IBIS Macromodel Task Group Meeting date: 24 January 2017 Members (asterisk for those attending): ANSYS: Dan Dvorscak * Curtis Clark Broadcom (Avago): Xingdong Dai * Bob Miller Cadence Design Systems: * Ambrish Varma Brad Brim Kumar Keshavan Ken Willis Cisco: Seungyong (Brian) Baek eASIC: David Banas Marc Kowalski Ericsson: Anders Ekholm GlobalFoundries: Steve Parker IBM Luis Armenta Trevor Timpane Intel: * Michael Mirmak Keysight Technologies: Fangyi Rao * Radek Biernacki Ming Yan Maxim Integrated Products: Hassan Rafat Mentor Graphics: John Angulo * Arpad Muranyi Micron Technology: * Randy Wolff Justin Butterfield QLogic Corp.: James Zhou Andy Joy SiSoft: * Walter Katz Todd Westerhoff * Mike LaBonte Synopsys: Rita Horner Kevin Li Teraspeed Consulting Group: Scott McMorrow Teraspeed Labs: * Bob Ross TI: Alfred Chong The meeting was led by Arpad Muranyi. -------------------------------------------------------------------------------- Opens: - Arpad noted that we will cancel the meeting on Jan 31st because it conflicts with DesignCon. ------------- Review of ARs: - Fangyi to email his Flow BIRD draft to Mike LaBonte for posting. - Done. - Bob Miller to update BIRD 147.5 to draft 4 and submit it to the Open Forum. - Done. -------------------------- Call for patent disclosure: - None. ------------------------- Review of Meeting Minutes: - Two sets of minutes needed review, as the January 10th minutes were not prepared and posted in time for the January 17th meeting. - January 10th minutes: - Arpad: Does anyone have any comments or corrections? [none] - Michael M.: Motion to approve the minutes. - Walter: Second. - Arpad: Anyone opposed? [none] - January 17th minutes: - Arpad: Does anyone have any comments or corrections? [none] - Michael M.: Motion to approve the minutes. - Walter: Second. - Arpad: Anyone opposed? [none] ------------- New Discussion: New Redriver flow BIRD: - Discussion: Walter noted that he had asked Fangyi to use some of the newer functional notation to detail the IR math that describes the flow. He said he felt it would be clearer and make the various flows easier to understand. Bob Ross asked if Fangyi's proposal would be a new BIRD competing with BIRD 166. Walter said it might replace the contents of BIRD 166. Walter said it was definitely a 7.0 feature (not for 6.2). Arpad said that he would contact Fangyi offline to see if he could help with editorial changes and other issues in order to speed up the process. Bob Ross and Radek's discussion of Editorial Task Group Issues: - Radek: Nothing new to report. - We've had no new meetings and probably won't until after DesignCon. - Arpad: Anything we can do here to help? - Bob R.: It's currently still between Radek and me. - Do we add zero, one, or two new keywords? - Do we define three distinct references or not? - Radek: We haven't really talked about [Pin Reference] yet. - [Local Reference] has nice features and is a good enhancement. - We've not yet agreed 100% on the implications. - We've agreed on its use for C_comp and [Package Model]. - Bob R.: I'm not inclined to add two more optional keywords. - IBIS already works. - We do distinguish references under different situations. - Walter: I sent out an email on this topic last week. - I got some flak for using "local ground" instead of "signal reference", but I believe the email's contents are substantive. - It breaks things down into three cases: - 1. If any of the [* Reference] keywords have value zero, then the corresponding *_ref terminals for all keywords with value zero are considered shorted together and serve as the local reference or signal reference for node voltage measurements. - 2. No [* Reference] keyword has the value zero, but there is some other [Pin] on the [Component] that is '0'. That [Pin] can be the reference. That's what IBIS assumes today. - 3. This is the only murky case. No [Pin] is identified as the the zero volt pin, and all the [* Reference] values are non-zero. In this case IBIS requires you to rely on the simulator global ground (node 0). - These rules tell you what the reference should be in all cases. - Radek: That's what I've been saying all along. - If all the [* Reference] values are non-zero, and we have no [Pin] defining the reference, then we need that node provided in the model. - As we've discussed, it's possible that node 0 is a legitimate node in the