================================================================================ IBIS INTERCONNECT TASK GROUP Mailing list: ibis-interconnect@freelists.org ================================================================================ Attendees from August 30, 2023 Meeting (* means attended at least using audio) ANSYS Curtis Clark Broadcom James Church Intel Corp. Michael Mirmak* Michael Brownell Keysight Technologies Ming Yan Marvell Steve Parker MathWorks Walter Katz Micron Technology Justin Butterfield Siemens EDA Arpad Muranyi*, Randy Wolff* ST Microelectronics Aurora Sanna Teraspeed Labs Bob Ross* University of Illinois Jose Schutt-Aine Zuken USA Lance Wang* Michael Mirmak called meeting to order. No patents were declared. Michael thanked Randy Wolff for running the previous meeting and taking the minutes. During the minutes review for the August 23 meeting, Arpad Muranyi noted that he had provided comments and suggested corrections through the e-mail reflector on August 24. He requested that the phrase: "Arpad added he was writing the TSIRD as if it could be in a separate specification. " should be changed to: "Arpad added he was writing the TSIRD as if it could be in a separate specification OR also included, either way." Arpad moved to approve the minutes with this change; Randy seconded. No objections and the minutes were approved with the changes. During the minutes review from August 16, Arpad noted that he had requested several changes through a note to the reflector. These are: Change: "but we could compress by a shorthand mentioning the subparameters once." to: "but we could compress by a shorthand mentioning the subparameter in the [Begin Common Poles Data] keyword only." Change: "Arpad stated that pole-residue is organized by i,j instead of by frequency; it's much harder to organize pole-residue by frequency." to: "Arpad stated that pole-residue is organized by i,j instead of by frequency; making it a lot easier to support sparse matrix or repeated matrix component shorthand formats." Change: "Arpad answered that this is the nature and use of port-map. In pre- vs. post- layout analysis one has two scenarios where this information would be used, and they are quite different. The first case is gate-drain-source for a transistor, with no idea how to plug this in. This is the "flexible" case; it's a description, not a map. Walter replied that relationships are being provided, as in measurements; these are called logical connections." to: "Arpad answered that there are two scenarios and use of port-maps, pre- vs. post-layout analysis. An example for the first case is gate-drain-source ports for a transistor, with no information on how it will be plugged into a design. This is the "flexible" case; it's a port description, not a map. Walter replied that relationships are being provided, as in measurements; these are called logical connections." Change: "Arpad noted that this is a map, what is connected "on the outside" - no group is needed." to: "Arpad noted that this is a map, describing what is connected to the ports 'on the outside' - no group is needed." Change: "for a three-point extraction" to: "for a three-port extraction" Change the following paragraphs from: "Walter suggested that the map represents the pin where one is measuring this Touchstone file's data; all you can do is to see how the connections are made. In a power-aware Touchstone file, there is one probe on the die, one on the bump. Arpad disagreed, suggesting an example starting with a 3-port network. Two dice, each with a ground and a VCC point for a three-point extraction. These three ports are like an ideal transformer, with individual windings isolated from each other. If you have 5 V on the ball side, 4.95 at the other end, you will see the 4.95 at port 2. Resistive losses in the package will be visible at ports 2 and 3 but the voltage between 1, 2 and 3 will be floating; there is no voltage. Walter agreed, adding that all voltages are relative to local ground. One cannot measure the voltage drop between the VCC ball and VCC pad. Arpad disagreed." ... to starting the 2nd paragraph without the "disagreed" word: "Arpad suggested to start with an example of a 3-port network." Arpad also asked to delete the last sentence from the third paragraph, as he would agree with that. Change: "Arpad suggested on cannot measure a 68 V difference between windings unless the ground is present." to: "Arpad suggested on cannot measure a 68 V difference between windings unless a common ground is present." Change: "If you plot them on the same chart, you do not know the reference for the windings." to: "If you plot them on the same chart, you will not know where to position the waveforms vertically with respect to each other because there is no common reference for the windings." Bob Ross noted that the