While saving a .s4p file seems very natural with a 4-port structure, the saving of .s2p and .s3p files raises some questions (such as, which ports?) as does the desire to save possible mixed-mode parameters in the same file format.

The .sNp setup dialog (see Figure: SnP Setup Dialog) helps in the matter of which ports are used for the lower order files. Since the system cannot know which ports are necessarily of the most interest to the user (particularly when a higher order calibration is applied), the pair or triplet to be saved must be specified on this dialog. Of course, the port used for .s1p file save must be similarly specified. If a port combination is selected that is not currently calibrated, available data for the relevant parameters will be saved but may not be physically meaningful. Parameters corresponding to a trace that is currently on-screen will have current data sent to the file but it will be uncorrected. If the parameter is not currently displayed, data in an existing buffer will be saved but that measurement may have occurred previously. As with .s2p files discussed earlier, the data format may also be specified on this dialog. Navigation to the dialog: Main | System | Setup | Misc Setup | SnP Files Setup

For mixed-mode parameters, another file save option is available: .mNp. The text file format is exactly the same as for the .sNp format but mixed mode parameters are used instead. The row-column orientation is what one would expect from the response menu dialogs but are summarized below.

As discussed in Enforced Passivity and Causality of this guide, enforced passivity and causality in saved .sNp files is possible. If either box is checked (Figure: SnP Setup Dialog), the data is coerced before saving so caution should be exercised. Enforced passivity uses a re-scaling of the S-parameter matrix eigenvalues, in a self-consistent way, to ensure that the 2-norm of the S-parameter matrix is less than unit (and the tolerance value correlates to how much less than unity).

Enforced causality invokes the Kramers-Kronig relations to minimize energy present before time t=0. With good calibrations and measurement practice (and no absurd de-embedding choices), these enforcement practices have minimal data impact but may aid convergence when using the data in large scale time domain simulations. When measurement issues are present, the changes to the data can be significant and the cause should be investigated.

Another .sNp saving choice is to include Trace Mem Math operations in the saved data. When this is selected and one or more traces have Trace Mem Math turned on, the modified data for the parameters with match applied will be saved to the .sNp file. This can be useful for saving normalized results. If trace math is not applied for every parameter (which would only happen for a .s4p file if 16 traces were active and all had trace math applied and each trace had a different S-parameter), the unmodified data for the uncovered parameters will be saved instead.