Minutes, IBIS Quality Teleconference May 20, 2003 9-10 AM EDT (6-7AM PDT). Upcoming Meetings No meeting on June 3 (week of DAC). June 10, 2003 9-10 AM EDT (6-7 AM PDT) Phone: 408-974-8478 or 800-249-5618 Meeting access: CONFERENCE SERVER 1, Conference ID: 54644 June 17, 2003 9-10 AM EDT (6-7 AM PDT) Phone: 408-974-8478 or 800-249-5618 Meeting access: CONFERENCE SERVER 1, Conference ID: 54644 --------------------------------------------- - Roll Call Adam Tambone adamt@maine.rr.com *Barry Katz bkatz@sisoft.com Benjamin P Silva benjamin.p.silva@intel.com *Bob Ross bob@teraspeed.com Brian Arsenault arsenault_brian@emc.com Eckhard Lenski Eckhard.Lenski@siemens.com Eric Brock ebrock@sisoft.com Gregory R Edlund gedlund@us.ibm.com Hazem Hegazy hazem_hegazy@mentorg.com John Figueroa jfigueroa@apple.com Katja Koller Katja.Koller@siemens.com Kevin Fisher kfisher@sisoft.com *Kim Helliwell kimgh@apple.com Lance Wang lwang@cadence.com *Lynne Green lgreen@cadence.com *Mike Labonte milabont@cisco.com Peter LaFlamme plaflamm@amcc.com *Robert Haller rhaller@sisoft.com Roy Leventhal Roy_Leventhal@3com.com Sherif Hammad sherif_hammad@mentorg.com *Todd Westerhoff twesterh@cisco.com Tom Dagostino tom@teraspeed.com Kazuyoshi Shoji kshohji@hitachi-ul.co.jp *Sadahiro Nonoyama nonoyama@hitachi-ul.co.jp Everyone in attendance marked by * Interested, but have not yet attended: Ralf Bruening Zuken ralf.bruening@zuken.de NOTE: "AR" =3D Action Required. -----------------------MINUTES --------------------------- - Review of minutes for May 6, 2003 Approved. - Announcements EIA Call for Patent Disclosure. No company declared a patent. - AR Review Bob H: IV curve reviewed from previous minutes. Kim H: Review parser for over-current limit check. (Only appears=20 to apply to clamps, parser Warning if >10 Amps in range 0V to -1.5V.) Bob H: IV Curve document updated and posted. - Opens for new items - DAC Summit and DesignCon East Summit. Barry will give Quality update at DesignCon East Summit. Kim will present Quality update at DAC Summit. AR: Kim to put together slides for the DAC Summit. - Detailed Documentation review IV_curve.txt Suggestions from previous meeting taken care of. A couple of detailed discussions were recorded in previous minutes. SPICE comparison: V-t waveforms should correlate. The transient=20 test should exercise the clamps. Companies generally do not do an I-V correlation unless problems=20 are seen in V-t curves. Usually do not check I-V correlation in the labs. Lack of clamp curves is a common problem. On-die termination thread from the SI reflector; in this case need=20 to sweep the clamps over the entire range -Vcc to +2Vcc. Since=20 simulators need to extrapolate over that entire range. Used Arpad's=20 notes on how to separate clamps. Bob R. noted that it is more correct=20 to take clamp currents that return to "ground" to the [GND Clamp] and=20 clamp currents that return to "power" to [Power Clamp], to allow=20 correct modeling of power/ground bounce. Separation is even harder if=20 termination is non-linear. It was noted that power/ground bounce analysis using IBIS models is not the best way to go, particularly for on-die terminations. s2ibis2 is known to incorrectly extract clamp currents. Documentation could show a diagram is a Thevenin structure for both clamps and terminations, and note s2ibis2 bug. Note that incorrect termination modeling might result in poor correlation in signal swings. Model should document when part has on-die termination, to alert user. Waveform_processing.doc This document should have keywords and quality levels added, and=20 separate the waveform and ramp into separate sections (separate=20 keywords). Note that EDA tools use Ramp values as estimate of how fast buffer=20 switches. This means value of dt should be within an order of=20 magnitude of expected dt under Rload conditions. (Order of Magnitude means factor of 10 or less.) Note implied Rload is 50 Ohms to opposite rail. dV should be 60% of signal swing. dt should be on the same order as Tr and Tf in data sheet for Level 0, and correlate to V-t tables for Level 1 for a similar test fixture. Note that [Ramp] should contain reasonable dt numbers for low-voltage swing parts (such as LVDS).=20 Note that both dV and dt must be positive values. s2ibis2 can generate negative values. This can happen if there are not enough points in the transition region, or the simulated model is in high-Z mode, or simulation did not finish. Vmeas does not need to be between 20% and 80% of the ramp voltage swing. Could combine bullets 8 and 10 (dt). AR Mike: update document and convert to text. - Plans and Schedule for next meeting June 10, 2003 9-10 AM EDT (6-7 AM PDT) Phone: 408-974-8478 or 800-249-5618 Meeting access: CONFERENCE SERVER 1, Conference ID: 54644 - Meeting adjourned at 10 AM EDT (7:00 AM PDT).