Subject: IBIS Futures/Cookbook sucommittee minutes - May5th 2005 From: Syed Huq Date: Fri, 06 May 2005 10:48:18 -0700 To: Bob Ross , Arpad Muranyi , Lance Wang , Randy Wolf , Michael Mirmak , "John' 'Angulo" , Ian Dodd , Otto.Bennig@samtec.com, katja.koller@siemens.com, Lynne Green , shuq@cisco.com ====================================================================== IBIS FUTURES/COOKBOOK SUBCOMMITTEE MEETING MINUTES Date: May 5, 2005 Previous Meeting Attendees: ---------- Cisco Systems - Syed Huq Intel - Arpad Muranyi Mentor - John Angulo, Ian Dodd Siemens (Munich) - Katja Koller Teraspeed - Bob Ross Green Streak Programs - Lynne Green ====================================================================== 1 PM - 3 PM US Pacific Time Telephone Bridge Passcode 916-356-2663 3 122-6310 Agenda: 1 - 1:05 PM Opens 1:10 - 1:15 International Review of ICEM and IMIC 1:15 - 2:00 Cookbook - Arpad 2:00 - 2:15 BIRD94.2 2:15 - 2:20 BIRD95.5 2:30 - 2:45 BIRD97 2:45 - 3:00 Other ------------------------------------------------------------------------- Upcoming Futures Meeting: Date: Thursday, May 19, 2005 Bridge: 4 Passcode: 747-9612 ====================================================================== Futures Minutes (abbreviated discussion) Syed Huq ran the meeting since Michael Mirmak was out. Katja Koller called in and introduced herself to the group. She is involved with IBIS modeling activities at Siemens. We moved BIRD97.1 as the first item to minimize her international telephone call time. OPENS Lynne Green added an Fabless Semiconductor Association (FSA) update as as a new item. BIRD97.1 Arpad Muranyi stated that BIRD97.1 added recently requested updates. He also had private discussions with Katja Koller regarding here derivation of a generalization to the power coefficient equation (sent out to the meeting participants. k = [1 + Pc*(Vinst-Vnom)/Vnom]**a Katja stated that the power a is between 1 (long channel) and 2 (short channel). Bob Ross commented that this is nice because this information can be added as an optional sub-parameter. So linear and non-linear power coefficients and corresponding scaling is supported. Bob mentioned that he also had some formatting comments. We had a long discussion on definitions, terms, and on derivation of the coefficients. Some material needs to be added in Other Notes to clarify how the coefficients are derived or determined for a common understanding. Katja and Bob will work with Arpad next week regarding an update BIRD97.2. It can be issued before the next IBIS teleconference meeting. FSA UPDATE (New Item) Lynne Green reported on a recent meeting and commented on some presentation material sent out earlier. A new group under FSA are working on a new, consolidated modeling approach to deal with the next generation smaller devices. FSA is taking equation or table based approach and wants to include multiple effects for Signal Integrity, timing, power and so forth in transistor modeling. This impacts IBIS because the IBIS accuracy is related to source model accuracy. FSA is in the very early stages and is investigating other organizations and documents dealing with modeling. It might integrate some of the existing results. Arpad Muranyi was concerned that the group might want to invent a new equation language format rather than use a standard format (e.g., *_AMS). We discussed various aspects. This activity is distinct from that of another group involved in a SPICE quality document. Lynne mentioned that FSA meetings with about 20 people are held over lunch, and at larger conferences bring in key industrial experts for formal presentations and discussions. She attends these meetings and has presented IBIS at FSA conference. Their next regular meeting is scheduled at the Design Automation Conference (DAC). We agreed that either or both groups are welcome to present their results at the next IBIS Summit at the DAC. We would like to see some substance (such as the Quality document still under preparation and not public) and also the modeling approach. We also plan to have FSA as a regular report under International/ External Activities. INTERNATIONAL REVIEW OF ICEM AND IMIC Bob Ross introduced this subject based on an e-mail request for comments sent to the IBIS officers on IMIC and ICEM. Bob plans to introduce this topic under the International/External Activities topic at the next IBIS teleconference meeting. Ross Carlton of Freescale, who is active in EMC activities and is a representative in IEC (International Electrotechnical Commission) working groups attended a working group meeting in 2004 in Tokyo under TC47A. Bob explained that some modeling standards were processed under TC93 where there still exist language activity including language based standards such a VHDL-AMS. Michael Mirmak had attended some DASC meetings for these groups, and TC93 hosts the related international working group. IMIC and ICEM were moved to a TC47A working group under a convener from Japan. Most likely any renewal or new version of IBIS would be assigned to the working group. ICEM appears to have been standardized as IEC62014-3. A new extension to it is being proposed as a new work item. At the same time IMIC is moving forward at the letter ballot comment stage for finalization as a standard. Ross Carlton had sent out the new ICEM work item and the draft IMIC document along with some related comment documents and meeting minutes. His major request was that IEC needs USA comments form an appropriate group in the US. IBIS could be the appropriate group. Also, Bob indicated that he and the IBIS committee had worked with both groups (ICEM in France, and IMIC in Japan) and we have had participation and presentations from both groups at IBIS SUmmits. The IEC working group is considering combining IMIC and ICEM into one docuemnt. So Bob stated that we should be aware of these activities and even get involved since both documents cover areas of interest to IBIS. Bob stated that Ross would be willing to serve as our spokesperson at International meetings to represent our positions. Bob closed by stating that this will be added as a topic to our IBIS telconference meetings and possibly Michael might want to be involved with this when he returns. Also, if anyone wants to provide technical comments, they are available from Bob (they cannot be uploaded per IEC rules). (Added note, Bob and Ross both were invited kickoff speakers at EMC Compo 02 in Toulouse, France), and J.L. Levant is the author of the new ICEM proposal - see pictures. http://see.midipyrenees.free.fr/page_see/texte/manifestations/compo_2002.html ) COOKBOOK Arpad Muranyi sent out a Cookbook update AM3 with pre-emphasis/driver schedule additions, and differential C_comp discussion. Aprad pointed out the changes in the documents. We discussed some of the changes. For pre-emphasis, Arpad included both the one inverted bit delay approach and [Driver Schedule]. Bob Ross commented that [Driver Schedule] was directly supported by IBIS and therefore important to document. There are issues with both approaches. Bob and Arpad discussed the differential capacitance assumptions and related IBIS approaches. Arpad commented that IBIS was missing the corresponding C_comp_diff element in IBIS (based on single-ended structures) and used [C Series]. The problem is that this addition degrades the V-T response under differential mode, but improves the impedance description, as a trade-off. He also considered that the differential extraction might be done with an "negative capacitance" in a manner similar to the way differential resistance might be "nulled" out for single-ended constructions of differential buffers. BIRD94.2 We deferred comments on BIRD94.2. BIRD95.5 Bob Ross commented that Lance Wang had some questions to the author of the Sigrity presentation on BIRD95.1. Syed Huq forwarded the question. The question related to the extracted crowbar currents under various load conditions and the algorithms to handle variations. John Angulo commented that he had similar questions. Bob Ross commented that he did not expect an exact or highly accurate result, but thought that BIRD95.5 [Composite Current] did provide significant improvement. Arpad Muranyi commented that this might cause improvements similar to Kumar's I-T table proposal several years ago. Bob commented that the proposals were different since the [Composite Current] contained pre-driver currents (not assumed in Kumar's proposal). So the algorithms were not as direct. However, the high frequency (fast edges) of the crow-bar, pre-driver currents are of most interest for SSN analysis through package parasitics. So even if not exact, BIRD95.5 provides major improvement for such analysis.