======================================================================
IBIS INTERCONNECT TASK GROUP
http://www.ibis.org/interconnect_wip/  
Mailing list: ibis-interconnect@freelists.org  
Archives at http://www.freelists.org/archive/ibis-interconn/  
======================================================================

Attendees from February 24 Meeting (* means attended at least using audio)
ANSYS                              Curtis Clark*
Cadence Design Systems             Bradley Brim*
Cisco                              David Siadat
Intel Corp.                        Michael Mirmak*
Keysight Technologies              Radek Biernacki*
Mentor Graphics                    Arpad Muranyi*
Micron Technology                  Justin Butterfield, Randy Wolff*
SAE ITC                            Maureen Lemankiewicz, Logen Johnson
Signal Integrity Software          Walter Katz*, Mike LaBonte*
Teraspeed Labs                     Bob Ross*
University of Aveiro in Portugal   Wael Dghais

Michael Mirmak convened the meeting.  No patents were declared.   Randy 
Wolff moved to approve the minutes from the February 17 meeting.  Curtis 
Clark seconded the motion.  The minutes were approved without objections.

Bob Ross added to agenda topic: a request to develop a final 
terminology consensus, particularly on Terminal_type and case 
sensitivity.  

Arpad Muranyi shared an S-parameter interconnect model ports/terminals 
presentation, with particular focus on the phrase, “the 
Number_of_terminals entry in the Interconnect Model shall be an integer 
equal to N+1.”  This sentence from the original document resolves 
several of Arpad’s original concerns, but has changed in recent proposal 
text.  Radek Biernacki commented that he still has a concern regarding 
this sentence, as it imposes a connection to the pin or signal where 
a connection has a POWER or GND type.  Imagine a traditional RLC package 
using R & L in series, with C=0; there, you have a two-terminal 
connection between the pin and pad without power or ground connection.  
For this kind of connection, we should probably not impose this rule.   

Industry presentations and diagrams often omit reference terminals.  
Walter Katz noted that wrapping these in IBIS-ISS models would be 
suitable.  

Arpad noted that Walter’s latest updates described N nodes and N 
terminals.  

Brad Brim replied that ports and references are two different things, 
and asked about the objectives of clarifying this in the proposal.  
Starting with slide 4’s material and ignoring text, six terminals are 
shown: VDD, DATA, VSS, on the pad side, with the same repeated on the 
pin side.  Slide 7 shows what people deliver as 4-port S-parameters, 
but where are the power and ground?  In that slide, Data_P and Data_N 
are shown both on the pad side and the pin side.  Without explicit 
power and ground, how are power and ground connections made between 
board and die?  When coding EDA software, does one code a short 
between VDD on left and right?  Should assume S-parameter connects the 
two?  Arpad replied that he would like to have non-ideal connection 
between left and right.

Walter noted that there *is* a connection already, but it is 
independent and not coupled to the network shown.   Brad replied that 
VSS parasitics tend to get put into every signal net rather than being 
made separate, if VSS, etc. are not included.  Parasitics become part 
of signal line.

Radek noted that Arpad’s slide 6 has an issue: it is not true that the 
only pins considered here are the black box pins.  He outlined an 
alternative approach, with only N ports, and N+1 terminals.  Bob 
suggested that IBIS-ISS supports a global-ground version or an N+1 
reference or a reference for each port.  

Walter suggested the team is converging on solution.  He added text to 
the proposal that the EDA tool shall choose the reference node and how 
it should be connected.  Two types of model makers exist: those like 
Brad, who want all the connections stated, and we want to support that; 
on the other hand, there are users  who don’t care about power delivery 
and we should not burden them.

Brad replied that Walter is assuming DC is ideal throughout the entire 
system, while Radek is not making that assumption.  What each Radek 
and Walter each are saying is correct.

Arpad will resolve this through update text for next time via discussion 
offline.  The issue is N+1 or N nodes and how referencing is to be 
addressed.  

Walter moved to adjourn the meeting.   Radek seconded the motion.  The 
meeting adjourned.