I have simulate the FET in ADS by using the nonlinear model provided by agilent for LNA (The Gaas FET is ATF54143). The s-parameter from the simulation does not agee well with the measurement s-parameters. I have some doubt about the nonlinear model. how it accurate? In the agilent application note, they use the measured s-parameter for design. Can I conclude that.
1. for small signal amp. LNA -> measured s-parameter is the best.
But how about P1, IP3? can i still use the nonlinear model to predict them?
2. for oscilator, VCO -> I think i still can use measured s-parameter for open loop analysis but I do need a nonlinear model for transient and HB to predict the harmornic and start up. ALso most nonlinear model does not provide the phase noise parameters so it renders the phase noise simulation useless.
3. PA .. I think only load pull work. I dont think any large signal model will work because the environment for parameter extraction is not same as the real environment (i.e. mechanical construction, heat sink, etc.)
I am very frustrated for the nonlinear model. even the V-I curve does not agree well with the data sheet. The model is Curtice 2. (I am frustrated becuase i am using Agilent simulator, Agilent FET and Agilent model, it still does not accurate how can I trust other vender model (such as CEL) )
Any experience which model will be used in what kind of design???