============================================================================== IBIS INTERCONNECT TASK GROUP Mailing list: ibis-interconnect@freelists.org ================================================================================ Attendees from January 8, 2025 Meeting (* means attended at least using audio) ANSYS Curtis Clark, Wei-hsing Huang, Juliano Mologni Arista Networks Jim Antonellis* Broadcom James Church Intel Corp. Michael Mirmak*, Xiaoning Ye Keysight Technologies Ming Yan Marvell Steve Parker MathWorks Walter Katz* Micron Technology Justin Butterfield Siemens EDA Weston Beal*, Arpad Muranyi*, Randy Wolff* Simberian Yuriy Shlepnev ST Microelectronics Aurora Sanna Synopsys Ted Mido, Edna Moreno University of Illinois Jose Schutt-Aine Zuken USA Lance Wang Michael Mirmak convened the meeting. No patents were declared. Michael displayed the Dec. 18 meeting minutes. Randy Wolff moved to approve them; Arpad Muranyi seconded. The minutes were approved without objection. Michael presented a brief slide set summarizing two proposed keywords for Touchstone 3.0 to ensure clear connection to IBIS Interconnect, EMD, and other circuit formats. One of these is [Reference Configuration], which would be required and would document which of three potential reference configurations was used in the original Touchstone data: 1) Ideal node zero for all ports (N nodes in a circuit) 2) A single reference node (not ideal) for all ports (N + 1 nodes in a circuit) 3) Individual references for each port (2N nodes in a circuit) Michael stated in addition that the number of ports in today's Touchstone data format (both 1.x and 2.x) could be used with all three and are therefore ambiguous; the usage in a SPICE netlist would constitute a "wrapper" associating the ports and referencing with specific node connections. Weston Beal raised an objection to case 3, stating that he could generate a file with 2N nodes, but the data looks exactly the same. Michael confirmed that this was exactly correct - a Touchstone file today can be connected in all three ways as the format does not state how the data was created (with what reference). Arpad requested clarification on the "wrapper" terminology. Clarification was provided by Randy and Michael that this was the S-element or SPICE circuit used to instantiate and connect the Touchstone data to other elements. Arpad asked whether an ideal node zero or an independent rail could be used for the first two cases. Michael noted that ideal node zero is assumed for the first case; the problem is in distinguishing "how to use" the data vs. extraction of the data. Michael also suggested an "origin" keyword, to distinguish between Touchstone 3.0 model and measurement data, for the whole data set. Arpad suggested that the origin keyword should be required. Walter Katz suggested that one could generate a data set without measurement or simulation. Walter replied that it is likely the team will not converge on a solution; industry uses ideal node zero for all ports, for signal and power integrity. This is no constraint on Touchstone files. Yes, there is a physical point which is the reference, but the only thing this is useful for, is to ensure that two mated Touchstone files use physically "close" reference points. Walter's recommendation is that the specification say that. For those used in IBIS, independent of what N+1 is set to, one should assume that simulators for the reference port is node zero. Anything else is informative, not normative. Arpad replied that if one has an S6P file, then one needs a reference A_gnd node instead of measurement of Vss as a reference. Walter noted that the format supports it, but simulators won't do it. Circuit elements would be used for power delivery, not S-parameters. Weston suggested that the ideal node zero is a tradition, not a correct method; some simulators do provide syntax with a specific reference for each port (two terminals) and solve all the nodes, including HSPICE. The standard would need to open the approach up. Arpad added that the S-element in HSPICE supports all three methods, and has for decades. Randy asked if we have port mapping that describes the two terminals per port, do we need an extra keyword? Michael observed that trying to put the entire wrapper inside Touchstone 3.0 would be too much work, and result in ambiguous connections. Arpad, in response to Randy's suggestion, stated that if we see A_gnd everywhere, we could decipher the connections and correct usage appropriately. The port mapping proposal may not have all the right information. Randy asked, if you have all this information in the port map; what combinations are OK? Arpad replied that the parser would have to check these. Weston suggested that this was overloading, putting too much into port mapping. Michael asked whether it was proper for the data generator to tell the system designer how to use the data. Weston suggested that the information on how the data was originally created could be included, but that, right now, this is not checked, doesn't mandate anything, and doesn't require anything. Arpad replied that this could be required information rather than merely useful; the EMD or package (IBIS Interconnect) structure could use that S-parameter in only one configuration. Michael noted that per-port reference data could used with ideal node zero reference connections. Michael restated his long-distance antenna example (Clam Lake, Wisconsin ELF transmitter), with Arpad restating his cable-to-the-moon thought experiment. Weston cited the simpler case of a coaxial cable 1 km long. Walter replied that these are all academic; the reference for both ends is ideal node 0, as they collect the returned currents. Weston added we have a nice, continuous reference to hide the differences. Arpad stated that he was not sure if this new reference keyword is necessary with all cases. Walter alternately suggested that the problem could be resolved by specifying the xyz location of the probes used, where the reference is either a bus label or nodes (noted through a reference designator on the top layer). Weston noted that, in a field solver, you can set a pin group as a terminal. Walter asked how one connects a probe to a pin group You can only validly make a measurement to an xyz location. Michael asked Arpad to take AR to cross-check the port mapping proposal versus this [Reference Configuration] proposal for Touchstone 3.0. [AR] Weston moved to adjourn; Arpad seconded. The meeting adjourned without objection. The next meeting will be held January 15, 2025. ================================================================================ Bin List: 1) Complete port naming proposal (Katz et al) 2) Complete/revise Touchstone 3.0 draft outline (Mirmak) 3) Complete ISS-IRD 1 Draft - enable cascading of S-parameters through W-element (Mirmak) - TABLED Tabled ARs: - Arpad to give an example of the physical connectivity needed for EMD automation.