Current feedback single-ended input stage + push-pull VAS?

[antonis]

one derivative is linear

More close to quadratic, I'd say..

[fryingpan]

I said derivative.

[antonis]

I said derivative.

Didn't know (till now) that Collector/Drain currents give a toss for differentiation..

[Rob Strand]

Didn't know (till now) that Collector/Drain currents give a toss for differentiation..

The derivative gives the small signal behavior.  gm = Ic/VT for a transistor comes from the diode equation.  The small signal idea is that with small enough signal swings the device is linear around some bias point.  So you can have a gm for a JFET or a BJT.

For large signal behaviour you need to consider the non-linearity.  You can think in rough terms that there's a Vin and a Iout for the device.  However the large signal ratio Iout/Vin doesn't give you gm unless the device is perfectly linear.   If you compare a small MOSFET to a BJT  the BJT doesn't do much until Vbe is 0.5V to 0.6V whereas a small MOSFET needs at least 2V.   Even if the small gm was the same at some bias point the large signal Vin is going to be larger for the MOSFET because of that offset.

[fryingpan]

It's more that with a linear derivative you can expect a more gradual transition from the ohmic region to the saturation region of a FET, compared to the more abrupt transition from the active region to cutoff for a BJT. I expect that the ideal double cascode configuration should be a combination of a BJT for the amplifier itself (more gain) and a FET for the common base/gate device, so that there is a slower and more gradual transition from an active load (FET on) to a passive load (FET in the ohmic region).

There will probably be more of a soft clipping action with less global feedback. If I weren't busy with stuff I'd model it myself.

[antonis]

It's more that with a linear derivative you can expect a more gradual transition from the ohmic region to the saturation region of a FET, compared to the more abrupt transition from the active region to cutoff for a BJT.

True & Correct but what is called FET's saturation region corresponds to BJT's active one..
Unless I got you wrongly, you compare reverse procedures..

[fryingpan]

It's more that with a linear derivative you can expect a more gradual transition from the ohmic region to the saturation region of a FET, compared to the more abrupt transition from the active region to cutoff for a BJT.

True & Correct but what is called FET's saturation region corresponds to BJT's active one..
Unless I got you wrongly, you compare reverse procedures..

The order was not important (if it's gradual one way it's gradual the other way  ), but of course saturation -> ohmic = active -> cutoff.