PROCEEDINGS

THE INSTITUTION OF ELECTRICAL ENGINEERS-,

volume 117 Electronics
Spectral transformations for digital filters
A. G. Constantinides, B.Sc.(Eng.), Ph.D.
Indexing term: Digital filters
Abstract
The paper describes certain general transformations for digital filters in the frequency domain. The term
digital filter is used to denote a processing unit operating on a sampled waveform, so that the input, output and
intermediate signals are only defined at discrete intervals of time; the signals may be either p.a.m. or p.c.m.
The transformations discussed operate on a lowpass-digital-filter prototype to give either another lowpass
or a highpass, bandpass or band-elimination characteristic. The transformations are carried out by mapping
the lowpass complex variable z"
1
[where z~
x
= exp (j'toT) and T is the time interval between samples]
. by functions of the form
known as unit functions.
a, b
/ =
F
s
-=
G() =
g() =
T =
z =
z~
l
=
<x =
a* =
co
List of principal symbols
real constant coefficients
frequency, Hz
sampling frequency, Hz
pulse-transfer function of a digital filter
unit function; spectral transformation
sampling period (T = 1/F
S
)
unit advance, z = exp (JcoT)
unit delay, z = exp (jcoT)
constant parameter
complex conjugate of a
cutoff (angular) frequency of lowpass-digital-filter
prototype
constant coefficients
angle of rotation
real constant coefficient
angular frequency
angular sampling frequency, Q,
s
= 2TTF
S
1 Introduction
This paper describes certain general transformations
for digital filters in the frequency domain. With the exception
of the restricted forms of the transformations published
elsewhere,
1
*
2
spectral transformations for digital filters have
been virtually nonexistent. A notable move towards establish-
ing a set of such transformations has been made by Broome,
3
but the result of that study was rather unsatisfactory, in that
certain distortion errors were introduced into the amplitude
characteristics.
The general transformations proposed in this paper are
compared with the restricted special cases of References 1
and 2, and Broome's translation formula
3
is critically
examined.
Paper 6206 E, first received 15th January and in revised form 17th
April 1970
Dr. Constantinides is with the UK Post Office Research Department,
Brook Road, Dollis Hill, London NW2, England
PROC. IEE, Vol. 117, No. 8, AUGUST 1970
75 F2?
The study included in this paper is completely theoretical,
and the various results on digital-filter theory are taken as
assumptions. An excellent introduction into the field of
digital filters is given in Reference 6.
2 Assumptions
The following assumptions will be made:
(i) The inputs, outputs and intermediate signals in the
systems that we shall be concerned with are sampled wave-
forms; sampling is carried out at constant frequency with the
sampling-theorem condition satisfied.
(ii) The implementation of the filtering devices is carried out
in a discrete form; i.e. processing of signals is either in p.c.m.
or p.a.m. form.
(iii) Quantisation effects and other 'noise' errors are negligible.
(These effects are pertinent to the operation of the devices, and
will have no bearing on the present theoretical study.)
(iv) The transfer function of a digital filter (called a pulse-
transfer function) is a real rational function in z"
1
(for a
proof see Reference 7.), where z~
l
= cos coT-Jsin toT, and
T is the sampling period.
(v) The digital filters that are dealt with in this paper are
stable;
7
i.e. they have their poles situated outside the unit
circle |^ * | = 1.
3 Projection of amplitude characteristic on
a cylindrical surface
Here we give a simple representation of the amplitude
characteristic of a given pulse-transfer function, on a
cylindrical surface. This representation will be helpful in later
Sections when we shall be dealing with the spectral trans-
formations.
