590 BELL SYSTEM TECHNICAL JOURNAL 



on these values and limits, and assuming that Toronto would be the 

 only gain switching center directly involved, a study was made of the 

 transmission possibilities for each group of circuits that was expected 

 to be routed through the cable. This study indicated that, provided 

 the return loss and crosstalk values required of the cable by the as- 

 sumed singing points and crosstalk gain could be met, all circuits could 

 be 2-wire between Toronto and Barrie with some transmission margin 

 and also that this type of cable could be extended at least another 

 20 miles to Orillia. 



A cost study, assuming a 101-pair cable, of this type, indicated that 

 while, due to the additional loading costs of the closer loading spacings, 

 the cable costs were very nearly the same as for a quadded 19-gauge 

 H88-50 cable, the considerably reduced repeater and repeater station 

 costs made this plan appreciably less costly than any open wire plan. 

 The elimination of any intermediate repeater station removed the 

 repeater station difficulties of the quadded cable plans. 



As no installation of such a length of this type of cable had been 

 made, some confirmation of the estimated values for the transmission 

 study, and particularly of the return loss and crosstalk, was considered 

 necessary. An 8-mile H-44 loaded 19-gauge exchange cable, which 

 had just been erected near Toronto, was chosen for study. Near-end 

 crosstalk measured on 286 combinations of pairs indicated 99 per cent 

 better than 81 db with an average of 91.7 db which, when modified 

 for impedance and length differences, indicated 99 per cent better than 

 72.5 db, and an average of 83.2 db for the proposed cable. While these 

 values were somewhat poorer than required, the size of the sample and 

 one or two other factors indicated that the proposed cable could be 

 erected to meet the crosstalk requirements. However, to obtain as 

 much crosstalk margin as possible, arrangements were made for the 

 manufacturer to use 6 lengths of twist, alternating 3 in each layer, 

 rather than the 4 twists which had previously been used for this type 

 of cable. It is felt that the excellent crosstalk results obtained as 

 outlined in more detail later are in large part due to this feature. 



Singing measurements on 10 pairs averaged 19.6 db. It was evident 

 from impedance frequency measurements that these singing points 

 could be raised to the desired value of 25 db by some modification in 

 the networks. Accordingly an adjustable precision type network 

 was developed. 



Also, four 1500-foot lengths of the proposed type of cable were 

 obtained from the manufacturer and tested for mutual capacitance of 

 pairs and capacitance unbalance between pairs. On statistical analy- 

 sis, these tests indicated a probable average near-end crosstalk of 79 db 



