Once upon a time, one could manually tune up the Pi-network of a class- C tube amp by minimizing plate current. I never really understood the math of this behavior (in fact, I never SAW it explained anywhere) and did not make intuitive sense to me either.
Why should plate current rise if e.g. the load Z was higher than the device output Z?
OTOH I can imagine that moving away from a parallel LC-resonance in the load might cause the load to absorb exponentialy more current on either the capacitive or the inductive node, reflected by increased plate current. This may mean that a well-matched Pi-network indeed looked like a parallel-resonant resistive load.
I now wonder if switched-mode RF power amplifiers behave the same way.
CON: SMPAs are at heart near-zero Z sources. Within I and V limits of switching devices, their output is determined by how much current a given-Z load will take at a given power supply. The lower the transformed Z seen by the amplifier, the higher the current. Changing the output transformer winding ratio changes the power. On the other hand a C-class amp did have an impedance defined by the amplifying device, which never reached near-zero resistance.
PRO: At a given output transformer winding ratio (or 1:1 in the case of E-class), a matching LC network may still somehow look like a parallel resonant load. The only time I played with a SMPA (push-pull into a balun uptransformer, followed by a low-pass T-network) I noticed a sharp and marked power maximum into a dummy load at a given capacity setting in the T - not just a rise as capacity decreased. I didn't check the input current, alas...
Enlightenment available, anyone? Pointers to the maths?