Integrated Microcomputer Division Folsom California Date: May 21, 1993 From: Donald Telian (916) 356-5029, fax -6710 Subject: Minutes from IBIS Open Forum 5/21/93 To: Attendees, list Attendees: Hyperlynx - Kellee Crisafulli Integrity Engineering - not present Intel Corporation - Will Hobbs, Stephen Peters, Gary Saunders, Arpad Muranyi, Donald Telian IntuSoft - Charles Hymowitz Meta-Software - Mei-Ling Wei MicroSim - Arthur Wong Quad-Design - Jon Powell Quantic Labs - Mike Ventham, Zhen Mu List: Intel - Randy Wilhelm Overall Summary: This document contains the minutes of the third IBIS Open Forum, held on May 21, 1993. The meeting was attended by six of the seven software simulation tool vendors who have announced support of IBIS modeling. Again, a very productive meeting. Complete agreement was reached by all attending regarding the "content" of the 1.0 IBIS ASCII template. Intel has the responsibility to correctly roll all the conclusions into a final 1.0 specification, and present at the next Open Forum. Open Issues: Various issues were brought up during the meeting. I'll use this section to track those issues, and site when they'll most likely be addressed # | Description | Owner | On Agenda When? | Done 1 | New Keyword under pullup/pulldown | Powell | During V/I disc. on 5/7 | 5/7 2 | tri-state v/i curve | GD/JP | 5/21, Quad presents | 5/21 3 | Hierarchical model/file structures | Doyle | During V/I disc. on 5/7 | 5/7 4 | How is v/i and ramp data derived | Wong | During V/I disc. on 5/7 | 5/7 5 | if & when to pass IBIS 1.0 to stds committee | Doyle | 5/21 or 6/4 | 6 | golden parser, what is it? who owns it? | Powell | 5/21, to 6/4 | 7 | format of PKG file (enhancements only) | Telian | 6/4 | 5/21 8 | thresholds for timing | Powell | 5/21, Doyle presents | 5/21 9 | buffer delay to ramp | Powell | please document | drop 10 | add examples with 1.0 release | Crisaful | Telian will do | 11 | ECL Model Extensions | GD/CM | | 12 | measurement guidelines | Crisaful | | 13 | IBIS model bulletin board | Crisaful | 6/4 | 14 | expand ramp definition for open-drain | Powell | 5/21 | 5/21 15 | sim. tech. for model vendors | Ventham | | Minutes: Announcements: Telian announced: 1. "IBIS Capable" waveforms are all in. Qualitative correlation across vendors is astounding, the best Telian has ever seen (when doing similar exercises). There are some issues with a couple of the vendors that Telian will discuss off-line during the week starting 5/24/93. More rigorous quantitative correlation will happen then too, and IBIS Capable vendors announced at the next Open Forum on 6/4. I. 5/7 Minutes review, Open time for new issues: There were no issues with the 5/7/93 minutes, and the new issues #12 through #15 were added to the table above. II. IBIS Simulation Tool Vendor Support discussion: With the fax for this meeting, Telian included a spreadsheet entitled "IBIS Simulation Tool Vendor Support". This spreadsheet will be distributed to the Intel salesforce world-wide to equip them to better assist Intel customers in selecting simulation tools that are IBIS capable. All vendors were encouraged to contact Telian with the data for any columns missing for their product. Though the spreadsheet offers a bit of a price comparison, everyone agreed that any benchmark attempt would not be fruitful. III. TEMPLATE.IBS closure (thresholds, input v/i, other...?): The following issues were identified by the group needed to reach closure at this meeting: 1. optionality of min/max columns 2. package model enhancement 3. method of deriving the ramp time for open-drain 4. input thresholds 5. input v/i curves Closure was reached on all topics, as follows: 1. Optionality of min/max columns: Issue: Various tables of data have the columns "typ", "min", and "max". "Typ" had come first, because we had previously agreed that at least a typical value must be given. We also agreed that, particularly when files are derived from device measurement only, "typ" may be the only quantity available. So the question was, what to do with the min and max columns. It is certainly desirable to have them in all cases, but optionality could make parser design complex. Conclusion: Full agreement was reached that the min and max columns should be optional, but to simplify parser development (and avoid confusion), not available data in any column must be marked with a new reserved word "NA." 2. Package Model Enhancement: Issue: The current template assumes a global range of package RLC values for an entire package. The request was for some provision to vary (or more accurately) specify the packaging data at the pin level. This is primarily desirable for larger, PGA packages, where the range of values is more substantial and influences simulation to a greater degree. Conclusion: It is now permissible to enter package data under the [Pin] keyword, on a pin by pin basis. New columns "R_pin", "L_pin", and "C_pin" are now included following "model_name". The rules for using these columns are: (1) columns are limited to 9 characters max, (2) if pin-level RLC is given, all three extra columns must be implemented (however NA can be used