (consistent
w i t h t h e i n t e l l i g e n c e bandwidth and audio S/N r a t i o r e q u i r e d , and with the
temperatme, shock, vibration, hunidity, etc. to which o s c i l l a t o r s , r e s o n a t o r s
e t c . w i l l b e s u b j e c t e d ) and then have t h e h i g h e s t p r a c t i c a l r a t e of cutoff from
t h i s a c c e p t bandwidth, i.e. have t h e s m a l l e s t p r a c t i c a l r e j e c t bandwidth
that adjacent channels can be placed as near as p o s s i b l e t o each other. Since
superheterodyne receivers are, of necessity, used, the above s e l e c t i v i t y would
be erected i n t h e IF amplifier portion of the receiver, and therefore the de-
signer must of reckon t h e w e l l known abrogating phenomena of spurious
responses,crossmodulation,intermodulation,desensitization,etc. However, it
i s s t i l l t r u e t h a t t h e I F s e l e c t i v i t y is t h e major r e c e i v e r c h a r a c t e r i s t i c which
definesthepossibleadjacentchannelspacing. One suggestionforthedesired
s e l e c t i v i t y o f t h e I F amplifier unit of these s o called split-channel receivers
is as follows: a t o t a l 6 db down bandmLdth of 17.5 Kc and a t o t a l 100 db
bandwidth of 30 kc, i.e. BWa 17.5 kc; and BIAr BW1- 30 kc.

Given t h e above requirement, the receiver designer must be able t o accur-

a t e l y answer the f o l l u w i n g three questions: huwmany resonators, and what
passband peak-to-valley ratio must be used in the receiver t o supply the above
r e q u i r e d s e l e c t i v i t y ; and, f o r givenpassbandmidfrequency, &at is the l ow -
e s t unloaded Q which these resonators my have; t h i s last question can of course
be s t a t e d in i t s inverted f o m as follows: f o r given available unloaded reson-
a t o r Q what is the highest passband midfrequency which can be used?

The answers tothesequestionsareobtained as follows: f i r s t s u b s t i t u t i n g

in Equation 6 t h e above f a c t s that(BYB10o&/BW&) 1.71 and t h a t ($/Va) 2,
and (Vp/Vr) 1 6 we obtain the answer t h a t t h e minimum number of resonators
which can beused t o s u p p l y t h i s d e s i r e d s e l e c t i v i t y , when the limiting case of
complete rippling, i.e. 6db r i p p l e s , is allowed in thepassband, is 9.75 resona-
tors. Substituting this and (Vp/Va) 2 i n Equation 11we s e e t h a t
the resonators usedmusthave unloaded Q's which are greater than 109 times
t h e f r a c t i o n a l midfrequency (fo/BWb&) which is used.

N e x t s u b s t i t u t i o n in Equation 7 shows that t h e minimum number of resonators

which can be used t o supply t h e r e q u i r e d s e l e c t i v i t y , with the other limiting
case of zero db r i p p l e s in the passband, is resonators.Substitutingthis

Authorized licensd use limted to: IE Xplore. Downlade on May 10,2 at 19:0354 UTC from IE Xplore. Restricon aply.
 and (Vp/Va) 2 i n Equation 1 2 see t h a t t h e r e s o n a t o r s used must
haveunloaded Q ~ s which a r e g r e a t e r t h a n l3.b times the fractional mid-
frequency ( f o ~ 6 d b which
) is t o used.
A graph of the type shown i n Fig.8 helps t o show t h e r e l a t i o n s h i p between
t h e four important variables considered i n t h i s paper, and t h e above two p a i r s
of values give the 6 db and zero db r i p p l e p o i x t s on the curve marked "1 stage."
The i.ntermediatepoints on this curve are obtained by i n s e r t i n g a s p e c i f i c
(V,/Vv) r a t i o i n Equation 8, s u b s t i t u t i n g v a l u e s o f n between 9.75 and and
constructing a l i t t l e a u x i l i a r y g r a p h o f the resulting (EWloodb/aJgdb) versus n;
where this curve crosses the (~1gr)db/~o16db) 1.71 l i n e we then accurately read
therequiredvalue o f nIC f o r this (Vp/Vv) r a t i o being used. S u b s t i t u t i n g t h i s
value of nIC and (Vp/Vv) i n Equation 13 gives the possible unloaded Q t s
which the nIC resonators must possess.

It w i l l be seen from t h e top curve of Fig.8 t h a t by c a l l i n g f o r a l m o s t

i n f i n i t e s i m a l l y small r i p p l e s f o r example le12 db (whichof course w i l l never
be seen in the response curve') instead of a Itflatt1passband, the designer can
use an appreciably smaller number of resonators i n h i s design, if he has avail-
able resonators with the correspondin requbed unloaded Q (andofcoursehe
must correctly synthesize the network

Irmering the Required Unloaded Q by Cascading Like Stages

The above section considered the numberof resonators and t h e i r unloaded
Q r e q u i r e d t o supply t h e d e s i r e d s e l e c t i v i t y in a one-stage filter, i.e.
no i s o l a t i o n between any i n t e r n a l r e s o n a t o r s of t h e f i l t e r .
A well-known fundamental procedure which enables a r e q u i r e d s e l e c t i v i t y
t o be produced w i t h l e s s unloaded Q f o r a given nmber of resonators (or with
fewer resonators for a given available unloaded Q) i s t o cascade N e s s e n t i a l l y
identical stages of selectivity, betweenwhich e s s e n t i a l l y complete i s o l a t i o n
exists, i.e. there is negligible %ouplinglf between the last resonator of one
and t h e first resonator of t h e stage. When t h i s is done t h e db down
defining the edge of the accept band and t h e db down defining the edge of the
r e j e c t band f o r , e a c h of the cascaded stages is of course the reqtiired overall
db down divided by t h e number ( N ) of cascaded stages, i.e. (6/N)db and (lOO/N)db
for our example, and similarly t h e db r i p p l e t o be used with each stage is the
o v e r a l l db ripple divided by N.

Equations 6 and U, Equations and 12, Equations 8 and 13 a r e t h e n

used exactly as a one-stage f i l t e r i n o r d e r t o find t h e number of required
resonators per stage and the required m inimum unloaded Qj t h e t o t a l number of
requlred resonators i s then of course N t h e n required per
The curves for 3, 4 and 5 stages on give the numerical values
which result, and see,the required unloaded resonator Q can be lonered
by m r e ' t h a n a n o r d e r of magnitude by this cascading procedure.

Authorized licensd use limted to: IE Xplore. Downlade on May 10,2 at 19:0354 UTC from IE Xplore. Restricon aply.
 The usual of i s o l a t i n g cascaded stages is t o s e p a r a t e themby vacuum
tubes and when, as is usual, it is necessary t o b u i l d up a required gain, as
well as supply a r e q u i r e d s e l e c t i v i t y , ' t h i s i a a desirable method ( i t should be
realized, however, t h a t t h e f i r s t tubes i n the chain will of course not be pro-
t e c t e d by t h e r e s u l t i n g o v e r a l l s e l e c t i v i t y ) . It is also mrth mentioning as
a p r a c t i c a l p o i n t t h a t h e n t h e magnitude of the voltage gain o r the transducer
loss of t h e f i l t e r is notimportant, one canproduce the e f f e c t of isolation
between s t a g e s by using a r e s i s t i v e o r r e a c t i v e a t t e n u a t o r between
which produces s u f f i c i e n t a t t e n u a t i o n so that there is no r e s u l t i n g i n t e r a c t i o n
betweenstages.This method even w i t h i t s r e s u l t a n t r a t h e r h i g h loss, is some-
tinies advantageous when it is undesirable to use vacuum tubes.

