[sdiy] Temperature stable lin-exp converter with a CA3086 or CA3046

[Ian Fritz]

At 06:26 AM 5/28/2002, Sowa Roman wrote:
>All this talk about exact compensation got me thinking.
>Does it really matter if overall VCO performance is 50
>or 100ppm/K because of not exact tempco, or other components
>temperature response? Standard crystal oscillators are specified
>to have 50ppm (not talking about those expensive 2ppm here)
>and they are the source for 'boring' DCOs. So are we going
>the right path? Someday one may say "those VCOs we have
>nowadays are so boring, DX7 - this had a drift!"

My experience is that it takes some very careful and tedious work to get 
below 200ppm drift. How much drift you can tolerate depends on how often 
you want to retune. If you just throw a tempco into a design you will most 
likely get at least 500ppm drift, more typically around 1000ppm.


>And now about Tim's idea of coreVCO. I'm thinking about
>same thing for some time now. I want to make it even more
>simple. Just expo converter, integrator, comparator, FET.
>CV with response 0.5V/oct able to cope with multiple inputs
>summed before getting into coreVCO and fitting inside +/-12V
>rails, one output (saw). Cap would be external, so one could
>use it for VeryLowLFO as well as RF VCO ;)
>All squeezed on small PCB size, say, 1x1cm. And dipped in epoxy.
>Premium components of course.
>
>Then why stop here. coreVCFs, and others, similar blocks
>with waveshapers, output buffers, input bipolar attenuators
>and we have LegoLikeSynth :)

Great ideas! I doubt that you will get this crew to agree on the designs, 
though.

   Ian