THEORETICAL FUNDAMENTALS OF PULSE TRANSMLSSION 779 



obtained from (9.08) would then l)e about U = /0//1 and thus in the 

 order of 10 per cent of the peak pulse amplitude for/o//i = 0.10. 



The bandwidth penalty ineurred in dipulse transmission can be 

 avoided by transmitting two identical pulse trains, one of which is de- 

 layed by one pulse inter^'al and re\'ersed in polarity with respect to the 

 other.* The combined pulse train will then be as indicated in Fig. 34, 

 and one or the other of the two original component pulse trains can be 

 restored at the receiving end by suitable conversion ecjuipment. In the 

 combined pulse train, a pulse of one polarity is always followed by a 

 pulse of opposite polarity, but not necessai'ily in the next pulse position. 

 For this reason the low-frequency cut-off compensation with the above 

 method of "dicode" transmission is not quite as effective as w^ith dipulse 

 transmission. Furthermore, since it is necessary to distinguish between 

 three pulse amplitudes (1, 0, —1), in the received pulse train, the maxi- 

 mum tolerable pulse distortion in relation to the peak pulse amplitude 

 is only half as great as with two pulse amplitudes (1, — 1) in an ordinary 

 code. 



10. TRANSMISSION DISTORTION FROM BAND-EDGE PHASE DEVIATIONS 



In pulse transmission systems where phase equalization is employed, 

 it may be impracticable or unnecessary to equalize over the entire trans- 

 mission band. There will then be residual phase distortion near the band- 

 edges, as indicated in Fig. 36. This type of phase deviation will give rise 

 to pulse distortion extending over appreciable time intervals if the band- 

 edge phase deviations are large, as indicated in the above figure, for the 

 reason that the frequency components outside the linear phase range 

 will be received with increased transmission delay. E\-aluation of the 

 pulse shape is in this case a rather elaborate procedure, but rms pulse 

 distortion or intersymbol interference resulting from such phase dis- 

 tortion can readily be determined as outlined below. In certain pulse 

 modulation systems, such as PAi\I time division systems, rms inter- 

 symbol interference is of principal interest. In other systems where peak 

 intersymbol interference is controlling, it may usuall}'' be estimated with 

 engineering accuracy by applying a peak factor. 



When the pulse shape is known, peak intersymbol interference may 

 be determined by methods outlined in Section 13. Comparison of peak 

 intersymbol interference evaluated in this manner with rms pulse dis- 

 tortion, for some cases in which the pulse shap(\s in the presence of phase 



* L. A. Mcacluiin oriiiiiiallx' |)r()|)()S('(l tliis iiictliod is an luipuljlishcd memoran- 

 dum. 



