[sdiy] uP stablized lin-exp converter

[Tim Ressel]

Yo,

I forget which wierdo suggested a uP for an expo
converter, but the more I think about it, the more
sense it makes. 

I got to thinking about all the sources of tempco in
you average expo conv, and it winds up being quite a
few parts. Some of which can't be remidied, like the
pots. So it ocurred to me that the uP tempco
stabilization scheme might be the wat to go.

This approach would require sampling the incoming CV
and generating a compensated out to sum into the
converter input. The down side is getting an A/D that
is fast enough and accurate enough to work.

But then I remembered the wierdo. Why not take it all
the way and generate the expo current out? There is
one probelm right away: a DAC is linear, so outputting
an expo curve will have resolution problems. This is
the classic probelm with linear DACs producing log
responses. The top end of the range will have many
'counts' per volt, the bottom will have not enough.
Now one could use range switching to equalize the
resolution. This approach has beeln applied
successfully. 

Now more questions pop up. First is: how fast does the
uP need to sample the incoming CV, and how fast does
the system need to respond? Is 100 updates per second
enough? Or is 1000 more like it? Second: if gain
switching is employed, there could be artifacts
produced that will translate into the oscillator
making wierd sounds. Glides would sound bad. Some kind
of sample and hold could be employed, but now things
are getting complicated.

So it looks like the uP expo converter is possible,
but needs some conceptual work.

--TR




--- Ian Fritz <ijfritz at earthlink.net> wrote:
> At 04:19 AM 5/28/2002, jhaible at debitel.net wrote:
> 
> >Very interesting - I didn't know that!
> >Which raises the question about the tempco of these
> metals approximately 
> >compensating the
> >tempco of a silicon pn junction. Is this just
> coincidence, or is there a 
> >system in it?
> 
> No special connection between the two phenomena.
> 
> >(To think of it, I don't even know if other
> semiconductors, Ge, GaAs, 
> >etc., have the same
> >tempco or not.)
> 
> Yes, if they make ideal junctions. If you go back
> and recheck the Ebers 
> Moll derivation you will se that there is no special
> use of Si parameters. 
> That stuff mostly goes into the prefactor that gets
> cancelled when you use 
> a differential pair.
> 
>    Ian
> 
> 
> 
> 
> 
> 
> >JH.
> 


__________________________________________________
Do You Yahoo!?
Yahoo! - Official partner of 2002 FIFA World Cup
http://fifaworldcup.yahoo.com