============================================================================== IBIS INTERCONNECT TASK GROUP Mailing list: ibis-interconnect@freelists.org ============================================================================== Attendees from September 9, 2026 Meeting (* means attended at least using audio) Amphenol Graham Kus* Cadence Kyle Lake* Chipletz Stephen Newberry Intel Corp. Michael Mirmak* MathWorks Walter Katz Siemens EDA Weston Beal, Arpad Muranyi*, Randy Wolff* Synopsys Ted Mido, Edna Moreno ============================================================================== Michael Mirmak called the meeting to order. No patents were declared. Michael reviewed the August 19, 2026 meeting minutes. Arpad Muranyi moved to approve; Randy Wolff seconded. The minutes were approved without objection. Michael noted that all ARs are either closed or obsolete. Michael summarized the state of Touchstone Port Mapping drafts and the challenges still remaining in completing Port Mapping and Touchstone 3.0 generally. Arpad added that he had produced significant commentary, including color-coding, on the recent drafts; Michael agreed, noting that many of these had been addressed but that some key rules needed explicit statements and explanation. Arpad showed a new, non-IBIS example he is developing. Few Touchstone examples exist for EMD in the IBIS documentation, so he is writing his own. Arpad suggested using three qualifiers in EMD, from the [EMD Model] declaration (not the Pin List): - Signal_name - Bus_label - Pin_name These would be separate parameters in the Port Mapping section, as well as part of the individual Port lines. If we have a Pin number alone, we don't have unambiguous connections back to the EMD Signal_name. A structure such as (Net Vss) could be used in this case on the Port line, for example. This gives rise to an additional rule: a pin name name or number alone should be accompanied by a Signal_name. This proposal and examples are changes from Draft 34. Example 15 is based on page 421 in IBIS, but that example is very minimalist, so it doesn't illustrate the Bus_label names, etc. very well. The new example will show what happens if you want to have a common reference terminal for all ports, not just A_gnd but an actual Signal_name. Michael asked whether this is translatable over to current IBIS. Arpad confirmed that it would be. Randy noted that, when we were talking about Signal_name_map as a parameter, we were just going to list which Bus_labels were part of a Signal_name. Now we are listing explicitly pins that are part of a Signal_name? Arpad replied that the format needs to address if you are missing the Bus_label column in EMD. You cannot have Bus_label alone in EMD. Bus_label can only exist with Signal_name. Therefore, the Bus_label parameter in Port Mapping must always have a Signal_name modifier. With no Bus_label, you can use Signal_name alone. Randy noted a typographical error: the last two Bus_label lines in the example should be U2:VDD1 and U2:VDD2. Arpad confirmed this. Randy and Arpad agreed that we only want to list Bus_label and Signal_name groupings. There are no subsets below this (in response to a question from Michael). Michael additionally asked whether these groupings would need to change to support SPIM clusters. Arpad noted that the EMD Pin_name refers to a top-level EMD pin OR could refer to a Designator pin name. He reiterated his favorite phrase, "if Bus_label is missing, Signal_name becomes the Bus_label". This is confusing to remember and should be made explicit. Michael asked whether we can we make it unambiguous by saying "Bus_label" in Touchstone 3 generates an EMD Bus_label. Arpad replied that these labels should refer to the column headers in the IBIS EMD. Names will not change going to these keywords. We aren't using the favorite phrase concept here. We are using the three qualifiers from IBIS. Randy added that Signal_name in TS3 now looks like the old Bus_label, but now Bus_label features the explicit Signal_name. There would be no overlapping pin names from IBIS. Michael suggested that Arpad's added example should be also introduced as a BIRD for EMD in IBIS. Arpad asked how we know that "Signal_name EMD: VTT (EMD.AA1)" is associated with a Signal_name except with Net. Alternatively, we could have a rule that, even if you have an explicit pin, you have to have a corresponding Signal_name. In the example Port 7 line, the "Signal_name: U1: VDD" (where VDD is a signal name, due to "Signal_name" usage) construction is explicit. If you have "Port 10 Pin_name EMD.AA1", you don't have a Signal_name. The rule becomes that you must have Signal_name for such cases. "Net VTT" would be present for the Port line and "EMD: VTT" for the Signal_name parameter. Lots of things are optional. Michael noted that, if you have Data_usage, you are committing to connectivity elsewhere. Arpad replied that then you can add a rule that makes the Net argument into the Signal_name. Net becomes required, and a new rule is needed. Net would be required when you have Pin_name in the Port definition. If Signal_name or Bus_label is used, Net is optional. This is added for automation of EMD generation. The relevant new rule becomes that Net is required with Pin_name or Bus_label (Net becomes Signal_name); Net is NOT required if Signal_name is present. The parameter Bus_label is always required when Port Bus_label is needed. The parameter Bus_label "Signal_name" column goes away with this approach. The use of *.2, etc. (wildcards) is ONLY to replace reference designators. Randy observed that the team has put a lot of automation into EMD; we may need to have something similar for IBIS Interconnect, but this would require similar level of effort. Arpad replied that he had also thought about this, and believes that the structures need for IBIS Interconnect are similar to EMD. These may easily apply, but after these rules are finished for EMD we can confirm. [AR] Arpad to send example statements and rules to the reflector for review. [AR] Michael will write up the rules for a new Port Mapping draft accordingly. Arpad moved to adjourn; Randy seconded. The meeting adjourned without objection.