I recommend to change all the LDMOS bias circuits to manual setting.
Also you get rid of this "doherty" feature that improves your eff. but reduces your pwr @2,4GHz noticeably. (At least on my amp)
I used 5,1v z-diodes in series with a 10k resistor against the 28V supply and a 10k Potentiometer across the Diode.
The set voltage (middle pin on pot) again with 10k to the gate of the LDMOS.
I have set the driver to 400ma and the finals to 500ma @ 28V.
Works great and linear to ca 90w for many months now. Got many responses on QO100 about my clean signal & also measured it a few weeks ago myself.
Of course you can implement a temperature control to the bias circuit but the IDq rises not that much if the LDMOSes are properly cooled.
The cooling fans on my amp starting at 37° and i never encountered problems with the IDq or signal quality.
I'm happy without all that "extra" stuff..
73s