42 



THE BELL SYSTEM TECHNICAL JOURNAL, JANUARY 1957 



A study was made of the expected intermodulation noise. This noise 

 is affected by repeater level differences, by talker volumes, and by circuit 

 activity. It has a time distribution, the most attention being given to 

 the root-mean-square modulation noise in the busy hour. 



Modulation noise was computed in two ways: by the Bennett method,^ 

 and by the Brockbank-Wass method.^ Results by the two methods are 

 in fairly good agreement. The values computed by the Bennett method 

 are shown in Table III. 



Because noise powers, not dba, are additive, these values of modula- 

 tion noise would contribute only a very small amount to the total noise 

 in the deep-sea cable system, as previously stated.* 



Other possible contributory sources of noise in the deep-sea cable sys- 

 tem are noise in the terminals, noise picked up in the shore lead-in cables, 

 and corona noise in the repeaters. The system has been designed so that 

 the expected total of these is small. Hence the estimate of total deep- 

 sea cable system noise, made before cable-laying, gave the values shown 

 in Table IV, to the nearest db. 



Misalignment Control 



Objective 



Since the noise objective for the deep sea cable link was set as 36 dba 

 in the joint meetings of the British Post Office and Bell System Repre- 

 sentatives^, the above estimate of system noise led in turn to the objec- 

 tive that the system be manufactured and laid in such a way that the 

 misalignment, including effects of seasonal temperature change, should 

 be no greater in the top channel than the value assumed in the estimate, 

 which was 12 db. Half of this was alloted to initial misalignment and 

 half to effects of temperature change. Greater misalignments were per- 

 missible at lower frequencies. 



* The following table for adding two noises expressed in dba shows the magni- 

 tudes of the increases: 



db difference lietween larger and smaller 2 4 6 8 10 15 20 

 noises 



Resulting db to be added to larger noise to 2.1 1.4 1.0 0.6 

 get sum of the two 



0.4 0.1 0+ 



