IBIS Macromodel Task Group Meeting date: 7 February 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: - Michael M: It would be good to discuss using AMI for single ended signals. - Bob Miller: BIRD 147.6 ------------- Review of ARs: - Fangyi to email his Flow BIRD draft to Mike LaBonte for posting. - Bob Miller to update BIRD 147.5 to draft 4 and submit it to the Open Forum. - Done. It was sent as 147.6. -------------------------- Call for patent disclosure: - None. ------------------------- Review of Meeting Minutes: - Arpad: Minutes from the 24 Jan meeting are not yet available. ------------- New Discussion: BIRD 147.6: - Bob Miller showed BIRD 147.6 - Arpad: I would like Vladimir to have a look at this. - Bob M: The main change is striking a paragraph about the EDA tool setting Ignore_Bits to BCI_Training_UI if necessary. - No one objected to the deletion. - Walter moved to submit BIRD 147.6 to the Open Forum. - Bob Ross seconded. - No one objected. IBIS 6.2: - Walter: We should restart the editorial group to produce IBIS 6.2. - We might not issue it, maybe IBIS 7.0 would be the next release. - Steve P: BIRD 189 can be in that? - Walter: Yes Filename BIRD 186.1: - Walter: Bob R. made many editorial changes to this. - It looks good. - We should decide to recommend it in our next meeting. - Arpad: Bob R. wrote that there is a conflict. - Bob R: IBIS specifies that all files are in the same directory as the .ibs file. - The BIRD says that means same directory or subdirectory. - The language needed cleanup. - Walter: Your changes address that. - Bob R: We define extension as the characters following the dot, not including it. - Walter: Bob R will be co-author. - It should be read carefully ahead of next week. Redriver Flow BIRD; - Walter: Fangyi was going to add equations similar to what Todd Westerhoff presented. - Also it should have more on requirements, like the packaging BIRD. - Arpad: I will help Fangyi with equations. AR: Arpad Muranyi to help Fangyi Rao with equations for Flow BIRD BIRD 158: - Arpad: Radek Biernacki may be working on changes to this. - Walter: We need to get closure on those. - I asked Ambrish Varma to try to get support within Cadence. - Bob R: Does this have Tx_Vol and Tx_Voh? - Walter: No. Radek suggested not doing common mode here. - His changes are mostly editorial. - Bob: It is limited to 4 port Touchstone models, that needs to be clear. - Michael M: We are excluding ISS? - Arpad: Yes. Single ended AMI: - Michael M: DDR5 has equalization, is AMI appropriate for this use? - Fangyi was concerned about Tx equalization. - Single ended buffers do not behave as differential. - Can this be corrected? - Walter: We do not need to do anything. - Bob: AMI assumes differential operation. - Walter: It assumes a single ended impulse response. - Bob: BIRD 158 assumes complementary inputs. - Walter: The analog section of the AMI model is differential. - Michael M: IBIS page 171 addresses single vs. differential models. - So the models may be different for single ended? - Walter: AMI_GetWave may be different. - Interpretation of waveform results may be different. - Arpad: Single ended can have a DC offset. - Walter: The EDA tool addresses that, not AMI. - In differential mode the waveform is symmetric around zero. - In DDR5 the waveform is symmetric around a Vref. - Arpad: The spec calls for -0.5 to +0.5 as the stimulus. - Walter: That can be relative to Vref_DQ. - Mike L: How does the tool know when a model is single ended? - Arpad: AMI assumes symmetric rise and fall, is that a problem? - Walter: For DDR5 they are symmetric. - Bob: The spec says it is LTI but vendors like to go beyond that. - Walter: We can add separate rise and fall responses but people might ask how to use it. - We should wait for model makers to ask for it. - Bob M: I generally agree with Walter. - It can be asymmetric but separate rise and fall responses may not work. - I'm not ready to propose it. - Michael M: I would like to see data about this. - Arpad: At the summit it seemed this would be a big issue. - Bob M: Fangyi was there. - Questions were asked and answered about DDR5 operation. - Randy: Will the tool determine what the DC level should be? - Walter: On-die termination will have to be determined for each channel. - At that point we will know Vref-DQ. - One question is how to display it to the user. - Arpad: Does this change during the simulation? - Randy: It has to be trained. - Walter: The JEDEC spec is still in debate. - Mike L moved to adjourn. - Bob R seconded. - Without objection he meeting adjourned. - Arpad: Thank you all for joining. ------------- Next meeting: 14 February 2017 12:00pm PT ------------- IBIS Interconnect SPICE Wish List: 1) Simulator directives