Re: gEDA-user: spice libs ( a little puzzled)

[John Doty <jpd@xxxxxxxxx>]

On May 3, 2010, at 9:09 AM, Kai-Martin Knaak wrote:

> On Wed, 28 Apr 2010 10:25:56 -0600, John Doty wrote:
> 
>> There's a special difficulty with SPICE libraries.
> 
> This is similar to hardware drivers for graphic Cards and the like. 
> 
> 
>> I cannot make my
>> private SPICE library available because the license terms of many of the
>> manufacturers' models contained in it forbid redistribution.
> 
> This shows, that free as in beer, but not as in freedom, is a major road 
> block in the long run.

Yep.

> 
> 
>> This isn't a problem that can be fixed easily.
> 
> Unfortunately, the only way to fix this in a future secure way is to 
> actually reinvent the Wheels in an open source way.

For simple parts it's not so bad. For opamps it's very difficult.

> 
> 
>> I did (at Kai-Martin's request) put some simple, generic opamp models in
>> my area at gedasymbols.org.
> 
> This is a great step forward to make simulation within the geda context a 
> viable alternative to the less stubborn. A simulation suite desperately 
> needs a set of generic models of standard parts. Else, potential users 
> will turn elsewhere before they become users.

The difficulty is that none of the three models I generated represents a particular real part: "opbw" and "opgain" are highly idealized (as is Al's parameterized gnucap model). "opmediocre" is close to something you could actually construct on Si, but is not (as far as I know) an accurate model of any particular standard device.

> 
> Of course, there are more essential ingredients to actually attract a 
> substantial user base. Next on the list is a workflow that is at least as 
> smooth and reliable like gschem->pcb->gerber. And of course, a complete 
> set of tutorial, examples, manual and in depth documentation. 
> 
> Are there plans to achieve these essentials with gnucap, or ngspice? Is 
> an active developer working on it? Is there a road map?

The difficulty of this is much harder than you imagine. 

But we can work on it. This weekend, I collected the subcircuit definitions for Professor Ikeda's "Open-IP" mixed-signal VLSI library. With his permission, they are now published under the GPL at http://www.gedasymbols.org/user/john_doty/models/openIP/ (you used to have to snip them out of the PDF docs). I've also made progress on symbols for them, but there's more to be done.

I'm also thinking about yet another SPICE back end for gnetlist. Stuart's "spice-sdb" works great for VLSI if you turn off the heuristics intended for simulating boards (--nomunge), and his documentation is superior, but I don't think the heuristics work very well. But the approach I have in mind will be able to handle slotting right (finally!).

John Doty              Noqsi Aerospace, Ltd.
http://www.noqsi.com/
jpd@xxxxxxxxx




_______________________________________________
geda-user mailing list
geda-user@xxxxxxxxxxxxxx
http://www.seul.org/cgi-bin/mailman/listinfo/geda-user