IBIS Macromodel Task Group Meeting date: 21 November 2017 Members (asterisk for those attending): ANSYS: Dan Dvorscak Curtis Clark Broadcom (Avago): Xingdong Dai Bob Miller Cadence Design Systems: * Ambrish Varma Brad Brim Kumar Keshavan * Ken Willis eASIC: David Banas Marc Kowalski Ericsson: Anders Ekholm GlobalFoundries: Steve Parker IBM Luis Armenta Trevor Timpane Intel: * Michael Mirmak Keysight Technologies: Fangyi Rao * Radek Biernacki Ming Yan Maxim Integrated Products: Hassan Rafat Mentor, A Siemens Business: John Angulo * Arpad Muranyi Micron Technology: * Randy Wolff Justin Butterfield SiSoft: * Walter Katz Todd Westerhoff * Mike LaBonte SPISim * Wei-hsing Huang Synopsys: Rita Horner Kevin Li Teraspeed Consulting Group: Scott McMorrow Teraspeed Labs: * Bob Ross TI: Alfred Chong The meeting was led by Arpad Muranyi. Randy Wolff took the minutes. -------------------------------------------------------------------------------- Opens: - Arpad reviewed the schedule for upcoming ATM meetings. - Walter asked if ATM should vote to recommend to the Open Forum that BIRD165.1 be approved. Arpad noted he took the BIRD off the agenda, and he did not think the ATM had to give approval. Bob confirmed that the BIRD will be up for discussion at the December 5 Open Forum meeting. ------------- Review of ARs: - Arpad to submit BIRD165.1 to the Open Forum. - Done. -------------------------- Call for patent disclosure: - Ken Willis reported that Cadence had a patent granted, #9798848, relating to channel simulations of parallel interfaces. Michael Mirmak asked if the patent would be licensed under reasonable and non-discriminatory terms. Ken noted he did not have any authority to speak for legal requirements. Bob asked if there was anything in the IBIS specification impacted by the patent. Ken said he needed to research that further. Bob noted we should report the patent in the Open Forum meeting as well. ------------------------- Review of Meeting Minutes: Arpad asked if anyone had any comments or corrections. Walter moved to approve the minutes. Bob seconded the motion. There were no objections. ------------- New Discussion: BIRD189 - File_TS0 restrictions. Arpad stated he would like a goal of finishing a proposal for what to put in the BIRD. Radek noted he has not read the last minutes, but discussions should be recorded there. We might bring up an email with options related to the issue. Bob noted he and Radek may not agree entirely. One proposal is to remove File_TS0 entirely. Bob thinks we've agreed to keep it in, although some vendors may prefer not to have it. He had two possible wordings that are actionable. One is if you use File_TS0, don't mix it with File_TS models. Also, don't mix use of node 0 in IBIS-ISS. This might be a caution message. He doesn't want to check looking for node 0 in IBIS-ISS subcircuits. Arpad asked if anyone wanted a short summary of the issue. In BIRD189, there are two subparameters, File_TS and File_TS0. For File_TS0, we don't explicitly list the N+1 terminal for the reference for Touchstone files. This reference is defined and connected for File_TS connections. Radek noted this implicit connection to node 0 in File_TS0 could be dangerous. Walter added in the common case where you have a component with pins connected to the databook named ground, typically 0 volts, it isn't an issue because you use File_TS and reference the ground pin on the component. The EDA tool can float that or tie it to node 0. The big problem is our unusual case of ECL/PECL/RS232 where there is no 0 volt pin on the buffer. In this case, there is no pin or signal pin for the reference for the Touchstone file. File_TS0 was created to cover this case. Bob noted it isn't limited to that, as buffers really do work where a reference of ground is outside the buffer and the supplies to the buffer exist relative to the reference outside the buffer. Walter asked about components not having a pin for 0 volts. Bob said this doesn't matter. Many EDA tools support those configurations where the supplies are supplied relative to an outside ground and simulations are done accurately relative to an external reference. Arpad noted this seems related to discussions on C_comp too. We have the same disconnect there, where capacitors are connected to node 0. Bob noted this has a fix to split C_comp to explicit references. Bob asked if there is a disconnect where any cascaded Touchstone files will work with one having a node 0 reference and one an implicit node 0. There is no current flow from the N port to the N+1 port. Arpad asked about the suggestion to remove the File_TS0 parameter option. Would we need a mechanism to define a