THE L3 SYSTEM — QUALITY CONTROL IN MANUFACTURE 997 



fine grid wire used in winding the grid might be from a lot of 300 to 1000 

 meters etched to size at one time. The mounts are generally sealed into 

 their glass envelopes in lots of 50 to 200 and pumped in lots of approxi- 

 mately 50 tubes. 



Since the individual batches of parts and processing lots are of such 

 different sizes, the final product does not result in groups of tubes that 

 can be readily segregated into distinctive lots. As the tubes are assembled 

 on a regular running basis the production is treated as being of a con- 

 tinuous nature and the Continuous Production method is applied. 

 Mounts made from each week's production are identified by a serial 

 number and, at test, samples from each week's production are used in 

 determining conformance to the requirements of the distribution specifi- 

 cation. 



In the application of control charts to the L3 carrier vacuum tubes, 

 the charts kept on the characteristics required by the test specification 

 have been plotted against the C, D and E limits derived from the specifi- 

 cations. The charts kept on the parts and on other test characteristics 

 have control limits derived from the process capabilities. In some cases 

 these natural limits have been compatible with the original drawing 

 limits and in others they have been at variance. Wherever incompati- 

 bility was established, an attempt was made to improve the process or 

 where correlation studies justified the action, agreement was reached 

 with the design engineers to increase the drawing tolerances. 



3.63 Application of Control Charts 



3.631 Cathodes 



The cathodes used in the 435A, 436A and 437A are purchased from 

 an outside supplier. A maximum variation from nominal of d=0.0002" 

 was specified for the minor axis outside diameter. These limits are tighter 

 than for any previous similar cathode used, the narrowest limits hereto- 

 fore being d=0.0005". A control chart established on the first lot of 

 cathodes produced for the 436A and 437A tubes indicated that a stand- 

 ard deviation (a) of 0.00021'' was obtainable. The 3-sigma limits thus 

 derived of ±0.00063'' in addition to the displacement in the average (X) 

 for the lot of +0.00026" represented a completely unsatisfactory con- 

 dition. By selection of cathode blanks the nominal was reduced to 

 0.03121", approximately equivalent to the maximum permitted by the 

 drawling. A much narrower range resulted from this selection. The stand- 

 ard deviation for selected cathodes amounted to 0.000066" corresponding 

 to 3-sigma limits of ±0.00020". However, this distorted distribution 



