[SI-LIST] Re: Draft Touchstone 2.0 document available

Michael,

One philosophy I have (weakly founded) is that many
(most?) of the existing/legacy tools run into problems when
reading in the existing Touchstone files.  For example, blank
lines and comment fields often cause many tools to misread
the data or completely fail.

Extending this experience a bit further, I suspect these same
legacy tools will have further difficulties when trying to
read 2.0 files.  The "message" back to the user is generallly
ungraceful, such as "bad input file" or total crash, but
there's some (minimal) comfort in the idea that legacy tools
will not mis-read the new 2.0 format I propose simply because
of the existance of the new features/keywords inherent in 2.0
will cause the tools to choke.

That said, this philosophy is flawed, in that there are
many tools that do deal with uncommon/nonstandard features just fine
(e.g., they generally ignore non-standard lines),=20
and there is some hurdle in having industry change over to
a new ordering/format. =20

I do not have sufficient exposure to code development to understand
the industry-wide impact of my suggestion, so I look to others to
fill in that data.  However, at the level of engineer-in-the-field,
it would be very helpful to have a consistent ordering.

One option proposed by Bob Ross makes sense, where the 2.0 format
offers backward compatibility through a new field.  e.g., [X12] or
[X21] per his note.  I'm open to alternative forms, but I can
envision a field that does two things:

        * When a legacy tool that is not upgraded to handle 2.0 format
        tries to read it in, the keyword or line format is uncommon/odd
        enough that the tool chokes (hopefully gracefully).  Even
        when ungraceful, it will give the user an indication that there
        is something wrong with the input file.

        * When a 2.0-compatible tool reads in the file, it recognizes
        the keywork and handles it appropriately.

One challenge with supporting two forms of the same file is the
extra burden it places on code developers during the transition period.
Again, my exposure to such issues/headaches is limited, so I look
to others for input/data.

In the end, industry will survive just fine with either format/option,
and we'll adapt as necessary.  My hope is simply to ease the burden
in terms of overall-effort and number of mistakes.  I wish I had some
concrete data on the subject or a poll from tool "developers" (e.g.,
simulation tools, measurement equipment vendors, in-the-field
hackers like me, ...), but I'm too far from such data to offer
anything solid.

