Re: what's the difference between soft switching and hard switching?
When you are talking Class E PAs, you are talking an
output that is switching (nonlinear) and using an output
filter to regain purity and also do an impedance transform.
So topologically you are very like a buck DC-DC converter
(minus the gross shunt-C of the output filter). Concepts
from DC-DC world have bled over to the RF cousin.
Soft switching means you're using the inductor's
"momentum" to pull the output low, when you open the
high side switch. The low side switch is engaged after
a "break before make" period that allows the output to
swing to ground under its own power, just in time to
prevent the current from taking parasitic diodes
forward (D-B diode, ESD diode, either is dissipative).
This can be tuned / "ballistic" or it can be actively
managed by a special control loop to optimize the
turnoff pedestal voltage where a slug of inefficiency
awaits.
At RF you probably don't have a good opportunity
to layer on a bunch of falling edge endpoint feedback
and most of what you will see is hard switched (and
besides, RF likes controlled impedances and changing
from low to high to low is probably not so clean,
distortion-wise). If the output stage looks like a
big fat CMOS inverter, and no obvious designed skew
in the high side, low side drive taper chains (always
some, to minimize shoot-through, of course, but you
would be talking delays approaching quarter-period
for soft switched RF if RF to you means GHz) then
call it hard switched.
In simulation you can make the call by just looking
at output port current on the HL edge, and see if
it's coming from the matching network (soft) or
the low side switch (hard).
A Class A, Class AB, etc. PA are not switching styles,
hard nor soft.