Re: [ng-spice] Who is Mike Seningen?
On Tue, 13 Jul 1999, Mike Seningen wrote:
> That will be very useful... I know a little spanish :-),
> I believe Italians close :-)
> Third: To make sure that everything is clean. Please
> let me know when I have said something or aluded to something
> which could expose your code to information that could be
> detrimental.
I am sorry, i do not understand, what do you mean ?
> sure that my industry experience is used to help the cause
> not hurt it by divulging something I know about the commercial
> products that could be considered proprietary.
Right, I did not aked you to divulge some secrets, I asked if you know
somenthing useful based on your experience. By the way, to avoid the
thunder and lightning of TCAD companies I would like to state that:
no one should divulge commercial secrets on this list when talking
about commercial products, so before writing, check tha manuals.
> That having been said:
>
> A couple of areas that I am most familiar with are:
>
> Measurement:
> All the commercial tools have various means for measuring
> data. I've seen some do it post-processing, but most
> do it internal to the run, some even use this information
> for future iterations.
Are you talking about a program like the Silvaco's UTMOST ?
>
> The second is the output. My understanding is that there
> is only a textual output which allows you to graph in
> ASCII, or Tabular formats... All comercial tools use
> some sort of binary output. Hspice is the defacto std. (the .tr0
> format) Smartspice has emulated this as a .raw file.
> I do not know if there is a actual standard used.
I think that most commercial tools uses binary output to compact the files
size and to discourage from take the results and import them in another
program.
>
> A third area is in the use of parameters. It is my understanding that a
> berkely spice deck requires hard coded
> numbers. I assume the commercial releases use a pre-processor
> to take in parameterized netlists and internally convert to
> a hard coded deck.
Yes, spice require hard coded parameters and that's bad. There is a perl
script thet partially overcome the problem, but an extension to the parser
is needed.
>
> A fourth exploits the ability to internally measure and
> the use of parameters. Different commercial versions
> name it differently, and they appear to handle slightly
> differently but in general its called "Optimization".
> It uses various Numerical Iterative Analysis methods
> to solve for some result by being able to vary certain
> given parameters. This could be classified as a frill,
> but it is a very powerful frill.
Yeah, I would like to see a .MODIF on Berkeley's Spice.
> A fifth feature we use a lot (but I do not know if
> it is in the Berkely spice) is the Voltage/Current
> controlled sources.
Spice3f4/5 is able to simulate all the four types of controlled sources,
plus an analog behavioral model. The analog behavioral model is an
arbitrary function I/V source. A sort of simple "a" device of SmartSpice.
> The last feature I know about is usually various
> proprietary spice model tweaks that are built into
> the programs. Obviously this would not be something
> you need to support as long as you support Bsim3v3.
Yes we support bsim3v3 and in plan is: BSIM3SOI.
Now some news, by the end of the week there will be a tarball ready for
testing. I will put the tarball on the ftp archive:
ftp://ieee.ing.uniroma1.it/pub/electronics
Plase note this is a really alpha version but it will support:
1) Bsim3v3.2.2
2) Temp sweep (to be tested) by Serban Popescu.
Bye,
Paolo
PS: How is the temp. there in Houston ?
Partial thread listing:
- Re: [ng-spice] Who is Mike Seningen?, (continued)