Wien Bridge Oscillator 2

This kind of "imperfect" amplifier (opamp) does not need an extra nonlinear gain control. Its transfer characteristic is already nonlinear enough. Its like a tangents hyperbolic transfer function which is doing a soft clipping.

Yes the clipping is soft indeed but it's better with no clipping ;)

You probably misunderstand clipping. Clipping in this circuit starts long before peak voltage. Thats why its called soft clipping. A tangents hyperbolic nonlinearity start at center position with distortion/clipping. It is there very little but it already starts. As more you increase the amplitude its getting stronger. Its only a matter if its stronger then the gain then the amplitude stops increasing far before ugly hard clipping by saturation. This should happens before its for you visible as hard clipping and can only result in low THD if gain is a little above 3.

I don't get it. Clipping is as nonlinear as it gets. No other amplitude stabilisation introduces so much distortion as clipping does.

You still think clipping is something which starts at around peak of amplitude. But it starts already much earlier, actually it starts as nonlinearity at center position / at operating point, and can be called "soft-clipping".

It doesn't matter where it starts. Clipping describes a bad form of nonlinear distortion. Harmonic distortion is one thing. In given boundaries it's tolerable, but nonlinear distortion such as clipping is very harsh and most unpleasant. Other types of nonlinear distotion are tolerable. For instance i don't know about you but crossover distortion i don't mind so much. Harmonic distortion is also tolerable for me at a given level, but boy clipping sounds horrible! One more distortion i personally hate very much is the intermodulation distortion.

It does matter

Clipping is always a nonlinear effect. There is no linear clipping

You talking about intermodulation and you have no idea what it is, right? Take any nonlinear black box and see how it inter modulates.

Your statement - "It doesn't matter where it starts" Tell us at which voltages you think this circuit starts clipping. Values - no more statements! http://everycircuit.com/circuit/4754740247265280

I don't know where it starts clipping and it doesn't matter. It's still clipping, which is one of the worst forms of distortion. And yes i know what intermodulation distortion is and it is quite harsh. For instance mixing 2 channels without an appropriate mixer circuit will cause the two frequencies to sum and substract and produce harmonics of the fundamental and harmonics of the sum and harmonics of the substract. It's a very ugly picture.

Then remove the diodes and see its still working

Yes it is. But it goes into clipping. And whether the clipping begins at the beginning or at the end, it still chops off part of the wave. I guess if the oscillator is with a fixed frequency, you could filter out the output, but with a variable oscillator you should take measures at the source ;)

You still dont get it. Remove the diodes and reduce the gain and there will be NO hard clipping. Because the soft clipping, the nonlinear behaviour of this amp is much enough to stabilize this oscillator. Boy you make me crazy...

Look i get it alright. This is not an op amp, so it won't have such a high gain, nor such a linear characteristic to smash it into hard clipping. I also know the clipping won't start at 0V and 20V. Due to the Re the clipping should be around 3V. BUT i also know that no matter the type of clipping, no other amplitude stabilization technique will introduce such distortion like a clipping will. A bulb stabilization may have a THD of less than 0.01% if properly constructed while even the slightest clipping will push the THD above 1%. If you won't take my word for it do the THD calculations yourself. I know you can. Compare even a diode stabilised circuit with a clipped one and see the results.

phew.. that was intense! lol