================================================================================ 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 February 23, 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 February 21 meeting. Mike LaBonte moved to approve the minutes. Randy Wolff seconded. The minutes were approved without objection. Review of ARs: - Mike to go through old minutes to determine why the N+1 terminal was added. - Michael noted that an email sent was out about this, and we can consider it closed. - Mike to post the BIRD189.5_draft17. - Mike noted that he took BIRD189.5_draft17_v4 and made the mark-up relative to BIRD189.4 and posted it. He noted that there are no comments remaining in the document. Opens: - Michael noted that the main topic for today is the N+1 topic. - Bob Ross raised an issue related to if an Interconnect Model contains only reference terminals, it still requires two interfaces. We may need to look into relaxing that requirement. Michael asked about the N+1 issue. Bob stated we may have cases where only references are modeled. Walter Katz suggested to discuss this in detail later, and asked if Bob could create a presentation detailing the issue. - Radek Biernacki noted that he still has an issue with BIRD158.7. Michael suggested for this issue to be taken up in ATM. Arpad Muranyi agreed. Radek stated that there may need to be some relative text about BIRD158 in BIRD189. Walter suggested for Radek to draft this proposed text and submit it to the group. - Michael noted there was some discussion in IEEE P370 about some strange issues in the Touchstone parser. He will send out an email when he has more details. N+1 Terminal Discussion: Mike stated that it appears the N+1 terminal was introduced in an initial Interconnect BIRD document from Walter. He noted that there were some missing meeting minutes. Mike will look for the missing meeting minutes [AR]. Mike stated that we did discuss with Brad Brim that some tools might not have a node 0. We also did have the 2*N terminal option. We also had the option for File_TS0 which assumed a node 0 reference. He was expecting to find more concrete arguments for one direction or the other. Radek noted that he has always been in favor of having the N+1 terminal for the reference. Walter referenced Vladimir's DesignCon IBIS Summit presentation which states that there is an implied return path, and there is no current going to the N+1 terminal. The return paths are implied, and it should not fundamentally matter what the N+1 terminal is connected to. Walter commented that in S-parameter models he has seen, the comments and documentation do not state what the reference terminal should be connected to. And, in S-parameter simulations he has seen, the reference terminal is tied to node 0. Walter suggested for the model maker to hook up the N+1 terminal how they see fit, and the EDA tool can determine how to connect the reference based on the tools requirements. Arpad stated the problem with the N+1 approach is that we only have one reference for the S-parameter. If we connect all ports to that reference, then all S-parameters have to use the same reference. If we allow the N+2 approach, we can have different references across the simulation. This would avoid problems where the Tx model maker and Rx model maker connect their references in different ways. Radek noted that it is not correct that there is no current in the N+1 terminal. Arpad stated that the positive and negative terminals currents must be equal. Vladimir's presentation does not stated this is true for all cases. Radek stated that if you have a two port case with separate references the currents might not be the same. If you have a three terminal case, the current can be split among the terminals. The question is how many equations you have. If you have N terminals and N equations, you do not have a determinate system of equations. You need the N+1 terminal to have a determinate system. Walter stated that he is okay with keeping the N+1 approach. Arpad noted that it is not quite the case and is shown on slide 8 of Vladimir's presentation. If we have multiple S-parameters with different references, this will start to violate the requirement that the current in N+1 is zero. Walter stated that we can debate this, but he suggested to follow Radek's suggestion and keep the N+1 terminal. Radek noted that you need to have determinate equations for the system or the S-parameter data is useless. Michael asked if anyone disagrees with Vladimir's presentation. Radek stated that he does not disagree, but he would say it differently. Michael asked if we have disagreement about using the N+1 reference terminal in BIRD189. Arpad commented that the issue is if we have different S-parameters the references have to all be the same. Bob stated that he is okay with different references, but the issue would be how the voltages are measured. Radek noted that this only applies when the S-parameters are cascaded. You may violate the regularity requirement if they are not cascaded. Arpad stated that if you have two different references, you may have different currents in the positive and negative terminals. This is true if you have some impedance between the positive and negative terminals. Walter asked if anyone has an issue with simulating node 0. He suggested that the EDA tool should be allowed to do what makes sense in their tool. Michael asked if we want to have the N+1 terminal restriction and lift the restriction later. Arpad stated that we want to simulation power delivery with BIRD189. Walter noted that every simulator has some node 0 reference node that they reference voltages. Michael stated that he would like to have crisp clean motions about the technical direction that we need to implement for next time. He would like to come to a decision next time and hold a vote if necessary. Radek noted that he will not be able to attend on Wednesday. Michael suggested he send his position in over email. Next Meeting: The next meeting will be February 28. Mike moved to adjourn. Walter seconded. The meeting adjourned without objection. Task List BIRD189.5 editorial additions/changes to be completed: