Interesting article on low-level distortion. JFET/BJT/valve

[brett]

Hi.
I just found this article concerning low-level distortion and using feedback to reduce distortion: http://www.its.caltech.edu/~musiclab/feedback-paper-acrobat.pdf

The feedback they are referring to involves adding a source/emitter/cathode resistor (to FET/BJT/valve circuits, respectively).

They included this summary of their conclusions:

For single-ended amplifiers
For all cases, input signals were at frequencies 3 and 5 (unscaled). Input
voltage levels were always equal for the two components and were as given
below. Unless otherwise stated, (a) Higher input level raises the relative
distortion and (b) emphasizes higher-order distortion products. (c) Adding
more feedback lowers all distortion products.
1. FET: 0.05 peak volts in each input component
Without feedback, distortion consists of only 2nd harmonics and first-
order sums and differences of the two input components; that is, only
the four frequencies 2, 6, 8, 10. Feedback creates complex new distor-
tion products extending over the full bandwidth of the amplifier and
thus constituting a kind of "noise floor."
2. BJT: 0.0004 peak volts in each input component
Distortion without feedback is complex, resembling that of the FET
amplifier with feedback.
3. Triode: 0.05 peak volts in each input components
Distortion without feedback is complex, resembling that of the FET
amplifier with feedback, but distortion products are distinctly lower in
level that in the no-feedback FET amplifier at the same input level, or
even the no-feedback BJT amp at its much lower level.

Should we be using less feedback in our JFET boosters.  Would small drain resistors (1k or less ?) and very small or no source resistors give us only 2nd harmonic and intermodulation distortion?

cheers

[davebungo]

There are a number of points to make - this is by no means a complete list:
Negative Feedback in an amplifier generally stabilises bias, gain, widens bandwidth, lowers output impedance, raises input impedance etc etc but as with most things there is always a price to be paid.  The question is what is the price and does it really matter?

The price generally is the addition of complex distortion products albeit at much lower levels than the primary distortion provided by the non-linear amplifier.  There are also stability issues to be considered but these can be dealt with.

Does it matter? For some hi-fi nuts, an op-amp would not be acceptable in a signal path due to the use of large amounts of negative feedback (even though the original recording would probably have been made on an expensive mixing console using large numbers of op-amps in series!)

For the class A FET case, the source resistor is the primary negative feedback mechanism, and you may have noticed that many circuits have this bypassed for frequencies of interest i.e. it just provides a DC bias setting/stabilisation function, so in these there isn't much negative feedback at all in the audio range in these cases.  The gain of such circuits is then a little unpredictable as the spread of gm (gain) of FETs is quite wide.  People seem to like the sound though and whether this has more to do with the 2nd harmonic distortion or the way the signal overloads the amplifier at high gain/high signal level is pretty much a matter of opinion although generally, even harmonic is considered good and odd = bad.

I don't know if this was any use but there you go.