[Don Pearce]

Eeyore wrote:

Vladimir Vassilevsky wrote:

4) FETs do not require high base currents of BJTs; that simplifies the
driver stage.

So what charges and discharges the damn gate capacitance you dozy bugger ?
The classic Class A voltage gain driver stage probably needs as much current
as if you were driving darlington output devices with a simple design..

In my ultra-low THD design I had a bipolar complementary Class A emitter
follower driving the mosfet gates ! It also removes the capacitive loading
from the voltage gain stage, increasing phase margin and loop HF response.

Graham


Where did you put the dominant pole cap? - still around the voltage
amplifier I'm guessing. It is vital that the dominant pole cap is the
only one that shows up in the phase response up to unity O/L gain, so
buffering of the fet gate caps is pretty much a given.

The big problem you need to overcome with mosfets is that they have an
essentially log response, resulting in (or from, depending how you look
at it) huge - maybe 10/1 Gm changes over the operating range of
currents. This is all easily taken care of by global nfb, of course, but
it does make the sewing together of the crossover point a bit trickier
than a bipolar design. The Gms are dropping at a much sharper angle and
the doubling spike consequently more obvious.

All unimportant, of course, given enough nfb.

d