Okay folks, I accept the use of Litz wire, thanks for the explanation.
Peter, most RF/HF power amplifier is designed to drive a 50 Ohm load but by default, this load needs to be purely resistive and not capacitive or inductive. Of course exceptions do occur when a load to be matched is a specific one, in those cases the matching network should be designed accordingly.
In your particular power amplifier schematic you surely recognize the Pi filter at the output, consisting of the two preset 100150 pF capacitors and the 3.5 turn air core coil, the purpose of this filter is to match the drain impedance of TR1 to a pure 50 Ohm resistive load.
What could be the drain impedance you may wonder? It can be pretty well approached by this formula Z=50V*50V/16W = 156 Ohm and this should be transformered down to 50 Ohm by the Pi filter, such filter network can transform impedance and also serve as a harmonic filter. Impedance transformation depends on the ratio of capacitors (C1 and C2) of such Pi filter, it can step impedance down or up too.
(You may wonder how low efficiency this amplifier may have: 16W/25W=0.64 where the 25 W comes from 50V*0.5A input DC power, i.e. 64% efficiency which is typical for such, nearly linearly biased AB class amplifiers.)
Yes, the amplifier's last stage includes a capacitor what could seemingly be used to tune a 'load' inductor of say 160 nH for a 45 MHz resonance but this then would represent a 50 Ohm pure resistive resonant impedance by a big luck only.
This may answer verpies's suggestion to use a parallel LC tank: matching could be obtained via the Pi filter if and when the final capacitor setting would insure a near 50 Ohm resonant impedance for the tank and this impedance would then be correctly transformed back towards the drain impedance of approximately 156 Ohm.
I think a better approach would be to use an additional capacitor at the output in series with your 160 nH 'load' coil, as I drew it in the schematic attached. This way C2 and C3 would constitute a capacitive (voltage, hence impedance) divider, their series resultant equivalent C value would tune the 160 nH to the 45 MHz resonance. Under such condition the transformed (stepped down) impedance would be pretty close to a resistivelike 50 Ohm what the amplifier transistor would need to see via the Pi filter.
Alternatively, the Pi filter could be omitted and the 160 nH drive coil would be matched by resonating it via C2 and C3 directly, this what I indicated in the second schematic below.
Sorry for confusion I may have caused for you, it was not my intention.
Gyula
