================================================================================ 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 April 11, 2018 Meeting (* means attended at least using audio) ANSYS Curtis Clark Cadence Design Systems Bradley Brim Intel Corp. Michael Mirmak* Keysight Technologies Radek Biernacki Mentor, A Siemens Business Arpad Muranyi* Micron Technology Justin Butterfield*, Randy Wolff* SiSoft Walter Katz*, Mike LaBonte* Teraspeed Labs Bob Ross* Michael Mirmak convened the meeting. No patents were declared. Justin Butterfield took minutes. Review of Minutes: - Michael called for review of the minutes from the April 6 meeting. Bob Ross noted that he intend "node 0" to be the preferred quoted text in BIRD189 rather than node "0" as stated in the minutes. Bob moved to approve the minutes with this correction. Mike LaBonte seconded. The minutes were approved without objection. Review of ARs: - Arpad Muranyi to review the "Important" paragraphs on page 30 for inclusion in BIRD189. - Michael noted that this is done. - Walter Katz to write up some text on the two types of models. - Walter reported he sent this out, but he is in the process of revising it. And, he would like to discuss the text today. - Radek to improve the language of the first and second "Important" notes on page 30. - Michael noted that this was sent out. Opens: - None BIRD189.5_draft18_v7 Review: Walter reviewed the text he sent out over email. He suggested that models which use node 0 inside an IBIS-ISS subcircuit are not suitable for power aware simulations. Michael asked if Radek was proposing a parser message for having node 0 at the ATM meeting, and if there was any opposition to this. Arpad stated that it was not agreed upon to add a parser message, but the EDA tool could do it. Bob commented that he does not want the parser to dig into the IBIS-ISS file. Walter stated he agreed and we should not do a message at all. Walter noted the historical usage of ground is implemented as node 0. Some power aware simulations need to account for the return path currents. The reference node needs to be physically close at 1/10 of a UI wavelength or less. His intent is to bring up the fundamental issues of accounting for the return current and the reference node. Michael asked how we would put this text in BIRD189. Walter thought it should go in the beginning of the IBIS specification or in a supporting document. Bob agreed with putting the text in a supporting document, and he noted that this is talking about physical measurements and simulations. But, we may need some fine tuning of the text to clarify. Arpad noted in simulations we can do all measurements in reference to the global ground, but physical measurements must be done with the reference close. He also suggested to change the first bullet to past tense. Randy Wolff asked if this text will be in BIRD189 and if discussion can be deferred. Arpad stated that Walter wanted it in the beginning of the IBIS specification. Michael noted it could be part of another BIRD. Arpad asked how this statement is going to help us. Walter thought that this could replace Radek's warning with a more general statement. Michael stated that the issue is if you have models with different assumptions for referencing, this could give the user wrong results. He asked if we should take the approach to educate the model maker or to warn the user they may get incorrect results. Michael asked Arpad if there were any changes to his Important note text. Arpad replied that there were no changes after our last meeting. Michael shared Radek's new Warning text that was sent out. Bob stated he has some questions about the text. A minor change he suggested was to change incorrect to less accurate. Also, he is assuming that an A_gnd pin exists at the pin interface. Michael commented that there is a terminal that will be the reference. Arpad asked about the phrase "shall be treated as", and if it means we should be bring out the node 0 as a terminal or if the simulator acts like this node is there. Arpad commented that a global node is a shortcut that you do not need to connected at each level of the subcircuit hierarchy. Michael noted the options for handling node 0 are to force the model maker to explicitly change the model, add an additional terminal when the user netlists, or the math behind the scenes changes. Arpad asked if this is a warning saying that the math is not changing. Michael noted that the math may not be consistent. Bob thought it could be less accurate or the effects could be negligible. Arpad asked if we could envision a situation where node 0 should not be brought out intentionally. He wondered how we would know when to connect node 0 and when to not. Michael stated the problem is when we mix the models with different references. He asked if we should change the text to reflect the mixed reference case. Arpad noted that the mixed cases might still work okay. Michael stated that this might be true, but it depends on the case. He sees simulation issues with mixing references and inadvertent use of node 0 frequently. Michael asked if changing the phrase "shall be treated as" would resolve the issue. Bob noted there are some other changes he would like. Mike asked if the second clause is backwards. The other question is if this text is asking the netlister to do something different based whether node 0 is found. Arpad asked if this is a warning to the user or if this is a statement to the EDA tool. Randy commented that this is saying that the tool should not leave node 0 floating. Michael will ask Radek if this is a tool instruction and if we can change the phrase "shall be treated as" [AR]. Michael asked if this will resolve the three comments in the BIRD. Bob thought yes, but there are still details to discuss. Michael asked if we have a resolution on the rail rule relaxation. Arpad asked if we can remove the text associated with this, as he does not want to delay BIRD189 further. Bob stated that he would like to discuss these changes. Unlike the IO terminals there is not an assumed connection for rails. He stated he was asked by this committee to put these changes in. Arpad said the text could be separated. Bob would like to discuss it before voting on the BIRD in the Open Forum. Michael suggested that we discuss the rail rules in the next meeting. Bob agreed. Next Meeting: The next meeting will be April 18. Mike moved to adjourn. Bob seconded. The meeting adjourned without objection. Task List BIRD189.5 editorial additions/changes to be completed: 1. Resolve inconsistency with using node 0 and node "0" by replacing with "node 0".