specification should be made compatible with mixed-mode data. Michael stated that minutes will capture the previous minutes' text changes. Arpad moved to approve the minutes with the changes reviewed. Randy seconded the motion. No objections were raised and the minutes were approved. Michael noted during the AR review that none was recorded from the meeting held August 16th. In the August 23rd meeting, Arpad accepted the AR to distribute the text of draft 9 to the reflector; this has been done. Michael summarized his knowledge of Sam Chitwood's availability; Sam has moved on to another industry and may not be available to contribute to the discussion. Michael had sent out his presentation on loop inductances to a smaller distribution list, not the reflector; however, replies to his note were on the reflector. Michael took the AR to contact Sam. [AR] Arpad suggested contacting other Sigrity/Cadence personnel. Michael summarized a recent conversation with Pete Pupalaikis, formerly associated with test-and-measurement and author of a recent book on S-parameters. He noted that Pete was very willing to provide his compression algorithm for network data, which should result in files ~99% smaller than the raw ASCII S-parameter data, legal restrictions permitting. Michael took the AR to get additional details from Pete. [AR] Arpad reviewed Draft 9 of his pole-residue document, noting that the team still had some decisions to make related to highlighted areas of the text. The rest of the document has now been formatted using an IBIS-specification-like syntax. Description text is often TBD; this will be populated after editing/compression. Changes in red are mostly new text, not moved text. Lots of changes have been made to the text describing referencing; after long discussions with Arpad's colleague Vladimir, "normalization" is not linked to the reference impedance but a frequency-based normalization (division by the pole, nothing to do with reference impedance). The Y(s) formula has now been explained in more detail in the new text. The Y in the equation does NOT refer to Y-parameters, but instead to a Laplace transform relationship. Some additional editorial/style decisions may be needed. All green highlighted text from previous drafts is now gone. The highlighted text at the end of the document remains, now including Walter's request for support of a large-connector shorthand. Bob replied that what Vladimir calls normalization is the Y(s) formula; the notation makes the content difficult to read. Conversion between S-, Y-, and Z-parameters is another type of normalization. Arpad suggested that these are different context; we do different normalization in Touchstone 1 (for S-parameters only) and no normalization at all in Touchstone 2. Bob disagrees; the commercial SPICE representation of Laplace data is different than what is shown here. The Ym(s) equation is what Bob views as normalization. Actual values per frequency are divided by a reference impedance. G-, H-, Y- and Z-parameters are not normalized in Touchstone 2 at all. The disagreement here is over the meaning and context of the word, "normalization". Michael brought up the Touchstone 2 specification. Normalization may occur in frequency or value, but we tend only to discuss the value aspect. There is some problem with the "when [Reference] is not present" language in Touchstone 2 currently. Randy and Arpad noted that the pole-residue data is a simple conversion, subject to "garbage-in, garbage-out" effects. S-parameters are not in question; Y-, Z-parameters are the issue. Arpad will discuss this issue internally. Bob noted that "=" should be present for subparameters taking only a single numerical value. Michael suggested moving the equations elsewhere in the specification (e.g., to an appendix). Next time the team will discuss port-mapping. Arpad moved to adjourn; Randy seconded. The meeting adjourned. The next meeting will be held Wednesday, September 5, 2023. ================================================================================ Bin List: 1) [Complete draft Touchstone document separating version 1.0 and 2.0 into their own chapters] - REMOVED 2) Create structures to encapsulate Touchstone 1.0 data in Touchstone 2+ specifications - TABLED 3) Complete draft Touchstone 2.0 document containing TSIRD3 and TSIRD4 draft (Muranyi) – COMPLETED IN DRAFT 10 4) Complete pole-residue format BIRD draft (Muranyi) 5) Complete port naming proposal (Katz) 6) Create alternatives to the Touchstone 1.0 option line before the "R" character - TABLED 7) Complete ISS-IRD 1 Draft - Enable Cascading of S-parameters Through W-element (Mirmak) - TABLED 8) Complete/revise Touchstone 3.0 draft outline (Mirmak) – dependent on several items above Tabled ARs: - Arpad to give an example of the physical connectivity needed for EMD automation.