The representation is effected as follows. If z~
[
is replaced
by cos coT j sin toT in a given pulse-transfer function, the
amplitude characteristic can be earily evaluated. To a
particular frequency co, there corresponds a definite point
1585
wT on the unit circle at which the amplitude of the given
function is calculated. Conceptually, we can represent the
amplitude characteristic on the surface of a cylinder, normal
to the z~
x
plane, having the unit circle as base and a height
of unity (i.e. normalised amplitude response). Then, for every
point <x)T,a. height is erected on the generator of the cylinder
equal in magnitude to the value of the amplitude character-
istic. On joining these points on the surface of the cylinder for
all arcs ooT(i.e. 2TT radians), we have produced the amplitude
characteristic. There is another advantage of this representa-
tion, in that the periodic amplitude characteristic has been
reduced to one period only. Such a representation of the
unit cylinder
amplitude characteristic
unit circle
Fig. 1
Representation of 2nd-order Chebyshev amplitude characteristic on
cylindrical surface
amplitude characteristic is shown in Fig. 1 for a 2nd-order
Chebyshev lowpass filter, and it will be used later on an
illustrating device.
4 Spectral-transformation problem
Statement of the problem: Given a pulse-transfer
function of a lowpass digital filter G(z~
x
), it is required to
obtain the unique transformations
for the following types of amplitude transformations:
(i) lowpass-lowpass
(ii) lowpass-highpass
(iii) lowpass-bandpass
(iv) lowpass-band-elimination
where the type of the amplitude characteristic is preserved.
Some preliminary points can be made concerning the
general class of required transformations g(z~
x
). Since
g(z~
l
) is to replace the variable z~
x
, and, since the resulting
pulse-transfer function must be real and rational in z~
l
, and
given that the original pulse-transfer function G(z~
x
) is real
and rational in z~
x
, it follows that g(z~
x
) must be a real
rational function in z~
l
.
/
Furthermore, the values assumed by the function g(z~
x
)
must correspond to those of z~
x
, and no other values must
be introduced. The above correspondence need not occur at
the same point.
From stability considerations, the mappings must be such
that the regions of stability and instability are preserved; i.e.
the inside of the unit circle must map into the inside of a
domain F, whereas the outside must map to the outside of the
domain.
For the domain F, let
g(z~
x
) = for z~
x
= ~
x
= e~
wherep(oS) is the amplitude and </>(cu) is the phase of g(z~
x
).
If stability regions are to be preserved,
f or | z- ' | 1 \g(z-
x
)\ $ 1
Hence p(aj) = 1. This means that the unit circle is mapped
1586
onto itself. In view of the above conditions, and from the
maximum-modulus theorem, it follows that the mapping
g(z~
l
) [where g(z~
x
) is not necessarily a real function] must
be given by
g(z-*)=e ft
a?z"
(1)
where | a
(
| < 1 and a? is the complex conjugate of a,-.
The functions given by eqn. 1 are called 'unit functions'.
Since g(z~
x
) is required to be real, it follows that the zeros
a, must occur in complex-conjugate pairs, and that
e#= 1
(For digital filters with complex coefficients,
5
such restrictions
on a,- and 6 are removed.); i.e. the general spectral trans-
formation for digital filters with real coefficients is given by
=
. . (2)
where the order is now 2n, and a,- are not necessarily all
complex.
In eqn. 1, there are two parameters that need further atten-
tion if the zeros a
(
- are to be set aside for the moment.
First, it was mentioned earlier that the angle 6 must be
such that e
i%
= 1. Therefore 6 must be a multiple of n. The
significance of the angle can be best described as a rotation
angle on the unit circle. Thus the only admissible rotations
are multiples of TT. The rotational nature of 6 will be explained
and used later.
A further parameter that needs to be considered is the order
of the unit function m when e
je
= 1. This order m dictates
the nature of the unit-function mapping.
Since the modulus of g(z~
x
) is unity for z"
1
on the unit
circle, the functional relationship between z~
x
and g(z~
x
)
can be taken as a mapping of the unit circle onto itself.
Furthermore, there will be m values of z~
x
for every value of
g(z~
x
); or, conversely, for every rotation of z"
1
on the unit
circle around the origin, the function g(z~
x
) rotates m times
on the same unit circle around the origin. This means that a
function/(z
-1
) of z~
x
assumes all its values m times for one
revolution of z~
x
when z~
x
is replaced by g(z~
x
).
