====================================================================== IBIS INTERCONNECT TASK GROUP http://www.ibis.org/interconnect_wip/ Mailing list: ibis-interconnect@freelists.org Archives at http://www.freelists.org/archive/ibis-interconn/ ====================================================================== Attendees from December 7 Meeting (* means attended at least using audio) ANSYS Curtis Clark Cadence Design Systems Bradley Brim Cisco David Siadat Intel Corp. Michael Mirmak* Keysight Technologies Radek Biernacki, Ming Yan Mentor Graphics Arpad Muranyi* Micron Technology Justin Butterfield*, Randy Wolff* SAE ITC Maureen Lemankiewicz, Logen Johnson Signal Integrity Software Walter Katz*, Mike LaBonte Teraspeed Labs Bob Ross* University of Aveiro in Portugal Wael Dghais Michael Mirmak convened the meeting. No patents were declared. Review of Minutes: - Michael called for review of the minutes from the November 30 meeting. Walter moved to approve the minutes. Arpad seconded. The minutes were approved without objection. Review of ARs: - Michael to summarize existing rules of precedence in the IBIS specification. - Michael reported he may have some update on this in the next meeting. Opens: - None Next Meeting Dates: Walter made a motion to have meetings as usual on December 14 and January 4, 2017, and cancel the meetings on December 21 and December 28. Bob seconded. The motion carried without objection. Mike Labonte has volunteered, via email, to take minutes for the December 14 meeting. Justin will take minutes for the January 4, 2017 meeting. Draft 43 Introductory Text: Michael presented a draft copy of the Chapter 12 introductory text with some proposed organizational changes. He suggests to label the first part of the text as "Introduction" and include up through the first three pictures in this section. He broke out a new section called "General Interconnect Model Syntax" with remainder of the introductory text. He commented that these are suggested changes, and he has not sent this copy out yet. Bob is concerned about using terms in the introduction that have not been defined yet. Michael stated that he did add some term definitions to the beginning of the introduction to address this. Bob asked about the term "Package pins". Michael noted that we do use "Package pins" elsewhere in the IBIS specification as well as "Component pins". Bob suggested to call them just "pins". Bob favors to introduce the three different interfaces as pins, pads, and buffers. Michael noted the word "interface" is not used in the Introduction, but the concept is introduced. Michael stated he is looking for feedback on the approach of having a basic introduction and separate general syntax section. If we choose this approach, then we can get into the details of the text. Bob asked about the difference and distinction between pads and buffers. Michael noted that we have three physical points, consisting of pins, pads and buffers. The rest of IBIS does need scrubbed for this point of confusion. Bob commented that existing IBIS assumes that the pad to buffer connections are shorts or ideal connections, and we can clarify in this section. Arpad added that one point of confusion is that in the IBIS specification we use "pad" when we really mean "buffer". Bob is concerned that we need to be consistent in the Introduction. Michael suggested that we also need a simple picture or diagram to show the interfaces in the Introduction. Incomplete Model Paths discussion: Michael thought that the way Bob's proposed text for [Interconnect Model Set]s is written, technically there is no disagreement. There is agreement that it is up to the EDA to connect these models. The sets are allowed to be incomplete, but the text recommends against this. Bob noted there are two different incomplete cases where you have incomplete pins or incomplete connections from pin to buffer. Walter asked if there is a difference when the model maker provides a model that is incomplete. Bob noted that incomplete cases are difficult to check. Michael showed a draft diagram of how the connections might look. Bob commented that in the case of S-parameters with VSS as the reference, there is no through path for VSS. Michael noted the more common scenario would be separate models for pin to pads and pads to buffer. He asked Randy how common this would be. Randy replied that different teams might create the separate models, but he envisions providing sets with complete paths to avoid any confusion. Bob agreed that separate interconnect models would be created, but the model maker should combine the models in a way that makes sense. Michael asked if that is something that can be a rule or is something that EDA tool needs to check with user input. Bob noted that only the EDA tool could match up terminals to check the connections, but this can not be checked in the IBIS parser. Walter commented that the most common scenario would be using VDD only based power deliver model, separate models for the IO signals and VSS as the reference. Randy agreed that this is his most common case. Bob recommends that each set goes from pin to buffer. Walter stated that we need to support the most common case. He commented that the EDA tool only needs to know is which sets the user needs. He has a simple algorithm to combine these models whether incomplete or not. Michael commented that the algorithm could go into the text and EDA tools could use the algorithm, but it would not be a requirement. Walter will send out his algorithm to the reflector. Bob asked about full path sets and how EDA tools would handle cases with multiple paths. Walter commented that if there are redundant pins, then it is up to the model maker to document which models to use in which cases. Michael suggested to combine the document draft with new diagrams, Bob's proposed text, Walter's algorithm, and the new introduction organization. Walter suggested to add some diagrams with some split rail cases. Arpad asked about the phrase "buffer pads" and if we should use "buffer terminals" instead. Walter agreed with the phrase "buffer terminals". Michael cautioned that we may have to use something else in a few cases before terminals is defined. Walter moved to adjourn. Arpad seconded. The meeting adjourned without objection.