Before you can write a Verilog-A model of an RF circuit object,
you need to create an rfmodel.rational object
to represent the component.
There are two ways to create an RF model object:
You can fit a rational function model to the component
data using the rationalfit function.
You can use the rfmodel.rational constructor
to specify the pole-residue representation of the component directly.
This section discusses using a rational function model. For more information on using the
constructor, see the rfmodel.rational reference page.
When you use the rationalfit function to create an rfmodel.rational object that represents an RF component, the arguments
you specify affect how quickly the resulting Verilog-A model runs in a circuit
simulator.
You can use the rationalfit function with
only the two required arguments. The syntax is:
model_obj = rationalfit(freq,data)
where
model_obj is a handle to the rational
function model object.
freq is a vector of frequency
values that correspond to the data values.
data is a vector that contains
the data to fit.
For faster simulation, create a model object with the smallest number of poles required to accurately represent the component. To control the number of poles, use the syntax:
model_obj = rationalfit(freq,data,tol,weight,delayfactor)
where
tol — the relative
error-fitting tolerance, in decibels. Specify the largest acceptable
tolerance for your application. Using tighter tolerance values may
force the rationalfit function to add more poles
to the model to achieve a better fit.
weight — a vector
that specifies the weighting of the fit at each frequency.
delayfactor — a
value that controls the amount of delay used to fit the data. Delay
introduces a phase shift in the frequency domain that may require
a large number of poles to fit using a rational function model. When
you specify the delay factor, the rationalfit function
represents the delay as an exponential phase shift. This phase shift
allows the function to fit the data using fewer poles.
These arguments are described in detail in the rationalfit function reference page.
Note
You can also specify the number of poles directly using the
npoles argument. The model accuracy is not guaranteed
with approach, so you should not specify npoles when accuracy
is critical. For more information on the npoles argument, see
the rationalfit reference page.
If you plan to integrate the Verilog-A module into a large design for simulation using detailed models, such as transistor-level circuit models, the simulation time consumed by a Verilog-A module may have a trivial impact on the overall simulation time. In this case, there is no reason to take the time to optimize the rational function model of the component.
For more information on the rationalfit function arguments, see the
rationalfit reference page.
You use the writeva method to create a
Verilog-A module that describes the RF model object. This method writes
the module to a specified file. Use the syntax:
status = writeva(model_obj,'obj1',{'inp','inn'},{'outp','outn'})to write a Verilog-A module for the model object model_obj to
the file obj1.va. The module has differential input
nets, inp and inn,
and differential output nets, outp and outn.
The method returns status, a logical value of true if
the operation is successful and false otherwise.
The write reference page describes the
method arguments in detail.
An example of exporting a Verilog-A module appears in the RF Toolbox™ example, Modeling a High-Speed Backplane (Part 5: Rational Function Model to a Verilog-A Module).