458 



BELL SYSTEM TECHNICAL JOURNAL 



horizontal and vertical scales. The measurements were made on a quan- 

 tizer with 32, 64, and 128 steps, and a sampHng rate of 8,000 cycles per sec- 



INPUT- SIGNAL POWER IN DECIBELS DOWN ON FULL LOAD 

 40 36 32 28 24 20 16 12 8 4 



Fig. 6 — Third harmonic in 64-step quantized output with bias at mid tread, 

 smooth curves represent computed values. 



The 



INPUT-SIGNAL POWER IN DECIBELS DOWN ON FULL LOAD 

 40 36 32 28 24 20 16 12 8 4 



^ z 

 o 



Q 

 10 to 



20 Q? 



30 



a. z 



OJ UJ 



O < 



a. Q 



z 



U =5 



40 Z 



50 



Fig. 7 — Third harmonic in 64-step quantized output with bias at mid-riser, 

 smooth curves represent computed values. 



o z 

 2 O 

 q: 

 < 



I 



The 



ond. The applied signal was confined to a range below 4,000 cycles per 

 second. With equal steps the distortion power is practically independent 

 of load as shown by the db-for-db straight lines. With tapered steps, the 

 distortion is less for weak signals, and only slightly greater for large signals. 