The usual requirements for lowpass, highpass, bandpass
and band-elimination characteristics, where one passband is
required in the region 0 < to < O
5
/2, can be translated in
terms of the mapping functions as shown in Table 1.
Table 1
REQUIREMENTS FOR FOUR TYPES OF FILTER
Required characteristic
Lowpass . . . .
Hi ghpass
Bandpass
Band-el i mi nat i on .
Number of passbands
in relation to a lowpass
prototype
1
1
2*
2*
Order
1
1
2
2
* Corresponding to the positive and the negative lowpass passbands
From this Table it can be seen that, for simple lowpass-
lowpass, lowpass-highpass, lowpass-bandpass and lowpass-
band-elimination transformations, we need only consider
powers of maximum order two; for multiple bands in the
Nyquist interval, however, the appropriate order must be
chosen in conjunction with the general transformation of
eqn. 2.
5 Lowpass-lowpass transformation
The lowpass-lowpass transformation is required to
keep the amplitude characteristic unaltered with the points
A and D shown in Fig. 1 fixed, but with the arc ABD (ACD)
compressed or expanded along the circumference of the
circle to result in a different cutoff frequency.
It is quite obvious that such a transformation is not a
linear one for the frequency o>, as is the case for the lowpass-
lowpass transformation in R, L, C and M filters where s is
replaced by ks (k > 0 and constant).
Mathematically, it is required to map a unit circle into
PROC. IEE, Vol. 117, No. 8, AUGUST 1970
itself in a 1:1 correspondence, with the points A and D
invariant.
Let the mapping be denoted by
so that, for one complete rotation of z~
x
around the unit
circle, g(z~
l
) also takes one complete rotation.
Furthermore,
= 1
(3)
The transformation given by eqn. 4 has no effect on the
amplitude characteristic except for the change in the position
of the frequencies on the unit circle. Thus, for negative a (i.e.
(D
C
> /3), the arc BDC of Fig. 1 is compressed on the circle,
and, since A is invariant, the arc BAC is stretched. For
positive a (i.e. co
c
< j8) the opposite effect occurs.
Fig. 2a shows the amplitude characteristic of a lowpass
digital filter whose cutoff frequency is 2-5kHz for a sampling
frequency of 10kHz. If tojlTr of eqn. 5 is chosen to be
1kHz, as a result of transformation (eqn. 4) the amplitude
characteristic is as shown in Fig. 2b.
10
= 05
2 3
frequency, kHz
10
05
10%
2 3
frequency. kHz
b
10
S 05
2 3
frequency, kHz
c
1-0
I-0-5
2 3
frequency. kHz
e
Fi g. 2
Amplitude characteristics from spectral transformations
a 4th-order lowpass Chebyshev characteristic
fc = 2-5kHz, F, = 10kHz
b Lowpass characteristic,/< = 1 kHz, resulting from lowpass-Iowpass transforma-
tion (prototype filter shown in Fig. 2a)
c Highpass characteristic, f
c
= 1 kHz, resulting from lowpass-highpass trans-
formation (prototype filter shown in Fig. 2a)
d Bandpass characteristic, f\ = 1 kHz, fi = 2 kHz, resulting from a Iowpass-
bandpass transformation (prototype filter shown in Fig. 2a)
e Band-elimination characteristic, f\ = 1kHz, fi = 3kHz, resulting from a
lowpass-band-elimination transformation (prototype filter shown in Fig. 2a)
and, since there is to be a 1:1 correspondence between z~
x
and g(z~
l
), it follows from eqn. 2 that
where \a\ < 1 and is real.
Let the cutoff frequency of the lowpass prototype digital
filter be ft radians per second and that of the resulting lowpass
digital filter be cu
c
radians per second. Then the following
relationship must hold:
, . _. exp (-jfiT) - a
from which we obtain
sin
(5)
6 Lowpass-highpass transformation
Consider the lowpass characteristic of a digital filter
as shown in Fig. 1. Now imagine that we have the freedom to
rotate the cylinder at another angle relative to the positive
direction of the real axis.
