Re: [Xcircuit-dev] XSPICE output possible?

[Svenn A. B. <sv...@bj...>]

On Monday 27 February 2006 21:22, R. Timothy Edwards wrote:
> And if people *don't* complain regularly, then nothing ever gets done abo=
ut
> it.  Aside from the development I do for my own work, my harshest critics
> are the ones who drive the direction and pace of development.  Generally,
> though, I don't apologize for my user interface.  One assumes that all
> schematic capture programs will generate a valid netlist, so I figured I'd
> write the user interface the way I wanted to be.  Sort of like "vi" vs.
> "emacs", you either like it or you don't.  If not, there are alternatives.

Tim, I would never expect that a simple complaint from me would make you=20
change from postscript to, say XML. For single-page documentation nothing=20
beats xcircuit in get-things-done speed.
Biggest problem for me is to create symbols of black boxes with a couple of=
=20
hundred pins (mainly control busses and some analog signals) by extracting=
=20
the pin list from the spice netlist and convert to an xcircuit symbol which=
=20
can be netlisted properly. If you had an idea how to get that done with a b=
it=20
of tcl (or perl (ugh)) then I would be happy for a while.

> I addressed the problem in xcircuit version 3.6, but as I have not had
> much time to write the 3rd tutorial, many of the new features are largely
> undocumented.  The goal of version 3.6 was to do a better job of handling
> large-scale projects with lots of subcircuits.  As long as each circuit
> symbol has a property called "link", xcircuit will look for the subcircuit
> in a file named by the link (by default, the name of the symbol + ".ps").
> You can load the top-level cell of the hierarchy, and then load in the
> entire hierarchy with "File->Load Dependencies".  Each subcircuit schemat=
ic
> is a separate file, though, and so can be dropped into a latex document.

At one point I had to stop using xcircuit because of a segmentation fault=20
which I think I have reported, but the following hunts with gdb were not=20
successfull, so latey I haven=B4t compiled as often as I used to. I=B4ll ha=
ve a=20
look at it.

>
> > I discovered qucs a while ago, and that one at least has a modern user
> > interface.
>
> Looks like its strength is primarily as a simulator.  I hope they'll add
> the harmonic balance simulation capability. . . the lack of it makes SPICE
> non-competitive with SpectreRF and Eldo.

I saw its strength as a schematic capture alternative to cadence as I have =
the=20
back-end simulators needed available. The idea of integrating waveform=20
outputs into the project is something cadence should have done years ago, b=
ut=20
because of their monopoly they don=B4t care. I was thinking of learning som=
e qt=20
anyway and maybe it would be possible to separate the front from the back a=
nd=20
write a configurable netlister. Maybe I get some time someday...

Kind regards,
=2D-=20
Svenn