CARRIER TELEGKAril .\fETHODS • 489 



Frcqiiouy-Shijl Terminal . I pparaius 



In most of tlic tests on the frequency-shift arrangement shown in Fig. 2, 

 the oscillator frequency was caused to vary abruptly by a relay modulator. 

 The sending relay was of the same type as used in the on-off arrangement and 

 varied the tuning capacity of the oscillator. For some other tests the fre- 

 quency variation was made more gradual by converting the sent signals into 

 polar signals and passing these through a low-pass filter in order to round the 

 wave so that the pulses had an approximately sinusoidal shape during 

 reversals and attained steady-state value only at the center of each pulse. 

 These polar signals were used to control the plate resistance of either of two 

 diodes, thereby connecting a positive or negative reactance across the tuned 

 circuit of the oscillator.'^ This caused the oscillator frequency to be either 

 increased or decreased in proportion to the amplitude change of the control 

 current. During most of the tests the spacing frequency was 1920 cycles 

 and the marking frequency was 1990 cycles, both equally spaced from the 

 midband frequency, 1955 cycles. The oscillator, when on marking fre- 

 quency, was set to produce one milliwatt into the 600-ohm resistance artiti- 

 cial line. After passing through the receiving filter connected to the output 

 of the artificial line, the signals entered a limiter (unless otherwise stated) 

 delivering an output current which was practically constant for input levels 

 between —55 and +25 dbm. From the limiter the signal passed into a 

 frequency discriminator circuit having two output branches, each of which 

 was connected to a diode detector tube followed by a low-pass filter. The 

 two discriminator branch circuits in combination with their detectors and 

 low-pass filters had output amplitude vs. input frequency characteristics of 

 opposite slopes. After differential recombination of the two low-pass filter 

 outputs, the resultant characteristic was linear over the range of fundamental 

 frequencies transmitted by the limiter. The differentially recombined wave 

 in the final d-c amplifier had an amplitude substantially proportional to the 

 instantaneous deviation from the average value of the received carrier fre- 

 quency over a range of ± 70 cycles. (Some calculations by one of the 

 writers indicate that the use of discriminators of this type is helpful in 

 reducing characteristic telegraph distortion.) In most of the tests the low- 

 pass filters associated with the detector output had a cut-off frequency (about 

 503 cycles) low enough to suppress the carrier but high enough not to affect 

 the telegraph transmission. The final d-c amplifier was substantially linear 

 and increased the d-c wave to a suitable value for operating the polar receiv- 

 ing relay. 



The wide-band frequency-shift arrangement was similar to that just de- 

 scribed, except for the change in filters and tuning of the sending oscillator 

 and the discriminator. The spacing frequency was 2055 cycles and the 



