Bootstrapping

[seanvn]

Bootstrapping is a well know technique to increase the gain of CE amplifier. With a near perfect voltage follower providing the bootstrapping the gain should be close to infinity. In actuality the gain is only about 2000. It is limited by the hoe (collector emitter resistance) of the transistor (about 40k). One of the few times hoe has a measurable effect. This is kind of fortunate because bootstrapping is a form of positive feedback. Positive feedback really exaggerates the frequency response curve of an amplifier. If you reduce the hoe effect by bootstrapping a cascode amplifier you get huge gain but the frequency response curve gets exaggerated to the point of being a narrow band filter.
I am interested in using controlled positive feedback, it is an under-explored topic. Using positive feedback locally and negative feedback globally can boost amplifier performance.

[jcx]

you still have to meet global stability requirements so internal positive feeedback isn't a "magic bullet" - you always have to roll off any loop gain by the intercept frequency

I think a useful way to look at positive feedback, particularly as seen with bootstrapping or Bob Cordell's implementation of Error Correction is as "unmasking" the negative feedback of the unity gain stage providing the bootstrapping

when this is the output stage there may be a helpful nonlinear stability consequence - when the output clips the positive feedback gain boost also goes away - protecting from nonlinear oscillation, bad clipping recovery dynamics that you may see ohterwise when using higher order loop gain slope needed to take advantage of higher loop gain

[Elvee]

Basically, when you use bootstrapping, you use the two stages to the maximum.
But it is only one way of doing that: you can use any other two-transistor configuration to the same effect: a mu amplifier for example.
You could object that the output impedance is very high, 1/2*h22, but if you want to use the full gain of the bootstrapped stage, you are not allowed to load it either.

Common positive feedback will not be usable practically, because of the exacerbation of the defects of the active elements.

Some forms of very limited and well controlled positive feedback can improve both gain and linearity, but you will generally need more than two devices.
The crossquad is an example of such a topology: it is possible to achieve insane gains and yet still improve the linearity.
But this kind of circuit is not for the faint-hearted, they are extremely temperamental.
They are usable though.
You can find alternative topologies using similar mechanisms in the Tringlinator/Tringlotron and Unigabuf threads. The Eva-quad is a more civilized version of the Xquad.

[Elvee]

You will probably find this thread interesting:
Needed: single transistor audio amplifier with positive feedback