=============================================================================== IBIS INTERCONNECT TASK GROUP Mailing list: ibis-interconnect@freelists.org ================================================================================ Attendees from July 17, 2024 Meeting (* means attended at least using audio) ANSYS Curtis Clark, Juliano Mologni* Broadcom James Church Intel Corp. Michael Mirmak* [Michael Brownell] Xiaoning Ye Keysight Technologies Ming Yan Marvell Steve Parker MathWorks Walter Katz* Micron Technology Justin Butterfield Siemens EDA Weston Beal*, Arpad Muranyi*, Randy Wolff* ST Microelectronics Aurora Sanna Synopsys Ted Mido, Edna Moreno Teraspeed Labs [Bob Ross] University of Illinois Jose Schutt-Aine Zuken USA Lance Wang Michael Mirmak called the meeting to order. No patents were declared. Michael reviewed the minutes of the July 3 meeting. Arpad Muranyi moved to approve the minutes; Randy Wolff seconded. The minutes were approved without objection. Two ARs to Michael are still in progress: - Follow up with remaining experts consulted on Touchstone 3.0 port-mapping changes - Ensure that all Open Forum and Task Group meetings have transcription enabled Michael reviewed a proposed list of TSIRD assignments for proposed new features in Touchstone 3.0. Walter Katz noted that, for generation of a schematic symbol, one would list the ports in the appropriate order according to the "sides" of the symbol; one needs a keyword for that. The list would contain the ports in order on the left side, right side, etc. sorted top to bottom, then top-to-bottom sorted left to right. Arpad suggested that a column could be added for schematic symbols as an alternative. Walter replied that the issue is that there's nothing visual (no reference physically) in the data there relative to the configuration for the symbol. Michael asked whether this data would be different from the propose port location (XY) data. Walter confirmed this. Walter asked whether the sub-parameters on the port are already defined for Physical Name (as opposed to Logical Name). Walter's earlier proposal encoded this in the string in the Physical Name, as x;y;z. Assuming that Touchstone file uses a board file database elsewhere mentioned (ODB, MCM, etc.). Units could be an issue in this approach - this is an important detail. Each one has a reference. Weston Beal asked whether location would be optional. This doesn't make sense for a connector model. Arpad asked, in this context, whether the reference locations would be with respect to board or to the part itself. Each part will have its own origin, which would be different than the system where the part would be used. Walter replied that, if this is an S-parameter for a PCB, then the reference would be to the PCB. Weston responded that this could be very useful if we know the origin and the units involved. Arpad asked whether, if a person wants to measure at a connector, they measure with respect to the connector or the PCB origin. Weston suggested this would involve the logical name instead of the physical name. Walter replied that the physical name could be coordinates or a physical name/pin number; this may involve probing on a via or test point. Michael asked whether an indicator keyword is needed or not for self-contained S-parameter data. Walter responded that the team cannot solve all problems in the world; there may simply be a change in meaning depending on context. It would be nice to do things in a consistent way; we at least need the information to be included to determine what to do. Michael asked about user defined parameters. Walter stated that user-defined parameters are really simple, in that the format is intentionally open-ended: ( ). The names and values cannot have parentheses, and would have no conflict with reserved parameters in the future. We may want to use a colon rather than a space as the delimiter. Arpad asked what the EDA vendor should do with it. Walter replied that it is for the model author to use as in-line documentation (e.g., is something measured or not). Weston suggested that we would need specific keyword for user variables, with a begin/end pair. Walter responded that we should put them in the port definition syntax. Weston suggested that we need to be careful with the parser behavior - what does it do with these? It may assume a user-defined parameter if the syntax of a keyword isn't recognized. Walter suggested using a "User:" string delimiter; Weston replied that this makes sense. Arpad noted that the industry could run into model-specific parameter support issues; we cannot force people to learn new things with old structures being used. Walter replied that the team has a choice on defining a strict format or addressing model-maker needs. The team discussed an IEEE 370 section. Weston asked whether we need to specify the sub-parameter names. Michael suggested this section would be a syntax synonym for the Begin/End Information section. It would amount to comments without the bang character. Weston suggested that people will use the files without looking into them. Walter added that each EDA tool is free to look in the IEEE 370 section and implement a parser accordingly. On the topic of combining simulated and measured data, Randy asked whether people combine data today. Walter replied that they do, yes. Michael agreed, and that often tools can use mirroring and symmetry to fill in missing data. Walter added that users often want accurate loss but don't need accurate crosstalk in this application. An S16P may be needed but your machine may only capture an S4P at one time. Arpad noted that, under frequency range information proposals, sampling data and configuration are not covered. Michael noted that a TSIRD 12 involving identifying both interconnect and passive expectations is not a high priority. Walter provided the example of a controller and three DQs, where one DQ net is defined to an on-board memory element; the controller connects to 2 memories. This is a passive system that is also an interconnect but is not a cable/connector assembly and is not necessarily symmetrical as the proposed keywords assume. On TSIRD13, Walter noted that one can model crosstalk on a connector model, associate with that a swathing name to map S-parameter data for three wafer sections into a whole connector. Do we want connector suppliers to support this format? There are two or three swathing methods possible and available. We can contribute to creating standardized methods across manufacturers. In this case, we would simply define "swath = ", with the name referring to a schema (file) supplied separately. Arpad moved to adjourn; Randy seconded. The meeting adjourned without objection. The next meeting will be held July 24, 2024. ================================================================================ Bin List: 1) Complete port naming proposal (Katz et al) 2) Complete ISS-IRD 1 Draft - enable cascading of S-parameters through W-element (Mirmak) - TABLED 3) Complete/revise Touchstone 3.0 draft outline (Mirmak) – dependent on items above Tabled ARs: - Arpad to give an example of the physical connectivity needed for EMD automation.