VACUUM TUBES AS HIGH-FREQUENCY OSCILLATORS 117 



to change with time, frequent readjustments of the cathode tempera- 

 ture are usually required. 



No completely satisfactory and generally accepted theory of the 

 positive grid oscillator has as yet been given. Many theoretical 

 papers dealing with the mechanism of oscillation have been published. 

 Some of these papers resort to pictorial explanations which, from their 

 very nature, must leave out certain basic factors. Readers interested 

 in a resume of the various theories are referred to the ex'cellent review 

 by Megaw ^* and to the original papers. It is now recognized that any 

 accurate theory must be based upon a general consideration of all the 

 forces acting upon the electrons in their flight between the electrodes. 

 This may take the form of either a particular solution of the classical 

 electromagnetic equations for the conditions within the tube or an 

 analysis of the energy contributions due to individual electrons in their 

 passage across the inter-electrode space. 



Construction of a Positive Grid Tube 



A representative positive grid tube of current design described by 

 Fay and Samuel ^ before the International Scientific Radio Union is 

 shown in Fig. 19. This tube differs from the conventional negative 

 grid tube primarily in the construction of the grid and in the arrange- 

 ment of the leads. While designed primarily for use in the frequency 

 range from 500 to 550 megacycles, it illustrates the general problems 

 encountered in the construction of the positive grid oscillator of this 

 type for any frequency range. 



The grid consists of a number of parallel wires supported by cooling 

 collars at each end, the so-called squirrel cage construction. It will 

 withstand 150 watts heat dissipation safely, and provides a minimum 

 of circuit inductance and resistance. The grid diameter is fixed by 

 the frequency for which the tube is designed and by the desired 

 operating potential, such that the relationship 



.. = ^ (1) 



n 



is approximately satisfied, where dg is the diameter of the grid, Ki is 

 a constant, n the frequency, and Eg the applied grid potential. 



An indefinite increase in output at a fixed frequency by the simul- 

 taneous increase in the grid diameter and in the applied grid potential 

 is not possible because of the limited permissible grid dissipation per 

 unit area. The optimum grid current is found to follow roughly a 3/2 



