Several recent technical reviews have been published on audio amplifiers which claim to use zero negative feedback, the objective of this article is to provide a limited insight into the application of negative feedback and demonstrate that it is most unlikely that any transistor amplifier is a truly zero feedback device.
The principles behind negative feedback are these:
1) Negative feedback reduces distortion. By taking a small portion of the output and subtracting it from the input any non linearity in the amplifier can be reduced, this is so because the feedback signal will produce an anti phase distortion signal at the amplifier input which will cancel( or reduce) the distortion caused in the amplifier. In the same way an amplifier with a poor frequency response will be improved by the application of negative feedback, because when the output is high more feedback is applied and the overall gain is reduced making a flatter or more even response.
2) Negative feedback is also used to change impedance at both the application and derivation points.
Negative feedback was first applied to amplifiers in the early days of valves, its main purpose being to reduce distortion. Valve amplifiers suffered from relatively low levels of distortion and had a frequency response dictated mainly by the characteristic of the output transformer. Hence small amounts of feedback were used; somewhere between 6 to 12dBs would be typical. As a rule of thumb the distortion would be reduced by the amount of feedback applied. This was of great advantage to the designer as the overall performance could be improved at the cost of providing a small amount of additional voltage gain.
The introduction of transistors to audio amplifiers caused a major change in design. Whereas in the valve era it was common practice to use only two stages(three if a phase splitter was used) of amplification, many more stages were used in semiconductor amplifiers. The open loop gain of a typical modern amplifier could be several thousand with the feedback being used to set the overall gain. Now this is the problem: if the open loop gain is so high and the semiconductors introduce a small time delay, then as the signal goes through the amplifier and transient distortion occurs, this will be amplified to overload levels before the feedback can act to correct the error. This causes is small but perceptible distortion and occurs in most semiconductor amplifiers. In this case, the introduction of single loop negative feedback to an amplifier will aggravate an initially small problem. This is why low level high order harmonics which are derived from crossover distortion in most types of transistor amps.are not improved by large amounts of single loop voltage negative feedback.
The important thing to remember is that negative feedback can only reduce an error once it has occurred.
In practice all the problems mentioned above can be eliminated by the application of current derived negative feedback at each individual stage in the amplifier. This removes the overall feedback loop and its disadvantages but allows the full advantages of negative feedback to be realized. When this is done the designer/manufacturer will claim a zero feedback design. This is untrue its just that the application of feedback has been changed.
The good news is that transistor ampflifiers do sound much more natural when distributed current feedback is applied. Conversely it is these amplifiers which will probably produce the poorest measured performance.
Dave McGhee
Ganymede Test & Measurement
email:
web: http://www.wavecor.co.uk
1st September 2001