B. h w r i c a l ESranrple f o r t h e &Derived Configuration Filters of Fig. 5

For t h e s e f i l t e r s , which supply n attenuation zeros andaneven number
(n-1 o r n ) o f attenuation poles, the three pairs of design equations are Equa-
t i o n s 22 and 32, Equations 23 and 33, and Equations 24 and 34.

numerical example of use of these equations we w i l l consider the

p r a c t i c a l problem given i n t h e secondand third paragraphs of PartII. 'Re
will thus be able t o see if t h e added complication of the m-derived configuration
r e s u l t s in any saving with reference to the number of required resonators or un-
loaded Q.

First, considering the case of completeChebishev accept and r e j e c t band

response, we s u b s t i t u t e in Equations 22 t h e f a c t s t h a t (Vp/Va) 2 and ( V p / V r ) = l d
and f -585,and after looking up the log qfobtain t h e an-swer t h a t " c C , t h e
number of arm which can be used, is 6.77 and therefore t h e minimum i n t e g r a l
number of resonators which can be used is 10. Then, using the number k c 6.77,
f .585 sin 35.80 therefore f ' sin 54.20, 1.73 therefore
1, Kf 1-74, Kv 1.57, therefore A9 .257 and Bc .OgO5. We obtain from
Equation 32 t h e f a c t t h a t the 6.77 arms used must have.unloaded Q's which
greater than 64.3 times the f r a c t i o n a l midfrequency (fo/W&jb) which is t o be
used.
N e x t , f o r t h e other limiting case of no r i p p l e s in the accept band, i.e.
Taylor accept-Chebishev reject-band behavior, substitution in Equation 23 pro-
duces as, 9.75, i.e. a minimum of 9.75 arm o r 15 resonators w i l l be required
with t h l s response shape; and using this 9.75 f o r Equation 33 that
Ai and Bt= 1.255, we f i n d t h a t t h e r e s o n a t o r s used must have an unloaded
Q which i s higher than 8.2 times t h e f r a c t i o n a l midfrequency (fo/BW6&) which
is t o beused.

The above two p a i r s o f values give the 6 db andzero db ripple points on

t h a t curve of Fig. 10 which is marked "1 stage".

The intermediate points on t h i s curve are obtained b solvingEquation 24

f o r nICC by the method given immediately b e l m Equation 2 l j and then using these
values o f nICC a n d t h e values of f f Kf, Kv obtained from Equation 24
t o solve Equation 34 f o r the corresponding requbed value of unloaded Q. Thus,

Authorized licensd use limted to: IE Xplore. Downlade on May 10,2 at 19:0354 UTC from IE Xplore. Restricon aply.
 f o r example, i f we use a response shape having a 0.1 db r i p p l e i n the pass band
t o SUPPlY required(m~lOOdb/~%db) 1.71, we f i n d t h a t 7.76 arms are re-
quired and the mloaded Q of t h e resonators used must be greater than 19.4 times
t h e f r a c t i o n a l midfrequency being used.

The discussion i n Section A-2 concerningthecascading of like stages i n

order t o reducetherequiredunloaded Q, of course, apDlFes t o m-derived
configurations. On the basis of theprocedurediscussedthere,the two and
three-stage curves on Fig. 10 were calculated and plotted.

111. The Vonstant Kt' Configuration

A . TheOptimum Response ShapeWhichCan Be Obtained With

the Inverse A r m (i.e. Constant K) Confieuration

For more than two decades2 it has been r e a l i z e d t h a t if, by straightforward

application of Kirchoff one s o l v e s f o r t h e t r a n s f e r impedance of t h e con-
s t a n t K configuration of FSg.2A, i.e. t h e r a t i o of (Vout/Ih); one obtains
numeratorwhich is a constant, L e . it i s not a function o f the frequency vari-
able, and a denominator which is simple polynomial i n consecutive descending
powersof t h e complex frequency variable

the highest powern of j (M/f w i l l be the number of arms used i n the configura-
t i o n o f Fig.28; and t h e c o e f f i c i e n t s of the polynomial be real, independent
of fkequency,andformedfrom variouscombinationsof the c i r c u i t c o n s t a n t s . The
above is true not only when lossless resonators are used, b u t a l s o when f i n i t e Q
resonators are used.(Seenotes 2, 3 . The peakresponse w i l l o f course be ob-
tained when t h e above polynomial has i t s minimum value and so the shape o f
attenuation curve, i.e. t h e r a t i o of V peak t o the complex V a t any frequency
will be in t h e form

where A&-, stands for the value of the magnitude of the bracketed
complex polynomial.
It is now well known t h a t even with dissipation present it i s p o s s i b l e t o
so adjust the coefficients of Equation 1 t h a t t h e shapeof t h e magnitudesquared
of Equation 1 is as shown in Fig.4 and i s given analytically by Equation

Authorized licensd use limted to: IE Xplore. Downlade on May 10,2 at 19:0354 UTC from IE Xplore. Restricon aply.
 l] cosh2 i n

When the configurations of a r e used f o r f i l t e r i n g purposes, the

important reason it i s desirable t o a d j u s t them so t h a t t h e i r amplitude re-
sponse is given by Equation 2 is as follows: for a given numberof resonators
(n), and a -given ripple (i.e. TJp/Vv) i n the accept band, Equation 2 gives the
sharpest rate of cutoff between the accept band and t h e r e j e c t band which can
physically be produced by the c i r c u i t s of Fig. 2 Thus through the medium of
Equation 2 t h e f i l t e r designer is i n the wonderful position of knowing t h a t when
constant K type networks a r e used, it is physically impossible t o produce a
sharper rate of cutoff than that produced by Equation 2, and a design whichhe
doesbased on t h i s equation w i l l c a l l f o r t h e s s t number of resonators it
is possible to i n the
configuration The of normalizing bandwidth
mv in Equation 2 is the skirt bandwidth where the attenuation is the (VpVv)
r a t i o , and in general t h i s Ifvalley bandwidth11 is not equal to the required
"accept bandwidth" given i n the selectivity specifications. It is therefore
necessary tn modify Equation t h a t our attenuation shape is expressed i n
terms of the <VpV and of the selectivity requirement. This modification B
1
1,
is performed i n t f e next section.

A - 1 . Complete Chebishev Behavior,"IncompleteChebishev"Behavior,

and Butterworth Behador
It is important for the reader to realize that as shown i n
EViv i n Equation 2 i s the bandwidth between the points on the skirts which a r e
t h a t number of db dorm equal to the peak-to-valley ratio. Now only for the
l h i t i n g case which we rill c a l l ncomplete Chebishevn behavior, as shown i n
is t h i s BWv identicalwiththeaccept bandwidth specified in f i l t e r
requirements. In many practical cases the edge of the accept bandwidth may be
s p e c i f i e d f o r example a s t h e 6 db down points whereas the allowable ripple i n
the passband may ,be limited t o one-tenth db ( f o r example This type of re-
sponse, shown in Fig.kB, we will c a l l "incomplete Chebishevn behavior. Finally
there is the other limiting case where, f o r a required accept bandwidth, the
peak-to-valley ratio is made t o just reachzero db. This type ofresponse,
s h m in w i l l caU. Butterworth response Fig..!$,
Evenwhen there is no limitation on the (V Vv) r a t i o i n s i d e t h e
accept bandwidth ma,there is very important praceical reason an "in-
c o q l e t e Chebishev" response may have t o be used this reason, which will be
considered in d e t a i l in the section, is as follows: f o r a required rate
of cutoff the unloaded Q requbed for the resonators of orill become less
and less. (And of course the required numberof resonators will become greater
and greater; exact number required be obtained from Equation 8.)

