98 THE BELL SYSTEM TECHNICAL JOURNAL, JANUARY 1951 



By a similar analysis, the locus of the shunt type of negative impedance 

 can be derived. With the proper choice of an impedance connected to 

 terminals 1 and 2 of the negative impedance converter the locus of the 

 impedance seen at terminals 3 and 4 can be made to follow a circle in the 

 clockwise direction (at least as long as k approximates a numeric). Thus, 

 over a portion of the frequency range the series and shunt type of negative 

 impedance can be made to have very similar impedance characteristics. 

 At very high or low frequencies where k has an appreciable angle there will 

 be a distinct difference between the locus of the series and the corresponding 

 shunt type of negative impedance. This would be expected because in one 

 case the network impedance is multiplied by — ^; and in the other it is 

 divided by — k. 



The El Converter 



The circuit of the negative impedance converter used in the El telephone 

 repeater is shown in Fig. 6(a). It consists of a transformer, two triode tubes, 

 an RC network and an inductor. The transformer T couples the cathodes of 

 the two tubes to terminals 1 and 2. The tubes while apparently in push-pull 

 are biased for Class A operation. The RC network couples the plate of each 

 tube to the grid of the other. The inductor L supplies plate current. 



The equivalent circuit of Fig. 6(a) is shown in Fig. 6(b). In obtaining this 

 equivalent circuit the two tubes have been assumed to be identical. Thus 

 the converter used with the El repeater can be reduced to a four- terminal 

 network consisting of (reading from left to right) : the equivalent circuit of 

 the line transformer T; the two biasing resistances R^', the two plate re- 

 sistances Rp divided by (1 + /X2) ; the ideal negative impedance converter C 

 of ratio — (mi — 1)/(m2 + 1) to 1; the elements of the RC coupling arrange- 

 ment which appear in shunt across terminals 3 and 4; the inductor L also 

 shunted across these terminals; and the capacitor Cx which has been 

 added to represent both the distributed capacitance of the windings of L 

 and the capacitance between vacuum tube plates. 



It should be noted that 1x2 is the amplification factor of each tube; and 

 that Ml equals j3ijU2 where jSi is a proportionality factor representing the 

 fraction of the voltage, between the plate of one tube and ground, which is 

 fed back to the grid of the other tube. The value of ^i depends upon the 

 values of Ci, i?3, -^6 and C2 of the RC coupling circuit. If ^i approaches 

 unity in value then /xi approximates /U2. If this is so and if both /ii and iU2 

 are relatively large in magnitude compared to unity then the ratio of 

 transformation, — (mi — 1)/(m2 + 1) to 1, approaches although it cannot 

 equal —1:1. 



As an illustration of how the elements in the El circuit may be propor- 



