============================================================================== IBIS INTERCONNECT TASK GROUP Mailing list: ibis-interconnect@freelists.org ================================================================================ Attendees from February 12, 2025 Meeting (* means attended at least using audio) ANSYS Curtis Clark*, Wei-hsing Huang, Juliano Mologni Arista Networks Jim Antonellis* Broadcom James Church Intel Corp. Michael Mirmak*, Xiaoning Ye Keysight Technologies Ming Yan Marvell Steve Parker MathWorks Walter Katz* Micron Technology Justin Butterfield Siemens EDA Weston Beal, Arpad Muranyi*, Randy Wolff Simberian Yuriy Shlepnev ST Microelectronics Aurora Sanna Synopsys Ted Mido, Edna Moreno University of Illinois Jose Schutt-Aine Zuken USA Lance Wang Michael Mirmak convened the meeting. No patents were declared. The previous meeting minutes were not available for review. Michael made a few general statements about coverage of related topics (SPIM support, Touchstone, IBIS and interconnect modeling, etc.) in multiple task groups, including IBIS-ATM and IBIS Interconnect. There is a lot of overlap, and the various teams are trying to separate the discussions to make efficient use of time. No other opens were raised. Michael introduced the issue of Touchstone and illegal characters, raised in a separate message from Barry Katz of MathWorks and discussed on teh reflector. The parser rejects extended ASCII characters in comments, which is somewhat limiting to documentation. Walter Katz noted that he had received files with degree symbol in the comment section. Michael agreed, noting that this was observable in IEEE 370 headers if the text was generated using a template from a source using extended ASCII. Walter suggested that the whole file need not support extended ASCII - the issue was simply whether comments could be exempt from the basic ASCII requirement. Arpad Muranyi asked how the change would affect those who use comments for other purposes. Walter replied that would be up to the person or tool parsing the data Michael asked whether EDA tools would care if there are two different formats of text in the same file. Arpad suggested that the IBIS parser and IBIS files be checked to see whether the same limitation existed. Michael asked whether supporting extended ASCII would be enough, or whether Unicode would be required. He also raised the concern about support for binary in the comments section, which could pose a security risk. He suggested the actual change to the specification would require modifying only one line of text in a TSIRD. Michael accepted the AR to write a draft TSIRD. [AR] He also accepted the AR to contact the IEEE 370 team to get their comments. Michael presented on the current port-mapping proposal, showing the original objectives plus each of the keywords and parameters, identifying each as related to connectivity, data origin (including for correlation), or both. He ended by stating that ending the confusion about port connectivity would require creating a table showing relationship between data generation and connectivity, showing which combinations would be illegal. This would also have an impact on other standards, including IBIS itself, as more IBIS features were integrated into Touchstone. Arpad suggested that identifying connectivity may not be clearly achievable in all cases. If a file is used for an EMD description, we know how to hook it up to other things, but if the Touchstone file describes a cable, connectivity could be done in nearly any way. Data origin implies connectivity too. Walter disagreed, stating that a Touchstone file should advertise how it should be hooked up to other things. Arpad asked whether there was still agreement that one needed to know how the files were generated in order for connectivity to be determined. Walter stated that this was not needed for the Sij status reporting and automatic generation of test probe locations Michael suggested that the objective enabling Touchstone file viewers to generate mixed-mode S-parameters for differential ports is unclear due to language around "differential". Arpad stated that he was not sure that Group was for referencing (the presentation suggested this was related to both connectivity and data origin, and therefore needed to be split into two parts). Walter added that the Physical subparameter can be a pin or list of pins, for example a metal plate that shorts a collection of pins, and the probe is on this plate. This affects how the data is measured, or how it is simulated. Walter added that he agreed that the presentation was correct. A specific case that needed addressing was to understand difference between the SPIM R-network model and the SPIM Touchstone model. Michael added that SPIM support was completely missing in the port-mapping proposal today, which is an example of the problem of Touchstone needing to incorporate "IBIS stuff" as IBIS is updated. Walter replied that the Relement in SPIM is limited to DC analysis; Touchstone is limited to AC analysis. Each uses "ground" in a totally different way. Arpad agreed, adding that supporting these should be relatively easy, as the analysis is completely either one or the other (no co-simulation). Interconnect and EMD don't do this, so simultaneous support for DC and AC is required. One never knows what combinations will be used in a simulation. Walter replied that, whenever you simulate, you may not be looking at the ground structures when you simulate for signal integrity. For instance, DDR4 connections would all use A_gnd. Any case where you wrap using N+1 nodes to a signal name will tend to use A_gnd as the reference. But for power integrity, "ground" concepts are more subtle. He suggested getting experts from additional EDA vendors. Michael proposed generating a table of data origins vs. a table of connectivity, to see what combinations would be allowed. Walter disagreed, stating that one still uses a measurement probe with a reference probe in 3D structures, whether MoM simulation or lab work. The reference used in a SPICE netlist may not matter. Michael noted that any table of cross-support would need to cover what other structures support (e.g., EMD, Ts4file, IBIS Interconnect, C_comp Model, and SPIM). Arpad briefly reviewed ball vs. pad probing concepts in an example. Walter replied that, in measurement, the port and reference on the silicon side be the same, and that on the board side the port and reference should be consistent as well. Michael took the AR to produce an initial table for next time. [AR] Arpad moved to adjourn; Walter seconded. The meeting adjourned without objection. The next meeting will take place on February 19, 2025. ================================================================================ Bin List: 1) Complete port naming proposal (Katz et al) 2) Complete/revise Touchstone 3.0 draft outline (Mirmak) 3) Complete ISS-IRD 1 Draft - enable cascading of S-parameters through W-element (Mirmak) - TABLED Tabled ARs: - Arpad to give an example of the physical connectivity needed for EMD automation.