difference between S-parameters of a nonlinear transistor model and its datasheet

Hello,
I must design an amplifier in ADS using two model of one transistor (NE321000) for my homework.
First I must use .S2P file of this transistor, and then I should use the nonlinear model of it.
But my problem is that the S-parameters of simulating the nonlinear model are very different with the datasheet S-parameters (or equivalently .s2p file).
I added bias circuit to my nonlinear model too.
Can anyone help me to correct my nonlinear model to work as its datasheet S-parameters?
Also I excuse for my week English.
Thanks in advance.

Hello, Are you sure you applied the same bias (voltage and current) for S-parameter extraction from non-linear model as used for the datasheet table (vds=2V at 10 mA)? Can there be influence from your bias circuit (or did you use ideal lumped sources)?

What do you mean with "very different" is it 10%, 50%, etc?

You may check your model with the one at http://www.cel.com/parts.do?command=...dRootPart=441#

Nonlinear model does probably not take the package parasitics into account, therefore there might be some differences.To converge these small signal s-parameter model and nonlinear model, you should consider package parasitics.

As well as making sure that the bias is set correctly and that any parasitic effects are modelled consistently the input power level is important too. S-Parameters are a measure of linear performance and that assumes that input power level does not alter the circuit's performance, i.e. small signal. To have any chance of replicating the same small signal performance with a non-linear model then that should be driven at small signal levels too. If the power is too large, above the linear region of operation, even just approaching compression then the data you see will be large signal performance and will not agree with s-parameter data.