Re: [Help-gnucap] can gnucap output impedance as a function?

[al davis]

On Monday 02 April 2007 17:32, Wen wrote:
> Is there a a possibility to make gnucap output the impedance
> of a part of an electric circuit as a function of the
> frequency?
> it would output something like
> z(freq)=
> R1/(1+(R1*2*pi*freq*C1)^2)+R2/(1+(R2*2*pi*freq*C2)^2) + j*
> (-(R1)^2*2*pi*freq*C1/(1+(R1*2*pi*freq*C1)^2)-(R2)^2*2*pi*fre
>q*C2/(1+(R2*2*pi*freq*C2)^2))

> Is gnucap able to give me such output? If not, can you
> recomend another free program thats able to do that?

No.

I have thought about a way to do it, based on something like the 
AC analysis with a "polynomial" type instead of "complex".   I 
think it is fairly simple in concept except that the complexity 
of the result builds up fast.  When you look at it you see 
common expressions that can be eliminated.  That's hard to do.  
Also, when you do it manually you will drop insignificant 
terms.  That also is difficult to do.

I think what you really want is a program that will extract 
poles and zeros from a frequency response.  If you have that 
you can easily make a circuit with the same poles and zeros.  
That is standard filter design.

If you look at numerical analysis programs, you should be able 
to find one that fits a rational function of polynomials to a 
curve.  Look for the "Pade approximation technique".  I think 
programs like octave will do that.

Make a table with measurements of the real system, then extract 
the Pade coefficients.  Factor it.  Then you have the poles and 
zeros.   Build the circuit in second order sections,