Output S22 is NOT Conjugately Matched to ZLoad in Class-C Broadband Amplifier

[Paul678]

It would appear most broadband amplifiers are conjugately matched on the input only, so that the S11's
inductive or capacitive reactances are cancelled out by the opposite reactances of the input matching network,
which is also called Z-source. This assured maximum power transfer into the transistor.

However, the output S22 is usually NOT conjugately matched to the Z-load, in a class C amplifier.
I have had at least one professional friend tell me that this means the S22 data from the S-parameter
files is pretty much worthless. I would therefore assume most of these output matching networks,
are created by the old "cut and try" empirical method, usually done on the actual, real bench.

So while we can usually simulate a good starting point for the input matching network,
in either ADS or Microwave Office ( or any other software package), it appears we cannot
do the same for the output matching network.

So how do you all find a good starting point for the output matching network, in a Class C, broadband
amplifier?

 

[BigBoss]

Your friends is right. In order to match a Power Amplifier 3 considerations should be taken into account.
-For Maximum Delivered Power OR
-For Maximum Efficiency OR
-For Maximum Gain.

Conjugate Matching is useless of the Output Circuit for PA Matching, instead Load Pull Technique that is generally obtained by measurements is most common method.
Load Pull can also be checked in Professional RF Simulators but the Models for hardly driven semiconductor devices are unfortunately not very accurate.
You have no option but Load Pull Measurement/Simulation because C-Class Power Amplifiers are driven by High Signal Levels ( in order to exceed threshold ) and s-parameters are already not meaningful under under-zero bias conditions.If your models is not sufficiently accurate, you have to match the PA manually for the targets which are mentioned above.But this is not easy..

 

[Paul678]

Good point. The Load Pull technique is really just a more sophisticated version
of the "cut and try", empirical method of finding an output impedance matching
network, that satisfies your needs.

Here is a video using load-pull in Microwave Office: