IBIS Macromodel Task Group Meeting date: 10 February 2015 Members (asterisk for those attending): Altera: David Banas ANSYS: * Dan Dvorscak Curtis Clark Avago (LSI) Xingdong Dai Cadence Design Systems: * Ambrish Varma Brad Brim Kumar Keshavan Ken Willis Ericsson: Anders Ekholm IBM Steve Parker Intel: * Michael Mirmak Keysight Technologies: Fangyi Rao * Radek Biernacki Maxim Integrated Products: Hassan Rafat Mentor Graphics: * John Angulo * Arpad Muranyi Micron Technology: * Randy Wolff Justin Butterfield QLogic Corp. James Zhou Andy Joy eASIC Marc Kowalski SiSoft: Walter Katz Todd Westerhoff * Mike LaBonte Synopsys Rita Horner Teraspeed Consulting Group: Scott McMorrow Teraspeed Labs: * Bob Ross (Note: Agilent has changed to Keysight) The meeting was led by Arpad Muranyi. ------------------------------------------------------------------------ Opens: - None -------------------------- Call for patent disclosure: - None ------------- Review of ARs: - Arpad to review IBIS spec for min max issues. - In progress. ------------- New Discussion: C_comp enhancements: - Michael M showed a proposed BIRD AMI Reserved Parameters for Buffer Directionality - Arpad: This should be posted in the TAM archive. - Michael M: This adds 2 new parameters. - It is implied that only Input-only and Output-only buffers can be associated with AMI. - The tool needs to know the capabilities and the current state. - Two parameters are added: - AMI_Model_Type - AMI_Model_Direction - The parser would flag models that are I/O but are missing these keywords. - This would help detect circuits with no TX or no RX. - AMI_Model_Type has 4 values. - This will not invalidate existing models. - A single model could handle both TX and RX. - Direction can not change during a simulation. - Tools would configure state at simulation time. - For 3-state Direction can be "Ignore". - Ambrish: Can multiple models be used with this by selector? - Michael M: The tool would have to adjudicate. - Radek: What about Res Par that are specific to TX or RX? - Michael M: The parser should check for TX or RX. - For I/O it might flag missing parameters, not sure. - Radek: We have not seen those parameters mixed so far. - Michael M: We could have a table of allowed/prohibited parameters by model type. - Ambrish: Is I/O one model or two models? - Michael M: IBIS assumes there is one block. - Maybe we could have [AMI Model Selector] - Ambrish: We have run into this question before. - Michael M: We are sneaking in an enable switch as part of Reserved_Parameters. - We could have a selector to point to different AMI files. - Should a single DLL family be state-aware? - Ambrish: We would like a single model to function for both. - Mike L: The Executable line could be extended to say when each line would be used. - Michael M: Or a selector might override the existing Executable line. - Mike L: That would be better. - Arpad: Two mechanisms are possible: - One for choosing the analog model - Another to select among algorithmic models. - The first one requires no GUI. - Michael M: One question is if the selection is implicit based on something else. - Michael M: We have 3 suggested methodologies. - This draft should be tabled until we have resolved some questions. AR: Michael M send BIRD proposal to Mike L for posting. C_comp enhancements: - Randy showed Improved C_comp Case Study. - Randy: This was shown at the summit. - slide 3: - Randy: One question is if we need a series element. - slides 4 to 10: - Randy: These show VNA measured vs. simulation of the enhanced model. - slide 13: - Randy: We need to look at voltage margin. - slide 14-16: - Randy: These are different loss and drive situations. - slide 17: - Randy: This shows a need for an improved model. - Arpad: Are the models IBIS or SPICE? - Randy: IBIS for die load, SPICE for package. - Arpad: Was the V-T compensation point moved? - Randy: This was an input model, it ignores that. - The series element question is important. - Michael M: The need for series resistance occurs in real devices. - This complicates I-V model extraction. - Bob: We assume it is a fixed resistance. - Arpad: The inductor on slide 5 is a regular inductor? - Randy: An AC resistance is calculated proportional to sqrt(freq) - It really is a skin effect model. - Arpad: Is the inductor in the ISS box for C_comp? - Randy: It is independent. - Bob: We could allow W element. - Arpad: What conclusions are we making here? - Randy: We need buy-in before going forward. - Michael M: We agree on the need for this. - Arpad: Walter had a simplifying proposal. - Radek: Did that address C_comp? - Michael M: It did but it assumed compensation would not be changed. - There was disagreement about that. - Bob: Correspondence with the V-T signal has to be checked. - Model makers would have to do an extra rigorous step. - Arpad; Did Walter post his BIRD? - Mike L: It was put in the archive Dec 2. - Bob: There should be an automatic mechanism to determine when to use the new model. - Also to say when C_comp should be ignored. - We would have to look at Typ/Min/Max issues. - Walter added terminals for differential connections. - Arpad: We might consider implementing proper differential termination. - Bob: It might be a C_diff keyword, separate from C_comp. - It probably would still be averaged for de-embedding. - That complicates the problem. - Arpad: Would Bob's presentation be helpful here? - Bob: It's not about compensation, it's about K(t) adjustment. AR: Mike L post link to Randy's presentation ------------- Next meeting: 17 Feb 2015 12:00pm PT ------------- IBIS Interconnect SPICE Wish List: 1) Simulator directives