A METHOD OF IMPEDANCE CORRECTION 797 



ment of its application to filters. Some of the probable limitations of 

 the method in other applications are suggested near the end of this 

 paper. 



The analysis used in impedance correction can also be applied to the 

 construction of networks having transmission properties somewhat 

 like those of the familiar wave filter. In contrast to the usual filter 

 theory, developed, after the analogy of wave propagation in contin- 

 uous media, from the conception of an infinite recurrent structure, how- 

 ever, it leads to networks which are not recurrent and are not divisible 

 into separate sections with matched image impedances. In its present 

 state of development the analysis is unquestionably much less powerful 

 than the established theory. Since it may be of interest as an example 

 showing at least the possibility of an alternative approach to filter de- 

 sign, however, it is discussed briefly at the conclusion of the paper. 



General Impedance Correcting Process 



If no transmission requirements were imposed upon electrical 

 structures, a wide variety of networks might be used for impedance 

 correction. For example, we might make up deficiencies of impedance 

 or admittance by a simple two-terminal network in series or in shunt 

 with the circuit. In almost all circuits, however, we are interested in 

 securing minimum transmission loss, that is to say, maximum energy 

 in the receiving impedance, throughout the frequency bands containing 

 the transmitted signals. The energy which goes into a system term- 

 inated by a correcting network depends only upon generator and the 

 corrected impedance, both of which are specified by the conditions of 

 the problem. We can increase the energy delivered to the receiving 

 device, therefore, only by reducing the amount absorbed in the correct- 

 ing network. Obviously the best possible condition is found when the 

 correcting network is composed of pure reactances. Unless either the 

 resistance or the conductance of the circuit happens to be ideal, how- 

 ever, impedance correction cannot be obtained by a simple two-term- 

 inal reactive network. For this reason, the impedance correcting 

 structures which have been developed are four-terminal networks of 

 pure reactances. Control of the resistance or conductance component 

 is gained, not by the direct addition of resistance, but rather through the 

 use of the network as a sort of variable transformer, whose impedance 

 ratio changes as we go over the frequency range. In such a circuit the 

 insertion loss of the network is determined entirely by the ratio of the 

 energy drawn from the generator by the original and the corrected im- 

 pedance. Ideal dissipationless network elements are, of course, not 

 available in practice. Except for the possible influence of this factor. 



