552 BEI.I. SYSTEM TECHMCM. JOCRXAL 



it is imperative that the difference frequency between the transmitter and 

 tlie beating oscillator should be maintained by automatic means. As an 

 illustration of the problem one may expect to have to correct frequency 

 shifts from all causes, in a 10,000 megacycle system, of the order of 20 mega- 

 cycles. Such correction may be demanded at rates of the order of 100 mega- 

 cycles per second per second. 



Although the frequency range of triode oscillators has since been some- 

 what extended, at the time that beating oscillators in the 10 centimeter 

 range were lirst required the triode oscillators available did not adequately 

 fullill all the requirements. In general the tuning and feedback adjust- 

 ments were complicated and hence did not adapt themselves to autcmatic 

 frequency control systems. \'elccity variation tubes of the multiple gap 

 type which gave more satisfactory performance than the tricdes existed in 

 this range. These, however, generally required operating voltages of the 

 order of a thousand volts and frequently required magnetic tields for focus- 

 sing the electron stream. The tuning range obtainable by electrical means 

 was considerably less than needed and, just as in the case of the tricde oscil- 

 lator, the mechanical tuning mechanism did not adapt itself to automatic 

 control. These dilTiculties fccussed attention on the refiex oscillator, whcse 

 properties are ideally suited to automatic frequency control. The feature 

 of a single resonant circuit is of considerable importance in a military applica- 

 tion, in which simple adjustments are of primary concern. The repeller 

 control of the phase of the negative electronic admittance which causes 

 oscillation provides a highly desirable vernier adjustment of the frequency, 

 and, since this control dissipates no power, it is particularly suited to auto- 

 matic frequency control. Furthermore, since the upper limit on the rate of 

 change of frequency is set by the time of transit of the electrons in the repeller 

 field and the time constant of the resonant circuit, both of which are gen- 

 erally very small fractions of a micro-second, very rapid frequency correction 

 is possible. 



As the frequency is varied with the repeller voltage, the amplitude of 

 oscillation also varies in a manner ])reviously described. The signal to noise 

 j)erformance cf a crystal mixer depends in part on the beating oscillator 

 level and has an c jitimum value with respect to this parameter. In conse- 

 quence, there are limitations on how much the beating oscillator power 

 may depart from this ( ptimum value. This has a bearing on the oscillator 

 design in that the amount of amplitude variation permitted for a given 

 frequency shift is limited. The usual criterion of perfomance adopted has 

 been the electronic tuning, i.e. the frequency difference, between points for 

 a given re])eller m( dc at which the i^ower has been reduced to half the maxi- 

 mum value. 



Reception of the wrong sideband by the receiver causes trouble in con- 



I 



