IBIS Macromodel Task Group Meeting date: 24 February 2009 Members (asterisk for those attending): * Adge Hawes, IBM Ambrish Varma, Cadence Design Systems * Anders Ekholm, Ericsson * Arpad Muranyi, Mentor Graphics Corp. Barry Katz, SiSoft * Bob Ross, Teraspeed Consulting Group Brad Brim, Sigrity Brad Griffin, Cadence Design Systems David Banas, Xilinx Donald Telian, consultant Doug White, Cisco Systems * Eckhard Lenski, Nokia-Siemens Networks Essaid Bensoudane, ST Microelectronics * Fangyi Rao, Agilent Ganesh Narayanaswamy, ST Micro Gang Kang, Sigrity Hemant Shah, Cadence Design Systems * Ian Dodd, Agilent * Jerry Chuang, Xilinx Joe Abler, IBM * John Angulo, Mentor Graphics John Shields, Mentor Graphics Ken Willis, Cadence Design Systems Kumar Keshavan, Sigrity Lance Wang, Cadence Design Systems Luis Boluna, Cisco Systems * Michael Mirmak, Intel Corp. * Mike LaBonte, Cisco Systems Mike Steinberger, SiSoft * Mustansir Fanaswalla, Xilinx Patrick O'Halloran, Tiburon Design Automation Paul Fernando, NCSU Pavani Jella, TI * Radek Biernacki, Agilent (EESof) * Randy Wolff, Micron Technology Ray Comeau, Cadence Design Systems Richard Mellitz, Intel Richard Ward, Texas Instruments Sam Chitwood, Sigrity Sanjeev Gupta, Agilent Shangli Wu, Cadence Design Systems Sid Singh, Extreme Networks Stephen Scearce, Cisco Systems Steve Pytel, Ansoft * Syed Huq, Cisco Systems Syed Sadeghi, ST Micro Terry Jernberg, Cadence Design Systems * Todd Westerhoff, SiSoft * Vladimir Dmitriev-Zdorov Vikas Gupta, Xilinx Vuk Borich, Agilent * Walter Katz, SiSoft Zhen Mu, Cadence Design Systems ------------------------------------------------------------------------ Opens: Arpad noted that LiveMeeting connections seem to be more troublesome when run through a VPN. -------------------------- Call for patent disclosure: - No one declared a patent. ------------- Review of ARs: - Walter: Invite IC vendors to meeting in 2 weeks using ibis-list - Arpad: Write parameter passing syntax proposal (BIRD draft) for *-AMS models in IBIS that is consistent with the parameter passing syntax of the AMI models - TBD - TBD: Propose a parameter passing syntax for the SPICE - [External ...] also? - TBD - Arpad: Review the documentation (annotation) in the macro libraries. - Deferred until a demand arises or we have nothing else to do ------------- New Discussion: Walter shared "Creating Broadband Analog Models for SerDes Applications": - This was presented by Adge at last IBIS summit - It is found on the IBIS website - http://www.vhdl.org/pub/ibis/summits/feb09/hawes.pdf - Page 3: - Analog can be modeled as IV/VT with C_comp, or with External Circuit - Page 4: - Michael Mirmak showed that frequency dependence of buffers is strong - Page 5: - We need a new model, but it should leverage existing work. - Models using s-params are interesting. - Page 7: - The driver can be an ideal source with an s4p filter model - Page 8: - We have no way to include s-param blocks in IBIS - Page 9: - Other suggestions include RC ladder, Final_Stage subckt - Page 10: - An alternative is a new TSTONEFILE IBIS keyword - Bob: What is the process for generating these s4p files? - Adge: Agilent uses standard s-param extraction tools - Create a differential model circuit into ideal loads. - Extract from 20MHz to 25GHz, linear spacing around 20MHz - IC tools are used - We can also extract Y-params using this technique - Sometimes several extractions are required: - For example, with DFE on and DFE off - Also one corner at a time - The Agilent tool extrapolates down to DC - Logarithmic files would be smaller than linear - Bob: How does one get additional input ports? - Adge: This needs to be discussed - Walter: The tool is given 4 nodes and told to extract - Bob: The input waveform has zero rise time? - Walter: In AMI, TX Init() produces this type of waveform Arpad showed TX waveforms: - The processed waveform has a DC shift relative to the input - Arpad is concerned that different vendors might process differently - Adge: The AMI spec should specify the processing more tightly - Arpad: How can this work without extracting down to DC? - Adge: Extraction can go lower than 20MHz - Todd: Knowing the resistance should be enough to generate DC - Arpad: But too many s-param models don't have DC, regardless - Todd: How many cycles do we have with a 5GHz signal into 20MHz? - Walter: There are lots of tools that work fine with existing s-params - Todd: Do we need to publish a guideline? - Michael M: There is a parallel to IBIS: - We put out only a format specification - The data can be crap - Should we have quality controls in the format? - Todd: The problem goes beyond just DC operating point - We would have to go out 500 symbols to see 20MHz effects on 5GHz - Also, differential circuits cancel out DC bias problems - Vladimir: The common mode is still important in differential signaling - Non-LTI effects affect the outcome - Adge: We do not have a good way to describe channels other than with s-params - Vladimir: But s-param has problems: - Non-causality - Insufficient resolution - Huge dynamic band makes linear spacing impractical - HSPICE only supports linear s-params - Adge: We should have quality controls for s-param models - Walter: These will be created by vendors producing serious models - Todd: We will run into hiccups bringing this to every tool - Putting this in IBIS we can add quality controls - Arpad: Some people are going beyond s-params to describe channels - Todd: IBIS AMI makes assumptions of channel characterization - Operating point is the only issue - Walter: At DesignCon Ian wanted to put passive and causal restrictions on these Tstonefile s-params Ian asked Walter to show Adge's slide 4 again: - Ian: Slide 4 shows both frequency and voltage dependence - Michael M's RC ladder proposal handles both of these - We should be able to convert between RC ladder and s-param - Also we can generate separate s-param models at different bias voltages - We should be able to support both without difficulty - Arpad: Can any simulators take multiple s-param sets? - Adge: Agilent has a coarse multi-voltage capability - But it is switched, not dynamic - Bob: Will an s-param at some mid-point voltage be good enough? - Adge: If the model scales well with voltage we can produce this - Bob: Only limited voltage ranges are needed - Adge: The discussion is about using s-params in IBIS, not finding a new model - Bob: A model may be a good frequency domain model, but not voltage domain - Todd: Linear links will work fine with this - A different analysis will be used to study bias points - We will not be able to run enough bits with that analysis, though - Mike L: Maybe frequency domain Golden Waveforms would help insure quality - Michael M: We would need good specs on how to do that - Mike L: It should help to at least verify DC accuracy - Arpad: Who would be willing to start a BIRD for IBIS s-params? AR: Todd start a BIRD for IBIS s-params Next meeting: 03 March 2009 12:00pm PT -----------