MAGNETRON AS GENERATOR OF CENTIMETER WAVES 337 



a heavy central support lead inside the Kovar cone through a hole pro- 

 vided in the cone. The input end of the heater chamber is flared to ac- 

 commodate the Kovar support cone to which it is brazed. All of the ma- 

 terials are thus of good structural strength. The tungsten and molybdenum 

 are good thermal conductors, the Kovar not. Some parts of the cathode 

 were grit blasted to increase their radiative emissivity. 



The total axial motion of the cathode by thermal expansion from cold 

 start to operation is 0.008 in. The cathode is offset axially by this amount 

 when installed. 



The cathode surface consists of a nickel matrix base, like that developed 

 for the 725A magnetron, into which the active coating is impregnated. 

 Attempts were made to increase the thermal conductivity through the 

 0.008 in. thickness of matrix. Larger particle size was used. Ball milling 

 the nickel powder before sintering resulted in more dense particles. A later 

 improvement was effected by making the matrix oversize by a mil or two, 

 impregnating it with the active coating, and then compressing it to size 

 in a mold. This procedure also resulted in a surface far superior in smooth- 

 ness and regularity to that previously obtained. 



The first output circuit used in the 3J21 magnetron, like that in the 

 Columbia model, involved a quarter wave length transforming section 

 of low characteristic impedance (about 20 ohms), extending from an iris 

 in the "back" of one of the large resonators directly into the output wave 

 guide. This section was 0.350 in. high and about 0.0125 in. wide. It 

 was made by milling a slot into solid bar stock, the top of which was closed 

 by brazing on a copper plate. A quarter wavelength section of this hne 

 was brazed between the anode body and the output wave guide piece which 

 carried the choke joints and wave guide window seal. This latter piece was 

 fabricated by hubbing a rectangular wave guide in a copper cylinder into 

 which the circular choke was later turned. The wave guide window con- 

 sisted of a circular glass disk sealed into a Kovar cup hke that used in the 

 4J50 magnetron (compare Figs. 77 and 75). In the 3J21, as in the 4J50, 

 the external choke facing the wave guide window and a short section of 

 wave guide were later incorporated as part of the magnetron. 



Holding the pulling figure of the operating magnetron to a specified value 

 presented a serious problem. The problem was one of securing uniformity 

 of the narrow transformer section and, to some extent, of the resonator to 

 which it is attached. A variation of 1 mil in the width of this transformer 

 produced 3 mc/s change in the pulling figure. Even with an improved 

 mechanical design, the spread in output characteristics was uncomfort- 

 ably large. Part of the difficulty resulted from the formation of solder 

 fillets in the transformer section during brazing, the size of which could not 

 easily be controlled. A rigid inspection of both electrical and mechanical 



