The Wien Bridge Oscillator problem (your Sunday ruined part1)

Gentlemen,

How critical is the choice of thermistor for a WB oscillator? Stability is
desired; we know that much (along with minimal distortion of course). The
usual problem with WB oscillators is they are not amplitude stable and
either die out or saturate. However, by means of a non-linear negative f/b
arrangement, they *can* be made stable. One common solution is to use a
thermistor, usually one with a positive temperature coefficient. The issue
is, that it must be 'nimble' enough to regulate the output level without
being *so* 'nimble' as to respond to peaks and troughs of the output
signal's cycles. I'm going to call this 'nimbleness' "tau" for the time
being. I know it's not the right Greek letter, but no doubt some helpful
soul will point out the correct one.
So the question is, how the hell do you select a thermistor with the
optimum 'tau' for any given wideband WB oscillator?

Your pal,

CD.