[Qucs-devel] Re: ASCO

[Stefan J. <st...@gr...>]

On Do, 24.11.2005, 16:20, asco developer wrote:

Hi there!

>> The program and it's usage seems straightforward.  I've not yet really
>> tested it.  My idea of it I got from your nice documentation.
>>
>> The "problem" I now also remember is the actual DE optimizer code in
>> the package which is not intended to be GPL'ed.  The people hold the
>> copyright with no indication about any license model.  This may be
>> problematicly...
>
> The only requirement the author has made is about giving him credit on
> de DE algorithm, which I did. If doubt exist I can ask him what he
> thinks about this.
>
>
>>
>> Anyway.  Have you ever tried to adapt asco to the qucs netlist
>> format?  If you ever try I could support you...
>
> The ASCO tool is netlist independent. What it requires is a coherent
> method of parameter extraction. If I'm not mistaken, at this moment it
> is not possible to:
> 1) measure a voltage at a given time
> 2) extract the Vgs/gm/Cgd... of a named transistor.
> 3) perform an fft of a specified waveform
> 4) ...

All this is in fact possible with Qucs.
  1) with transient analysis
  2) with an option of the analysis to output operating points
  3) this is a new feature in cvs
  4) what else?

> The tool proposes component values, but later on it must be able to
> see how they influenced the circuit functionality. For example, what
> happens to the output power if I increase/decrease this capacitance.
> So, Qucs must allow give him the output power.
>
> Commercial simulators implement this in a .extract (Eldo), .meas
> (Eldo/HSPICE/LTSpice) or using a script language (Spectre MDL).
> Because everyone likes to have their own format, you can imagine that
> its a mess.
>
> Although Spectre does not implement SPICE (like Qucs itself) it is
> widely used in the semiconductor industry. So, Qucs can implement its
> own format. If Qucs provide a method of obtaining a physical quantity,
> I can add support Qucs in a few days.

Cheers, Stefan.