Re: IBIS tools

Hi Raghu,

At 12:20 PM 10/24/97 -0700, Raghu wrote:
>With all this talk about SPICE and IBIS, I felt compelled to say
>something on this issue. Also SPICE algorithms are well known and the
>results are better understood.

I don't wish to get into a war about SPICE problems, but your above
statement is simply not true. The transistor behavior models in SPICE
packages vary all over the place. For example I personally know
of two very well known SPICE packages both simulating the exact same
SPICE model and getting slew rates that where more than 2x different.
The problem was chased down to differences in the transistor behavioral
models used by the two SPICE packages. These are apparently
NOT well known algorithms. This is not an unusual problem with SPICE
IC models contrary to the propaganda. If someone creates a SPICE model and
runs it on the exact same companies SPICE simulator the transistor models
do match very well. If the transistor model is from one SPICE company
and the simulator from another than the matching erodes.
Often people attempt to translate transistor parameters from one package
to another by hand because the simulations will not run and again the
transistor model accuracy is gone. There is a very serious lack of standards
in the SPICE community regarding transistor models. (My opinion) And I
have used HSPICE, PSPICE, Intusoft SPICE, Berkley SPICE, and Contec SPICE,
AMI SPICE, and DR SPICE.

I suggest we drop this thread for now and refocus
on the purpose of this group (to develop and promote IBIS models).

 
Received on Fri Oct 24 14:37:43 1997