IETE Journal of Education

ISSN: (Print) 0974-7338 (Online) Journal homepage: [Link]

The Lattice Filters

V. Ramachandran

To cite this article: V. Ramachandran (1967) The Lattice Filters, IETE Journal of Education, 8:1,
52-62, DOI: 10.1080/09747338.1967.11467902

To link to this article: [Link]

Published online: 11 Aug 2015.

Submit your article to this journal

View related articles

Full Terms & Conditions of access and use can be found at

[Link]

Download by: [University of California, San Diego] Date: 09 April 2016, At: 02:38
 THE LATTICE FILTERS
V, Ramachandran*
Introduction
So far, the discussion centred around filters having T-and .,._
sections or L-sections derived from above. Another type of structure
known as the Lattice-structure (Fig. 1), can also yield the properties
of the filter. Such a network is so named because of its structure. It
Downloaded by [University of California, San Diego] at 02:38 09 April 2016

can also be noticed that this is equivalent to a \Vheatstone's bridge.

z,

FIGr.l
The lattice work

This is the case when the arms of the network of Fig. 1 are rearranged
as shown in Fig. 2. If the relationship Z1 Z,, =Z 2 Z3 , no output will be
obtained across the terminals 2-2'. Another feature which can be im-
mediately noticed is the absence of the common terminal between the
input and the output. Now, the properties of such a lattice structure
is studied.
The symmetrical lattice network
In general, the unsymmetrical lattice network, shown in Fig.
is not used. But, the symmetrical lattice network, in which Z 1 =Z, =
Za and Z2 =Z 3 =Zb, is in great use. This is as shown in Fig. 3.
* Department of Electrical Engineering, Nova Scotia Technical College, Halifax,
Canada.

52
 THE LATTICE FILTERS

In the writing of figures, normally the other two are shown by dotted
lines, denoting that the structure is symmetrical.
2
Downloaded by [University of California, San Diego] at 02:38 09 April 2016

I'
Fl~.2

c:!'

L----- Souyce ----~

The Wheat•tones bridge-equivalent of the lattice network of Fig. I

Our primary interest is in the determination and the behaviour

of the characteristic impedance and the propagation constant of such
Zo.
a---~-{==::r---r---o 2.

F1~.3.
"

•'~------~..__-Cl2.'
The repre•eutation of the lattice network

networks. For this purpose, the open-circuit and the short-circuit

input impedances are required.

53
 STUDENTS' JOURNAL VOL 8 NO. I, 196 7

Accordingly,
Z 00 = (Z 2 +Zb) in parallel with (Zb+Za.), (Fig. 4a)
Za+Zb .. (Ia)
= --2--
z.c =Series combination of two impedances, each
Zct
Downloaded by [University of California, San Diego] at 02:38 09 April 2016

2 2'

FIG.+(c:t) FIGt.+(lr)
Detennination of Zoe Determination of Zsc
being a parallel combination ofZa. and Zb
(Fig. 4b).
2ZaZb .. (lb)
= :t~+z~
Hence,
ZoL=Characteristic impedance of the symmetrical lattice
= VZoc·Zsc
= o/[Link] .. (2)
and
'YoL--propagation constant of the symmetrical lattice
= Tanh-1 ~ ~sc l t
(Zoe~
= Tanh -1 { ""~}
2
.. (3)
Za+Zb
=2Tanh-1 v'Z~
zb
54
 THE LATTICE FILTERS

From Eqs. (2) and (3), we can immediately get the impedances
for the branches as
z. = ZoL Tanh ; .. (4a)
"'I
Zb=ZoL coth .. (4b)
2
These relations enable us to write the equivalent-T or "lr-structure
Downloaded by [University of California, San Diego] at 02:38 09 April 2016

of the lattice. From the artificial line study, it is known that

'Y
ZlT=Series-arm of the T-section=Z 0 Tanh - - .. (5a)
2

FIG.?(o.)
The cquivalent-T network of the lattice

and Z 2t==Shunt-arm of the T-section=Z 0 cosech "'/ ( .. 5b)

Comparison of Eqs. (5a) and (4a) yields Za=Z 1t·
Similarly, Z:rr= zb-;f•. The equivalent-T is as shown in Fig. 5 (a).

