10w rf amplifier

Dears,
I am working on RF power amplifier. DC bias, small-signal frequency response and stability are just fine.
The problem is that the output power above certain limit it reaches the expected value then immediately drops down by several dBs, however cooling is OK (heat sink and air flow). Any suggestions?
Thanks,
WM3

i just care about the stability now.
suppose that p1dB is at -15dBm , it will be instability if the input power is -20dBm. i am working on it for weeks.
any one can help me?
fuyunjifan@126.com

I guess that the operating point of the power amplifer is chnaged by "self biasing effect". It apeears whan the input power level sufficiently high to drive and change the bias point of the amplifier.This will result in changing of operating class of the amplifer.

Second thing is that might be when high level is injected, a feedback can result a oscillation at higher frequencies. Thus, the amplifier should be checked up at least 100GHz in terms of stability..

But, BigBoss it's already class-AB and yes the input drive level influence the Q-point, this is expected. Regarding stability it's stable and no oscillation observed at any frequency during measurement.
If I turned the input signal on and observed the maximum output it shows the expected value. But if I left it running continuously, this output level gets lowered in front of my eyes.

Does this power amplifier have very high level of RF power such 10W or more ?
In this case load may change by temperature or optimum power matching might be shifting by temperature if it's critically matched..
Another case, bias circuit may not sufficiently provide the current neccesary..
Take care of decoupling capacitors and supply inductors..

Try to measure the DC drop exactly on the collector/drain of the final stage at maximum output power. Also check if there is no DC voltage drop at full power through the ground near the final stage.

You didn’t mention the output power, but some low-Q capacitors starts to behave poorly at high RF power, and this could be other cause of power dropping.

@ BigBoss
Yes this PA outputs nearly 10W, I measured 38dBm. When I supply the input, I see the 38dBm then gets lowered to about 33dBm then If resupplied the input again the same scenario repeats.
And if the problem is that the temperature changes the load, so how can I account something like that in my design and simulations?

@ vfone
The weird thing is that there is no drop on the voltages at gate and drain nor drop in the current drawn from the supply when the output power gets lowered.

Thanks a lot.
WM3