Fig. 16.14-1 (a) Constant-k T-Section Band-Pass Filter (b) Constant-k Π

Download as docx, pdf, or txt
Download as docx, pdf, or txt
You are on page 1of 9

16.

14 CONSTANT-K BAND-PASS FILTER

Constant-k T-section and Π-Section band-pass filters are shown in Fig.


16.14-1.

Fig. 16.14-1 (a) Constant-k T-Section Band-Pass Filter (b) Constant-k Π-


Section Band-Pass Section

‘Constant-k’ implies that the product of underlying


impedances Z1 and Z2 must be a constant independent of frequency.

Therefore, L1 C1 has to be equal to L2 C2 if the filter is to be of a constant-k


type. This constraint is imposed on band-pass filter design.
This product is defined as Ro2 of the filter.

Note that the constraint L1C1 = L2C2 resonating at the same frequency. We
define this frequency as ω rad/s.

Consider the T-section in Fig. 16.14-1. The series branch is open at DC and
the shunt branch is short at DC. Therefore, the DC steady-state gain of the
filter is zero. The series branch becomes a short-circuit and the shunt
branch becomes an open-circuit at fo. Therefore, the gain at that frequency
is unity. The series branch is open at high frequencies and the shunt branch
is short at high frequencies. Therefore, the gain of the filter goes to zero
as f → ∞. Thus, this circuit is indeed of a band-pass nature.

It is clear that the centre frequency of the filter is ao. We proceed to


determine the upper and lower cut-off frequencies of the filter by analysing
its propagation constant. A constant-k filter terminated at characteristic
impedance has a propagation constant that is decided by the following
equation.
Both Z1 and Z2 are pure reactances at all ω, and hence, we can express them
as Z1 = jX1 and Z2 = jX2. X1and X2 can be positive or negative depending on
the ω value. For instance, X1 – the series branch reactance – is negative
for ω < ωo and positive for ω > ωo. X2 – the shunt branch reactance – is
positive for ω < ωo and negative for ω > ωo.

Therefore, the only way in which eγ can be a complex number is by the

term becoming imaginary.

This term can become a pure imaginary number only if is


positive. X1 is the reactance of series combination of L1 and C1. It starts at –
∞ at ω = 0, crosses zero at ω = ω and increase further towards ∞ as ω → ∞

Then, there must be a range of frequency values f1 < fo < f2 in which


In that frequency range, eγ will be a complex number and we can eγ express

as . The attenuation suffered by a signal in


this frequency range is given by the magnitude of eγ(i.e., ea).
This implies that, in the frequency range f1 < fo < f2 in which there is
no attenuation in the filter (provided it is always Zo-terminated). The
attenuation constant α is zero in that frequency range. Hence, that range
must be the pass-band of the filter. We obtain the edges of pass-band –

i.e., f1 andf by solving with the equality sign.

Pass-band edge frequency values for any constant-k filter can be obtained
by solving the equation X12 = 4Ro2 where X1 is the series arm reactance
and Ro2 = X1X2, where X2 is the shunt arm reactance.
Cut-off frequencies of a constant-k prototype band-pass filter.

These two values give the lower cut-off angular frequency and the higher
cut-off angular frequency. The bandwidth is the difference between the two.

We develop an interesting relationship between the three


frequencies a1, ωo and a2 by multiplying ω1and ω2.

Thus, the centre frequency of this band-pass filter is the geometric mean of
its cutoff frequencies.

16.14.1 Design Equations of Prototype Band-Pass Filter

Design specifications will be the value of load resistance RL and the lower
and upper cut-off frequenciesf and f2.

Ro will be the value of characteristic impedance at ωo and the usual design


choice is to make the filter zo-terminated at that frequency. Therefore, Ro is
chosen to be equal to RL. Therefore,
Therefore, the design equations are

Design equations for constant-k prototype band-pass filter.


16.15 CONSTANT-K BAND-STOP FILTER

Constant-k T-section and Π-section prototype band-stop filters are shown


in Fig. 16.15-1.

Fig. 16.15-1 (a) Prototype Band-Stop T-Section (b) Prototype Band-Stop


Π-Section

Therefore, L1C1 has to be equal to L2 C2 if the filter is to be of constant-k type.


This constraint is imposed on band-pass filter design.
Note that the constraint L1C1 = L2C2 results in both series branch and shunt
branch resonating at the same frequency. We define this frequency
as ωo rad/s.

Consider the T-section in Fig. 16.15-1. The series branch is short at DC and
the shunt branch is open at DC. Therefore, the DC steady-state gain of the
filter is unity. The series branch becomes an open-circuit and the shunt
branch becomes a short-circuit at fo. Therefore, the gain at that frequency is
zero. The series branch is short at high frequencies and the shunt branch is
open at high frequencies. Therefore, the gain of the filter goes to unity
as f → ∞. Thus, this circuit is indeed of a band-stop nature.

A frequency is in the pass-band of this filter if X1 at that frequency

satisfies condition as in the case of band-pass filter. X1 starts at zero


for ω = 0 (due to inductor), goes to ∞ as ω → ωo from left, starts at on the
right of ωo and ends up at zero as ω → ∞. Therefore, 0 < f < f1 and f2 < f must
be the pass-band of the filter, where f1 and f2 are solutions of the

equation Solving this equation, we get,

Cut-off frequencies of a constant-k prototype band-stop filter.


Design equations for a constant-k prototype band-stop filter.

The design specifications will be the values of load resistance RL, f and f2.
The design equations can be derived using the same procedure we
employed in the case of a band-pass filter.

The design equations are

The filter will be Zo-terminated only at ω = 0 and ω = ∞ with this design. In


practice, m-derived sections and m-derived half-sections will be needed to
tailor the pass-band and stop-band attenuation characteristics in the case
of band-pass filters and band-stop filters too. We skip these topics, as they
will represent too great a detail for an introductory textbook.

However, we can not close this chapter without a brief mention on an


important circuit element that we have neglected in our analysis of filters. It
is the winding and core loss resistance of inductors used in the filter
designs. Obviously, they will affect the pass-band attenuation. The
attenuation constant will no longer be correctly described by the compact
expression that we derived for it in this chapter. A practical design can not
ignore the winding resistance of inductors and this ugly reality will often
dictate the computer simulation for design validation.

You might also like