To obtain the shape of the ncomplete Chebishev'l response i n terms of

the specified bandwidth, B TR, we make the valley bandwidth in Equa-
t i o n be equal to the required accept bandwidth Solving f o r t h e bandwidth

93

Authorized licensd use limted to: IE Xplore. Downlade on May 10,2 at 19:0354 UTC from IE Xplore. Restricon aply.
ratio we g e t Equation 3,

Ln order to get the shape of attenuation curve terms of the

accept bandwidth BW a f o r t h e "incomplete Chebishev" response shape must use
Equaticn 2 twice, f i r s t a t and then a t any other BW on t h e sMrt. Dividing
the resulting two equations we obtain Equation for the bandwidth r a t i o ,

cosh cosh-l

Finally r a t i o i n s i d e the accept band

approachzero, infinity, we see that Equa-
t i o n 4 becomes i n

Bw

solving Equation 5 for (Vpfi) obtain Equation sa, which will be

recognized t h e well known Butterworth o r maxin~llyflat response shape,

Thus we that when the edge of the accept band i s by

points outside t h e valley-db-dm points, then the limiting of Cheb-
ishev response is the B u t t e m r t h response.

Authorized licensd use limted to: IE Xplore. Downlade on May 10,2 at 19:0354 UTC from IE Xplore. Restricon aply.
A - 2 . The Hmber of Resonators Required t o Produce t h e
Attenuation Shapes of the Preceding Section
We a r e now i n a p o s i t i o n to o'ctain our desired design equations for
constant K configurations. Solving Equation 3 f o r t h e numberof arms, and
tinerefore resonators, required t o produce t h e d e s i r e d r a t e of cutoff when t h e
limiting case of "c3mplete Chebishevtt behavior can be used, we o b t a i n ~ q u a t i o n 6 ,

where i n p l a c e of the bandwidth BW a t any attenuation ratio (V#),,We

have used the spec if&( V p h r ) and (WPr) of t h e s e l e c t i v i t y requireme*.
This equation (Equation 6 ) is t h e f i r s t o f our desired design equations f o r
constant K type networks. When completeChebishevbehavior is allowed, the
obtained from Equation 6 is the smallest n which it is p o s s i b l e t o in
order t o obtain the required selectivity.

N e x t let us f i n d t h e lxrmber of arms (and therefore resonators) re-

quired t o produce the desired rate of c u t o f f when other limiting case,
Butterworth behavior, is used. Solving Equation f o r n we obtain Equation

This i s t h e second of our desired design equations for constant IC

type networks. The required number of arms and therefore resonators,
obtained from it w i l l be appreciably greater than t h a t c a l l e d f o r by Equation
6, but remember t h a t , as be shown i n t h e section, the unloaded Q
which the resonators must possess rill be appreciably less than that required
f o r t h e r e s o n a t o r s of Equation 6
Equation 6 and Equation 7 give us respectively the smallest number
and t h e l a r g e s t number of resonators which can be used t o produce a required
s e l e c t i v i t y with t h e optimum Chebishev type of response described by Equation 2.

To find the passband r i p p l e which must be accepted when the numberof

resonators is between and a, must useEquation and solve for "IC
with specific value of p e a k - t o - v a l l e y r a t i o i n s e r t e d i n Equation 4; (or solve

95

Authorized licensd use limted to: IE Xplore. Downlade on May 10,2 at 19:0354 UTC from IE Xplore. Restricon aply.
 f o r t h e r e s u l t i n g p e a k - t o - v a l l e y r a t i o with a specific number of arms nIc
more than Q~ but less than IQ i n s e r t e d in Equation Unfortunately, it
does not seem possible t o solve Equation 4 and obtain an e x p l i c i t e q u a t i o n f o r
n o r (Vpflv) A procedure is not very elegant but which i s actually not
too time-consuming i s t o merely r e m i t e Equation 4 as Equation 8,

Then, i n s e r t a specific value o f ( V p / o ~ )i n t h i s equation; pick values

of n between nc and and construct a small auxiliary graph of t h e r e s u l t i n g
(Bwr/BWa) versus n; t h i r d , where this curve crosses the required value of (Bwr/Bpya)
accurately read the corresponding necessary value of nTc,

This Equation 8 i s t h e t h i r d of our desired design equations f o r constant-::

typenetworks, and together w i t h Equations 6 and gives u s t h e a b i l i t y t o answer
the f i r s t two p a r t s of the question stated in Section I - C of t h i s paper, f o r the
constant-K configurations of Fig. 2.

B. The h e s t Unloaded Resonator WhichCan Be Used

To Produce a Desired Attenuation Shape
The answer to the Very i m o r t a n t t h i r d p a r t of the question stated i n
Section I - C of this paper (For a given mid-frequency, what is the minimum un-
loaded Q which these resonators must possess?), i s obtained from the following
reasoning.

As is well known, in o r d e r t o write t h e complex attenuation expression

which, together the generator-load-resistance ratio, specifies a lossless
network that will prqduce a desired attenuation magnitude shape, we must solve
f o r the 2n r o o t s of t h e d e s S e d magnitude equation, and then those n of the
2n r o o t s which are in t h e l e f t half plane, t o form the required complex. p l y n o -
mid. Fig.7 shows a t y p i c a l r o o t p o s i t i o n p a t t e r n f o r constant K type filters
f o r the of .n

However, wish to produce our desired attenuation shape not by

lossless ( L e . i n f i n i t e Q) resonators but with resonators having f i n i t e Q. It
is therefore necessary to make the attenuation zeros o f the d i s s i p a t i v e network
coincide with those of the desired attenuation shape. The next paragraph shows
how t h i s c a n be accomplished.
Darlingto& and Bode(s have shown t h a t if one is luckily dealing net-
works having a t1unifonnf1 type of dissipation, i.e if a l l the resonators of Fig.2
have the same unloaded Q, then by a simple procedure it is possible t o o b t a i n t h e

Authorized licensd use limted to: IE Xplore. Downlade on May 10,2 at 19:0354 UTC from IE Xplore. Restricon aply.
 complex polynomial which, w i t h t h e r a t i o of generator resistance t o load resist-
ance, specifies the reactances of the required network.

The procedure is t o simply move the attenuation zeros o f Equstion 9 and

Fig.6 closer t o t h e r e a l frequency by an amount equal to the inverse of the
unloaded Q possessed by the resonators; then when these newly positioned left
half plane zeros are used they w i l l define the reactances of the network,
which w i l l produce our desired attenuation shape when the reactance elements
used have the assumed f i n i t e unloaded Q.

Thus t h e well known requirement that attenuation zeros of a physically

r e a l i z a b l e network cannot appear in the right half plane plus the principle of
the above paragraphimmediately fixes the lowest allowable resonator which
can beused t o obtain a desired attenuation shape; f o r , r e f e r r i n g t o Fig.7 we
see that the smallest allowable resonator unloaded Q which can be used t o produce
the attenuation defined by this root plot is equal t o the inverse of the perpen-
dicular distance between t h e jF axis and that zero which is c l o s e s t t o t h i s axis.
h e next s e c t i o n a p p l i e s t h i s p r i n c i p l e t o find the design equations giving the
minimum unloaded resonator 0's which can be used when the desired attenuation
shape is the optimum Chebishev attenuation shape.

