754 



THE BELL SYSTEM TECHNICAL JOURNAL, MAY 1954 



into a resultant pulse or signal as in Fig. 20((') the equivalent and less 

 laborious method shown in (d) can be employed. With the latter method 

 the pulse Po is plotted with reversed time scale and its peak made to 

 coincide with the point for which the amplitude of the resultant pulse P 

 is to be determined. The amplitude of P is determined as indicated in 

 the figure. In the particular case where the original phase characteristic 

 ^0 is linear, the pulses Po(t) will l)e symmetrical with respect to their 

 peak amplitude, and this assumption will be made in the following 

 applications. 



For amplitudes 6 « 1, the Bessel Functions become negligible except 

 for Jo and Ji , which are given by JqQ)) ^ 1 and /i(6) ^ b/2, so that 

 (6.05) becomes 



Pit) = Poit) + I Poit + r) 



I Poit - r). 



(6.06) 



p (oj) = coTd - h SIN cor 



FREQUENCY, U) —»' 

 (a) LOW-PASS CHARACTERISTIC 



S'Cul^urd-bsiN ur 



(b) BANDPASS CHARACTERISTIC 



Fig. 19 — Low-pass and band-pass characteristics with sinusoidal phase dis- 

 tortion. 



