.IX F.LF.CTRICAI. FREQUENCY ANALYZER 30J 



The s[HH-ial fi-atures of design of this an.ilyzer circuit can Ik; better 

 explaineil \i\ reference to a typical recoril made by the apparatus. 

 Fig. 2 is the reooril of analysis of the current from a buzzer which' 

 vibrates with a frequency sliglitly under KiO per se<:ond and gives 

 an irregularK- sli.iped vva\e which is shown in tin- accomjianying 

 oscillogram. In taking this record the windings of the tuning in- 

 ductance were in parallel so as to give the fre(iuency range 80-5000 

 cycles. The vertical scale gives appro.ximatcly the r. m. s. current 

 in milliampi>res at each frequency (as read on the horizontal scale) 

 at which a {)eak occurs. It will be seen that a peak occurs at each 

 multiple of the frequency of the buzzer. The r. m. s. v^ilues of input 

 current at the corresponding frequencies as read from the peaks on 

 the record are: 160, 1.6 milliampercs; 320, 1.6 milliamperes; 480, 

 1.2.1 milliamperes; 640, 1.2 milliamperes; 800, 1.4.") milliamperes; 

 i)6(). 1.2") milliamperes; 1120. 1.2 milliamperes; 1280, 1.1 milli- 

 amperes; 1440, l.Oo milliamperes; KiOO, 1.0 milliamperes; ITtiO, 

 1.0 milliam[X'rcs; etc. The root square sum of all components 

 shows that 4.7 milliamperes was the effective value of the complex 

 current fed into the anahzer. 



The fact that the 80-5000 cycle records read directly the current 

 at each frequency component is due to the special design of the 

 input network. A small correction is still necessary but can be 

 neglected except where maximum obtainable accuracy is desired. 

 If the input network were a pure resistance the higher frequency 

 components would produce relatively lower peaks because of the 

 falling off of etTftciency with frequency of the amplifier-rectifier cir- 

 cuit and the increase in resistance of the tuning coil. The input 

 network was designed empirically so that with constant input cur- 

 rent the voltage drop across the input terminals increases with fre- 

 quency in such a way as to compensate for these high-frequency 

 losses. The tests to determine this were matle by taking records 

 of single frequencies of known amounts. 



It will be seen that the frequency scale is gradually contracted 

 as the upper end of the record is approached. Owing to the in- 

 crease in resistance of the coil with frequency, the sharpness of tuning 

 of the analyzing circuit decrea.ses with frecjuency. Each peak on 

 the record corresponding to a single frequency is a plot of the res- 

 onance curve of the variable tuned circuit. The sizes of the capa- 

 citance steps are so adjusted that a sufTficient number of points, 

 necessary to trace a resonance peak at all frequencies, is recorded. 

 The length of the record and the time required for an analysis are 

 determined by the number of points needed. 