B-1. The Unloaded iiesonator Q's WhichCan Be Used

To Obtain "Complete Chebishevtl, lKhcomplete Chebishev"
and Butterworth Attenuation Shanes
As outlined in the previous section, we must f b s t solve for the 2n r o o t s
i.e. attenuation zeros of our desired attenuation magnitude equation.
For general Chebishevshape o f Equation 2 obtain Equation 9, wbich
gives the r e a l ( r ) and imaginary (i)coordinates o f these attenuation zeros, on
t h e complex frequency plane whose have the dimensions of (m)
35

rm sin (2m -1) 900

Bwv/fo n

where

97

Authorized licensd use limted to: IE Xplore. Downlade on May 10,2 at 19:0354 UTC from IE Xplore. Restricon aply.
 By solvingEquation directly f o r i t s zeros(or by s u b s t i t u t i n gf o r
i n Equation 9 i t s value in terms of BOrr, as given by Equation and then
l e t t i n g t h e <Vp/Vv) r a t i o approach u n i t y ) we obtain Equation 10 giving the re-
quired root locations for Butterworth attenuation shape,

The e l l i p t i c a l p a t t e r n formed by plotting the zero positions given by

Equation 9 is now well known and is shown in F i g . 7 f o r case of n
Equation 10 gives the limiting case which is a circular pattern.

From Fig. 7 and Equations 9 and 10, f i n d t h a t s e t t i n g m equal to unity

gives the root nearest the j F and therefore, i n accord w i t h the p r i n c i p l e of
Section I I I - B y Equations 9a and loa, with m s e t e q u a l t o u n i t y , Kill give directly
the inverse of the lowest unloaded resonator Q whichcan be used to produce the
optimum Chebishev attenuation shape.
For t h e l i m i t h g case of t h e complete Chebishev response wherein valley
bandw3dth BWv is identical with the required accept bandwidth BRa Ire directly ob-
tain Equation 11 *om Equation

is obtained f r o m Equation 6. This i s the f o u r t h of our desb-ed design

-
n
s f o r c o n s t a n t K type networks and it gives the lowest unloaded resonator
Q which can be t o produce t h e “complete Chebishevfl attenuation shape.

.For the other limiting case of the B u t t e m r t h response wherein the (Vp/Vv)
r a t i o is unity Ire directly obtain Equation 1 2 from Equation loa,

Authorized licensd use limted to: IE Xplore. Downlade on May 10,2 at 19:0354 UTC from IE Xplore. Restricon aply.
when is obtained f r o m Equation This is t h e f i f t h of our desired design
equations for constant K type networks.

Finally f o r the incompleteChebishev attenuation shape where the accept

barsdwidth ma is outside the valley bandwidth must s u b s t i t u t e for BWv,
in t h e root location equations Equation 9 , i t s value in terms of EWa; t h i s rela-
tionship i s of co&se given by Equation and when t h e s u b s t i t u t i o n is made
obtain Equation 13, the s i x t h and last o f our constant K network design equa-
tions,

where nIc and I/ (VpDv)2 1 are not independent of each other, but are
obtained from the solution of Equation 8.

These t h r e e pairs of equations, Equations 6 and 11, f o r t h e l l m i t i n g comp-

l e t e Chebishev shape, Equation 7 and 12 f o r the other limiting case of Butter-
response, and ions 8 and 13 f o r the general case of t h e incomplete muat
Chebishev response, give us the a b i l i t y t o answer completely and exactly
three parts of t h e question stated in Section I - C , when use t h e configura-
t i o n s of Fig. t o produce required selectivity. The numerical example
given previously i n Section 11-A-1 i l l u s t r a t e s t h e use of these three pairs of
equations.

IV. The ~"4lerivednConfigurations

The Opmit um Attenuation Shape Which Can Be Obtained

With t h e "&Derived" Configuration
By the straightforward application of Kirchoff Laws t o t h e b a s i c m-derived
configuration of F'ig.sA f i n d that lossless elements r a t i o of t h e peak
output voltage t o t h e complex output voltage a t any frequency, L e . the shapeof
the atternration cume i s given by t h e r a t i o o f tm polynomials as in Equation U,

where A& is t h e minimum value of the magnitude of the bracketed expression;

99

Authorized licensd use limted to: IE Xplore. Downlade on May 10,2 at 19:0354 UTC from IE Xplore. Restricon aply.
F stands f o r our bandpass frequency variable (EC;V/f and c o e f f i c i e n t s ( U ) of
t h e numeratorpolynomial real, independent of frequency, and formed from
various combinations of the circuit constants; and t h e s p e c i f i c f r a c t i o n a l
bandwidths F F etc. are thefl-actional bandwidths a t which infinite
attenuation i s produced.

It is important f o r the reader t o r e a l i z e t h a t n i n Equation is not a

t h e number of resonators being used, b u t is t h e number of arms i n t h e b a s r
ladder of Fig.bA. To f i n d t h e number o f resonators required use t h e rela-
tionships given a t the end of Section I-A.

If t h e number n of arms being used i s odd, then there will be (n-1)

denominator factors; i f n i s even,there n denominator factors. wiln
In Darlingtonfs classic paper published in 1939,k it is pointed out that
it i s p o s s i b l e t o a d j u s t t h e c o e f f i c i e n t s and infinite attenuation frequemies
ofEquation t h a t t h e shape of t h e magnitude squared of Equation is as shuwn
i n n g . 6 and is given analytically by Equation 15, (9,

where

Thenumber of arms n which m u s t be used in Equation 15 cannot be picked

independently of and f ; but must always satisfy the relationship between t h e
t h r e e quantities given by Equation 18,

where q is the so-called modular constant o f t h e modulus given by the subscript.

The function l o g qk is tabulated on pages 49 51 of t h e 1945’e d i t i o n of Jahnke
Emde; and the Smithsonian E l l i p t i c Function Tables compiled by RM Spenceley
contains a very useful short Appendix dealing with the numerical computation of
the various elliptic functions. (It will be noted that f o r modulus values (k)
less than 0.1 (say) the corresponding modular constant is simply qk (k/L)2.
In Equation 15 t h e symbol%) s t a n d s f o r t h e r a t i o of t h e two e l l i p t i c f u n c t i o n s
cn over dn, where the subscript o r f is t h e modulus of the function; and KT
and Kf the complete e l l i p t i c integrals of the first kind, evaluated for the
modulus value given by the subscript.

100

Authorized licensd use limted to: IE Xplore. Downlade on May 10,2 at 19:0354 UTC from IE Xplore. Restricon aply.
 As we see, the modulusv (Equation 16 i s immediately set by w t o voltage
r a t i o s , i.e the m m
iu m allowable ripple in t h e passband (V/V,) and t h e minimum
required attenuation in t h e r e j e c t band (V /Vh). Similarly the modulus f (9qua-
t i o n 17 is immediately s e t by t h e r a t i o
bandwidth(whichexcept
two bandwidths, i.e. the ripple
f o r completeChebishevpassbandbehavior will not be
OF
the required accept bandwidth) and the tthilltt bmdmidth (which w i l l be the re-
q u i r e d r e j e c t bandwidth only if completeChebishev r e j e c t band behavior i s used.

When the attenuation requirements are as s t a t e d in the introduction, and

the configurations of Fig.5 a r e used t o s a t i s f y t h i s t y p e of at.termation re-
quirement, the reason f o r a d j u s t i n g t h e network so t h a t t h e amplitude response
of Fig.6 and Equation 15 is O b t a b d , is exactly si;nilar to that given for using
Equation 2; i.o f o r a given number of arm (n) and a given allowable peak-to-
V a l l e y r a t i o ('Jp/Vv) i n the accept band,and a given required minimum attenua-
tion, i.0. peak-to-hill ratio (Vp/Vh) i n t h e r e j e c t band,Equation 15 gives t h e
s h a r p e s t r a t a o f cutoff which can physically be produced by t h e c i r c u i t s of Fig.s4

Equation 1Sa takes on two d i f f e r e n t f o r m depending upon t h e p a r t o f t h e

a t t e n u a t i o nc h a r a c t e r i s t i c i n h i c h one is i n t w e s t e d . -%om t h e n i d d l e o f the
passbandout t o the valley bandwidth,i.e. when (3?/3iiv) i s less than uni-ty,
Equation 1Sa s t a y s as given. I n thocutoffregionbetveon ths of valley
bandwidth and t h e edge of the h i l l bandwidth, i . 2 . whare varies betwee:l
u n i t y and l/f, Equation turns i n t o Equation 1%,

When t h e rime indicatesthecomplinentary modulm, i . 0 . v' JT

and f '
bracketedexpression).
(Note also theinversion of the bandwidth r a t i o i n the
Making use o f t h e d e f i n i t i o n s of Equations 16 and 17 we
,/d.
canimmediately change Equation 1% t o Zquation l j c , which gives attenuation i n
terms of the h i l l bandwidth Bffh and the peak-to-hill r a t i o (Vp/Vv).

