Fitting complex functions with real fitting parameters

[CoCra]

Hi All:

I have a problem which should be easy to solve, however I cannot manage to do it.

I am trying for the first time to fit complex impedance data with equivalent circuits using mathcad programs like Complex_Fit.mcd, or Least_Squared_Error.mcd. The fitting procedure itself works, however I always obtain complex fitting parameters which does not make sense, e.g. an Ohmic resistance should always be real. I defined the guess values as reals, e.g. R=30+i*0, but the ouput is still complex. Any suggestions how to avoid complex solutions?

 

[Skogsgurra]

It sounds like you have to accept that the real world is complex!

Apart from being a good pun, it means that real world impedances are complex and if you get complex solutions, you have to use them.

 

[nbucska]

A real number e.g. 5 ohm can be represented as 5+0*J
In reality small R-s have +,large - immaginary component.

Complex result in case of real date may be due to
calculation/rounding errors.

<nbucska@pcperipherals DOT com> subj: eng-tips
read FAQ240-1032

 

[IRstuff]

Additionally, in Mathcad, you should be able to add constraints to force R to be real.

TTFN

 

[nbucska]

For a sanity check, calculate the angle -- it may be negligeable.

<nbucska@pcperipherals DOT com> subj: eng-tips
read FAQ240-1032