CARRIER TELEGRAPH METHODS 501 



band, except that the distortion at the higher speeds is greater with curve E 

 of Fig. 6 because sideband components in the middle portion of the total 

 frequency band were considerably attenuated by the channel filters of the 

 one-source two-band arrangement, but not by those of the wide-band fre- 

 ([uency-shift arrangement. The distortion vs. speed characteristic of the 

 two-source arrangement having paths with loss characteristic similar to 

 curve C of Fig. 1, is shown by curve F of Fig. 6. The large distortion is due 

 to the narrower sidebands transmitted and to the combined use of two inde- 

 pendent oscillators and a limiter, the effect of which is discussed above. 



Single-Sideband A rrangement 



The same linear detector, receiving relay, and effective relay operating 

 current were used for the single-sideband tests as for the linear on-off tests. 



An ideal single-sideband arrangement should operate at twice the speed of 

 the on-oflf arrangement for the same pass band, if the quadrature component- 

 is eliminated. The cost of a phase discrimination method of reception' for 

 this purpose would probably be prohibitive in practice. If the quadrature 

 component is allowed to remain, it is a principal cause of distortion, so that 

 the single-sideband method gives only a slight increase in speed. The effect 

 of the quadrature component on telegraph distortion can be reduced- by the 

 transmission of a certain amount of spacing carrier current. Curve B of 

 Fig. 7, measured with a spacing current 6 db below the marking current, 

 shows less distortion than curve A of Fig. T, measured with no spacing cur- 

 rent; and, in the range of speeds investigated, does not differ greatly from 

 curve C, taken on the linear on-off arrangement without channel filters. 

 Thus, it is apparent that the single-sideband arrangement is capable of higher 

 speeds, for a given distortion and band width, than the other arrangements 

 here considered. 



Tests of Carrier Frequency \'ariations 



When a carrier telegraph circuit contains a radio or carrier telephone link, 

 some instability may occur in the average received carrier frequency. In 

 order to investigate the effect of varying the mean carrier frequency, the 

 carrier supply frequency was varied as a matter of convenience. Since the 

 signals were transmitted through both sending and receiving channel tilters 

 the effects observed were doubtless about twice as bad as if only the received 

 carrier frequency had been varied, except j^erhaps in the frequency-shift 

 arrangements where the discriminator produced a large effect. 



Distortion at 60 Words per Minute 



In Figs. 8 and 9, the distortion obtained over the various arrangements is 

 shown as a function of carrier frequency variation from the nominal value, 