2
vP
V

Finalxv o u t s i d e t h e h i l l band-xidth, i.s where (E;ii3JV) is greater than

l/f, Equation 1sa turns into Equation lsd,

L? 1

Authorized licensd use limted to: IE Xplore. Downlade on May 10,2 at 19:0354 UTC from IE Xplore. Restricon aply.
 vP
V

andof course the relationships o f Equations 16, 17, and 18 apply t o lsa, 1.%,
l s c , and 1sd.

It i s important for the reader to r e a l i z e that the two normalizing band-

widths BYv and BWh i n t h e above equations are the two bandwidths,on t h e skirts
of the response, where the attenuation i s t h a t db d m equal t o the (Vp/Vv) ratio
and (Vp/Vh) r a t i o r e s p e c t i v e l y . I n general,thesenorrnalizingbandwidths will
not equal the accept and r e j e c t bandwidths specified i n a s e l e c t i v i t y requirement
and it i s therefore necessary to modifyEquation 15 t h a t our attenuation shape
is expressed in terms of the (Vp/Va) and (VD/Vr) and BWa and B Ur of the select-
i v i t y requirement.Thismodification i s performed in the next section.

A-1. Complete Chebishev Accept Band -Complete Chebishev Reject Band Behavior;
Incomplete Chebishev Accept Band -Complete Chebishev Reject Band Behavior;
and Taylor Accept Band-Conplete Chebishev Reject Band Behavior
As shown in fig.sb, EWv in Equation 15 i s t h e bandwidthbetween t h e p o h t s
on t h e skirt where the attenuation is equal t o the peak-to-valley ratio (Vp/Vv);
and BW'h i n Equation 15 i s the bandwidthbetween the points on the skirt where
the attenuation is equal t o the peak-to-hill ratio (Vp/Vh). Now only f o r the
case shown i n Fig.6a which will c a l l "complete Chebishev accept and r e j e c t
band behavior" does the valley bandwidth BWv equal the required accept banrfffidth
Ma; and t h e h i l l bandwidth EWh equal the required reject bandwidth BWr. For
t h i s l i m i t i n g c a s e we therefore obtain our shape equation in terms of the speci-
f i e d bandwidths b y s i n p l y s e t t i n g i n Equation 15b equal to the required
accept bandwidth ma, o r by s e t t i n g BTfilh in Equation 1sc equal to the required
r e j e c t bardwidth BWr. Solving f o r bandwidth r a t i o obtain Equations 19s
o r 19b,

where v, f , ' n are obtained fromEquations 16, 17, and 18, with (Vp/Vv) 3
(Vp/V,>, and <vp/V,> (Vp/Vr>

10

Authorized licensd use limted to: IE Xplore. Downlade on May 10,2 at 19:0354 UTC from IE Xplore. Restricon aply.
 The next type of response considered i s t h a t shown i n Fig.6b,where t h e
peak-to-valley ratio is l e s s t h a n t h e db down defining the edge of the accept
band; b u t t h e h i l l bandwidth i s still t h e same as the required reject band-
width. IncompleteChebishevreject-bandbehavior notbeconsidered in
t h i s paper because i n most cases it i s p r a c t i c a l t o demand completeChebishev
r e j e c t band behavior, with i t s attendant higher rate o f cutoff.
In order t o obtain the shape of the attenuation curve i n term of t h e
specified accept bandwidth B91a f o r t h i s case of incomplete Chebishev passband-
complete Chebishev r e j e c t band behavior it is necessary to use Equation 1s
twice, first a t Bwa and t h e n a t a n y p o i n t on the curve BW; dividing the result-
i n g two equations w i l l give the equation equivalent to Fig.4 for the constant K
configuration. However, because we consideringcompleteChebishev reject
band behavior a simpler equation results if we express the attenuation shape
i n terms of t h e r e q u i r e d r e j e c t bandwidth r a t h e r t h a n in terms of the required
accept bandwidth. Setting i n Equation 1%e q u a l t o t h e r e q u i r e d r e j e c t
bandwidth BWr we obtain Equation 20,

The f i n a l type of m-derived configuration attenuation considered is that

obtained when, while s t i l l keeping t h e h i l l bandwidth equal t o the required
r e j e c t bandwidth, make t h e p e a k - t o - v a l l e y r a t i o i n t h e a c c e p t band approach
and just reach zero db. For this case v approaches 2er0 and therefore, no
matter how many arms ( n ) a r e used, f must approach zero due t o t h e r e q u i r e d
relationship Equation 18. Therefore f and a roachunity, and therefore
Equation 20 becomes, i n t h e limit, Equation fB

cosh
cosh 1

Solving Equation 21 f o r (Vp/V) obtain Equation ( 2 l a ) which is t h e

attenuation equation for the lirmting case of zero db r i p p l e in t h e passband,
an attenuation shape which will say has Taylor accept band- complete Chebi-
shev r e j e c t band behavior.

Authorized licensd use limted to: IE Xplore. Downlade on May 10,2 at 19:0354 UTC from IE Xplore. Restricon aply.
 As matter of academic i n t e r e s t it shouldbenoted that if one a lies
Equation 2 1 t o tuo points on the skFrt and then allows the (Vp/Vhyratio t o
become infinite, then the Butterworth equation (Equation sa) r e s u l t s . Simi-
larly Equation 15 would t u r n i n t o t h e Chebishev Equation 2 if t h e (Vp/Vh)
r a t i o is c o r r e c t l y made to approach infinity.

A-2. The M e r of Arms Required t o R o d u c t t h e A t t e n u a t i o n

Shapes o f the Preceding Section

We are now r e a d y t o o b t a i n our desired design equations for m-derived

configuration f i l t e r
For t h e l i m i t i n g case of complete Chebishev accept band and reject band
behavior Equations 19a and l9b apply, with the number of required arms n ob-
tained directly Equations 18, 16, and 17. These equations are rewritten
belw the proper subscripts as Equation 22,

and previously noted log qf c a n b e d i r e c t l y r e a d from the pages 49 51 of

Jahnlce and Equation 22 is t h e first of our desired design equations for En&.
m-derived configurations.

N e x t l e t us f i n d t h e number of arms r e q u i r e d t o produce t h e d e s i r e d

s e l e c t i v i t y when we use the other limiting case of Taylor pass band behavior
completeChebishev r e j e c t band behavior, as shown i n F i g . 6 ~ . Here Equation 21
or Equation 21a applies, and solving for n t h e number of arms required ob-
tain Equation 23,

cosh-I
ma
This is the second o f our desired design equations for m-derived configu-
r a t i o n s . The required number of arms obtained from it be greater t h a n t h a t
called for by Equation 22, but as w i l l be shown i n t h e next section the unloaded
Q which the resonators must possess be less t h a n t h a t r e q u i r e d f o r t h e re-
sonators of Equation

Authorized licensd use limted to: IE Xplore. Downlade on May 10,2 at 19:0354 UTC from IE Xplore. Restricon aply.
 Equations and 23 give us respectively the smallest number and the
l a r g e s t number of in Fig.5A which can be used t o produce a required sel-
ectivity t h e optimum a t t e n u a t i o n shape of gquation 15.
Mnally, we wish t o f i n d t h e number of arms n r e q u i r e d f o r t h e intermed-
iate case o f Fig.6B wherein the (Vp/Vv) in t h e passband is l e s s t h a n t h e db down
d e f i n h g t h e edge of t h e passband,butgreaterthanzero. For t h i s c a s e , we must
Equation 20 with Equations 16, 17, and 18.

