I should stay away, since this stuff tends to become mysticized anyway, for reasons which I can't fathom. But I can't help myself.
For those unaware, class AB amps are just a pair of class B amps run in opposing compliment to each other.
I feel there is some descriptive sense to this, but it isn't literally correct for most audio amps, and it doesn't at all address the fundamental difference between B and AB, which is that the former has devices in conduction for 180 degrees of phase, and the latter has devices in conduction for more than 180 degrees (but less than 360). (Where "phase" is used in the context of a sine wave, even though music isn't sine waves.) The output devices of most solid state push-pull amps are in conduction for slightly more than 180 degrees, because they run a small amount of idle current. (There are some audio amps where the
output devices are class B - perhaps even better described as class C - because they run no idle current. But in these amps the drivers will be running class AB and there will be a relatively low-impedance path between the drivers and the amp's output point.)
The benefit to class A is the nonlinearity of each half of the amp cancel each other out.
I'm afraid I don't see any way to interpret this as correct. The benefit to class A is that the devices operate in a
more linear region because they are always in conduction.
The cancellation mentioned works for complementary push-pull stages of either class A or class AB. (Edit
: I now believe I'm wrong about the italicized part. Still think that cancellation is not the most important benefit of push-pull class A, though.)
As has been commented above, there seems to be conflation between class and topology. They are really pretty independent concepts.
Sorry to get all hard-core but, well, this is pretty hard-core stuff. There is no mystery to it at all, and damn little "wiggle room".
Descending from high horse, and yet still riding into the sunset,
chazix