================================================================================ IBIS INTERCONNECT TASK GROUP Mailing list: ibis-interconnect@freelists.org ================================================================================ Attendees from March 22, 2023 Meeting (* means attended at least using audio) ANSYS Curtis Clark Intel Corp. Michael Mirmak* Michael Brownell Keysight Technologies Ming Yan Marvell Steve Parker MathWorks Walter Katz* Micron Technology Justin Butterfield Randy Wolff* Siemens EDA Arpad Muranyi* ST Microelectronics Aurora Sanna Teraspeed Labs Bob Ross* University of Illinois Jose Schutt-Aine Zuken USA Lance Wang* ---------------------------------------------------------------- Michael Mirmak convened the meeting and took minutes. No patents were declared. Review of Minutes: - Michael called for review of the minutes from the March 15, 2023 meeting. Michael displayed the minutes. Bob Ross moved to approve the minutes. Randy Wolff seconded. The minutes were approved without objection. Review of ARs: - Michael to check if the Network Analyzer manuals documenting the original Touchstone can be posted to the IBIS website [AR]. - Michael noted that Keysight's legal department had contacted him with questions, which he answered. He suggested keeping this open until there are further updates. Opens: - Michael noted that he has not had any responses regarding his IEEE 370 team inquiries. Discussion: Arpad Muranyi posed a question on the minutes for Walter Katz, regarding the phrase "since parsers don't know the number of frequency points". Arpad looked in the specification and sees the Number of Ports and Number of Frequency Points keywords, which suggests the comment was in reference to old parsers for Touchstone 1.x, not new parsers for Touchstone 2.0. Walter responded that parsers for Touchstone 1.x have to read the entire file to find the number of frequency points, and allocate memory accordingly. Michael wondered whether this was a potential typo in the minutes. Arpad noted that he was thinking this as well. Randy clarified that it is a complex operation to support data wrapping across lines and an "open" file format in terms of port organization and arbitrary wrapping locations. Walter replied that the reference is to Touchstone 1.x. A Touchstone 1.x parser can read Touchstone 2.0 files by skipping keywords, reading the file extension, and reading in data in accordance with this. Free-form data placement makes reading in difficult, in Touchstone 2.0, with number of fields per line varying. Updating older code is also tricky. Arpad stated, as he noted in an e-mail, that there still appears to be a desire to make old parsers work with new files. He expressed concern about how long that can be made to work; at some point, the downconversion becomes impossible. Bob commented that the TSCHK2 parser performs downconversion with reformatting without looking at keywords. Arpad added that recalculating data values would be the only requirement if downconversion was needed. Walter countered that we may not need to do that; a parser could just copy the existing data and then add multiple impedance references to the options line. This would mean a minor change to existing parsers; some mathematical conversion would be needed with potential loss of accuracy. Arpad observed that the next Touchstone version could do nothing more than update the reference impedances. According to experiments with EDA software by Arpad, additional impedances are not fully supported; no error messages were generated for Siemens EDA products, but the rest of the numbers were ignored. He added that he would be OK with a Touchstone 2.1 (not 1.1); this would expand the option line to allow multiple reference impedances and legalize the existing Synopsys HSPICE syntax. We could then release Touchstone 3.0 with pole-residue, etc. features. Bob replied that he does not believe that option line reliance is the way to go, as too many existing parsers will be affected. We are talking about a vendor-specific addition to Touchstone 1.x. We already have per-port support in Touchstone 2.0. Arpad asked whether, if we allowed HSPICE syntax in the option line, it would that break old parsers. Walter replied that it would be a minor break, but easy to fix. Bob stated that it consumes resources to document changes in Touchstone; there would be other technical issues to address as well. His position is that Touchstone 1.0 is effectively fixed in place. Walter replied that impedance per port is supported by HSPICE. He expects this to be the next de facto standard Michael summarized and shared comments from Ted Mido of Synopsys, explaining the history of HSPICE's multiple impedance support; the feature was needed before the formal approval of Touchstone 2.0. Arpad noted that the values for the option line are allowed to be in any order. Can this be permitted for 1.x? Walter asked whether there was a limit to the number of characters per line. Randy replied that there was not. Michael confirmed that no specific limit is shown. Bob added that there are some format restrictions, but we assumed people would be reasonable in file reation. Bob noted that many EDA vendors release their own versions of formats before IBIS, etc. is finalized. Synopsys is one of the few vendors that does not own the parser source code for TSCHK2. Arpad asked whether HSPICE supporst arbitrary ordering of options line elements. Michael took an action to find out. [AR]. He noted for the record that he has invited Synopsys to participate directly in the IBIS Interconnect meetings. Arpad formally proposes creation of a Touchstone 2.1 with multiple impedances. Walter moved that Arpad write the TSIRD for 1.1 to support up to three features: - multiple impedances on option line - number of frequencies on option line - two port order specified on the option line The format may also identify port names with comments. Arpad asked whether this should in fact be called Touchstone 2.1. Walter accepted this alteration to the motion. Bob noted that several of these features are already in 2.0. Arpad replied that the [Reference] keyword is optional in 2.0. Bob stated that many changes would still be needed, which may create confusion in the industry. A new option line syntax would likely break existing parsers or create bogus data interpretations. He would instead prefer to close the 1.x format, and let private formats remain private. Arpad replied that reference changes can be fairly simply added; the number of frequencies and two-port order changes would demand many more document alterations. He would prefer to keep only the multiple reference impedances change in the proposal. Bob agreed, adding that a fairly large number of parser changes would be needed. Randy added that these would generally be easy. Walter agreed to modify his motion to only support multiple impedances; Bob seconded. No objections were raised. Arpad took the AR to write up the TSIRD change. [AR] Randy noted that we have a new document with new sections on Touchstone 1.x and 2.0; the resulting update will cause it to look different. Arpad replied that he was assuming we woudl be using a template; the separation document may look different. Arpad moved to adjourn. Randy seconded. The meeting adjourned. Next Meeting: The next meeting will be March 29, 2023. ================================================================================ Bin List: 1. Draft Touchstone document separating version 1.0 and 2.0 2. Encapsulating Touchstone 1.0 data 3. Touchstone 2.0 draft with TSIRD3 and TSIRD4 4. Pole-residue format 5. Port naming 6. Alternatives to the Touchstone 1.0 option line 7. Enable Cascading of S-parameters Through W-element 8. Touchstone 3.0 draft outline (dependent on several items above) Tabled ARs: - Arpad to give an example of the physical connectivity needed for EMD automation.