Equation 20 a p p l i e d t o t h e edge of the accept band gives Equation a,

ma f

and the reader should realize that Equation 24 i s meaningless without Equations
16, 17, 18. Unfortunately, it doesnot seem possible t o obtain fromthese equa-
t i o n s a n e x p l i c i t e q u a t i o n f o r nEC, but the following procedure can be s t r a i g h t -
f omardly used.

Pick a s p e c i f i c (Vp/Vv) r a t i o . SolveEquation 16 f o r v; v 1

is then immediately known and from t a b l e s KV is obtained. Now f o r this v
prepare a small a u x i l i a r y graph o f n versus the resulting (BWr/BWa).The points
for the curve are obtained as follows: f i r s t pick an n ( l a r g e r t h a n but
smaller than and useEquation 18 t o s o l v e f o r f; f i s then
immediately known, and from t a b l e s Icf i s obtained; then second, use these values
of n, v', Kv, Kf, and f 8 in Equation 24 t o o b t a i n t h e r e s u l t i n g (aWr/BWa). ,Three
points usually s u f f i c i e n t , andwhere t h i s curve crosses the desired value of
(BWr/BWa) one reads the correct required value of nEC f o r t h e (Vp/V,) r a t i o
chosen.

Equation (with Equations 16, 17, and 18) i s t h e t h i r d of our desired

design equations for m-derived corfigurations, and, together with Equations 22
and 23, gives us t h e a b i l i t y t o answer the first two parts of t h e q u e s t i o n s t a t e d
i n S e c t i o n 1 4 , f o r t h e m-derived configurations of Fig. 4.

B. The Approximation Involved When F i n i t e Q Resonators AreUsed

i n the *Derived Configuration
m e n t h e c o n s t a n t K configurations of Fig.2 a r e ueed, t h e optimum attenua-
t i o n shape of Fig.4 and Equation can be exactly produced when f i n i t e Q resona-
tors used long as the unloaded Q1s higher than the limiting values
given by EQuations 11, 12, and 13 Unfortunately, when t h e m-derived config-
urations of Flg.4 a r e used, t h e optinnun attenuation shape of Equation 15 cannot
be e x a c t l y produced when f i n i t e Q resonators are used.
The t r a n s f e r impedance e q u a t i o n , f o r t h e f i n i t e Q network, Equation
s h m why t h i s is unfortunately the case; the numerator of the resulting trans-
f e r m e d a n c e i s s t i l l in e x a c t l y t h e f o r n shown in Equation but the denomi-

Authorized licensd use limted to: IE Xplore. Downlade on May 10,2 at 19:0354 UTC from IE Xplore. Restricon aply.
nator o f t h e t r a n s f e r impedance w i l l now contain both even and odd p m r s of
the frequency variable,

Another o f s t a t i n g t h i s latter point i s to say that the conjugate pairs of

r o o t s of the denominator (i.e. the conjugate attenuation poles whish were
e x a c t l y on the real frequency axis when infinite Q resonators are used) can
never be placed on t h e r e a l f r e q u e n c y axis when f i n i t e Q resonators are used
i n the configurations of Fig .5, but will always appear to the left of t h e r e a l
frequency axis by a distance exactly equal to the unloaded decrement of the
resonators; i.e. can be seen the root locations of t h e denominator o f Equa-
tion a r e (-d j F e ) . It is thereforeimpossibletoobtainexactlythe
optimum attenuation shape of Equation 15,for as we see from Equation 1Sd the
conjugate attenuation poles o f t h i s shape occur exactly on the real frequency
axis.
However, it should be realized that when t h e d e r i v e d response shape
i s used t o s a t i s f y a s e l e c t i v i t y requirement, the extreme attenuation produced
i n t h e r e g i o n of the Wdinite1’ attenuation frequencies is usually not of much
p r a c t i c a l importance f o r two reasons; first, the width of these l%nfiniten
attenuation cusps is usually a very small percentage of the t o t a l r e q u i r e d
width of r e j e c t band; and second, it is points where the attenuation
i n t h e r e j e c t band is a minimum, i . e . t h e % i l l f l p o i n t s , that t h e designer m u s t
worry about, with reference to meting his attenuation specifications.

Thus, if,with f i n i t e Q r e s o n a t o r s , i t is s t i l l possible to preserve the

passband shape and the reject band attenuation points, i.e. t h e h i l l
points, then the rounding off of t h e % n f i n i t e l l attenuation points is usually
o f negligible importance

We f i n d experimentally that i n most p r a c t i c a l c a s e s , if it is possible t0

a d j u s t t h e d i s s i p a t i v e network t h a t t h e numerator roots of Equation CO-
incide with the attenuation zeros o f the d e s i r e d , shape equation Equation 15,
and t h e denominator r o o t s o f Equation have the real frequency coordinates
as the attenuation poles of Equation 15 then, even though these poles are unfor-
tunately displaced to t h e l e f t by a n amount d, the resulting response shape
d i f f e r from the optimum shape mainly by having the infinite attenuation points
round o f f t o som f i n i t e attenuation which is almost always greater than the
required h i l l attenuation; while a t t e n u a t i o n a t t h e h i l l points, i.e. at
the attenuation points i n t h e r e j e c t band, will i n most p r a c t i c a l c a s e s
be negligibly reduced from the value given by Equation 1s.
This then i s the approximation t h a t we make in calculating the minimum
unloaded resonator which can be used in m-derived bandpass configurations
designed t o produce the optimum shape of Equation 15, i . e . place only t h e

106

Authorized licensd use limted to: IE Xplore. Downlade on May 10,2 at 19:0354 UTC from IE Xplore. Restricon aply.
 network poles in exactly the correct position, and alow t h e network z e r o s t o
be displaced a distance t o t h e l e f t from their correct position.

Section 111-B shows us how t o make the zero positions of a uniformly

d i s s i p a t i v e network coincide with the zero posifions of a desired attenuation
shape. As shown i n t h a t section,thelimitingvalue o f unloaded Q f o r which
i t i s possible to obtain the desired zero positions, i s given by Lhe hori-
zontal distance r l between the j F axis and the zero which i s c l o s e s t t o this
Thus given a desired attenuation shape, we must solvefor i t s zeros
i n order t o find the location of t h a t zero which is nearest the j F This
i s done in the next section for the optimum attenuation shape of Equation 15.

The bwest Unloaded Resonator WhichCan Be Used To Obtain

B-la the Attenuation Shapes of Section

S e t t i n g our desired optimum a t t e n u a t i o n shape equation, Equation lsa,

equal t o zero and solving we f i n d t h a t the r e a l (r and imaginary (i)coordin-
ates of the atteriuation zeros are given by Equation 26,

where the angles Ac Ek are given by Equation 27

It should be r e a l i z e d t h a t f o r modulus values as c l o s e t o unity is i n

practical cases one canusetheapproximation(sn/cn)-l sin h-1

In the design of the networks it i s necessary t o the required loca-

t i o n of t h e i n f i n i t e a t t e n u a t i o n points, and f o r t h e sake of completeness t h e
real and imaginary coordinates of the poles are given by Equation 28,

Authorized licensd use limted to: IE Xplore. Downlade on May 10,2 at 19:0354 UTC from IE Xplore. Restricon aply.
 Equation is obtained by simply s e t t i n g t h e denominator o f Equation 1Sd
equal to zero and solving f o r Bw g/fo.
1% should be realized that (cn/dn)k u snk(Kk 11 that values of
the (cn/dn) can be obtained simply from t h e s n t a b l e s .

