Spice2 to spice3 library converter

"sppp" seems dead-ended, no googling gets me anywhere.

Is it true that spice3 won't run POLY?

...Jim Thompson

Hello!!! I'm looking for program which can be use to conversion POLY directive from spice2 to spice3. (may by linux version??)

-- *** Krzysztof Piecuch *** kris snipped-for-privacy@poczta.onet.pl GG 7084611 *** Linux User #132375 ***

W artykule Jim Thompson napisa³(a):

It is OK, but ...

where (in 1998 year):

I'm look for this program (sppp) or other. The Author didn't answer me.

W artykule Jim Thompson napisa³(a):

spice3f5 for linux

-- **** Krzysztof Piecuch -- SP9RQA **** snipped-for-privacy@radio.org.pl **********Linux User #132375*********

Krzysztof Piecuch wrote: : Hello!!! : I'm looking for program which can be use to conversion POLY directive : from spice2 to spice3. : (may by linux version??)

For Linux, use ngspice:

Download the sources (either rework or rework-devel are OK), and configure using --enable-xspice. That is, build like this:

../configure --enable-xspice --with-readline make make install

Ngspice with XSpice enabled will properly handle netlists with SPICE 2 POLY constructs.

There is a SPICE on Linux HOWTO at:

And this HOWTO talks about ngspice here:

(scroll down a ways to get to ngspice.)

Good Luck!

Stuart

Why don't you write a program that works on a PC ?:-)

...Jim Thompson

W artykule Stuart Brorson napisa³(a):

Thanks for your answer.

I can not compile this program correctly. Is somebody which has similar problem (debian-3.0, gcc version 2.95.4 20011002).

In file included from ../../../src/include/memory.h:20, from ipc.c:80: .../../../src/include/../misc/stringutil.h:28: conflicting types for `index' /usr/include/string.h:252: previous declaration of `index' .../../../src/include/../misc/stringutil.h:29: conflicting types for `rindex' /usr/include/string.h:255: previous declaration of `rindex' make[4]: *** [ipc.o] Error 1

MacSpice

can translate spice2 poly into spice3 B sources rather well. If you don't have access to a Mac just collect all the polys into a fake netlist email it me and I'll send you back a translated version. There is obviously a limit to the number of times I'll do this before I get bored...

Charles

Krzysztof Piecuch wrote: : W artykule Stuart Brorson napisa?(a): :> :> ./configure --enable-xspice --with-readline :> make :> make install : I can not compile this program correctly. Is somebody which has similar : problem (debian-3.0, gcc version 2.95.4 20011002).

: In file included from ../../../src/include/memory.h:20, : from ipc.c:80: : ../../../src/include/../misc/stringutil.h:28: : conflicting types for `index' : /usr/include/string.h:252: previous declaration of `index' : ../../../src/include/../misc/stringutil.h:29: : conflicting types for `rindex' : /usr/include/string.h:255: previous declaration of `rindex' : make[4]: *** [ipc.o] Error 1 :

Hmmmm. . . . not good. Can you please e-mail me your config.log file? E-mail to: sdb (AT) cloud9 (DOT) net.

I will look to see if the variables HAVE_STRCHR or HAVE_INDEX are defined.

Also, what version of glibc do you have?

Stuart

Because I don't own a PC :-)

Charles.

Hi,

I ran into the spice3 not handling POLY() statements issue.. looked for a converter but couldn't find one so I wrote one. It works on a very simple op-amp model, but is still _very_ untested, so watch out.

This gave me considerably more trouble to write than I thought it would. Best way to handle the poly() is almost certainly to compile ngspice with the right options. But if anyone can use a (probably still very buggy) converter... here it is:

spice2to3-2005-03-14.tar.gz on

BTW, anyone know of a good Eagle netlist to Spice format converter? Any interest in one?

regards,

Alex

Alex Faveluke alex at dee en el en kay dot com

Krzysztof Piecuch wrote:

directive

Truly strange. EVEN PSpice can read POLY extensions.

...Jim Thompson

Jim,

I've always had the impression that MicroSIM invented the POLY() syntax because it was easier to parse and differentiate then general syntax like V=3*V(x)*I(E1). I see POLY() popularly accredited to SPICE2 in usenet, but the string "POLY" doesn't seem to exist in the the FORTRAN code of 2g6.

Nowadays, even the PSpice doc's frowns on POLY() syntax as obsolete.

--Mike

POLY CAN certainly be confusing.

You sure about POLY not in 2G6? I sure seem to remember using it.

...Jim Thompson

Yep, the MicroSim folks hated that function. It always seemed to invite convergence errors and difficulties in setting up the matrix properly. But, I think it was one of those things that they couldn't really 're-code' from the original...

Jim,

Opps, you're right. Faulty memory. I just grep'ed the 2g6 source and now I do see the function evpoly(result,itype,lcoef,ncoef,larg, narg,lexp) in the file evpoly.f

--Mike

IIRC Charlie IS one of the original MicroSim folks.

And some of the originals are involved in...

...Jim Thompson

Charlie,

I thought you told me you didn't know the original MicroSim folks, one founder of which I used to retain. No matter. Anyway, the POLY() syntax is much better wrt to convergence then the alternative, arbitrary arithmetic expressions. The advantage is that the POLY() syntax limits you to a polynomial with all positive powers. This is everywhere differentiable and is only singular at infinite input. The arbitrary arithmetic expressions can have negative powers and various functions that aren't differentiable everywhere. The reason that the POLY() is obsolete is that it's a terrible syntax for the user, not for the convenience of the simulator author.

--Mike

Jim,

I used to retain one of the original MicroSim founders. Charlie tried to guess which one and said he hardly know so-and-so -- that that predated his association. That's what I based the first sentence on. Anyway, POLY() is much easier to deal with with respect to convergence then arbitrary arithmetic expressions like

E1 1 2 VALUE { if(V(x) < 1, 1/V(x), 1/V(x)/V(x)) }

--Mike