switch ip3 power level

Ip3 is supposed to be a weak function versus power level as long as we have enough backup from the compression. But I see in both simulation and test that actually the IP3 gets better with higher two tone power.

I've tried simulation on SiGe and GaAs process, and I've test some RF switches and DSAs in GaAs process. Measurement always has dynamic range limitation and switches normally have high linearity, so it's hard to get enough data point to prove that. But from simulation, I see this all the time.

BTW, I use ADS/ dynamic link. The two tone power is from -30 to 10dBm in 1MHz space. I've change the number of harmonic from 5 to 15, no changes.

Thanks.

Normally switches has fixed IP3. so better check with Amplifiers.