[sdiy] expo accuracy? or integrator accuracy?, or both?

[Ian Fritz]

Hi Magnus --

> > The objection seems to come from soakage, not from tempco.
> >
> >
> > Fortunetaly it seems that soakage has not such a great impact
> > in vcos then on s&hs.
>
>Since it is hysteresis just look how the counterpart of the capacitor, the
>inductor work in this sense... the hysteresis causes losses, but frequency
>error? No, I don't think so.

Oh, I think it definitely could.

The origin of soakage/hysteresis is the slow dynamics of molecular or 
domain reorientation.  If there is a relaxation component in the audio 
frequency range, then you will get more reorientation (wider loops) at low 
frequencies than at high ones.  Furthermore, these processes are usually 
thermally activated, which means potential exponential increases in their 
effects with temperature.

Even worse, there can be memory effects that will cause the exact shape of 
a loops at one frequency to vary in time depending on previous history.  So 
you might not even get the correct initial pitch if you jump from low to 
high frequency.

I spent some time studying these effects in ferroelectric ceramics.  The 
time scales range from microseconds to years in this case.  Hopefully mica 
isn't so bad!

   Ian