It can easily be seen that, when the cylinder has been
rotated by 180, we obtain a highpass amplitude character-
istic (i.e. the passband and stopband of the lowpass filter have
interchanged their relative positions) so that, if the cutoff
frequency of the lowpass filter is /? and that of the resulting
highpass is co
c
,
a)
c
T + flT = IT (angle of rotation)
or
where
(6)
T
PROC. IEE, Vol. 117, No. 8, AUGUST 1970
The above method of achieving the required transformation
1587
is rather restrictive, because it means that one has to replace
z~
l
in the lowpass-filter pulse-transfer function by z"
1
where the two cutoff frequencies are restricted to satisfy the
relationship given by eqn. 6.
In view of the lowpass-lowpass transformation given by
eqn. 4, the above difficulty can be resolved by first applying
the lowpass-lowpass transformation to obtain the particular
lowpass cutoff frequency which, in conjunction with the
required highpass cutoff frequency, satisfies eqn. 6, and then
replacing z~
l
by z~
l
(i.e. rotate by 180). These two steps,
however, can be combined into one by replacing z~
x
by z~
l
in eqn. 4.
Thus the required transformation is given by
(7)
It is quite simple to deduce the above transformation, how-
ever, from the general form of the unit function. The argu-
ments proposed for the lowpass case are equally applicable
here, except that we must take the angle of rotation 9 of eqn. 2
to be 180, which gives us the negative sign; and since there is
to be a 1:1 correspondence between the lowpass prototype
and the resulting highpass digital filter, we immediately
obtain eqn. 7.
If the cutoff frequency of the highpass filter is co
c
and that
of the prototype lowpass digital filter jS, similarly to eqn. 5
we have
a =
COS
(8)
We can state the above result formally in the form of a
theorem.
Theorem 1: Given the pulse-transfer function G(z~
{
) of a
lowpass digital filter of cutoff frequency /8, the pulse-transfer
function of a highpass digital filter having the same type of
amplitude characteristics as G(z~
x
), and the cutoff frequency
w
c
is given by replacing z~
l
of G(z~
i
) by the unit function of
eqn. 7, where a is given by eqn. 8.
Fig. 2c shows the highpass amplitude characteristic that
resulted from the lowpass one of Fig. 2a. In this case /8/27T =
2 5 kHz, and OJJITT = 1 kHz.
7 Lowpass-bandpass and lowpass-band-
elimination transformations
The order of the unit function in this case will be two, as
was shown earlier. The required mapping has the effect of
reproducing the cylindrical surface of Fig. 1 twice on another
cylindrical surface in such a way that the resulting two sur-
faces are symmetrical about a diameter of the base.
This cylindrical surface can be rotated in relation to the
point z~' = 1 on the circumference of the unit circle about
its axis at any angle. This is equivalent to saying that 6 (the
angle of rotation) can take any value, but it was seen earlier
that 6 must be a multiple of IT for real functions g(z~
l
). It
can easily be seen that, if 8 =0, we have the band-elimination
transformation, whereas for 6 = TT (or odd multiples of TT)
we have the bandpass transformation. From these conceptual
considerations, let us return to the mathematical descriptions.
Consider the mapping requirements for the lowpass-
bandpass transformation given in Table 2.
Table 2
LOWPASS-BANDPASS REQUIREMENTS
Lowpass
frequency
0
4
Lowpass
variable z~
l
1
exp (jpT)
exp (-y/97")
- 1
Bandpass
frequency
CO0
COl
CO2
Qs/2
0
Bandpass
variable
exp (-j(o
0
T)
exp (-jcoiT)
exp (-j(o
2
T)
- 1
+ 1
Since the order of the required transformation is 2, we can
write
<x
2
z
2
+ a,z
(9)
By imposing the requirements of Table 2 on eqn. 9, we
obtain
(10)
, 2ak . k - 1
l
z~
x
-\
where a = cos w
Q
T =
cos
. . . . (11)
~ (12)
We therefore have the following theorem.
