84 BELL SYSTEM TECHNICAL JOURNAL 



fied frequency. It is quite easy to tune tubes in this way so that they are 

 uniform to within ± 0.25%. 



Unless the tube is properly designed, changes in ambient temperature may 

 seriously afifect its resonant frequency. The part of the disc which is inside 

 the glass envelope may be considered as a diaphragm supported around its 

 periphery by the glass which has a temperature coefficient of expansion 

 negligibly small compared to that of the copper. An increase in tempera- 

 ture, which causes the copper to expand, will force the cone tip to move 

 toward or away from the gap, depending on the initial slope of the nearly 

 flat portion of the disc. The temperature coeflBcient of frequency may be 

 either positive or negative, and will have extreme variations in magnitude 

 from tube to tube if consideration is not given in disc design to avoid 

 such diflficulties. 



A cavity made wholly of copper will have a fractional change in wave- 

 length with temperature the same as the fractional change in length of 

 copper (approximately fourteen parts in a million per degree centigrade). 

 As the temperature increases, the frequency decreases. At a frequency of 

 1000 mc, the approximate temperature coefficient of frequency is —.014 

 megacycles per degree centigrade; at 3000 mc it is —.042 mc/°C; and at 

 10,000 mc it is —.14 mc/°C. Magnetrons normally have temperature 

 coefficients of about these magnitudes. The ideal TR tube would have the 

 same coefficient as the magnetron; practically speaking, any coefficient be- 

 tween zero and twice the value for copper is satisfactory. 



It is practical to make a copper disc structure which has the required tem- 

 perature coefficient. Fig. H, which is a cross-section of a 721A tube, shows 

 how temperature compensation within the tube is effected. The disc is 

 slanted away from the center portion of the tube, so that as temperature 

 rises the cone is carried away from the gap. At the same time the cone 

 itself expands; the net effect is to increase the gap between the two cone 

 tips. The angle of the slanted part of the disc must be such that the gap 

 increases with temperature at the same rate that it would in an all-copper 

 cavity. If this condition is fulfilled, the net result of the expansion of all 

 the tube parts, and of the cavity itself, will be the same as if it were all made 

 of copper. This result is achieved by an experimental series of successive 

 approximations. A number of models are built until the angles are found 

 which give the desired temperature coefficient of frequency. 



The gas content of the tube was the subject of considerable study. As 

 stated in the section on Recovery Time, gases which readily form negative 

 ions are invariably the most satisfactory from that viewpoint. Gases of 

 low ionization voltage, such as the rare gases, give excellent protection but 

 usually have extremely poor recovery. The choice of a TR gas must of 

 necessity be a compromise between the two requirements. Some otherwise 



