564 BELL SYSTEM TECHNICAL JOURNAL 



frequency has drifted so that the oscillator is operating in a range between 

 A and H of Fig. 19. If now the operation of the system is momentarily 

 interrupted the a.f.c. system will start hunting. This is done by returning 

 to the non-oscillating repeller voltage just as when operation is initiated. 

 When the hunting repeller voltage passes through the value between A 

 and B from the non-oscillating state no oscillation occurs and hence the 

 a.f.c. cannot lock in and the system becomes inoperative. Thus it is im- 

 perative that hysteresis be kept, as a minimum requirement, outside the 

 useful electronic tuning range. 



As indicated in Section \TII it was found that the electronic hysteresis 

 occurred when the electron stream made more than two transits across the 

 gap. Thus an added objective of the design of the 10 centimeter metal 

 oscillator became the achievement of an electron optical system which 

 would limit the number of transits to two while insuring that the maximum 

 number of electrons leaving the cathode would make the two transits with 

 a minimum spread in zero signal transit time. 



Figure 64 shows a sectional view of the final design adopted. The elec- 

 tron optical structure differs from that of the 723A/B in a number of 

 respects. The first grid of the 723x'\/B has been eliminated and one of the 

 cavity grids now plays a dual role in simultaneously serving as an accelerat- 

 ing grid. The grids are curved towards the cathode, which has a central 

 spike. This arrangement is intended to produce a hollow cylindrical elec- 

 tron stream. It will be observed that the second grid is larger in diameter 

 than the first and that the repeller has a central spike. The design is such 

 that the cylindrical beam entering the repeller region is caused to diverge 

 radially, so that in re-traversing the gap after its reversal in direction it 

 impinges on and is captured by the frame supporting the first grid. 



The repeller design was determined by using an electrolytic trough to 

 determine the potential plots for a number of trial configurations. Then 

 by making point by point calculations of the electron paths the best con- 

 figuration was chosen. A typical example of such path tracing is shown in 

 Fig. 65. This figure shows the equipotential lines and the trajectories 

 computed for electrons on the inner and outer boundaries of the outgoing 

 stream. The method of calculating the trajectories has been described by 

 Zworykin and Rajchman.^" It assumes that space charge may be neg- 

 lected. Fig. 65 shows that the repeller design is such that the cylindrical 

 outgoing stream is focussed upon its return onto the frame supporting the 

 first grid. The cathode spike prevents emission from the central portion 

 of the cathode, since it would be difficult to prevent electrons from this 

 portion from returning into the cathode space. A second requirement on 



12 V. K. Zworykin and J. A. Rajchman, Proc. of I.R.E., Sept. 1939, Vol. 17, No. 9, pp. 

 558-566. 