Theorem 2: Given the pulse-transfer function G(z~
x
) of a
lowpass digital filter of cutoff frequency /8, the pulse-transfer
function of a bandpass digital filter having the same type of
amplitude characteristic as G(z~
l
), centre frequency co
0
, and
upper and lower cutoff frequencies co
2
and coj, is given by
replacing z~
l
of G{z~
x
) by the unit function of eqn. 10, where
a and k are given by eqns. 11 and 12, respectively.
The transformation given by eqn. 10 can take different
forms depending on the values of a and k.
In particular, if A: = 1,
z-\z~
x
- a)
1 - a z
- l
and from eqn. 12 we have, with k = 1,
a>
2
= cu! = fi (14)
The form of the transformation given by eqn. 13, along with
the relationship of eqn. 14, has been given elsewhere.
1
This
form is restricted in its application, because one has to use a
lowpass filter of a particular cutoff frequency jS given by
eqn. 14 before applying eqn. 13. With the unrestricted form
of eqn. 10 there is no particular difficulty, since the real
coefficient k can be adjusted to satisfy eqn. 12.
Another form that the transformation can take corresponds
to the case when a = 0 and k = 1. This form is particularly
important, since the resulting bandpass digital filter has an
amplitude characteristic which is arithmetically symmetrical
about the centre frequency. In this case,
~
x
) = - z -
2
and
g(z~
x
) =
0)2 =
"4
+
2
with co
0
= -
(15)
(16)
Fig. 2d shows the bandpass amplitude characteristic when
OJ
X
12TT = 1 kHz and co
2
/27r = 2 kHz derived from the lowpass
characteristic of Fig. 2a.
Now consider the lowpass-band-elimination transform-
ation. The requirements for this case are tabulated in Table 3.
Table 3
LOWPASS-BAND-ELIMINATION REQUIREMENTS
Lowpass
frequency
0
a/2
Lowpass
variable z~
l
+ 1
exp (jpT)
exp (-7/570
Band-elimination
frequency
0
a/2
CO2
COl
co
0
Band-elimination
variable
+ 1
- 1
exp (-jo)
2
T)
exp (-jcoiT)
exp {-JCOQT)
1588 PROC. IEE, Vol. 117, No. 8, AUGUST 1970
Table 4
SPECTRAL TRANSFORMATIONS FROM A LOWPASS-DIGITAL-FILTER PROTOTYPE OF CUTOFF FREQUENCY jS
Filter type
Lowpass
Highpass
Bandpass
Band-elimination
Transformation
z - l - a
i.
z-l + a
1 + ocz
/ i 2ock ,
2 2
k + l
z l
k+ 1 k +
1 1 +k" '
I l - *__, 2 .
M + k~ 1 + k"
1 +A: \
a =
a =
a =
a =
k =
Associated design formulas
sin ( -
sin f
cos |
COS
- COS O)Q1
/at
fa
COS ( -
/CO
COS f
= tan I -
2 >
</ ? - <
; 2
^ +
I 2
CO
H
CO
2
2 + CO
2
2 CO
2
2 - CO
2
\T
\T

c
y
to
c
y
/co
2
+ coA
" \ 2 y
n
/oiz <\\
T
\ 2 /
) otan
2
-^COS (OnT
A j87
On imposing the constraints given in Table 3 on the unit
function of eqn. 9, we obtain
r-2 _
2a
1 -k
TT~k'
.-2 _
2a
COS
where a = cos OJ
0
T =
1 - f r
1 +A:
+ 1
(17)
cos
(2
. . . . (18)
and
(23)
Fig. 2e shows the band-elimination amplitude characteristic
when O)
X
I2TT = 1kHz and ai-Jlir = 3kHz, derived from the
lowpass characteristic of Fig. la. All the transformations are
summarised in Table 4.
^ . . . . . (19)
From the above result, we can formulate a theorem similar to
theorem 2.
