“An amplifier, with or without negative feedback, having the greatest fidelity in faithfully reproducing the input with the least distortion. It is however the least efficient, in as much the power delivered to the load is only a small percentage of the d.c. power used up in the amplification process”.
That was not the most elegant definition admittedly, yet it covers the basics:
(a) might use feedback.
(b) good fidelity.
(c) lousy efficiency.
Feedback – This is where part of the output signal is fedback to the input BUT 180o out-of-phase (i.e. partially cancels the input). If it were in-phase feedback then we would have an oscillator – which in this case we definitely do not want.
Fidelity – This means many things to many people but to us it means the output must be an exact replica of the input but only magnified or amplified.
Efficiency – The theo
retical limit to this amplifier’s efficiency is 50%, meaning for every watt of output we will use up at least 2 watts of d.c. power input in to the amplifier. Depending upon the application this may or may not be significant. If we have a large power transformer available to us and power, literally to burn, who cares?.
On the other hand if miniaturisation were the keyword and precious battery supply is the only power source available thennnnnn! that’s another matter.
Most experts will agree with me when I say you will only see this 50% efficiency in your dreams. So why use class A?. At low levels of signal and amplification, the losses through inefficiency are not significant and are far outweighed by the goal of fidelity or in the r.f. game, linearity, i.e. a linear amplifier.
negative feedback amplifier
– Negative Feedback –
In the circuit described 20db of negative feedback is taken from the output transformer secondary 8 ohm tap to the cathode of the input pentode. The circuit is set up by adjustment of the feedback resistor bypass capacitor to produce a critically damped 20kc square wave into an 8 ohm non-inductive load. The feedback network is a step circuit to prevent excessive high-frequency feedback in a range where internal amplifier and transformer phase shifts could produce oscillation, especially with capacitor (electrostatic loudspeaker) loading.
With all other low-frequency time constants as indicated a low-frequency 2 c.p.s. d.c. function generator is used for the adjustment of the screen bypass of the input pentode. Perfect reproduction of the low-frequency square wave is what is desired. In this circuit with 20db of overall negative feedback this capacitor should be adjusted to reproduce the 2 c.p.s. square wave at the output without slope or tilt. Too large a screen bypass capacitor will produce overshoot, indicating an undesirable peak in low-frequency response. These tests must be done at a low signal level into an 8 ohm load to avoid saturation of the core of the output transformer.
In some respects circuit one is derived from the earlier (1948) Williamson amplifier design. The Williamson amplifier could exhibit a constant low-frequency oscillation resulting in a “breathing” action of woofer cones moving in and out, triggered by wow or rumble in the reproduction of long-play vinyl discs.
The Williamson amplifier was stabilised by the use of “step” network coupling of the driver plates to the output tube grids. This modification corrected the low-frequency phase response and was originated by Norman Crowhurst.
Circuit one incorporates step circuit coupling to the output tube grids as well as the low-frequency phase and gain adjustment afforded by variation of the screen bypass capacitor of the input pentode.
In a vacuum tube amplifier there is essentially zero time delay between the signal grid drive and a current response at the plate. This not the case with transistors where the flow of charge carriers through solid semi-conductors is much slower than electrons moving through a vacuum.
The limitation in the use of negative feedback to reduce distortion occurs because of phase shifts in the circuit elements adding up to 180 degrees at subsonic and ultrasonic frequencies and thus turning negative feedback into positive feedback. This is especially true when an output transformer is one of the circuit elements. Experience has shown that 20db of feedback is optimum in a circuit where it takes 30 – 35db to make the circuit oscillate. Negative feedback is very useful in distortion reduction if not overdone.