Minutes of the Combined IBIS Connector/Futures Group Meeting  4/4

Agenda
------
Roll Call                         5min      Peters

Lynne Green
Arpad Muranyi
John Angulo
Stephen Peters
Bob Ross

Minutes and A/R Review            5min      Peters
  -- review of minutes

  No corrections

 *AR Gus, release latest update with [Cn Number of Conductors] removed,
     clarification text added to [Derivation Method] keyword and Arpad's
     latest write-up included. [Not Done] 
 *AR Stephen, review sub-parameter and argument capitalization for 
     IBIS vs. spec [Not Done]
 *AR Gus, explain method whereby pins on both sides of a connector are
mapped
     to the rows of an s-parameter matrix.  Give us a syntax that shows
     how an S-parameter matrix is mapped to the pins of a connector.
[Pending]
 *AR Lynne, quarry tool developers and see if the current spec contains
enough info to
     cascaded s-parameter matrixes (touchstone format). [Pending]
 *AR Arpad, Final write-up of Chapter 7 [Done].

Opens For New Issues              5min      All

None.

Old Business                      60min     Peters/Arpad
 -- Touchstone Spec status
    No update from last week.  Stephen is still waiting to hear back
from
    Agilent contact.

 -- Close on even-odd number of nodes questions
    Stephen read minutes from last week.  Arpad says that in totally
general
    case, an odd numbers of nodes is allowable.  Arpad noted an
underling assumption
    with general nodal case wrt RLGC rows.  Series elements L11 go from
a single
    goes-ina port to a single goes-outa port.   This requires an even
    number of rows.  However, assume a 2 input, 3 output topology.  The
way
    to express this is put series elements from goes-ina port 1 to all
three
    goes-outa ports -- but there will be NO L33 term (this does not
exist).
    L12 & L13 account for coupling between line 1 and 2&3.  Therefore
there
    will be an RLGC matrix with an odd number of rows.  We agree that
this
    is a legit case, but in the larger context do we really want to
allow 
    these weird matrixes?  There is a workaround for the "T" or "Y" 
    topology case -- describe it using three RLGC matrixes.  A topology
that
    has a different number of goes-ina and goes-outa can be handled by
adding
    extra nodes.  After discussion we agreed to the following rules:

      There must be an even number of nodes in a node list when using an
RLGC 
      matrix to describe a section.  There will be half as many rows in
an RLGC
      matrix as there are nodes.  There can be even or odd number of
nodes in
      a node list when dealing with s-parameter matrixes.
      Further, we will adopt this convention:

      pair-of-nodes => row in an RLGC matrix
      each node     => row in an s-parameter matrix.

    
 -- FINAL review of chapter 7
    First, we reviewed and answered the questions that were at the top
of the 
    write-up:
    for items 1 and 2, we settled on [Tree Path Descriription]
    for Item 4, the document will use "RLGC"
    for item 5, the document will use double quotes where appropriate
    for item 7, The last sentence of the first paragraph under Usage
Rules should
    be changed to:

       "This keyword MUST be present if number of rows in the 
        matrix does not equal the number of pin names as
        given by the [ICM Pin Map] keyword used for this model."

    Also, just eliminate reference to [Number of Conductors] in the
following
    paragraph.

    For items 8, 9 and 10, we agreed that the node name length will be
20,
    the allowable characters are the alphanumeric character set
    (lower case a thru z, upper case A thru Z, digits 1 thru 9, and
underscore),
    and this will be documented directly in the [Nodal Path Description]
area.

    There was a long discussion on the usage/meaning of the Mated,
Unmated_Side_A
    and Unmated_Side_B subparameters under the [ICM Model List] keyword.
Stephen
    emphasized that these subparameters impart information to the user
on how the
    model is to be used, and are not really simulation data.  No
consensus reached
    on if any wording needs to be changed.

    Agreed to strike the following sentence from paragraph 8 of the
ICM_Model_Type
    subparmeter description under the [Begin ICM Model] keyword.

      "The connection pins used as the return 
       path should be specified as "GND" in the matrix."
  

  We will do more review next week.
New Business                      45min     All
 -- Start review of section 8

Next Meeting Plans
 We will meet again next Thursday, April 11.  Stephen will supply the
number.