1'o-.......JL..------'---o a'

The equi'VIIlent-network of the lattice

55
 STUDENTS' JOURNAL VOL. 8 NO. ] , I 967
The same could be inverted to give the equivalent-'IT which is as
shown in Fig. 5(b). This shows that a physically realisable equiva-
lent-T or 'IT- structure is possible only when Za. >
Zh. This is an
advantage of the lattice, in that, such a restriction does not arise in
the treatment of the lattice and it is. the most general network consi-
dered. Let us consider the frequency charactristics of such a symme-
trical lattice network.
Downloaded by [University of California, San Diego] at 02:38 09 April 2016

The propagation characteristics of the lattice network

From (3), it is known that the propagation constant is depen-
dent on the ratio Zso or ._za. . First, the discussion will proceed
Zoe Zb
. I1 t he ratio
Wit . -z-
Zse , as 1t
. 1s
· more genera,
I an d t h en 1t . app 1'1ed to
. 1s
•bo
Za. ·
zb YoL WI'11 b e real, when both Zse an d Zoe are of t h e same

sign; and will be imaginary, when they are of opposite signs.

Foster's reactance theorem can be conveniently applied for the
determination of the pass band and the attenuation band. For
any physically realisable passive reactance function, the poles

The constant-k low-pass T-network

and zeros interlace. Between a pole and an adjacent zero,

the function is negative; and between a zero and the adjacent pole,
the function is positive. It is clear, then, that, in the pass band, the
poles and zeros of Z80 and Zoe are not identical. That is, if a pole
of Z 80 exists at a particular frequency, in order that this comes within

56
 THE LATTICE FILTERS

the pass-band, Zoe should be a zero; and vice versa. If the poles and
zeros of Zsc and Zcc are identical, the region constitutes the attenua-
tion bani. The cut-off frequencies are tho~e points where the signs
of Zsc az,d Zsc are the same. An illustration clarifies the discussion
considerably. Consider a constant-k T-section low-pass filter (Fig. 6) .
.Zse = ~!-_~ +- 1- - s~~C::~_:±-_3 (6 )
Downloaded by [University of California, San Diego] at 02:38 09 April 2016

2 sCk - 2sCb ·· a
sLk I
z sLk -2- sCk sLk (s8LkCk + 4) .. (6b)
se = - 2-- + s~~-; ~=- 2(s 2LkCk+2)
2 sCk
0 Oc
._------------~~~----------------------------------x
II _ _ ..ta
:[Link]
I
I
ICVT-oFF
I
Zsc

FIGt-. 7
The pole-zero configuration of Zoe and Zsc of Fig. 6

The pole-zero configurations of Zoe and Zsc are as shown in Fig. 7.

Only one-half is as shown, as the other half is a mirror-image of this.
By the above argument, the frequencies between the zero and

Wc = / ~C constitute the pass-band and those between / - 2 ..

../ Lk k ../ LkCk
and infinity constitute the attenuation band.
The above results can be applied to the Lattice network. The
functions Z 80 and Zoe need not be computed, since the ratio
Zzse_ is equal to the ratio ZZa. . If the poles and zeros of Za. and Zb
oc 0
are plotted, the frequency behaviour can be obtained. In fact,

57
 STUDENTS' JOURNAL VOL. 8 NO. J, 1967

even if zb and zb are to be interchanged, the pass-band and the

attenuation band still remain the same, even though the frequency
characteristics may vary. Let us develop all the four types i)f filters,
using this reasoning.

(a) Low-pass filter

In order that the low-pass action exists, there should h:: a pole
Downloaded by [University of California, San Diego] at 02:38 09 April 2016

in one of the arms and a zero in the other at the origin. This means
that if an inductance is contained in one of the arms, the other arm
0

(~L,+ -tc,) +-----<>-,- - - - -

JL,Co

L,

Foe, 6

The four basic types of lattice )ow-pass networks and the corresponding
pole-zero configurations.

has to contain a capacitance. At infinity, both the arms should have

a zero or a pole. If an inductance is contained in one of the arms, the
other arm is a series combination of an inductance and a capacitance,
so that both the arms have poles at infinity. If a capacitance is con-
tained in one of the arms, the other arm is a parallel combination of
an inductance and a capacitance, so that, both the arms have zeros at
infinity. The cut-off frequency, in both the cases, is the resonant

58
 THE LATTICE FILTERS

cy of the inductance-capacitance combination. These four

vmbinations areas shown in Fig. 8. Complicated structures
puilt up similarly.

~h-pass filter
r the high-pass action to exist, there can be a pole or a zero
.1, in both the arms simultaneously. In addition, at infiinity,
Downloaded by [University of California, San Diego] at 02:38 09 April 2016

•f -- ..- - - -~ ...~·'.

-' ...... :.....__..-

'•

"'· . .:L-
"'-"""_,-,-...-,·- ~-··

' ( >:..
...... p.