For t h e l i m i t i n g of no r i p p l e s i n the pass band solve for the zeros

of Equation 21a and obtain Equation

BW,/fO s i n h2 Bt A$

i, cosh Bt Ai

m,.,/fo sin h Bt A$

where theangles At and Bt givenbyEquation 30

r 1

and a g a i n f o r the sake of completeness solving for t h e i n f i n i t i e s of Equation 21a

obtain Equation 31,

108

Authorized licensd use limted to: IE Xplore. Downlade on May 10,2 at 19:0354 UTC from IE Xplore. Restricon aply.
 The general pattern formed d e n t h e zero and i n f i n i t y locations are
p l o t t e d on t h e complex frequency plane are shown i n Fig.!?.

From Fig, 9 and Equations 26 and we f i n d t h a t s e t t i n g m equal t o

unitygives t h a t r o o t which i s c l o s e s t t o the j F axis. Therefore,following
t h e principle of Section 111-B, Squations 26 and w i t h m s e t equal t o
unity, will give directly t h e h i g h e s t unloadeddecrement (i,e. the inverse
o f the lowest unloaded Q) which can be used t o produce t h e zero position
p a t t e r n c a l l e d for by t h e optimum attenuation shapes o f Fig, 6 and Equations
15 and 21a.
For limiting response of Fig.Sa, i.e. completeChebishev accept and
r e j e c t band behavior where the v a l l e y bandwidth B#v is i d e n t i c a l w i t h the re-
quired accept bandwidth BWa and t h e h i l l bandxidth B'& i s i d e n t i c a l with t h e
required reject bandwidth BWr we o b t a i n d i r e c t l y f i o m Equation 26a t h e
Equation 32,

where the angles A$ and Bc givenbyequation 27 w i t h m s e t e q u a l t o u n i t y

and with f , and v obtained from Equation 22. Equation 32 is t h e f o u r t h of
our desired design equations for xu-derived configurations and gives the lowest
unloaded resonator Q which can be used t o o b t a i n t h e correct zero positions re-
quired by the completeChebishev accept and r e j e c t band attenuation shape of
FSg .6a.
For t h e other limiting case o f Taylor accept band behavior shown in F i g . 6 ~
we d i r e c t l y o b t a i n fromEquation 2% our next desired equation, Equation 33,

t h e angles A> and B t are given by Equation 30 w i t h m s e t e q u a l t o unity

and nt is obtained from Equation 23. This i s t h e f i f t h of our desired design
e q u a t i o n s f o r m-derived configurations

Finally f o r the incomplete Chebishev accept band -complete Chebishev r e j e c t

band behavior wherein the required accept bandwidth M a is outside the valley
bandwidth BWv ( b u t t h e r e q u i r e d r e j e c t bandwidth B Wr is s t i l l e q u a l t o t h e h i l l
bandwidth BWh) must s u b s t i t u t e Equation 26a the value of ENv in terms of
the specified acceptbandxidth ENa. With the understanding t h a t t h i s relation-
ship is given by Equation Equation 26a gives us Equation

109

Authorized licensd use limted to: IE Xplore. Downlade on May 10,2 at 19:0354 UTC from IE Xplore. Restricon aply.
where the values of f , f t , v, V I , Kv, Kf, and n are notindependentofeachother
b u t are the values obtained when Equation 24 i s solved by t h e method given im-
mediately below that eauation
mediatelv equation and the angles A and B are then obtained from Eaua- Equa-
t i o n 27. Equation 34 is t h e s i x t h ar,d last of our desired design equations for
m-derived configurations and gives the minimum unloaded resonator Q which can
beused t o o b t a i n t h e z e r o l o c a t i o n s r e q u i r e d by the response shape of Fig.%.

The numerical example given previously i n Part 11-3 i l l u s t r a t e s t h e u s e

o f these three pairs o f equations.

References Notes

1. I n figures SA, C, and D, the inductance of t h e n u l l producingresonators

(a,b,etc.) can be supplied by using magnetic coupling between the two reson-
a t o r s t o which t h i s i n d u c t o r is connected. The propervalue of coupling co-
e f f i c i e n t i s s o large, however and up), t h a t it seems proper t o c a l l t h e
a,b,etc.circuits %ruet1 resonators even i f t h e L of these resonators i s
supplied by magnetic coupling.

3. Dishal, Wesign of Dissipative Band-Pass f i l t e r s ProducingDesiredExact

Amplitude Frequency Characteristics, It Proc. BE, Vo1.37,pp.1050-1Cxj9, Sept.
1949; and Electrical Cornmication, Vo1.27, pp.56-81, March 1950.
4. P. I. Sichards,YJniversalOpthum Resonance Curves f o r A r b i t r a r i l y Coupled
Resonators," Proc. IRE, Vol.32, pp.624-629, Sept. 1946.

It is a s s m e d t h a t tcle reader realizes that the wgnitude equations given

by c l a s s i c a l f i l t e r theory are approximation equations which cannot be pro-
duced by physical networks.

6. Equation 5 is obtained from Equation by u s b g t h e f a c t s t h a t

cosh'l x log n (x>lC)

cosh y (Y eY 3)

7. S. Darlington,"SynthesisofReactanceFour-PolesNhichProduce a Prescribed
b e r t i o n Iass Characteristic," J o u r . Xath. Physics, Vo1.18, pp.257-353,
Sept 1939

Authorized licensd use limted to: IE Xplore. Downlade on May 10,2 at 19:0354 UTC from IE Xplore. Restricon aply.
 8. H. Bode, "Network Analysis andFeedback Amplifier Design," D. Van
Nostrand Co., pp.216-218.
9. The m i t e r has used the (cn/dn) elliptic function rather than the (sn)
e l l i p t i c f u n c t i o n because the former function one equation (i.e. Equa-
t i o n ls) can be written w-hich holds for both n odd and n even. Similarly,
the important zero and pole location equations, Equations 26 and 28,
apply t o both n odd andeven. Insofar as numerical work is concernedthe
f a c t t h a t (cn/dn) u sn (K-u) enables the (cn/dn)valuzs to be obtained
f r o m the more common sn tables.

It is important for the reader to realize that when the network is to

be synthesized WB wish t o produce t h e optimum shape of Equation 15
with an even number of arms, then in the basic node configuration of Fig.
SA we must use a s h o r t c i r c u i t t e r m i n a t i o n a t t h a t end of t h e network a t
which t h e s e r i e s arm occurs. And i f the msh network d i c h i s t h e dual
of Fig.SA i s to be used, then must use an open circuit termination a t
which t h e shunt a r m occurs. The reason f o r t h i s requirement is that a n
even n i n Equation 15 produces f i n i t e attenuation a t i n f i n i t e fPequency,
and t h e above d e s c r i b e d s h o r t c i r c u i t (or open c i r c u i t ) t e r m t i o n
satisfy this requirement
10. Equation 2 1 i s obtained from Equation 20 by r e a l i z i n g that when the angle
involved is v e r y c l o s e t o 1 Kk, and k is very c l o s e t o u n i t y , then

It is a l s o n e c e s s a r y t o r e a l h e t h a t when f and v both approach aero, then

Authorized licensd use limted to: IE Xplore. Downlade on May 10,2 at 19:0354 UTC from IE Xplore. Restricon aply.
n number
of arms i n thebasic networks of Fig.2A and Fig.SA. Also
equals the number of attenuation zeros produced by these networks.

