Wi W/ I: Active Notch Filters
Wi W/ I: Active Notch Filters
A drawback of the above two networks is that in order to vary the centre frequency and still maintain the infinite null, two or three closely matching ganged variable components must be used. Several ! bridge networks are known in which a single component is sufficient to control the re"ection frequency. #owever, their practical significance is limited because the frequency response becomes sever$ ely asymmetric as the re"ection fre-. quency is varied. A more acceptable variable network was proposed by #all7. It is shown in +ig. 52c3. 4his network can be tuned by means of a single potentiometer and the tuning law is wo8l9 !7which in theory spans the whole frequency range. &n practice the tuning range is quite limited due to the e(treme non$ linear dependence of Wo on a. #owever, this network has unity gain on both sides of the null frequency, irrespective of the tuning. #owever, unlike the twin$tee and the -ien bridge it is asymmetric on a logarithmic frequency scale. 4his fol$ lows from the fact that the transfer function of this network is not given by e(pression (1) but contains an addi$ tional real pole and real %ero. A similar potentiometer tuned null network based on the twin$tee was proposed by Andreyev$, All the networks discussed so far are
characteri%ed by low selectivity. &n fact', no passive ! notch network, however comple(, is capable of achiev$ ing )o higher than 0.*. If the notch filter must be passive, a relatively high QQ may be achieved by. including an inductance as in the bridged$tee network shown in +ig. 3. &n order to achieve a complete null this network must satisfy the two conditions,
and
253
the network tends to maintain a v oltage $ gain of Ao/ 2/ + Ao). #owever, it fails to do so where the forward gain is low, i.e. in the vicinity of W (t As a result, the response curve is compressed as shown in curve 2b3 and the re"ection band is narrower. As an additional benefit, the active filter can now be cascaded without being sub"ected to loading. 4he calculated transfer function of the active notch filter is,
Ao 62s3 = -. 1 +Ao' s2+UJi S2+UJoS/(1 +AoIQo+(J)i
&S
"
233
of merit will then be i.e. proportional to the quality factor of the coil.
)o = 5 woLfr,
4he figure
A different reah%auon
shown in +ig,
As has been mentioned above, passive ! notch filters suffer from a low selectivity. A theoretically unlimited selectivity can be obtained by the use of active notch filters. 4hese can be built by various active reali%ations of the transfer function given by e(pression: 2/3. Simple active circuits are based on passive null networks in which the selectivity is raised by means of nega$ tive feedback. ;ne such scheme is shown in +ig. < and the effect of feedback can be e(plained as follows, -hen the feed$ back loop is open the network is simply a passive null network with a passband gain of Ao represented by curve 2a3 in +ig. 1. -hen the feedback loop is closed,
ity gain amplifier. &t can be seen that there are two feedback paths in the configuration, a positive unity$gain feedback which renders the effective gain of the amplifier equal to K/(l-K) instead of K, and a negative feedback which subtracts the output voltage from the input. &f K/(l-K)=A(). this method is equivalent to the former and the transfer function is,
=ractical circuits
4he simplest amplifier for the above method is the emitter follower. #ow$
c
Vo
(1+k)(1+2k)
.Fig. 5. Three
! n !! networ"s# ($) the s%&&etri' twin(tee) (b) the Wien *ri+ge) 2c3 a ,otentio&eter(t ne+
networ".
null
n~twork
+ig. 3. -ri+ge(tee .CL n !! networ") the se!e'ti/it% of whi'h +e,en+s on the Q 0$'tor of the 'oiL
Fig. <. >csic active 'on0ig r$tion 0or enh$n'ing the se!e'ti/it% o0 ,$ssi/e not'h 0i!ters.
Fig. 1. Fre1 en'% 'h$r$'t!iristi's o0 the networ" in Fig. < (a) o,en !oo,) (b) '!ose+ loop.