I think that the problem is not within the driver.
The new driver which you tested proves that
If the problem is linearity - which
seems to be based on the description you showed,
then the problem lies within the balanced amplifier.
The significant stage for cascaded IP3 calculation
is the last stage of the Tx chain.
as the first stage for cascaded NF calculation).
Try to focus on the balanced amplifier and
In many data sheets of power
transistors, there are measured results for different
--> Have you looked on what are
the expected performance of the transistor (within
the balanced amplifier) for the type of signal you
are using (i.e. modulation type, filter applied,
It might be the case that
you need to consider a final stage with more output