here too), consequently, there will be either 3 or 6 columns on each line under [Pin], (3) if RLC_pin values are not given, the values for those pins are assumed to be the default/global values specified under [Package], this is also true of any column with an NA. It was also decided that, if pin-level data is given, it does not need to include a range. Range on packaging parameters is only intended to cover numerous pins, not a range of values for a single pin. 3. Method of deriving the ramp time for open-drain driver: Issue: Jon Powell pointed out that the current methodology for determining a ramp time will not work for an open-drain pull up. Conclusion: An open-drain rise time ramp rate will be determined by (1) driving low with a 50 Ohm pullup to VCC attached to the output, (2) driving the output high, (turning off the pull-down), (3) measuring the 20-80% "dv" and "dt" as usual. This methodology will give a measure of how the pull-down turns off. 4. Input thresholds: Issue: Both Greg Doyle and Jon Powell had sent data to suggest that there should be provision (perhaps requirement) in the file to specify data such as the test load assumed by the AC specifications, as well as thresholds in addition to V_il and V_ih (e.g., hysteresis data). Having this type of information in the file greatly enhances the user's ability to perform automated flight time and signal quality measurements (typically the reason why he wants a model in the first place). Conclusion: The group decided collectively to drop support for these enhancements in 1.0. Specifically, 1.0 will include "those parameters required to build a model" and not necessarily those desired to use one. This was primarily done in the interest of getting 1.0 finalized, and will certainly be addressed in the future when time allows the issue to be addressed more carefully. 5. Input V/I Curves: Issue: There is currently no provision for input leakage (or greater, as with older TTL structures) currents. This could be addressed by adding in a simple thevenin source, or another V/I curve. Conclusion: Provision was made to include input currents (or tri-state output currents) within the diode to ground v/i curve already present. This was accomplished by expanding the permissible range on the diode to ground curve from [-POWER->GND] to [-POWER->POWER], if desired. Since many devices (such as CMOS) do not need this effect, it is still acceptable to take the negative diode curve only to GND. IV. Golden Parser Discussion: At the 4/23/93 meeting there was a short discussion on the creation of a "golden parser" to become the mechanism to prove IBIS file compatibility. At that time, Meta-Software had expressed an interest in owning and delivering that parser. The group is awaiting a response from Meta- Software affirming their desire to own the parser, otherwise there was interest in pooling resources (money) to contract a third party to write, support, and maintain it. ACTION ITEM: Meta-Software. Clearly formulate and articulate your interest in owning and driving the golden parser, and present to the IBIS Open Forum at the meeting on 6/4/93. V. IBIS ASCII Release to Standards Committee Ideas: Kellee Crisafulli from Hyperlynx reported on his findings after contacting IEEE, and finding out what is involved in turning IBIS ASCII into an IEEE standard. From IEEE staff contact Anne O'Neill, the following steps were outlined: 1. Request that an IEEE society sponsor our project by submitting a project authorization from (PAR). The IEEE suggested that the computer society might be a good choice for IBIS (they sponsor VHDL). 2. Once approved, the next step is to prepare a draft of the proposed standard and have the IBIS members approve it, (we are working on this already). The draft must meet IEEE's style guide requirements. The draft should also be captured in MS Word format. 3. The draft must then be approved by the IEEE membership. This is done by ballot by a group of about 75 members. The IBIS committee must find these members. IEEE will assist with the balloting. They do not require a consensus to approve the standard. The 75 IEEE members should represent a cross section of the IEEE that would be interested in the standard. 4. The IEEE also expressed a desire to ensure that there would not be overlap with the work the VHDL committee is doing. Typical time to create a standard is 2-3 years. However, if the standard is already drafted and approved by the working committee, the process might be completed in less than a year. IEEE contacts: Anne O'Neill (908) 562-3809, John Hines (503) 255-4448 (Chair of Design Automation Standards Subcommittee). In general, the process was much larger and more exhaustive than expected. There was little interest in pursuing this course, the general feeling being that there was enough representation involved to ensure its success as a de facto standard. This item will be discussed at future meetings. IBIS Open Forum: Minutes of 5/21/93 Meeting, Page 2 IBIS Open Forum: Minutes of 5/21/93 Meeting, Page 4 IBIS Open Forum: Minutes of 5/7/93 Meeting, Page 1 IBIS Open Forum: Minutes of 4/23/93 Meeting, Page 2