Theorem 3: Given the pulse-transfer function G(z~
l
) of a
lowpass digital filter of cutoff frequency j8, the pulse-transfer
function of a bandpass digital filter having the same type of
amplitude characteristic as G(z~
1
), centre frequency a>
0
, and
upper and lower cutoff frequencies a>
2
and oi, is given by
replacing z~
x
of G(z~
l
) by the unit function of eqn. 17, where
a and k are given by eqns. 18 and 19, respectively.
There are again the following two restricted forms of the
transformation. When k = 1, we have
~\z~
x
-
2-
J
(z
a)
-OLZ~
X
and, from eqn. 19,
. . . (20)
(21)
The above restricted form of eqn. 20, along with the relation-
ship of eqn. 21, has been given elsewhere,
1
and the restrictions
are precisely those stated for the equivalent bandpass case.
Now, when a = 0 in eqn. 20, we have the arithmetically
symmetrical band-elimination case, where
g(z~
l
) =-z~
2
PROC. IEE, Vol. 117, No. 8, AUGUST 1970
(22)
8 Comparison of transformations to
Broome's translation formula
Broome
3
gave the following formula as a means of
transforming lowpass pulse-transfer functions into highpass,
> bandpass and band-elimination pulse-transfer functions.
Let G(z~
l
) be the pulse-transfer function of a lowpass
digital filter. Then
H(z~
x
) =G{z~
1
exp (-Ja>
0
T)} + G{z~
l
exp (+jto
0
T)}
is a translated version of G(z~
l
). Several comments have been
produced by Kaiser in Reference 6.
Basically, the above formula represents two rotations on
the z"
1
plane, one in the clockwise and the other in the
counter-clockwise direction. The sum of the two rotations of
functions of the complex variable z~
l
is a real rational
function z~
l
if the rotations are at equal angles. Because the
pulse-transfer function resulting from the above formula is
the sum of two individual pulse-transfer functions rotated to
the angles co
0
T, it will exhibit what Broome
3
and Kaiser
6
term a 'distortion error' which results from the tails of the
image of G(z~
l
) centred at OJ
0
. It is suggested in these
References that a good lowpass-filter design will reduce the
distortion error to a very small level.
No distortion will occur when the lowpass pulse-transfer
function is rotated by 180 or multiples thereof, because,
when o)
0
T= 180,
exp(ja)
0
T) = - 1
1589
and hence
H(z~
l
) = 2GC-Z-
1
)
which is recognised as the restricted case of the general low-
pass-highpass transformation of eqn. 7.
If COQT = 0, the pulse-transfer function remains unaltered
except for a multiplier of value 2. This is not, however, a
lowpass-lowpass transformation, because no change in cut-
off frequency occurs.
Thus the Broome translation formula is only satisfactory
under very special conditions where the distortion errors are
small and when co$T = 180, so that the special case of the
lowpass-highpass transformation is obtained, and, in this
restricted sense, the translation formula is exact.
The general transformations given in this paper, however,
are therefore superior to the above translation formula, sWe
they not only introduce no distortion errors but they are also
unrestricted in their application, be it for wideband or narrow-
band digital filters.
9 Conclusions
We have given in this paper the complete set of
transformations on the z~
l
plane necessary to transform a
given lowpass-digital-filter pulse-transfer function into a pulse-
transfer function having the same type of amplitude char-
acteristic and belonging to one of the following classes:
(i) lowpass
(ii) highpass
(iii) bandpass
(iv) band-elimination.
Furthermore, along with the transformations, the necessary
design formulas were given which relate the critical frequencies
of the required digital filter to the cutoff frequency of the
lowpass-digital-filter prototype.
The transformations are quite general, and, their restricted
forms that have been published elsewhere
1
*
2
are easily
obtainable as indicated.
Proofs of the most important theorems are given in
Appendix 12.
The spectral transformations possess the following im-
portant and useful features:
(a) They are allpass functions, and hence, because of their
inherent form, the multipliers in the digital implementation
can be reduced by a factor of 2 by multiplexing.
