IBIS Macromodel Task Group Meeting date: 26 July 2016 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 Cisco: Seungyong (Brian) Baek eASIC: * David Banas Marc Kowalski Ericsson: Anders Ekholm GlobalFoundries: Steve Parker IBM * Luis Armenta Intel: * Michael Mirmak Keysight Technologies: * Fangyi Rao * Radek Biernacki Ming Yan Maxim Integrated Products: Hassan Rafat Mentor Graphics: John Angulo * Arpad Muranyi Micron Technology: * Randy Wolff Justin Butterfield QLogic Corp.: James Zhou Andy Joy SiSoft: * Walter Katz Todd Westerhoff * Mike LaBonte Synopsys: Rita Horner Kevin Li Teraspeed Consulting Group: Scott McMorrow Teraspeed Labs: * Bob Ross TI: Alfred Chong The meeting was led by Arpad Muranyi. -------------------------------------------------------------------------------- Opens: - None. ------------- Review of ARs: - Randy send C_comp presentation to Mike for posting - Done -------------------------- Call for patent disclosure: - None. ------------------------- Review of Meeting Minutes: - None were available to review. ------------- New Discussion: Next Steps - Arpad: Are there any new agenda items or existing ones that should be higher priority? - Walter: I assembled a list of what should go into IBIS 6.2. - None of the items are yet approved. - Bob said we need to have some number of BIRDs to be approved. - We think we know what those BIRDs are. - Arpad: What are these BIRDs? - Walter: The list is not readily available now. - Mike: I intend to send a document about this by Friday. AR: Mike produce list of candidate BIRDs to write for IBIS 6.2 6) IO Buffer Reference Terminals - Walter: What we were calling Device In Action is now Device In Simulation. - When the simulation rail voltage is different from that tested it gets complex. - For drivers, power aware simulation adds more complexity. - Return path is another issue, that we probably can't deal with. The model maker must get that right. - IBIS can't say how to scale when the rail to rail voltage difference is not what was measured. - Bob: I support that. We should have a generic statement like the one Michael Mirmak proposed, and then Walter tuned. - Walter: We discussed IEEE P370 by email. - In their system an upside down diamond symbol was signal ground, a local ground. - The other symbols in P370 are earth and chassis grounds. - The IEEE people agree their wording could be cleaned up. - We should use the signal ground symbol in IBIS 6.2. - I will circulate their new text when they have an update. - Arpad: We should have a written definition of the meaning of their symbol. - Are all of those die grounds shorted together? - Radek: The implication is that they are shorted. - Agree we could use that symbol, replacing "GND" in our figures. - Walter: We need to put words around it. - Arpad: There is a disconnect. - V/T tables are measured from the test fixture ground to the signal. - The IBIS model may not have that ground if there is no 0 voltage terminal. - In that case the measured voltages are shifted from anything that can be measured. - Radek: We have the [* Reference] keywords for that, to give terminal voltages relative to the unavailable reference. - Arpad: Those are referenced to the test fixture ground. - Walter: Test fixture ground may not be a pin on the component. - A voltage is supplied to each rail from that reference node. - A trace on a plane layer must be imaged to some rail attached to package. - Arpad: For a bouncing device the question is what does the V/T table voltage mean? - Bob: V_Fixture is also referenced to local ground, which we assume is perfect. - Rising and Falling waveforms are also relative to that local ground. - The waveform is transformed by K table using the reference node. - In simulation the reference nodes can be modulated. - I don't envision recalculating K tables based on a changing voltage. - The IV tables however are referenced to the changing voltage. - Radek: Without V/T tables we calculate simple K table multipliers. - But we do have V/T tables, so we should accommodate that. - Bob: The V/T is used once under static conditions. - Walter: I once wrote a simulator that finds the K/T function under fixed supply conditions, then it just runs it as calculated. - Arpad: We can expect a few related BIRDs to be written and reviewed in this group. - Walter moved to remove his item 6 "IO Buffer Reference Terminal proposal" from the weekly agenda. - There was some discussion about this: - Radek: There is useful information here. - Walter: But we are limited to using just the buffer terminals. - The reference point is not always a buffer terminal. - Radek: But the reference exists somewhere. - It may be the same pin used for [Pulldown Reference], for example. - Walter: An input has just three terminals. - Arpad: I agree with Walter. - Radek: This is more of an issue for the driver. - Walter: I/V table current is a function only of the difference between I/O and reference terminals. - Radek: Where is the local ground? - Arpad: That doesn't matter. - Radek: We need to answer the question of which terminal the I/O driving against. - Mike: We seem to have part of the answer to how to supply buffers but not all of it. - Having started the discussion and feeling we understand it as well as we can, we should not now decide that the specification needs no change. - Other people will need to understand this. - Arpad: An example mother board might have a ground plane and 2 supply planes, +2.5V and -2.5V. - How do you get a connection to the 0v ground plane? - The tool will connect a complete circuit, and once the connection is completed you know the reference point. - That reference cannot be defined by the buffer, it doesn't know how it's connected. - Radek: We must know what to measure against. - Walter: It is measured relative to the rails that supply the device. - You can't reference against a node that's not on the chip. - Bob: Note that this discussion is about the Pin_Reference keyword, listed further down on the agenda. - Bob seconded the motion. - With no objection the motion passed. - Arpad: We look forward to agenda recommendations from the editorial group. - Arpad: Thank you all for joining. ------------- Next meeting: 2 August 2016 12:00pm PT ------------- IBIS Interconnect SPICE Wish List: 1) Simulator directives