504 



BELL SYSTEM TECHNICAL JOURNAL 



variations when the distortion was about 20 per cent: two-source without 

 Hmiter on a par with two-source with Umiter, one-source two-band with 

 limiter, level compensated on-off, linear on-off, frequency-shift, and single- 

 sideband with — 6 db spacing carrier. This comparison includes only those 

 arrangements using the same type of filter (with loss per curve A, Fig. 1). 

 Figure 9 shows the results obtained by making similar tests on the wide-band 

 frequency-shift (with loss per curve B, Fig. 1) and on the narrow-band two- 



-30 -20 -10 10 20 30 40 50 



MEAN CARRIER FREQUENCY IN CYCLES PER SECOND 

 (DEPARTURE FROM NOMINAL VALUE) 



Fig. 9 — Distortion vs. mean carrier frequency characteristics of narrow- and wide- 

 band arrangements at 60 w.i).m. (23 d.p.s.). 



source arrangements (with loss per curve C\ Fig. 1). The distortion curves 

 for the wide-band frequency-shift arrangement in the latter figure are more 

 favorable than the curve for the normal-band frequency-shift arrangement 

 shown in Fig. 8 because the wide-band arrangement had a discriminator with 

 only half the slope of that used in the normal-band arrangement. The wide- 

 band arrangement with the higher cut-off low-pass filter could tolerate a 

 greater change in carrier frequency than the arrangement with the lower 