Q0 the unloaded Q of the resonators used in the arms of Fig.28 and

Fig.SA.
number of arms required i n the network of Fig.2A t o satisfy a shape-
f a c t o r s p e c i f i c a t i o n when the Itcomplete Chebishev behavior" of fig.
i s used. See Equation 6 .
number of arms required in the network of Fig.2.A t o satisfy a shape-
factor specflication when the "Butterworth behavior11 of E'ig.4C i s
used See Equation
number of arms required in the network of Fig.28 t o satisfy a shape
" f a c t o rs p e c i f i c a t i o n the "incompleteSbebishev behavior11 of
F i g . b i s used. See Equation 8.
C 'B IC
the smallest unloaded Q which canbeallowed in t h e
resonators of Mg.2 if g:, nE, arms a r e t o satisfy t h e shape
factorspecification. See Equations ll, 12, and 13.
q.c number of arms required i n the network of Fig.% t o s a t i s f y a shape
factor specification when bomplete Chebishev accept and r e j e c t
band behavior" of is used. Equation

number o f arms required i n the network of Mg.sA t o satisfy a shape

factor specification when the ItTaylor accept complete Chebishev
r e j e c t band behavior" of Fig.6B is used. Equation

number of arms required in the network of -.SA t o satisfy a shape

'
3
3' fChebishev
a c t o rs p e c i f i c a t i o n d e n the tlincompleteChebishev accept-complete
r e j e c t band behavior't of Fig.6B is used. See Equation a.
T ICC
h-Q the smallest unloaded Q which can'bealloffed in the
resonators of Fig.6 ncc, nt, or nICC anas (respectively) are to
satisfy the shape factor specification. SeeEquations 32, 33 34.

f0 mid-frequency, i n cps,oftheattenuation shape.

BWf/ F fractional bandwidth.

B
W
, the bandwidth, i n cps, between thosepoints on the s k i r t of the
response which a r e the same number of db down as the "valleysrri n
theaccept band. See Figures h and 6 .
mh the bandwidth, i n cps, between thosepoints on the s k i r t o f there-
sponse which are the same number of db down as the " h i l l s 1 I i n the
r e j e c t band. See Fig.6.

Authorized licensd use limted to: IE Xplore. Downlade on May 10,2 at 19:0354 UTC from IE Xplore. Restricon aply.
 ma thespecifiedwidth i n cpsofthe"accept band," i.e,therequired
bandwidthbetween those points in the s k i r t where the attenuation i s
(Vp See Figures and 6 .

mr thespecified width i n cpsofthereject band, i.8. therequired

bandwidthbetween those points on the s k i r t where the attenuation i s
(Vp/Vr). See Figures 4 and 6.

(Vp/V) the attenuation ratio a t some f r a c t i o n a l bandwidth (W/f0).

(V,/Vv) the peak-to-"vallegn ratio allowed in the accept band.See Figures
and 6.
(Vp/Vh) - t h e p e a k - t o - % i l l f t r a t i o o c c u r r i n g i n t h e r e j e c t band.SeeFig.6.

(V#Jr) the attenuation given in s p e c f l i c a t i o n defining t h e

accept bandwidthand t h e r e j e c t bandwidth respectively. See Figures
4
r, i t h e real and i m a g i n a r y coordinates, on t h e complex f r a c t i o n a l band-
width plane, of the attenuation zeros of d e s i r e d a t t e n u a t i o n shape.

r, i t h e r e a l and imaginary coordinates on t h e complex f r a c t i o n a l band-

width plane of the attenuation poles of a desired response shape.
F,= m ~ - f r a c t i m a l bandwidth between i n f i n i t e a t t e n u a t i o n p o i n t s of the
5 response shown i n fig.6

02-l t h e s p e c i f i c modulus o f t h e e l l i p t i c f u n c t i o n s t o which it

(vP/vh)*-1 is attached as a subscript.

f' t h e complementary moduli of M and f respectively.

Kf,Ic, the complete e l l i p t i c i n t e s a l s of first ldnd evaluated for the

modulus value gimn by subscript f o r example Dwight s
Mathematical Tables, pp J99-203

log qf log t o t h e baae 10 of c a l l e d modular constant of t h a t modulus

given by the subscript. See for example Jahnke and Rude, pp.49-51.

f t h e s p e c i f i c modulus o f t h e e l l i p t i c f u n c t i o n s t o which i t is
attached as subscript.
wh

Authorized licensd use limted to: IE Xplore. Downlade on May 10,2 at 19:0354 UTC from IE Xplore. Restricon aply.
 b
Fig. 2
Fig. 1

I 3

XI

-I---_-~

Fig.3

Fig. 1
The bandpass selectivity specification which i s considered i n this paper. The
question marks outside the reject bandwidth indicate that in general wedo not, care
what shape i s used i n this region s o long as the minirmvn attenuation i s always greatm!
than (vdvh) f o r abandwidthwhich i s a t l e a s t 10, say, r e j e c t bandwidthswide.

Fig.
The basic invers'e arm, or llconstant k", configuration and i t s small percerkage
bandwidthnodenetwork equivalent. The correctelenentvalues are not thesane as
those obtained from classical f i l t e r theory; the configamtion i s the same.
A l l discussions apply, of course, to the dual meshRetworkswhich are not shown.

Fig. 3
This figure i s t o show that f i l t e r s emplof.ing mechanical resonators are fundmentally
the same those employing electrical resonators. TheEode network equivalent i s
obtained i f we say voltage i s equivalent t o velocity and current t o force; and the
mesh network equivalent is obtained if we say current i s equivalent t o velocity and
voltage to force.

Authorized licensd use limted to: IE Xplore. Downlade on May 10,2 at 19:0354 UTC from IE Xplore. Restricon aply.
 u
Fig. 5

Fig. b
The opthum attenuation shapes which can be prodmed by the constant k type
networks of Fig, 2 , See Section 111-A-1.

Fig. 5
The basic "m-derivedlf ccnfiguration and i t s node typeequivalepts. The correct
element values are not the sane those obtained frorr. c l a s s i c a l f i l t e r theory;
or-& the configuration i s t h e same. All discussions apply, of course, t o t h e
dual mesh networks which are not

Authorized licensd use limted to: IE Xplore. Downlade on May 10,2 at 19:0354 UTC from IE Xplore. Restricon aply.
 Fig 6

6 POLES

Fig.

20

Fig. 6
The opthum attenuation shapes which can be produced by the +derived type networks
of Fig. 5. See Section IV-A-1.

Fig. 7
The networks of Fig. must be adjusted s o t h a t t h e i r a t t e n u a t i o n zeros are i n t h e
above positions as given by Eqs. and 10. The perpendicular distame do of t h a t
attepuation zero marest the jF axis s e t s t h e lowest allowable resonator unloaded
Q which can be used. See Section 111-B.

Fig. 8
The graph which i s obtained when t h e t h r e e pairs of constar;t K typo design equations
Eqs 6 and 11, Eqs. 12, and Eqs 8 and 13 are applied t o a s p e c i f i c
requiremnt

116

Authorized licensd use limted to: IE Xplore. Downlade on May 10,2 at 19:0354 UTC from IE Xplore. Restricon aply.
 0

Fig.9

Fig. 10

Fig. 9
The networks of Fig. 5' must be adjusted t h a t t h e i r a t t e n u a t i o n z e r o s and poles
are in the positions as given by Eqs. 26, 28 and Eqs. 31. The perpendicular
distance do of that attenuation zero nearest the j F axis sets the lowest allck-able
Q which can be used. SeeSections 111-B and N - B .

117

Authorized licensd use limted to: IE Xplore. Downlade on May 10,2 at 19:0354 UTC from IE Xplore. Restricon aply.