THE COMPANDOR 327 



this limit, somewhat greater improvements are obtained from the 

 compandor, ranging up to at least 10 db. 



The particular values of compression and expansion ratio were 

 chosen initially for the relative ease in the design of the system with 

 commercially available vacuum tubes whose characteristics closely 

 approximated a parabola. Tests of the equipment have shown that 

 this degree is sufficient for present telephone circuit intensity range 

 requirements. Increasing the amount of compression is limited by 

 increase in quality distortion and by increased variation in the in- 

 tensity of radio noise as heard by the listener. A noise which is con- 

 stant at the input to the expandor varies on the output as the speech 

 intensity changes. Also variations in attenuation equivalent between 

 the compandor terminals are multiplied by the expandor. Herein 

 lies a reason for having a constant compression and expansion ratio 

 over the working range. If it were different at different intensities, 

 attenuation changes would distort the reproduced speech as well as 

 appearing as a somewhat increased change in intensity. This change 

 in intensity is n times the attenuation change in front of the expandor 

 in db. 



The degree of compression may obviously be controlled in a variety 

 of ways: such as, using different values for the indices 5 and /, applying 

 control voltages upon more than one variable stage in tandem, the use 

 of variable jj. vacuum tubes, etc. The circuits as shown use variable 

 shunt control for the compressor and variable series control for the 

 expandor. Either or both may be changed to the other by inverting 

 the polarity of the control potential and properly designing the rectifier 

 characteristics of the control circuits. 



There are two major sources of possible speech distortion which 

 must be considered in the design and use of these devices in addition 

 to those ordinarily present. The first is due to the non-linear char- 

 acteristics of the vacuum tubes used for controlling. The even order 

 distortion terms are largely balanced out by using two tubes in a 

 push-pull arrangement. The remaining distortion is minimized 

 by having speech pass through the control tubes at a sufficiently low 

 level. In the operating ranges for the device shown on Fig. 6, the 

 harmonics of a single-frequency tone are 30 db or more below the 

 fundamental. 



The second major source of distortion is the time lag in the control 

 circuits due to the presence of the filters after the linear rectifier. 

 However, with a complete compandor circuit using the compressor 

 circuit No. 1, it was found on careful laboratory tests with expert 

 listeners that it was almost impossible to distinguish whether the device 



