Volume 9, Issue 5, 01 May 1938
Index of content:
9(1938); http://dx.doi.org/10.1063/1.1710418View Description Hide Description
9(1938); http://dx.doi.org/10.1063/1.1710421View Description Hide Description
9(1938); http://dx.doi.org/10.1063/1.1710422View Description Hide Description
9(1938); http://dx.doi.org/10.1063/1.1710423View Description Hide Description
It has been found in rectifier and ignitron practice that arcback can occur after deionization of the discharge path at such vapor pressures where no self‐sustaining glow can develop. Experiments to study this kind of an arc breakdown were made in Hg vapor vessels between cold electrodes at voltages between 10 and 60 kv. It appeared that the start of arc breakdown was not dependent on the Paschen law. At the voltages used and distances between the electrodes of less than 50 cm arc breakdown always started at PD values which were much lower than those required to sustain a glow discharge. This is contrary to earlier experience at lower voltages. With rising voltage arc breakdown frequency increased in an exponential manner. Changes in vapor pressure and electrode spacing had no definite effect on arc breakdown frequency as long as glow discharge conditions were not approached. Changes in partial air pressure did not affect arc breakdown frequency if air pressure was below 10 percent of mercuryvapor pressure.Heat treatment of the cathode(graphite``anode'') has a large effect in reducing arc breakdown. Even well heat‐treated ``anodes'' show seasoning effect, i.e. steady decrease of arc breakdown frequency after initial maximum. Phenomena on cathode surface are believed to be responsible for formation of arc breakdown.
9(1938); http://dx.doi.org/10.1063/1.1710424View Description Hide Description
The shape of the cut‐off curve is explained on the basis of a component of high energy random motion superimposed on the electron orbital motion (see references). Electron temperatures determined from the cut‐off curve shape agree with those determined by probe methods. The possible effects of the random motion on other magnetron characteristics are pointed out.
9(1938); http://dx.doi.org/10.1063/1.1710426View Description Hide Description
This paper considers in detail numerical methods of solving Laplace's equation in an arbitrary two‐dimensional region with given boundary values. The methods involve the solution of approximating difference equations by iterative procedures. Modifications of the standard Liebmann procedure are developed which lead to a great increase in the convenience and rapidity of obtaining such a numerical solution. These modifications involve the use of formulas which simultaneously improve a block of points in place of a single point; methods of operating on the differences of trial functions in place of the functions themselves; and also a method of extrapolating to the final solution of the difference equations. The theory underlying these procedures is considered in detail by a new method which involves the expansion of the error and difference functions in terms of eigenfunctions. This permits definite comparison of rates of convergence of various procedures. The techniques of handling practical problems are considered in detail.