530 BELL SYSTEM TECHNICAL JOURNAL 



are also tapered in the same way, the signal-to-interference ratio is inde- 

 pendent of the frequency of the disturbing CW. Varying the CW frequency 

 only changes the number of the channel into which the interference falls. 

 Loading the channels distributes the interference over several channels 

 instead of concentrating it in one, but we have plotted in Fig. 20 the more 

 severe case in which all channels are idle. 



We have not undertaken to compute curves for similar system inter- 

 ference in the case of FDM-FM, but estimates for two extreme conditions 

 can be made. In the case of low index FM systems the carrier frequency 

 component of the spectrum is not affected by the modulating signal and the 

 FM wave is, in fact, like an AM wave with the carrier displaced 90 degrees 

 in phase. A similar interfering FM wave combines with the wanted FM 

 to produce frequency or amplitude variations and does this cyclically as 

 the r-J phase between the systems varies. When the phases are appropriate 

 for the production of frequency variation, crosstalk appears in the wanted 

 reception at a level lower by the ratio of FM wave amplitude. Averaging 

 over all r-/ phases should reduce the crosstalk by 3 db. The actual amount 

 of interference received in a channel is less than would be predicted from 

 replacement of the interfering FM wave by fluctuation noise of the same 

 mean power spread over the r-f band, because the bulk of the interfering 

 power is contained in the carrier component located at a frequency which 

 does no harm. Increase of the frequency swing in both systems produces 

 significant reduction in crosstalk when the carrier amplitude diminishes 

 appreciably and important higher order sidebands appear, i.e. when the 

 interfering system has its spectrum spread out more or less uniformly, like 

 noise. Systems designed for wide swings under full load may, however, 

 operate with only a few channels active; in such cases the low index situation 

 may exist and the received interference will be down approximately by the 

 ratio of the FM waves, without the benefit of FM advantage. While in 

 this situation the bulk of the interfering power is again contained in the 

 harmless carrier, the received interference is concentrated in a few channels 

 and is greater than if the interfering wave power were spread, like noise, 

 thinly over the r-f band, which in this case is many times wider than the 

 band occupied by the low index signal. For such adverse loading conditions, 

 the curve for similar system interference, while starting at the left above 

 the corresj)onding noise curve of Fig. 19, may actually cross over and finally 

 approach it from the lower side. 



In the case of systems of very wide swing such as are involved in Table II 

 we regard the interfering system as equivalent, under all common load con- 

 ditions, to noise spread uniformly over the bandwidth and having the same 

 power as the interfering wave. The entry in Table II is obtained by reading 



