S-parameters --> Z-Parameters --> single frequency Spice FET model?

[mrln55]

I am designing a microwave oscillator for use in a compact device with an overall length of less than one inch.
Since the device and its components can be treated as lumped elements, I would like to use LTSpice to simulate it.
My problem is that although there are Spice models for transistors operating at less than 1GHz, there are not many
models for devices that operate at higher frequencies. Since I am only interested in behavior at one frequency, I
hope to create an "application specific" device model from the Z-parameters obtained from the S-Parameters that are
available for a FET at a given frequency under specified bias conditions. My question is how do I derive an
equivalent circuit for a FET based upon Z-parameters? Is it reasonable to assume that the simulation would be valid
if the passive feedback network were tuned so that oscillation occurred under the specified bias conditions and
frequency for which the S-parameters were obtained?

TIA

 

[BigBoss]

Z and S-parameters are small signal parameters and you can obtain hybrid H parameters to model your transistor.But that is not enough..
But a real nonlinear model that you need can not be obtained by those parameters.
You can just obtain a small signal model which can be used to predict the oscillation frequency (Maybe!!) but in the other hand nonlinear model is a must to work with an oscillator.

 

[mrln55]

Hi BigBoss,

Thanks for your quick response. It appears that I would be best served by using an external off-the-shelf signal source (e.g., VCO) and restrict my efforts to providing gain and matching for a linear amplifier. With regard to the device model, would it be correct to assume that I would be limited to a model with four parameters since I would have four Z-parameters to work with, solving four equations/four unkowns?

 

[BigBoss]

Hi BigBoss,

Thanks for your quick response. It appears that I would be best served by using an external off-the-shelf signal source (e.g., VCO) and restrict my efforts to providing gain and matching for a linear amplifier. With regard to the device model, would it be correct to assume that I would be limited to a model with four parameters since I would have four Z-parameters to work with, solving four equations/four unkowns?

If you got Z-parameters, they will be sufficient to obtain S-parameters that is more convenient to design an linear amplifier with gain and acceptable I/O return losses.