circuit, and it's up to the user to hook up this reference node to node zero or not. The reference node is actually floating. - Walter: I think my short description is good enough for 6.2 "ground" cleanup. - I agree with Bob in the sense that what you're proposing would require changes in EDA tool functionality. - I think adding the explicit reference node could be a separate BIRD for 7.0. - Discussion: Michael M. commented that if a new reference parameter were added then its "value" should be zero. Arpad asked why it was so important to find the node that was "zero"? Michael M. said we had painted ourselves into a corner years ago, when we defined the [* Reference] keywords as having values instead of being nodes. He said we really needed to declare a point (node) that was not associated with a value but defined the local reference point. Walter noted that the issue is dealing with what happens when people say they are doing "floating ground" simulations, all measurements are relative to this local reference node. Michael M. noted that if this new reference parameter had a value other than zero, then we have implicitly smuggled in some other reference point. Arpad noted that his interpretation of the [* Reference] keywords had always been that they simply defined the reference value for the I/V keyword lookup, but those locations could have their own external DUT reference. For example [GND Clamp Reference] could be -5V. Walter and Radek asked, -5V relative to what? Walter noted that the -5V value could be used to map the threshold and I/V table values. However, at DUT time there was some point on the device that served as the 0V reference. Walter noted that Bob had provided some unique PECL and ECL examples in which none of the [* Reference] values were zero, and no [Pin] on the component provided 0V. He noted that these rare cases were the only ones that fell through to the murky 3rd case of his rules. In these rare cases, we would need to refer to the mythical "node 0". Bob R. noted that the [Local Reference] keyword he and Radek had been considering was like the others and would not have required the value to be 0. Radek noted that this had been allowed so that, for example, if you had an ECL example and you wanted C_comp connected to the pc_ref, then it could be done with this keyword. Walter noted that this could be done with the C_comp_xxx keywords anyway, and said that he thought it wasn't worth spending so much time on the ECL/PECL/MECL cases. Bob R. noted that he thought we might be starting to converge on not needing to add any new keywords. But he noted that he still considered Walter's three rules to be too restrictive. Michael M. noted that if Bob and Radek's keyword discussions would need to appear as a BIRD for 7.0 anyway, then we could table this discussion. Michael M. moved to table this discussion. Walter seconded. No one opposed. BIRD 186.1 File Naming Rules: - Walter: I proposed some wording changes so that we could add one up front statement as opposed to modifying every location in the document where a file name is referenced. - If Bob R. is happy with them then that's great, if not then what's left is merely a bunch of editing. - Bob R.: I did my homework and sent out an email listing all the affected locations in the specification. My position is that we should do all the individual editing. - I'm willing to make the editorial changes and produce a new draft after DesignCon. - Walter's approach was one global statement that said reference to a file can also be a reference to a directory path. - I want to: 1. Explicitly change every location where the original intent was "in the same directory as the .ibs file" but we are now allowing "in a directory defined by the path inside the file name." 2. Clean up and clarify that file extension is always without the "." (ibs, pkg, ami, etc.) - Walter: Let's give Bob the AR to make those edits. - Bob R.: Okay. - Is this targeted for 7.0? - Walter: It's targeted for the same revision that includes BIRD 147. - Michael M.: Motion to adjourn. - Radek: Second. - Arpad: Thank you all for joining. AR: Arpad to contact Fangyi to see about helping with the Redriver Flow BIRD. AR: Bob Ross to update BIRD 186.1 with his proposed edits. ------------- Next meeting: 07 February 2017 12:00pm PT ------------- IBIS Interconnect SPICE Wish List: 1) Simulator directives