Where To Find This Example
AWR Version 14
AWR Version 13
This example shows how to use the SModel measurement to make the model parameters from one device match the model parameters from another. The circuits involved are all built up in the MWO / VSS Getting Started Guide Nonlinear Example.
Simulating the design shows the two different sets of s-parameters and the “SModel” measurement that quantifies the difference between them.
To see how the “SModel” measurement allows for easy optimization when the goal is to have two sets of s-parameters match, open the Optimization control window (Simulate -> Optimize), and then press the “Start” button.
While the optimization is running, note how the two sets of s-parameters are converging (the “modeled” set is being driven towards the “measured set). Also note that there is only a single optimization goal (which utilizes the “SModel” measurement), as opposed to 8 optimization goal (one for the magnitude and phase of each s-parameter). Of course, if desired, the 8 different goals could still be setup.
After running an optimization, the s-parameters should all align very closely, as well as the actual model parameters as shown on the circuit schematics.
Note: It is very important to constrain the BJT parameter values. If not constrained the parameters can take on any values during optimization and this can cause the device to fail the convergence.
Sparam Part A Schematic
This schematic represents the “measured" device. Note the realistic values of RB, RE, RC, CJE, and CJC.
Sparam Part B Schematic
This schematic represents the “modeled” device shows the "current" device - note the unrealistic values of RB, RE, RC, CJE, and CJC. These parasitics will be optimized to make the circuit response match the measured device.
This graph shows S11 and S22 for both circuits - note the initial differences between Part A and Part B.
This graph shows S12 and S21 for both circuits - note the initial differences between Part A and Part B.
SModel Equations Graph
This graph shows (in dB) the computed differences between the s-parameter matrices for Part A and Part B. This difference accounts for all eight numbers in the s-parameter matrix (4 real, and 4 complex).
The “SModel” measurement has four different error functions - each of the traces on this graph show the results from a different error function. More information on the 4 different "SModel" error functions can be found in the Microwave Office help pages.
There is one optimization goal pointing to the “SModel" measurement with the default error function (“Average L1 Nom”). Note that for the two circuits to match, the "SModel" output should be driven to zero (no difference in the two matrices), but since it's plotted in dB the optimizer goal specifies that it should be less than -100 dB.