Vocoder design

[Scott Bernardi]

I've placed the formulas on the website, http://members.home.net/sbernardi/elec/vocoder/vocoder_design.html, under the "bandpass calculations" link.
Actually, what I did to get my values was to write a computer program that took desired parameters of fc (1st stage 1/2 octave below desired fc, 2nd stage 1/2 octave above desired fc), Av (1 for 1st
stage, 2.4 for 2nd stage), and Q of 6. Then assuming standard capacitance values, it calculated the "actual" resistor values. Then it found the nearest 5% resistor values, and recalculated the Av,
Q, and fc with these, then the % error from the design values. Lastest I calculated a weighted error. All this was printed out in a big table (see the "table" link), and I picked the row that had
the lowest weighted error. I did one table for each of the bandpass channels.
Unfortunately, I've lost the source code to the program. If you're interested, I could scan the whole set of tables. They use cap multipliers  of 1, 12, 15, 18, 22, 27, 33, 39, 47, 56, 68, 82.

jorgen.bergfors at idg.se wrote:

> >One thing I was going to try is alternate the polarity of the inputs on each channel so the responses between the channels
> >would add, and therefore flatten out the overall response.
>
> Yes, that's a common trick among loudspeaker builders.
> In the vocoder, you could feed every other channel from a non-inverting buffer and the rest from an inverting one.
>
> By the way, I would like to have the formulas for calculating the resistors/capacitors, if possible.
> I'm thinking of designing a PCB for the vocoder. But one problem is that may favourite capacitors are only available in E6-values.
> Is it possible to use only E6 capacitors and just change the resistor values accordingly? Otherwise I'll have to provide space for two parallell capacitors everywhere, to get the correct values.
>
> /Jorgen

--
Scott Bernardi
sbernardi at home.net