================================================================================ IBIS INTERCONNECT TASK GROUP Mailing list: ibis-interconnect@freelists.org ================================================================================ Attendees from January 17, 2024 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 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 convened the meeting and took minutes. No patents were declared. During review of the January 10 minutes, Arpad Muranyi noted a typographical error in the phrase: "Her further stated two points:" ... which should be "He further stated two points:" Arpad moved to approve the minutes with this change; Lance Wang seconded. No objections were raised and the minutes were approved. During AR review, Arpad noted that he completed his AR to research whether the option line reference impedance is used in PLS for pole-residue data: it is. However, the current draft does not mention the need for the reference. Michael mentioned his closed AR regarding IEEE 370 reciprocity and symmetry; the document addresses the matrix data using these terms, not circuit data. Michael still needs to look at how IEEE 370 handles sampling, if at all. [AR] The team reviewed the latest updates to the draft. Arpad highlighted the new Source Data Conditions keyword group. He suggested the draft could alternately combine these with other source information (date, original filename, etc.) already in the draft. Bob Ross replied that combining this keyword data could work. However, it's not clear which new subparameters would be required vs. optional. Bob added that Weston Beal had stated that in some data he received from a customer he had to truncate data due to instabilities. Arpad suggested that the point-to-point step type (relationship) was mentioned in an e-mail to the reflector from Walter Katz. "Adaptive" spacing was not yet defined in the step type field. This would mean that the spacing between frequency points is not linear, not logarithmic, but change as the data demands. This should be captured in the draft. Arpad noted that the current draft places the source frequency list at the end of the data to make irregular spacing explicit. This list could be quite long. Randy Wolff noted the importance of maximum fitted frequency. This could relate to how EDA tools extract parts of a channel. Tools might issue some sort of warning if the model has less bandwidth than expected. Otherwise, as Bob noted, pole-residue data has theoretically an infinite bandwidth. Walter asked whether maximum frequency lists is what is valid data, or just what was fitted. Arpad stated that the maximum frequency that the model is good for is what "max fitted" refers to. Walter replied that we may use the higher frequencies to fit the data, but may restrict the result to lower frequencies. He added that he will internally distribute the draft when definitions are available. Arpad asked whether the frequency list and the concept of extrapolating to DC are mutually exclusive. Bob replied that noting extrapolation to DC is informational. Some matrix formats do not include zero frequency. Listing frequencies may be unnecessary. Walter added that the phrase "these subparameters are required" needs resolution. Arpad agreed, noting that not all rules have been written up. Randy stated that, when the original data does not have 0, does the extrapolated data have meaning? From experience, have many times seen tools save off fitting information and the tools will even generate an S-parameter from the fit, but the DC point can be totally different. Bob agreed. Arpad asked whether he should put his feedback in e-mail. Michael replied, yes, to save time in the next meeting. Michael asked whether the unfitted port term means per-port referencing. Randy replied that, yes, this could include short or open. This is similar to IBIS Interconnect models. Bob asked why we need the number of ports. The team needs to review all of these features in gory detail. Bob added that source ports vs. fitted ports may be a concern. Arpad asked whether one can regenerate the original data with more ports using pole-residue data. Bob stated that you cannot. Numerical syntax should use "="; non-numerical should not use "=". Several keywords are informational. The format allows the model-maker to create frequencies not in the original list. Arpad replied that the keywords document what was used to create the model data. Randy agreed that the team needs to decide what is required and what is optional. Michael noted that he is reserving additional comments (e.g., on quality metrics) for offline or later discussion. Arpad stated that Ted Mido's comments were agreed to/confirmed by Vladimir Dmitrev-Zdorov of Siemens EDA offline. Vladimir would prefer not to use preconditioned data; unbounded data issues with divide by zero should be handled by the tool. He prefers keeping the original unconditioned data. Michael asked whether PRECFAC applies to data as received or instead refers to how the tool should use the data. Arpad replied that this was a good question, wondering whether PRECFAC applies to S-parameters. Some questions should be addressed to the reflector. PRECFAC = 0 means original data is used. Michael noted that the team is just now discussing syntax niceties; RFM and PLS are convertible assuming feature sets are identical now. Arpad noted that Ted seemed to be very amenable to taking on features in RFM if PLS or other formats cover useful material. Bob replied that he tentatively favors PLS. He wants an appendix showing how easy it is to convert to/from the RFM format. Arpad asked whether this includes Foster format. Bob replied, yes. Bob added that IBIS-ISS includes Foster, so having conversions would provide closure. This could be in an "Other Notes" section. Arpad noted that the PLS format contains common poles, which should be equivalent to Ted's proposed POLES_FROM feature. Michael replied that having feature side-by-side listing is still a very good idea. Arpad suggested that symmetry of matrices is not explicitly mentioned but can be addressed through (i,j) syntax. Randy added that sparse matrices are already allowed. Bob replied that there is no need to list (11,1) (1,11) in matrix format; just use upper & lower. Randy likes this idea, suggesting that tools could read the upper/lower half keyword and say they apply to pole-residue data, if we can reuse them. Arpad asked whether two separate keywords should be used for source data and source file. Arpad moved to adjourn; Bob seconded. The meeting adjourned. The next meeting will be held on January 24, 2024. ================================================================================ 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.