498 BELL SYSTEM TECHNICAL JOURNAL 



have zero bias while transmitting the test sentence because the characteristic 

 of the level compensator was such that reversals could not be used to adjust 

 for zero bias of the received signals. The distortion vs. speed characteristics 

 of the level compensated on-off and 40B1 arrangements are given by curves 

 A and C, respectively, in Fig. 5. This figure indicates that at high speeds the 

 frequency-shift arrangement is subject to somewhat greater distortion than 

 the on-off method. 



It may be argued that Fig. 5 is not a fair comparison between frequency- 

 shift and on-off methods because of the difference in detector characteristics. 

 In order to overcome this objection an experimental on-off arrangement was 

 set up utilizing a linear detector and the same receiving relay as in the fre- 

 quency-shift arrangement. The effective operating ampere-turns in the 

 255A receiving relay were kept the same for both frequency-shift and linear 

 on-oflf methods of transmission. The line winding current varied from about 

 10 mils during spacing signals to 50 mils during marking signals, and the 

 biasing winding current tending to move the armature toward spacing was 

 about 30 mils for the on-off arrangement. The effective relay operating 

 current in the frequency-shift arrangement was +20 mils in the marking 

 condition and — 20 mils in the spacing condition. 



The distortion measurements are shown in Fig. 6. In order to compare 

 frequency-shift with the linear on-off arrangement, consider curves A and B 

 of Fig. 6. There is not much difference between them, but the frequency- 

 shift characteristic shows slightly higher distortion over part of the speed 

 range, as in Fig. 5. 



In these tests the channel loss characteristic used was that of Fig. 1 , curve A. 



Two-Source and One-Source Two-Band Arrangements 

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

 current were used for these two-band arrangements as for the frequency-shift 

 arrangement. Curves C, D, and E of Fig. 6 show the speed characteristics 

 of various two-source and one-source two-band arrangements in which the 

 marking and spacing paths had loss characteristics similar to curve A of Fig. 

 1. Curve C of Fig. 6 applies to the arrangement using two oscillators and 

 no limiter and does not differ greatly from the characteristic for the on-off 

 method, curve B. Curve D applies to the arrangement using two oscillators 

 and limiter, and shows greater distortion than curve C because of modulation 

 products arising in the limiter between the sidebands of the marking and 

 spacing carriers. This type of interference was due to discontinuities in 

 phase of the carrier wave at the signal transitions, and was eliminated by 

 the use of a frequency modulated oscillator in place of the two independent 

 oscillators, as indicated by curve E. The latter is somewhat similar to curve 

 C on Fig. 4 measured on the frequency-shift arrangement with the wide 