The four basic types of lattice high-pass

filters and the corresponding pole-zero
configurations

if one of the arms contains a pole, the other arm should contain a
zero. Thus, the four basic high-pass lattice structures are as shown
in Fig. 9.

(c) Band-pass Filter

For the band-pass filter, the arms Za. and Zb should contain
either a pole or a zero at both the origin and at infinity simultaneously.
This means that if a series resonant circuit is contained in one of the
arms, the other arm shall also contain a series resonant circuit; and
the same case exists for the parallel resonant circuit also. Thus the
two basic band-pass structures are as shown in Fig. 10. An interesting
situation exists when L 1C 1 =L 2C 2 in both the cases. This means
that the resonant frequencies of both the arms are identical. In this
case, according to the earlier discussion, no pass-band can exist and
they behave like all-rejection filters.

(d) Band-elhnination Filter

For the band-elimination filter, the arms Za and Zb should

59
 STUDENTS' JOURNAL VOL. 8 NO. J, 1967

~c~
: PASS

---~~Z__
1 BAND

~ca~~J+-----~----~,----------
,JLec..
Downloaded by [University of California, San Diego] at 02:38 09 April 2016

I
~
I
I

: PA55
I BAN»

~La~~----~----~,-------------
4L.a+ .&Ca -
J~aCa

FIG,.IO

The two basic lattice band-pass filters and the corresponding pole·
[Link] configurations

contain alternating poles and zeros at both the origin and at infinity;
that is, if Za contains a pole at the origin, Zb should have a zero there,
and vice versa. Hence, if one of the arms contains a series resonant
circuit, the other arm contains a parallel resonant frequency. Thus,

.,..
~·~----------

"I

Fig. 11. The two-basic lattice band·

elbnination filters and the corresponding
pole-zero configurations

the two basic structures are as shown in Fig. 11. In this case, if both
the resonant frequencies are identical, the network behaves as an all-
rejection filter.

60
 THE LATTICE FILTERS

All-pass Fllter
Another type of filter known as an 'all-pass filter' is possible.
•rimary characteristics of such a filter .are: (i) the frequency

~
of the output is alsways unity, irrespective of the frequency,
the phase exists between the output and the input. That is,
d or bands of frequencies are stopped by this filter. Thus,
Downloaded by [University of California, San Diego] at 02:38 09 April 2016

filter, if the series arm contains an inductance, the shunt arm

.~.:----

__; ....'• __ ,
~

"))·~~;:;-::
... ·, ..
_L .... :~
J-··-·.
•.•.•• ;r-:".

Fig. 12. The four basic all-pass lattice

filters aDd the corresponding pole-zero
configurations

shall contain a capacitance and vice versa; and if the series arm con-
tains a series resonant circuit, the parallel arm contains a parallel
resonant circuit and vice versa. These are as shown in Fig. 12. Such
a structure is not possible in a T or a '11'-structure and this is made use
of in phase-equaliser.

Constant-resistance networks
From (2), Z L =characteristic impedance
0

= ,.;[Link]
This will always be a constant-resistance, when Za and Zb are inverse
with respect to each other or Za=kYb· Under this condition, the
-1 z.
propagation constant becomes equal to 2tanh --- which means
zoL'
that the desired propagation constant and the constant-resistance
property can be achieved simultaneously, with this type of structure.

Conclusions
In conclusion, the foregoing discussion enables one to list the

61
 STUDENTS' JOURNAL VOL. 8 NO. 1, 1967
merits and de-merits of a lattice network.
(i) This is the most general form of structure, which can a
be realised.
(ii) One of the foremost advantages of such a network is t
characteristic impedance can be made a constant
and the propagation constant can always be oll
Downloaded by [University of California, San Diego] at 02:38 09 April 2016

almost independently of this, whereas, in the earlier

the two were inter-dependent.
(iii) In addition to the earlier four basic filters obtair:
namely, low-pass, high-pass, band-pass and band-elirl
nation, another type of filter known as the all-pass fil~
also possible.
(iv) A constant-resistance network is also possible, which is q
the case with the T- or the .,._ netv:ork.
Regarding the de-merits,
(a) The lattice network, in general, involve more mnnber of
elements. This is not such a big disadvantage, as more
number of clements have to be used for a better response.
(b) The structure is always balanced.
(c) But the most important factor to he horne in mind is that
the elements obtained should have very dose tolerances.
The network, being in the form of a bridge, may become
balanced, if this is not adhered to. Specially in narrow-
hand filters, th;s is very important; otherwise, as we saw
earlier, the network may behave like an all-rejection tilter.

62