(b) Since the application of the transformations involves a
substitution of z~
l
in a lowpass-digital-filter transfer function
by a unit function, it follows that the structure of the lowpass
digital filter remains unaltered (i.e. the adders and multi-
pliers are the same), and only an extra elementary transfer
function, representing the transformation, is inserted in
place of z~K
(c) As a consequence of property (b) above, certain forms of
the transformations (e.g. eqns. 13 and 20) that have one
variable parameter exhibit simple characteristics where the
centre frequency can be varied over the Nyquist range but
the bandwidth is kept'constant.
10 Acknowledgments
The author wishes to thank the authorities of North-
ampton College of Advanced Technology, London (The
City University, London) for financial support, and the
Senior Director of Development, UK Post Office Tele-
communications Headquarters, for facilities provided.
11 References
1 CONSTANTINIDES, A. c. : 'Frequency transformations for digital
filters', Electron. Lett., 1967, 3, pp. 487-489
2 CONSTANTINIDES, A. c : 'Frequency transformations for digital
filters', ibid., 1968, 4, pp. 115-116
3 BROOME, P. : 'A frequency transformation for numerical filters',
Proc. Inst. Elect. Electron. Engrs., 1966, 54, pp. 326-327
4 JURY, E. i.: 'Theory and application of the z-transform method'
(Wiley, 1964)
5 CRYSTAL, T. H. : 'The design and applications of digital filters with
complex coefficients', IEEE Trans., 1968, AU-16, pp. 330-335
6 KAISER, J. F. : 'Digital filters' in KUO, P. F., and KAISER, J. F. (Eds.):
'System analysis by digital computer' (Wiley, 1966), pp. 218-253
7 CONSTANTINIDES, A. G.: 'Synthesis of recursive digital filters from
prescribed amplitude characteristics'. Ph.D. thesis, University of
London,1968
1590
12 Appendix
Proof of theorem 2: Rewriting eqn. 9 in a different
form, we have
y
2
z + yjz i + l
From Table 2 we see that #(1) = 1.
Hence, by using this constraint on the above form of eqn. 9,
we obtain e
JQ
= 1.
Furthermore, since ^{exp (Jco
0
T)} = 1, we have
(y
2
+ 1) + 2y
t
exp (-Jco
0
T) + (y
2
+ 1) exp (-j2co
0
T) = 0
and hence (1 + y
2
) cos co
0
T + y
x
=0 .
Let cos COQT = a, so that
(
u
=
_
8
y
2
y
2
)z ~
2
and also let y
2
= y. Hence we can write the above equation
in the form
and, since a = cos coT, |a| < 1, so that the function
i / z~
x
a \ . . . .
z
l
[ - . ) is a unit function.
\ 1 az~W
Furthermore \y\ < 1, since this quantity represents the
product of two zeros of the unit function which are inside the
unit circle (^ * j = 1. Hence the quantity (- ^-, J
\1 + yz
is a unit function.
El
= -
yz
and E
2
= , -
so that g(z !) = E
X
(E
2
).
Now let us form
i.e.
+ g(z ') 1 + yE
2
y E
2
1 + y 1 +E
2
i -r gK* ) 1 y 1 E
2
_ 1 + y z~
2
- 2az~
l
+ 1
~ 1 - y 1 - z -
2
When o> = cj[ in the bandpass characteristic, the function
g(z~
l
) corresponds to exp (j^T), and, when co = co
2
,
g(z~
1
)= exp (-/j8r). Hence
6 r y + 1 coscuiT a
2 y 1 sin co\T
BT y + 1 cos co
2
T - a
and tan ; = ^ =
2 y - 1 sin co
2
T
Therefore a =
cos
y + i
=
_,
Let ^ ' = k, so that
y - 1
and y =
k - 1
k + 1
and the transformation becomes
, - 2 _
* - . - 2
Jfc + 1 * + 1 .
Hence the theorem is proved. The proof for theorem 3 is
similar to the above.
PROC. IEE, Vol. 117, No. 8, AUGUST 1970