[ibis-macro] Re: Important question, please reply!

  • From: Bob Ross <bob@xxxxxxxxxxxxx>
  • To: ibis-macro@xxxxxxxxxxxxx
  • Date: Wed, 02 Nov 2005 17:11:08 -0800

Arpad:

I now support the general preference of SE for all tables
except those with time as the independent variable.

The ibischk2/3/4 parsers use these assumptions for waveform
end-point consistency.

The main benefit of this interpretation is that ibischk
catches non-zero slope {Power Clamp] tables truncated at 0 V
because they would create problems with many tools.
With some Spice to IBIS methodologies truncate
both the [Gnd Clamp] and [Power Clamp] tables at Vcc,
and the ibischk2/3/4 utility will probably
issue errors for generated on-die terminators  because the
build-in extrapolation algorithm will force double counting
of such internal elements split into two tables.

As I understand it, the macro element simply provides
the common syntax to hook into any tool.  If the tool
offers options regarding the type of extrapolation,
then the statement provides the expected interpretation
to set up the internal element.

If some tools cannot support SE, and still want to
provide the closely related macro-elements with HE
internally, then they can do so assuming the data
is defined over an extended range.  This might apply
to some SPICE tools.

As with IBIS, there is always a possibility of a wrong
interpretation of the data by the modeler or tool.  To
be conservative the general recommendation should still
be to provide data over the ranges similar to IBIS, and
not rely strictly on SE.

Bob



Muranyi, Arpad wrote:

Bob,

I have thought about that possibility myself.
However, as they say, if it can happen it will
happen, I believe we cannot enforce that the
table will always be long enough to cover all
possibilitites. And if it is not long enough,
my model has to do something about it. I don't
see a way around that in my code. It will either do HE or SE or I can even put code in
there that would stop the simulator, or hang
the system (infinite loop), but I can't do
nothing...


Arpad
===============================================

-----Original Message-----
From: ibis-macro-bounce@xxxxxxxxxxxxx [mailto:ibis-macro-bounce@xxxxxxxxxxxxx] 
On Behalf Of Bob Ross
Sent: Tuesday, November 01, 2005 4:45 PM
To: ibis-macro@xxxxxxxxxxxxx
Subject: [ibis-macro] Re: Important question, please reply!

Arpad:

After thinking about the discussion and problem, I have
some more thoughts.

Why do we need it defined outside the range?

Like with IBIS models, the burden should fall on the
model developer to provide the data over a reasonable
range of operation.  The tool using the macro library
will implement the macros according to their actual
operation.  Some tools will do HE,, others will do SE.

This would occur whether or not the macro is officially
defined for one mode and a non-compliant vendor still
uses the macro approach.

So, I would now leave it as undefined, and encourage
that the data cover a reaonable range of operation,
as defined in the IBIS Spec.   (HE of time data seems
seems to automatically occur in all tools to the
point that truncating the time data is a good
practice.)

Bob


Muranyi, Arpad wrote:


All,

I would like to get everyone's input on this question,
because we could not decide in today's meeting which
way the library should implement the PWL sources.  The
difficulty in making a decision is due the conflict of
various aspirations:  tool compatibility (everyone seems
to be doing something different, even multiple tools of
the same company are some times different), maximum
flexibility, and consistency.


1) What should the PWL sources do when the input is outside the range of the table definition? The options I can think of are:

a) Repeat the first or last points, which amounts to a
horizontal extrapolation (HE), or
b) use the first and last slopes and extrapolate with that,
which I call slope extrapolation (SE), or
c) use HE for PWL-s in which the independent axis is time,
and use SE for PWL-s in which the independent axis is voltage or current
d) add a parameter to the model and let the user choose.




2) The second question is regarding the event triggered
PWL sources. Does anyone else besides Cadence's have such PWL sources? If so, could you please describe it to me so
I could choose an implementation that matches most tools.



You response would be very important. If you feel
uncomfortable to reply publicly, please send an email to me directly. I know we have at lest 10 companies
on this list, so I would like to get at least 10
responses. (I hope I am not going to get 11 or more
different answers, though :-)


Thanks,

Arpad
==========================================================
---------------------------------------------------------------------
IBIS Macro website:  http://www.sisoft.com/ibis-macro
IBIS Macro archives: http://www.freelists.org/archives/ibis-macro
To unsubscribe send an email:
 To: ibis-macro-request@xxxxxxxxxxxxx
 Subject: unsubscribe




-- Bob Ross Teraspeed Consulting Group LLC Teraspeed Labs 121 North River Drive 13610 SW Harness Lane Narragansett, RI 02882 Beaverton, OR 97008 401-284-1827 503-430-1065 http://www.teraspeed.com 503-246-8048 Direct bob@xxxxxxxxxxxxx

Teraspeed is a registered service mark of Teraspeed Consulting Group LLC



---------------------------------------------------------------------
IBIS Macro website:  http://www.sisoft.com/ibis-macro
IBIS Macro archives: http://www.freelists.org/archives/ibis-macro
To unsubscribe send an email:
 To: ibis-macro-request@xxxxxxxxxxxxx
 Subject: unsubscribe

Other related posts: