================================================================================ IBIS INTERCONNECT TASK GROUP Mailing list: ibis-interconnect@freelists.org ================================================================================ Attendees from February 28, 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 called the meeting to order. No patents were declared. Michael noted that minutes were yet not complete from the meeting of February 21st. He will distribute them for review in time for the next meeting. He also noted that all ARs had been completed except his own to bring a formal request to the Quality Task Group regarding generic parser outputs from IBISCHK7 and TSCHK2. Arpad Muranyi reviewed draft 2 of the options line removal TSIRD. He noted that under this TSIRD, the option line itself becomes optional instead of required. Importantly, if the options line is not present, no defaults will be assumed or in effect related to it. Randy Wolff asked whether multiple # lines allowed under this TSIRD or the previous specification. Arpad replied that the specification assumes the last option line seen is the one that takes effect for the file. Some surprise was expressed by multiple attendees about this rule. Weston Beal asked whether it is permitted to have a # line in the middle of the data. In the TSIRD, [Reference] is now required if the option line is not present. Arpad added that, in his text, he is trying to stay away from mentioning Version 3.0 specifically in the context of new rules and interpretations. Michael noted that the changes to [Reference] should allow similar practices to the way IBIS supports "instant" updates of [Version]-like keyword fields (e.g., from 2.0 to 3.0), as raised by Walter Katz last time. Some discussion resulted, focused on whether the complete if/then rules for [Reference] have been defined. Michael mentioned the potential use of Karnaugh maps in finding missing situations. Arpad noted that the override rules in Touchstone 2.1 are in the option line definition. If both [Reference] and the option line are present, the keyword will "win". Arpad took the AR to add this relationship to the text to make it explicit. [AR] Walter asked why the specification should include the version 3.0, 2.0, etc. rules. He suggested just referencing earlier version documents for history. Arpad replied that he is trying to make minimal changes versus what is already written. Arpad also reviewed the [Frequency Unit] and [Complex Number Format] keyword new rules. Michael asked whether the text was complete, in that only the need to clarify overrides was still outstanding. Arpad replied that he is assuming the Editorial Task Group will go through a major reformatting of the text. Walter asked why the document specifies each rule relating to option and [Reference]. Michael replied that the team is assuming that some data files will be updated by authors, rather than being created from nothing. Randy added that the Editorial Task Group will be finding things (combinations) that the team may have missed. Arpad took the AR to add a new sentence clarifying the [Reference] and option line relationship, and to send out draft 3 for review [AR]. Next time, the team will review the document and vote on forwarding the document to the Open Forum. Walter reviewed the formatting of the two port mapping proposals being considered by the team. He provided updated slides including comparison examples with reference ports both present and removed for clarity. He noted that the team agrees that they are equivalent - which do we want? Arpad asked whether we have the capability to define the reference terminal of each port. He expressed a desire to "open up" the reference terminal in terms of access to permit pad-side reference, pin-side reference, etc. . Weston Beal noted the correct usage of "port" and "terminal". Walter asked whether this was already done for the first Group Port definition in the example. Weston suggested it wasn't; Randy noted that the + and - characters on a port line denote referencing. Walter stated that the name "group" links the terminals for the ports. Walter's example calls "groups" "pindefs" in contrast to Arpad's. Weston asked what happens if the port reference is a single pin. Is this a pindef? Walter affirmed this. Weston noted that this is similar to having the model name as a selector or a single-buffer definition. Randy added that EMD also does this. Weston observed that Arpad's format is more explicit with the names, while Walter's is more concise. Randy added that this is like EMD vs. SPICE syntax. Michael asked whether a full system topology definition is still separate and outside the model structures here. In other words, connections between models must still be handled by some other netlisting or connection approach outside the individual models themselves, to pair up different model port names. Walter suggested this is like an extended CAD net. For example, J1 connecting J2 between a childboard and parentboard. The format tells you exactly how to make the connections. Weston asked whether the port and net are explicitly connected in each file. Walter affirmed. Arpad noted that the port has to define how it is connected to the outside. U1.94 refers to pin 94 of U1; port 6 assigned to BGA.K2 is not telling us where the port is connected to the outside, but is the name of the EMD pin. This could be connected anywhere outside. Weston suggested that the phrase "no reference ports" should be "no reference terminals". The reference terminals need to be included in Walter's format example. Walter stated that S-parameter references are usually node zero; it's rarely otherwise. Arpad replied that ground pins in IBIS are not always shorted to node zero. Walter stated that modeling usually assumes a reference of node zero. Weston noted that this is somewhat true; the reference terminal for each port has to be close by for connectors; the text here is trying to correct a frequent error in the industry. Walter agreed, noting that there is a lot of industry guessing at whether the data uses a local reference or not. In this format, we can specify the actual reference terminal. Michael observed that Sam Chitwood documented these kinds of issues previously: in the best cases, models will be calculated to use inductive loops with respect to a single reference that may or may not be node zero. The same system could be modeled in two different equivalent ways with different values and different targeted reference nodes. Michael noted that, next time, the meeting agenda will start with port mapping first, then move to closing the reference TSIRD. Arpad moved to adjourn. Randy seconded. The meeting adjourned without objection. The next meeting will take place on March 6, 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 4) Complete pole-residue format BIRD draft (Muranyi) - COMPLETED 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.