510 BELL SYSTEM TECHNICAL JOURNAL 



Limitations in Choosing Optimum Parameters 



Generally, there are mechanical, thermal, emission and specification 

 limits which prevent the realization of optimum values for all of the above 

 parameters simultaneously. A good design is one in which a nice balance 

 is effected between these various optima and their limitations. 



Limitations on Emission Current Density, / 



It is generally true that the life of a thermionic electron tube varies 

 inversely as the average cathode current density in a complicated fashion. 

 The maximum permissible value of j is therefore always a compromise 

 between our desire for highest figure of merit and long life. In the present 

 state of the cathode art as it has been evolved for the 1553 triode it is 

 possible to operate at a current density of 180 ma/cm- and obtain an 

 average life of several thousands of hours. It is perhaps of interest to 

 note that it was necessary to develop much more dense and smooth oxide 

 coatings in order to make possible such life in the thin coatings necessary 

 for operation at such close spacings. 



Limitations on Cathode- Grid Spacing, xi 



Consider the limitations in reaching the optimum in xi. There is, of 

 course, the obvious one that it is mechanically and electrically not pos- 

 sible at present to make Xi equal to zero and still retain the essential 

 features of unilateral controlled space charge flow. Granting then that the 

 spacing cannot be zero, we must choose the smallest value of xx for which 

 parallelism and reasonable tolerances can be maintained. To this end in 

 the 1553 a value of x\ = .0006" is very near this limit with present 

 structures. 



There is, however, at present another limitation which is essentially 

 mechanical in nature but makes itself felt electrically in a way not indi- 

 cated in the above simplified theory. This theory has assumed that the 

 grid dimensions are infinitesimally thin compared to the electrode spac- 

 ings. However, if this is not the case then the grid has less control action 

 than an ideal fine grid, and the intrinsic gain band product must be 

 reduced by still another factor F^ which is a function of the grid trans- 

 mission factor a = and the ratio '- where p is the pitch distance be- 



P P 



tween grid wire centers and d is the diameter of grid wires. This function 



has the form shown in Fig. 9.* 



Thus if the grid pitch and wire diameter are mechanically limited to 

 some finite though small values, the optimum in input spacing Xi will 



* Data transmitted informally from C. T. Goddarti and G. T. Ford. 



