Question about large signal modelling of HEMT in ADS

[a_ronagh]

Hello

I am trying to model a microwave transistor using Agilent ADS built-in models. The model name I have chosen is EE_HEMT model from the GaAs library. I studied the model details from the help files provided by the Agilent and after separation of the parameters into DC and ac ones, I tried first to fit the model I-V curves onto the measured I-V characteristics of the device by optimization. The result was excellent.
Then, taking the DC parameters constant and tuning the ac parameters, I tried to fit the S-parameter results from DC to 20GHz in a specific operating point. Again a reasonable match between model and measurement results could be achieved. BUT the problem is that the fitting is valid only for that operating point or in other words most of the ac parameters seem to be bias dependent. Due to the fact that I am trying to find a large signal model, this discrepancy is not acceptable.
Does anybody know where I am making a mistake? Is this not a way of finding the model parameters? What is wrong? Do you any paper or thesis about modeling?

Thank you all in advance

 

[kspalla]

I guess even large signal models depend on bias.
May I ask you to share the design.

BUT the problem is that the fitting is valid only for that operating point or in other words most of the ac parameters seem to be bias dependent.

The observation is correct only

 

[a_ronagh]

It seems that I have find a way improve my model and I am now working on it. I will inform you of any possible improvements as soon as possible. Thank you for your attention.