Tables in IBIS
--------------
Material from Arpad Muranyi, July 4, 2010


We can discuss the tables that may be
parsed by this new capability we are working on.

http://eda.org/pub/ibis/ver5.0/ver5_0.pdf

The primary interest should revolve around the [Pullup], [Pulldown]
[GND Clamp], POWER Clamp], [Rising Waveform] and [Falling Waveform]
keywords (pg. 55 and 71).  There are a few other keywords which use
tables, but the above are the most fundamental to IBIS models.

These tables follow the same general format.

The first column is the independent variable, voltage for the I-V tables
and time for the waveform tables.  The first column is followed by three
columns, current for the I-V tables, and voltage for the waveform
tables.

The three columns represent three corner cases, typ. min. and max.  Only
the typ column is required.  The min and max columns may be all "NA"
(=not available).  If there are values in the columns, the first and
last position must not have "NA", but any other location may have "NA".  The
points are expected to be interpolated with a straight line (linearly).
The extrapolation for the I-V curves should continue with the slope that
is between the last two points, but for the waveform tables the
extrapolation should be done horizontally (repeat the last point).

The I-V tables are limited to 100 points and the waveform tables are
limited to 1000 point in the specification.

The tables can be found as follows:
1)  File name (***.ibs)
2)  [Component] name
3)  [Model] name
    - optionally [Pin] name which points to [Model Selector] or [Model]
4)  name of I-V table [Pullup], [Pulldown], etc...
    or
    name of waveform table [Rising Waveform], [Falling Waveform]

The problem with the waveform tables is that there can be multiple
tables within a model for [Rising Waveform] or [Falling Waveform] but they have
no identifying names by which they can be selected.  They only have a
few subparameters by which they may be identified.  These are R_fixture,
V_fixture (these two are required) and L_fixture, C_fixture (optional).
But even with these, it is hard to search for a specific waveform table
because the values of these subparameters are not predetermined or
predictable.  An index number might work best, but in such a way that
the subparameters would also be made available with the table so that
these would also be available when processing  because these parameters
are also needed for the processing.  I will try to explain this in one
of the meetings when you are addressing this topic.