Thanks much for your consideration,
Pat

> -----Original Message-----
> From: Mirmak, Michael [mailto:michael.mirmak@xxxxxxxxx]=20
> Sent: Sunday, December 23, 2007 12:47 PM
> To: Zabinski, Patrick; si-list@xxxxxxxxxxxxx
> Subject: RE: [SI-LIST] Re: Draft Touchstone 2.0 document available
>=20
> Patrick,
>=20
> Thank you for the kind comments and suggestions!
>=20
> The specification is now being considered by the IBIS Open Forum for
> standardization and your request was discussed at our most recent
> meeting, this past Friday.  We can certainly consider making=20
> this change
> in version 2.0 files.  It would not be a difficult adjustment=20
> to make to
> the draft and would make the format more consistent.
>=20
> One concern is simply whether changing the order for version 2.0 files
> would increase confusion in switching to the new format, particularly
> when converting existing files to 2.0 would be convenient. =20
> We tried to
> navigate a middle path between making radical improvements in 2.0 and
> keeping existing Touchstone files, tools and methods almost unchanged.
> In addition to maintaining backward compatibility, we elected to
> maintain most of the existing network data format, in order=20
> to make the
> 2.0 transition reasonably easy.  The exception is the removal of
> normalization, which was required to support per-port impedances.
>=20
> So, in your opinion (and in that of the SI-List readers) would
> correcting the ordering of S12, S21 in 2.0 files outweigh the=20
> resulting
> inconsistency with the previous ordering and existing files, tools,
> etc.?
>=20
> Thanks again for your comments, and also thanks to those who provided
> comments off-line.  Anyone interested in the topic is=20
> certainly welcome
> to attend our Open Forum teleconferences and participate in the
> Touchstone 2.0 discussions.
>=20
> - Michael Mirmak
>   Intel Corp.
>   Chair, IBIS Open Forum
>=20
>   http://www.eigroup.org/ibis/
>   http://www.eda.org/ibis/
>=20
> -----Original Message-----
> From: Zabinski, Patrick [mailto:zabinski.patrick@xxxxxxxx]=20
> Sent: Sunday, December 16, 2007 7:33 AM
> To: Mirmak, Michael; si-list@xxxxxxxxxxxxx
> Subject: RE: [SI-LIST] Re: Draft Touchstone 2.0 document available
>=20
> Michael,
>=20
> Thank you and the IBIS Open Forum for supporting and updating
> the Touchstone Spec.  The spec has become a de facto industry
> standard, and it is in much need of updating.
>=20
> Overall, I really like the improvements suggested in Draft Version 2.
> They should enhance the utility of the file format and ease
> tool development that supports the format.
>=20
> I spoke with a few folks about the draft, and we would appreciate
> consideration of one change.  Under Network Parameter Data, the
> ordering of the S-parameter matrix elements are consistent for
> most - but not all - matrix sizes.
>=20
> More specifically, the 1-, 3-, 4-, 5-, ... port matrices are
> presented in Ascending-Row then Ascending-Column order, which
> matches most (all?) common matrix representations.
>=20
> In contrast, the 2-port matrix breaks this order, in that S21
> (Row 2) is listed before S12 (Row 1). =20
>=20
> Although we can generally adapt to the 2-port matrices having their
> own unique element ordering, it has historically caused many
> issues for several of us.  Several analysis and measurement tools
> do not follow this ordering, and many engineers are unaware of
> the uncommon ordering.
>=20
> If at all possible, we would much appreciate the 2-port matrix
> element ordering to comply with the rest of the spec.  By doing so,
> it would make Touchstone format consistent with many
> commercially-available
> tools and make it easier (i.e., less error prone) for folks to
> develop tools around it.
>=20
> Much appreciate your consideration.
>=20
> Pat
>=20
> >=20
> >=20
> > -----Original Message-----
> > From: Mirmak, Michael=3D20
> > Sent: Wednesday, December 12, 2007 5:26 PM
> > To: si-list@xxxxxxxxxxxxx
> > Subject: Draft Touchstone 2.0 document available
> >=20
> >=20
> > The IBIS Open Forum has completed a draft of an improved Touchstone
> > specification, entitled Touchstone 2.0, to expand the=20
> capabilities of
> > the original format and eventually release it as an=20
> industry standard.
> > Some of the changes in the new version include:
> >=20
> > - support for independent per-port reference impedances
> > - the number of ports that may be represented is unlimited
> > - support for lower- or upper-half matrix representations, to reduce
> > file size
> > - the version number, the number of ports and the number of=20
> > frequencies
> > in any file have all been made explicit
> >=20
> > Backward compatibility with 1.0 models has been preserved.
> >=20
> > Touchstone 2.0 will be formally introduced at this week's IBIS Open
> > Forum teleconference, for review and comment in preparation for
> > standardization.
> >=20
> > The latest review draft may be obtained at the URL below.  =20
> > Comments are
> > welcome.
> >=20
> > http://www.eda.org/ibis/docs/touchstone2_review_draft.pdf=3D20
> >=20
> > - Michael Mirmak
> >   Intel Corp.=3D20
> >   Chair, IBIS Open Forum
> >=20
> >   http://www.eigroup.org/ibis/
> >   http://www.eda.org/ibis/
> >=20
>=20
------------------------------------------------------------------
To unsubscribe from si-list:
si-list-request@xxxxxxxxxxxxx with 'unsubscribe' in the Subject field

or to administer your membership from a web page, go to:
http://www.freelists.org/webpage/si-list

For help:
si-list-request@xxxxxxxxxxxxx with 'help' in the Subject field


List technical documents are available at:
                http://www.si-list.net

List archives are viewable at:     
                http://www.freelists.org/archives/si-list
or at our remote archives:
                http://groups.yahoo.com/group/si-list/messages
Old (prior to June 6, 2001) list archives are viewable at:
                http://www.qsl.net/wb6tpu

Other related posts: