1480 THE BELL SYSTEM TECHNICAL JOURNAL, NOVEMBER 1957 



In such cases the 2B meter reading is also adjusted to the effective mes- 

 sage circuit noise which would have existed in the presence of speech on 

 the tested channel. 



Open triangles connected by fine dotted lines indicate summary plots 

 for the idle channels, and for the busy channels. 



Examination of the plot shows that there is only a general correlation 

 between the 2B set and level distribution recorder readings, as sum- 

 marized by the dotted lines. Thus the 2B meter is not a reliable instru- 

 ment to denote how noisy a circuit is for data transmission of the speed 

 used in the proposed project. 



The telephone objective for Nl circuits, as read on the 2B meter, is 

 indicated by the vertical dotted line. This show^s that most of the cir- 

 cuits (actually 55 out of 62) met the objective. 



The suggested requirements for data are shown by the horizontal 

 dotted lines. The "through" or noncompandored channel has a 3 db more 

 lenient requirement (as indicated during some of the tests) than the 

 compandored one. This results from the penalty mentioned earlier which 

 compandors impose on in a data circuit, as compared with the 23 db or 

 so advantage that it introduces for voice transmission. Only a few" of the 

 circuits measured (actually 10 out of 62) met the suggested require- 

 ments. 



The principal conclusion reached from these measurements is that 

 where used for a data service of the type considered, with vestigial 

 sideband transmission, most of the hitherto installed Nl circuits (and 

 probably other compandored circuits) will require modification to reduce 

 noise exposure. Also noise measurement will be more complicated for 

 such a service than for telephony. 



3.4 Envelope Delay Distortion 



Some simple theoretical considerations'' have shown that the envelope 

 delay distortion limits for a telephotograph circuit generally also hold for 

 a data transmission circuit of the same speed. The principal difference is 

 that less emphasis need be placed for data circuits upon fine structure 

 deviations of the envelope delay as plotted on a frequency scale. In 

 general, distortion of ±0.4 signal element in the important part of the 

 band has been found to give a signal-to-noise impairment of some 3 db 

 in signal reception. This has been assumed here as a tentative engineer- 

 ing objective. 



In accordance with this, the envelope delay requirements for service 

 with the vestigial sideband signal consideration have been set at not to 

 exceed 500 microseconds (±250 microseconds) between 1,000 and 2,500 



