RE: SPICE compatibility issues

Sri,

I think your assumption is wishful thinking, far from reality...

The SPICE differences are not only syntactical, but also very
deeply technical.  I mean the model equations may be completely
different.  Even for a simple thing, such as a resistor, for
example, HSPICE has options for expressing it as a sheet
resistance, using geometric description, or the R=value,
or R='any_valid_HSPICE_syntax_expression' syntax.  I am
not sure how many other SPICE flavors can do the sheet
resistance or free expression format.  But when it comes
to the diode and transistor models, the picture gets even
worse.  For one, HSPICE has a bunch of MOSFET levels
which may not be available in all other SPICE tools (or
vice versa), but add to that the proprietary company
SPICE flavors, such as Intel's or IBM's own home grown
SPICE tools and their own highly specialized and proprietary
equations.

There is also a significant variation in the world of behavioral
capabilities, the controlled sources.  Some can do derivatives
and integrals, others can't.  Some have event triggered sources,
others don't.  And guess what, my relentless questions on the
`define and string parameters came from my desire to implement
an HSPICE compatible syntax, in which a voltage or current source
could be written this way:

 E1  n1  n2  VOL='any_valid_HSPICE_syntax_expression'
 I1  n1  n2  CUR='any_valid_HSPICE_syntax_expression'

In this syntax you can reference any node voltage, branch current,
and/or simulation variable (time, frequency, etc...).  Not too
many SPICE flavors can do that, although one may argue that this
is just a syntax feature.

Well, in short, it is a big jungle...

On the other hand we need to ask ourselves the fundamental
question:  What are we trying to achieve with the library?
I don't think we can "standardize" SPICE, but we may be
able to provide bridges of some sort if we define what we
are trying to bridge...

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



-----Original Message-----
From: Chandrasekaran Srikanth-A12788 [mailto:Srikanth.Chandrasekaran@freescale.com] 
Sent: Tuesday, July 26, 2005 3:25 PM
To: Muranyi, Arpad; verilog-ams@eda.org
Subject: RE: SPICE compatibility issues


Arpad,

I am assuming the various flavours of spice will differ not much in the behaviour but more so in the name of the primitives in the different spice implementation including the parameter names that they take in etc. If that is the case then we can hopefully standardize on a Verilog-AMS library for the spice models. The reason we had to include the tool dependency when including spice primitives inside Verilog-AMS for this reason. We didn't want the language to be seen as standardizing spice names that have already been implemented in various circuit simulators. 

But, if you think there is lot of difference in the different spice flavours apart from syntax, names and naming conventions then it might become difficult to have a standardized Verilog-AMS library. Having  multiple libraries corresponding to various flavours of spice would be very unmaintanable.

Regards,
Sri