PHASE DISTORTION IN TELEPHONE APPARATUS 



499 



distortion would be the same even though the delays might be different 

 due to different slopes of the subtracted portion.^" 



In the case of the band pass filter of Fig. 4 a distortionless portion 

 may be subtracted having the slope of the original at the mid-frequency 

 but only for special cases will it be tangent to the curve. Fig. 4 shows 

 curve A broken into two parts a distortionless part Ai giving delay 

 and a part Ai responsible for the distortion, i.e., A = Ai -\- Ao. It 

 is doubtful if effects of ^2 on speech and music would be noticeably 

 different than any other curve obtained by adding a constant angle 

 to it at all frequencies. This type of variation occurs to the same ex- 



360 



£ 320 

 y 280 



UJ 



'='240 



Z 



,_ 200 

 I 160 



in 



^ 120 

 < 



I 80 



40 

 



0.5 



1.0 



1.5 



3.0 



2.0 2.5 



FREQUENCY {^^ 



3.5 



4.0 



4.5 4.5 



Fig. 6 — Image transfer phase characteristics of typical all-pass lattice type network 



sections. 



tent in apparatus observed to produce negligible phase distortion as 

 where distortion is apparent. It is for this reason that the slope of the 

 phase curve rather than the curve itself is generally taken as the 

 criterion for determining the amount 0/ phase or delay distortion in a 

 network. In other words if two phase curves give the same dBjdoi or 

 delay characteristics (sometimes called "envelope" delay characteris- 

 tics) they are regarded as introducing the same phase distortion par- 

 ticularly for speech and music though such an assumption is not 

 rigorously true. 



In the case of the high pass filters and all pass networks of Figs. 5 and 



^'^ This procedure of breaking the phase characteristic into two parts one part 

 introducing no distortion but a delay To given by the minimum slope in the transmit- 

 ting range and another part responsible for the distortion is perfectly rigorous, but it 

 must not be interpreted to mean that for a signal starting at < = absolutely nothing 

 will be received before t = To. In the first place there is generally a small amount 

 of energy outside the transmitting range that will come through earlier and in the 

 second place, unless the residual phase characteristic is of a kind that can be produced 

 by physical apparatus the distortion characteristic alone will cause the received wave 

 to differ from zero prior to / = T^. 



