1000 THE BELL SYSTEM TECHNICAL JOURNAL, JULY 1953 



The difficulties with instrumentation encountered with measurements 

 of gold plating thickness make it more difficult to establish the standard 

 deviation of the process. By determining the standard deviation of the 

 measuring method used, it appears that a value of 4 to 5 micro inches 

 may be realized. This corresponds to 3-sigma limits of 12 to 15 micro 

 inches. Fortunately, correlation studies have established that this control 

 has resulted in tube characteristics which have more than met the test 

 requirement. 



3.634 Screen Grids 



The screen grids used in the 435A and 436A tubes are of conventional 

 design. The tolerance for minor axis diameter is ±0.001''. The grids are 

 first wound in strips on a mandrel which is shaped to the approximate 

 dimensions of the desired finished grid. The strips are next degreased and 

 given a preliminary heat treatment at 700° C for 5 minutes and then 

 stretched longitudinally to straighten the side rods. The strips are then 

 cut into individual grids and the loose grid lateral turns are trimmed from 

 the ends. A heat treatment at 925° C for 15 minutes follows and finally 

 the grids are sized, which consists of stretching the major axis on an 

 expanding set of sizing blades to obtain the desired shape and size. 



It was felt that it would be difficult to control the minor axis around a 

 desired nominal since so many operations took place between the winding 

 of the grid and the final sizing operation. Corrections for deviation of the 

 center of the distribution usually consisted of slight changes in the 

 amount of stretch imparted at the sizing operation and major corrections 

 were usually attempted by a change in winding location on the tapered 

 winding mandrel. 



Process capability studies revealed a parent distribution of approxi- 

 mately ±0.001" which is within the specified tolerance but provides no 

 allowance for shift in average from one lot to another. Successive lots of 

 grids processed in the same manner showed a considerable shift in center 

 of distribution, amounting to as much as ±0.0007". 



Studies were then made to determine the important variables which 

 needed to be controlled in order to minimize the shift in average value. 

 It was found that for a given spool of grid wire, considerable control of 

 the process average could be obtained by careful attention to two points 

 in the processing. The grid minor axis size varied directly as the tension 

 of the lateral wire was increased at the winding operation. By variation 

 in tension, as much as 0.0005" shift in process average could be obtained. 

 Secondly, when a more precisely controlled heat treating oven was used 



