CORRECTED: Verilog-AMS Committee Meeting Minutes - Oct 12 2006

Martin O'leary - Cadence
Graham Helwig - ASTC
Jim Barby - University of Waterloo
Patrick O'Halloran - Tiburon
Marq Kole - NXP Semiconductors
David Miller - Freescale Semiconductor
Geoffrey Coram - Analog Devices

Continued review of Chapter 4 - Expressions. Reviewed some of the more
key rework changes identified.

Section 4.2 Math Operators - Make sure that the font size is consistent
for all the cells in the tables. Also check that the tables are aligned
and indented consistently.
No need to have the column for the C equivalent functions as the
existing Verilog-AMS syntax is the same apart from ln(), and log(). Juts
make a mention in the initial paragraph how log() and log10() differ
between Verilog-AMS and C.

Section 4.4.9 Slew - The rewording is clearer w.r.t what happens when
using slew in small signal analyses. Just add an additional sentence to
the last sentence of the last paragraph "e.g for a small signal analysis
  performed on a transient operating point."

Section 4.5.4.3 noise_table - Decided to allow the third optional
argument to noise_table(). This will be used to identify the
interpolation method used on the frequency / power pairs. We will still
clamp the frequency at the lower and upper bounds.
The third optional argument will be a string literal.
"LIN" == linear interpolation
"LOG" == logarithmic interpolation.

The argument will be case insensitive, so LIN == lin.
If the user doesn't wish to specify the name, they can use the ,, style
of NULL argument (similar to laplace and zi.).
i.e.
noise_table({f1,p1,f2,p2,f3,p3,f4,p4},,"LOG");

By default the interpolation will be linear.
Geoffrey will post the logarithmic formula that should be used.
Graham will update the BNF as necessary.

Section 4.6 - Analog User Defined functions.
Decided to allow constant analog user defined functions.
Will have the same restrictions that are applied to digital constant
UDF's (whichever ones actually apply in the analog context).
Will also restrict recursive analog udf's although the reason why this
is done in digital doesn't seem to be valid.

Will also allow recursive non-constant analog user defined functions.

Graham will make the necessary changes to the BNF to support this.

Section 4.6.2.1 should start with a capital letter, and in the first
paragraph of this section there is a 'd' all on it's own.

There was also a comment on the domain ownership of parameters.
Currently they are treated domainless, but this may need to be revisited
since we now allow constant analog and digital udf's to be assigned to
parameter initialisations. This can be covered in the Mixed Signal chapter.

Globally - Make sure that DC, AC, etc are all lower case: dc, ac. This
is an IEEE style. Also initialised should be American style: initialized.

Section 4. Analog Operators. There was a request after the meeting to
refer to them as Analog Filter Operators (chapter heading etc.). This
seems to be a reasonable compromise between the BNF: analog filter
functions and the existing LRM: analog operators. If no objections to
this, I will go ahead and make the necessary changes.

Section 4 - Expressions is now finished. For next meeting, Martin would
like to go through some changes to disciplines.vams - logic. Document
will be posted shortly.

Next meeting scheduled for Date & Time: 19 Oct 2006, 7:00am Pacific
Call-In Details:
USA Toll Free Number: 877-346-8823
USA Toll Number: +1-203-320-0407 (for intl)
Participant Passcode: 602538

Cheers...
Dave

-- 
=====================================
-- David Miller
-- Design Technology (Austin)
-- Freescale Semiconductor
-- Ph : 512 996-7377 Fax: x7755
=====================================