This is the famous analysis of class B amplifier cross-over distortion by the then head of HP Reasearch Labs, Dr. Barney Oliver, published in the February 1971 edition of the HP Journal. The bias current Iq for a class B emitter follower amplifier is shown to be approximately Iq = .026/(Re+re+(rb/hFE)). In practice, – the ideal value ending up somewhere between a 13 to 26mV drop across the output transistor emitter degeneration resistors. This paper provides the theoretical underpinings for that relationship. So, when you hear about the ‘Oliver’ voltage, you know where it came from. You can read more about Barney Oliver here and here
Download the PDF here: Cross Over Distortion in Class B Amplifiers
See the comments below. My practical experience over many amplifier builds is that indeed somewhere 13 and 26 mV is ideal.
This belief seems to be widespread, but the paper does not contain that formula, and does not make any such claim.
Esmond, see equations 14-23 and the accompanying text which point to an optimum bias voltage of .026/Retot. He then goes on to discuss the effects fo feedback on distortion. The ‘Oliver voltage’ was a term coined later by people writing in the field. See Bob Cordells book where he also discusses the ‘Oliver voltage’.
Yes what Bernard Oliver actually says is “somewhere between 13 and 26 millivolts” , see the last two paragraphs of the 3rd page of the article. Then he says “the most practical solution is to make the voltage dropped across the emitter resistor to be MANY times larger than (13 to 26 mV) and then rely on negative feedback to reduce the resulting distortion”. And after making the emitter resistor voltage drops big, he adds high current diodes in parallel with the emitter resistors (a la James Bongiorno / Marantz) and observes they do not increase distortion, they decrease it.
Like many sacred texts, people often erroneously misquote Barney Oliver, and then pretend their errors are heaven-sent, axiomatic, truths.
Thanks for your comments Mark. I am finding that the lowest distortion (measured using a QuantAssylum DAC/A-D set-up) is actually well below the Oliver voltage and typically at 0.5 to 0.66 of the calculated value. I use 0.33 Ohm resistors which would indicate 78mA bias current per pair. The distortion minima is actually occurring between 40 and 50mA. Even at 30mA, it is in the very low double digit ppm level (1kHz).