connection to node 0 on the N+1 terminal? Bob noted this is the same thing as the shortcut. Radek noted you still have to understand that the node connected to the N+1 terminal must be within the node span. You can't connect to a node outside the span of the defined terminals. Arpad asked if we allow the reference terminal to be connected to node 0 or not. If we have no pin suitable for the connection, what do we do then? Walter asked how it could be that you have an interconnect model that does not have a reference of a pin of the package. If you have no node 0 on a pin, then how could you have that package? If there is no current going through that N+1 terminal, then it doesn't matter. Radek said it makes the requirement that the model makers must understand. Walter said then we should add a sentence that for File_TS0, there should be no current flowing through the terminal N+1. Radek stated that this was a suggestion he had in his email. Arpad said he agreed with what Walter said. The package model is pinned out to the component pins. The ground pin could be 1000V away from a reference node, and if there is still 5V across the ground pin to the power pin, why does it matter? Radek noted his three options shared in an email to Bob and Arpad: 1. Scrap File_TS0, or 2. Allow File_TS0 under the condition that node 0 (which, for example, can be named “GND!” following the IBIS-ISS specification, or some other name) is used (either following Bob’s text, or making an equivalent statement) throughout for all interconnect paths, or 3. Allow File_TS0 and specify the restrictions on the Touchstone file numerical data needed for unambiguous simulation setup (repeating my previous text): If the same Touchstone file containing N-port data is used in an S-element with N+1 nodes than for any loading conditions (the external circuitry outside of that S-element) and at all frequencies present in the file the current entering (or leaving) the (N+1)st node has to be zero. Bob said whatever current flow rule for node 0 applies. Walter said we make a choice if we have File_TS0 or not. If we get rid of it, every Touchstone file must use a reference available as a pin of the component. Walter noted if you ask anyone writing a package model, they do it to node 0 commonly. They may also do it to VSS that is nominally 0V. Bob noted you can use an N+1 reference in these cases. Arpad asked what you do with a 1 port model. Bob noted the second terminal happens to connect to node 0. Arpad noted there would be current through node 0 in that case. Radek noted this is his second case, and the other node must exist in the component pins. Bob described the ideal transmission line model where the reference nodes at each end of the transmission line connect to node 0. Walter said we need to decide what to do, either keep File_TS0 or require File_TS to have a reference N+1 node that is an existing pin. Bob doesn't support the notion that there is a single reference for everything. Walter noted you can have complex interconnect without single references for every pin. In that case you can't use the shortcut, you have to use the Touchstone file in an IBIS-ISS subcircuit. Arpad moved to remove File_TS0. Walter seconded the motion. Bob said he saw a case where the N+1 terminal is irrelevant. Walter said if you measure one or create one from simulation of a layout, you know what the reference is, and you can define the reference without the shortcut. Arpad noted the shortcut states you can only have a single common reference for all ports. Mike LaBonte asked if there was anything in the motion to allow the last terminal to not be present with File_TS. Walter clarified that you must have an N+1 terminal that must be defined. Michael Mirmak moved to close the debate. Ken Willis seconded the motion. The vote tally was: Cadence - abstain (Ken asked if there was another committee responsible for Touchstone usage. Walter responded that there is no agreement in an IEEE measurement committee on node 0 usage. Ken asked if this was supposed to be voted on in another task group. Arpad responded that the discussion was moved to ATM) Intel - yes Keysight - yes Mentor - yes Micron - yes SiSoft - yes Teraspeed Labs - yes (Bob noted he doesn't agree with a common reference for everything.) The vote passed with 6 yes votes and one abstain vote. Arpad noted the BIRD will need to be updated to remove the parameter. The syntax and rules will need to be defined to handle the description of the reference. Arpad thanked everyone for joining the call. ------------- Next meeting: 28 November 2017 12:00pm PT ------------- IBIS Interconnect SPICE Wish List: 1) Simulator directives