=============================================================================== IBIS INTERCONNECT TASK GROUP Mailing list: ibis-interconnect@freelists.org ================================================================================ Attendees from October 30, 2024 Meeting (* means attended at least using audio) ANSYS Curtis Clark, 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 called the meeting to order. No patents were declared. During review of the October 23 minutes, Arpad Muranyi recommended that the minutes strike the following phrase: "Arpad showed his IBIS EMD package interfacing drawings, stating that his preference is for Touchstone to use a node-zero-independent way of referencing, but recognizing that measurements using such a scheme may be impossible." Arpad added that he does not mind having a node zero connection to all paths, so long as Vss paths are also available. Michael noted that the word "S-parameters" was incorrectly spelled in two places. The word "reference" is also misspelled as is the phrase "power-delivery". Arpad moved to approve the minutes with these changes; Randy Wolff seconded. The minutes were approved without objection. Michael noted that he has an open AR to distribute the minutes of his conversation with Scott McMorrow, with his approval. [AR]. The team discussed the referencing concept. Arpad stated that the term "reference" has been used loosely so far, possibly confusing the concepts of potential vs. return path. Michael agreed, mentioning Scott's contrast of field or wave theory vs. circuit theory, where the latter is a special case of the former. Arpad mentioned that Yuriy Schlepnev had commented that he might generate slides summarizing the requirements of and differences between the two approaches. Michael took the AR to contact him about this [AR]. Some questions still remain to be answered. The team discussed a set of e-mails between Walter and Arpad regarding potential reference clarifications. Michael noted that one could add an opening statement to the document, but this may not be particularly useful and will likely be ignored. Walter noted that he reviewed IBIS documentation and showed a presentation with IBIS 7.2 diagrams and examples. The terminator model is a particularly interesting one. Asking what voltage means in an IO buffer, he noted that each has a GND. What is the voltage measurement relative to, as a point? All IO buffers have a local point to it called GND. Weston Beal stated that "GND" may or may not be same as SPICE node 0. Jim Antonellis noted that the ground symbol is not labeled "GND" in several IBIS figures, but there are GND or [GND Clamp Reference] labeled nodes in the same diagrams. Michael explained the unfortunate confusion between the concepts, for "Pulldown Reference" of rail location vs. voltage as a number in IBIS so far. Walter replied that one can have GND Clamp Reference that is not zero. He added that he is working on an opening paragraph statement to clear all this up. [AR] Arpad: in the beginning, only had C_comp and Pullup reference; started to "wise up" after that point, and introduced reference keywords, due to need for noise treatments, split up C_comp. Associate C_comp parts with IV tables. Didn't fix C_comp words for referencing and connections. Also never fixed language/definitions for cap in packages RLC. By default, every one may be connected to ideal node zero. Weston added that the IBIS authors didn't even use GND initially; this issue was ignored until IBIS had to support more exotic structures, including packages. It does matter what we call the reference for each port, and it's not node zero. Walter replied that a buffer on die has a local reference node; when you measure a voltage on the I-V table, etc., that is the reference for that table. All voltages in this area are measured with respect to a local ground. The pin voltage is measured with respect to its local ground. Weston and Arpad voiced disagreement with the "grounds" being the same in all these cases. Michael suggested that part of the challenge was moving from undistributed to distributed approaches for "Vss"; as soon as a "Vss rail" became anything other than an ideal short, the assumptions fall apart. Arpad noted that, in IBIS 7.2, Figure 18 on page 106 shows part of the issue. What is the ground symbol of the package actually connected to? Weston replied that it was valid to use node zero in IBIS 1; this is not correct anymore for more complex structures. He added that the team should not reference the U-element anymore due to deprecation of the model. Randy Wolff added that one can't model power delivery with it. Michael asked whether the ground rail in the U-element is actually the same physical ground node point. Randy replied that there was no way this diagram showed physical distribution. Michael cited the physically-possible example of a system where a pin is miles away from a buffer, but the buffer pad is close to a "local" reference point. Could this same ground node be used for a reference? Walter replied that everything was pin referenced in original IBIS, but today it is meaningless to use pins; references now are to pads. Michael asked whether this was still true when combining different files representing different parts of a system. Walter responded that where the models meet is where one needs to be careful about connection points Arpad showed the U-element diagram containing both a Vss rail and an ideal ground. Weston noted that capacitances would normally be between the Vss rail line and the signal line. Jim also recalled capacitors between lines. Walter took the AR to send out his prefatory statement [AR]. Michael will also send out his Touchstone prefatory statement [AR]. Arpad moved to adjourn; Weston seconded. The meeting adjourned without objection. The next meeting will be held November 6, 2024. ================================================================================ 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.