============================================================================== IBIS INTERCONNECT TASK GROUP Mailing list: ibis-interconnect@freelists.org ================================================================================ Attendees from January 15, 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. During the review of the January 8, 2025 meeting minutes, Arpad Muranyi noted that he observed the word "by" was missing in one sentence, but was not able to identify the specific location during the live review. Arpad moved to approve the minutes as written; Randy Wolff seconded. The minutes were approved without objection. During the AR review, Arpad noted that his AR to capture port mapping differences between the draft port mapping document and Michael's Reference Configuration proposal was not yet done. Michael suggested deferring this until later, as it would likely be discussed during the meeting. Michael reviewed a presentation showing how IBIS-ISS addresses S-parameter instantiation in the S-element, contrasting and comparing its reference approach to the proposed [Reference Configuration] concept earlier. Arpad raised objections to the "global" vs. "local" terminology used, as global references need not also be node 0 or "ground". Similarly, one could have an explicit node 0, even as node 0 being omitted is implied. Other language should be used. Michael took the AR to check how IBIS-ISS uses terms like "GND", "ground", etc. [AR] Arpad also expressed concerns about the use of "ideal" referring to node 0. He wondered if GND needs to be replaced with A_gnd. Walter Katz disagreed, noting that on cannot do this replacement, as subcircuits in IBIS-ISS use node GND. One could however say in the text that A_gnd and GND are the same. Arpad agreed with this direction. Walter agreed with this approach, suggesting use of the names "Vss", "Vss1" and "Vss2". No objections were raised to correction of the example text. Walter stated that we are allowing Touchstone to be very general; we are allowing IBIS-ISS to do unusual things with reference nodes, but to define the way the nodes were used to make the measurements. Fundamentally, any use of Touchstone files in IBIS models assumes that the referencing of those Touchstone files can be used with the A_gnd node. This is the way the industry is using IBIS Interconnect. One may be performing fancy PI analyses, but you are opening up the Pandora's box when simulators are using something other than A_gnd for IBIS simulations; this is a mistake. Arpad replied that [Reference Configuration] is not really the issue. He cited examples from his earlier e-mail involving a coaxial cable between the moon and Earth, but also a buffer model with separate package model, etc. In both cases, VNAs are connected to their local references: negative on the shield of coaxial cable, positive the on center connector. Arpad suggested one would use per-port connections in this instance, but in usage one would connect the references to the same node. The data does not tell you how to use it. Imagine having a PDN buffer with its Vss terminal and signal separated. If you are making a package model for this with the Vss included (see Arpad's upcoming DesignCon IBIS Summit presentation), should we use Vss as reference? Walter stated that measurements are conducted using two wires with locations specified. Michael suggested that using Earth and Moon as location labels explicitly could enable tools to identify extreme distances and warn the user that references were widely distant. A parser could provide a caution, assuming that units were provided, when physical references were very far apart. Arpad asked whether the location-label approach indicate whether a given node is ideal node zero; we have to document the meaning of XYZ location relative to other locations. Walter suggested that the term "ideal" in our context means "trust me, the references are close". Jim Antonellis suggested that word "ideal" was used earlier in a slightly different way. It can mean "was done properly" but also "ideal GND node" in other contexts. Arpad noted that IBIS Interconnect models can have two separate connections. Imagine that two people extract two S-parameter models: one addresses the buffer side, while the other addresses the package. The two people are totally isolated. One may choose to model a structure with an S4P using reference of Vss. The other uses an S6P where the negative terminals are all connected to node zero. You cannot connect these two structures, but you don't know it without the meaning being known about both. There are two ways to measure these; (a) positive on signal bump pad and the reference on a nearby Vss; or (b) with the same positive node, but the other node on imaginary/fictious metal. Walter stated that measurement always involves real locations, not fictional metals on packages. If they both use "close" probing locations, the references can be treated as identical. Arpad proposed a second "crazy" example: imagine an on-die interconnect between buffer and pads, using Vdd as measurement reference point. This could also be done with a floating reference. Walter replied that he would fire the person who did that kind of measurement. Electrical engineering is based on a reference concept. Walter added that would like to see a real example of the actual engineering problem these syntax changes were being proposed to solve. Michael accepted the AR to compare the port mapping scheme syntax against the Reference Configuration approach, to see if location information can be required to solve this issue. [AR] Arpad accepted an AR to generate 1-2 drawings showing the issue. [AR] Arpad moved to adjourn the meeting; Jim seconded. The meeting adjourned without objection. The next meeting will take place on January 22, 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.