Subject: 1076.1 Meeting Minutes
From: Ernst Christen (christen@analogy.com)
Date: Wed Mar 15 2000 - 13:46:19 PST
1076.1 WG Meeting Minutes 3/7/2000
==================================
Meeting attendees:
Tom Kazmierski (chair) tjk@ecs.soton.ac.uk
Ken Bakalar Kenneth_Bakalar@mentorg.com
David Barton dlb@wash.averstar.com
John Michael Williams jwill@pacbell.net
John Willis jwillis@ftlsys.com
Ernst Christen (note taker) christen@analogy.com
Meeting agenda:
1. Introduction
2. Group policies
3. Organizational matters
4. Study group activities
5. Other standardization activities
6. AOB
7. Next meeting
Tom opened the meeting at 1:40pm.
2. After the welcome Ernst summarized the newly updated WG policies, which
have been updated to reflect the changing focus of the working group
and to be compliant with the DASC bylaws. After review by the DASC chair
the policies have been sent to the reflector in January and are now
considered approved since no objections were raised. Tom moves to
recognize the work, John Williams seconds. The policies are recognized.
3. One of the major questions that the group has to answer is how to track
developments of the VHDL 1076 standard. After some discussion the following
ideas cristallized:
- 1076.1 should track 1076, but there is no rush to start with any
updates until a new version of 1076 has been approved.
- The conversion of the LRM to FrameMaker will make it possible
to create a single LRM for both standards using conditional text.
This will make tracking language changes much easier in the future.
David offered to investigate whether he could provide the services
to do the initial work to make this happen.
Tom then listed the various projects:
a) Standard packages
b) Mixed netlists
c) Real time subset
d) RF extensions
e) SPICE models
f) Dimensional analysis
Tom suggests to prioritize. In the discussion it was pointed out that
some prioritization has already happened since only a) through e) are
active. A project doesn't make progress unless there is a champion.
John Willis pointed out that the project he is involved in deals with
microwave extensions, not RF extensions. No decisions were made.
4. Ernst gave an overview of the Mixed Netlist issue and outlined three
different approaches. The presentation is available as a pdf file.
The discussion first focused on complete understanding of the three
approaches. Ken preferred the first, which localizes the problem such
that the behavior a single connection association element can be defined.
David had an idea about a fourth approach and promised to send it to
Ernst.
John Willis reported on VHDL-MW, the study group on microwave extensions
to VHDL-AMS. A PAR is forthcoming, proposing the development of the
following:
modeling in frequency domain (s-paramaters)
distributed elements
harmonic balance
Parasitics were dropped from the PAR. The goal is to have a single
description language for a whole design up to the 4GHz range. Ken asked
whether RF issues at lower frequencies could be included as well, where
elements can still be considered lumped. Ken has an action item to find
out whether there is a lumped RF problem besides the distributed one.
Tom summarized Eduard's document on a real time subset of VHDL-AMS that
was sent to the reflector in January. In the discussion Tom posed the
question how projects like this one could be handled by the 1076.1 WG.
They are not language extensions, and there must be enough interest to
make it its own standard. John Willis suggested the Recommended Practices
approach. With respect to Eduard's document, Ken and Ernst would like to
see a rationale for some of the decisions.
5. Ken reported that there has been no activity on standard natures, but that
he will have discussions later in the week with Paul Menchini on forming
a study group or even a working group.
John Willis reported on activities to deal with SPICE netlists. He mentioned
three major issues:
algorithm control parameters (e.g. SPICE .OPTIONS card)
The group proposed to ignore these because such parameters
are very tightly linked to the SPICE algorithms and have
nothing to do with the model to simulate
undefined values
SPICE uses the absence of a model parameter as an indication
to select a slightly different model. Default values are
not suitable, because they may be indistinguishable from
specified values. With the move of VHDL to support IEEE 754
floating point it might be possible to use NaN for this
purpose. However, VASG hasn't considered extending the
expression semantics to include IEEE Inf and NAN. John
proposed to suggest to VASG to do this. Ken and Ernst welcome
this proposal.
mutual inductance
A SPICE coupling element is not a stand-alone model and
cannot be expressed in VHDL-AMS. The approach to be taken
is to recognize transformers in the SPICE netlist and convert
them to a VHDL-AMS transformer model. There are some potential
problems with hierarchy, but it was pointed out that designs
that spread a transformer across the design hierarchy are not
even portable across different versions of SPICE.
6. No other business.
7. The next WG meeting will be held at FDL'2000.
The meeting was adjourned at 5pm.
This archive was generated by hypermail 2b28 : Wed Mar 15 2000 - 13:50:05 PST