A comparison of different techniques (i.e. - windowing, vector fitting, etc.) for extracting the impulse response from S-parameters.
Note: You can use the contents widget (just to the right of the pencil widget, above right) to navigate this content.
On August 25, 2020 I had an article published in Signal Integrity Journal (SIJ) titled, Impulse Response from Insertion Loss.
In my article I explained how the raw, positive frequency only, data contained in a Touchstone® file needed to be conjugate duplicated and reordered, in order to get a correct impulse response from a typical ifft() function.
I also went on to explain how applying windowing to the data, before calling ifft(), can help eliminate the non-causal ringing, which typically occurs as a result of the incomplete data in most Touchstone files.
Later, on May 12, 2022, Professors Morales and Agili, of Penn State Harrisburg's Center for Signal Integrity, presented a SIJ webinar titled, The Use of DSP Techniques to Analyze S-Parameter Sampled Data. Their webinar was partly a rebuttal to my own previous article. In fact, they used some of my figures unaltered, presumably to make it quite clear that they were rebutting my suggestion of the use of windowing techniques to smooth non-causal ringing. In their webinar, they propose, as an alternative to windowing, the use of certain DSP techniques, in particular the Bilinear Transform, to better transform the raw frequency domain S-parameter data in a Touchstone file into the discrete time domain, for further processing there.
I responded, on January 16, 2023, with a Jupyter notebook published as a Gist, which partially rebutted Morales and Agili's critique of my original article and pointed out some flaws in their understanding of certain relevant concepts. Professor Morales was kind enough to take the time to respond, in the form of a comment on this Gist. I then posted a notification of that fact to the SI List e-mail reflector, inviting others to join in the discussion. Since then several other individuals have contacted me about this topic, expressing interest in participating in the debate but unsure of how best to do so. So, I created this new Github repository, to provide a central and well organized platform for that discussion. The intent of this repo. is twofold:
-
To foster focused discussion by topic of:
- my original SIJ article,
- the SIJ webinar rebutting my article, by Prof. Morales & Agili,
- my Gist critiquing some of the reasoning behind their rebuttal, and
- Professor Morales' comments on that Gist. (Scroll to bottom if not initially visible.)
- Note: Please, do not add to this commentary, but instead use an Issue in this new repo. to further discussion. And, please, try to keep a 1:1 relationship between Issues and discussion topics. That will keep the various topics being debated well separated and assist new participants in acclimating quickly.
-
To provide a central public space for developing our communal understanding of the pros & cons of the various techniques available for extracting the impulse response from the S-parameters:
- Inverse Fourier transforming, with and without pre-windowing,
- Vector fitting, using the Bilinear Transform,
- others?
This Github repository contains the following files:
README.md: (This file).gitignore: Contains a list of filename patterns to ignore when reporting status and/or suggesting files to add.ImpulseResponseFromSparams.ipynb: My critique of Professors Morales and Agili's rebuttal of my SIJ article, as a Jupyter notebook.LICENSE: Contains the details of the BSD-3 license, which governs the use of this work.
Note: References 1-7 are the same as in Prof. Morales' original comment on my Gist, for convenience.
[1] John Choma, "Scattering Parameters: Concept, Theory, and Applications," Lecture notes, https://www.ieee.li/pdf/essay/scattering_parameters_concept_theory_applications.pdf. Last accessed July 23, 2021.
[2] A. Morales and S. Agili, "Analysis of Analog Sampled S-Parameters Data Using DSP Techniques," IEEE Transactions on Instrumentation and Measurement, Vol. 71, January 2022.
[3] P. J. Pupalaikis "The Relationship Between Discrete-Frequency S-parameters and Continuous-Frequency Responses," Proceedings of 2012 Designcon, Santa Clara, CA, February 2012.
[4] Alan V. Oppenheim and Ronald W. Schafer, Discrete-Time Signal Processing, Prentice Hall Signal Processing Series, 1999.
[5] A. Morales, S. Agili and T. Meklachi, "S-Parameter Sampling in the Frequency Domain and its Time Domain Response," IEEE Transactions on Instrumentation and Measurement, Vol. 70 September 2020.
[6] B. Gustavsen and A. Semlyen, "Simulation of transmission line transients using vector fitting and modal decomposition", Power Delivery IEEE Trans, vol. 13, no. 2, pp. 605-614, Apr. 1998.
[7] B. Gustavsen, and A. Semlyen, "Rational Approximation of Frequency Domain Responses by Vector Fitting," IEEE Trans. Power Delivery, Vol. 14, no. 3, pp. 1052-1061, July 1999.