IBIS Macromodel Task Group Meeting date: 12 September 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 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, A Siemens Business: John Angulo * Arpad Muranyi Micron Technology: * Randy Wolff Justin Butterfield 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. Randy Wolff took meeting minutes. -------------------------------------------------------------------------------- Opens: - None. ------------- Review of ARs: - Radek write BIRD 158.6 draft 5 as discussed. - It is not finished but BIRD is modified based on comments from 2 weeks ago. -------------------------- Call for patent disclosure: - None. ------------------------- Review of Meeting Minutes: Arpad stated there were minutes from two weeks ago by Mike to review. Dan moved to approve. Walter seconded the motion. There were no objections. Arpad stated there were minutes from last week's meeting distributed by Curtis to review. Dan moved to approve. Walter seconded the motion. There were no objections. ------------- New Discussion: BIRD158.6: Arpad noted we are mostly done with technical discussions. Radek was asked to clean up some of the drawings in the BIRD. Radek shared draft 5 of the BIRD, still a work in progress. Radek accepted all the changes form Walter's version. The document now shows only Radek's changes. The transmitter analog circuit drawing was updated based on Bob Miller's comments. Arpad asked if the decision was to label buffer or pad terminals. Radek noted we are differentiating between buffer and pad terminals in the BIRD. Radek did not think a final decision was made on naming the terminals. Radek stated that another change is no numbering of the package subcircuit terminals in the diagram on the top of page 7. A statement was added about inclusion of on-die interconnect parasitics. The Ts4file_Boundary text was removed following the removal of the keyword and associated keywords. Guidelines for using different package models in the .ibs file were added, including two options provided to the user. The last option mentions [Interconnect Model Set Selector]. Bob commented that Radek was documenting all combinations of package models existing in IBIS. Arpad noted that [Interconnect Model Set Selector] will need to change due to the latest discussions on BIRD189. Arpad asked why all these package precedence rules were being discussed here in this BIRD. This BIRD is somewhat unrelated to package models in IBIS. There are other places where this hierarchy is described in the spec. Radek responded that we are documenting how 4-port differential data must be connected properly by the user. Consistency should be expected by the tools when they use the package model from the .ibs file. He is not describing precedence rules, he is describing what the model should look like. Arpad asked for a high level overview of section (b). What is new in this section that doesn't exist in the specification yet? Radek responded that let's say you have a 50x50 matrix in the package model. The text says how you should create the package model for two pins. No text was added yet for handling the power supply pins that may be included in a new Interconnect Model. Arpad asked again if any new information on package modeling in IBIS was new. He did see some new text related to coupling in matrix models. Bob stated that for multiple channels, the coupled matrix package model could include all of the coupled channels, not just a single channel. Also, there is language defining a reference. Bob wasn't sure what the triangle symbol means in the Ts4file picture. A terminal on the buffer may not be the reference. Arpad asked if the triangle was implying the Node 0 connection. Walter said the ground symbol is the local reference point for the measurement, which is the local reference point. It is not Node 0, it would be pulldown_ref or gnd_clamp_ref for a buffer. Radek said that the "Ref" node is all the same point, but not necessarily node 0. Walter stated that if you are doing ground based simulation, then all the points are node 0. If you are doing local grounds, then the ground symbol represents local ground for those measurements. Arpad said that considering a Touchstone file that represents a circuit trace, with a common reference for all its ports, then Walter's statement can't be true because ports are many inches apart and referencing at both ends can't be the same. Walter responded that either you assume ground based simulations where they are all tied together and Node 0, or you should wrap that S-parameter and use a local reference port per port. The grounds on one side are not the same as the grounds on another side. If you use the shortcut, then it doesn't matter. Walter confirmed that there are three referencing options in IBIS-ISS. You can have no reference terminal assuming Node 0 as the reference, an N+1 terminal that is the reference, or a local reference per port, for a total of Nx2 terminals. Radek noted that the "Ref" symbols in his diagram should be the same node. Walter agreed with Arpad that the description of the package modeling options did not need to be in the BIRD. The language does not exist for [External Model], so it is not needed for this. This BIRD describes a shortcut similar to [External Model]. Bob noted we don't support package models in [External Model], but the EDA tool can bring in package information external to the [External Model] any way it wants to. Arpad said that in terms of the BIRD, this information should not be in this BIRD. If it needs to be in the specification, then it should be a separate BIRD. Bob noted in this BIRD there is the word package, and the package model is restricted in the way it hooks to the Ts4file. He thinks you could use standard language defining how to use IBIS package models. Arpad stated that section (a) talks about how to zero out package data in the IBIS file. He sees the EDA tools handling of the package model as outside this BIRD. Radek noted the power supply pins are not talked about in this BIRD. Bob noted that the package model is connected by pin numbers. Bob thought EDA vendors already have solutions for package model inclusions with IBIS-AMI. He doesn't know if this is a generic IBIS file or an outside Touchstone file. Do we get into this now, since we don't have a specific syntax for bringing in Touchstone files? Arpad asked if the triangle in between the Vp and Vn voltage sources is Node 0. Radek noted that a differential port requires three terminals, and the third node is typically Node 0. Bob concurred that this should be an ideal Node 0 voltage connection. Arpad wanted to make sure it wasn't a noisy node. Arpad noted that he thought the text on the package model options should not be in the BIRD. Walter moved to remove the text related to the package model options from the BIRD. Arpad noted the package text could be included in a separate BIRD. Arpad seconded the motion. The vote tally was: ANSYS - abstain Cadence - abstain Keysight - no Mentor - yes Micron - yes SiSoft - yes Teraspeed Labs - yes Arpad asked Radek to remove the text. Radek questioned which text should be removed. Walter noted that he wanted all the text related to package models removed. Walter asked Radek to send him the BIRD with the corrected pictures, and he would finish the editing [AR]. AR: Walter to finish editing of BIRD158.6 Draft 5. ------------- Next meeting: 19 September 2017 12:00pm PT ------------- IBIS Interconnect SPICE Wish List: 1) Simulator directives