[sdiy] What's going on in the M12 filter circuit?

[The Old Crow]

On Fri, 29 Mar 2002, DJ Marmalade wrote:

> http://www.planetp.org/xpvoice.jpg
> 
> The 3372 LPF is used as a multimode filter.  The circuit on the right
> apparantly sums at different levels, the signals available from the filter
> caps.
> 
> I'm interested to know what is going on here, why does this work?  I also
> want to know if there is something unique about the 3372 or could this idea
> be applied to any number of filter chips, say for example, the 2044?

  The buffers (chips labeled UX13, UX15) are all summed into channels of a
4051 MUX, using different 'weightings' (the row of resistors groups
attached to each MUX channel input) for each cell output.  The 'even'
cells (cell 2 and cell 4 which taken from the VCF OUT) are
polarity-inverted so as to provide the necessary negative feedback to make
a given filter topology work properly.  These 'weighting' resistors each 
form part of a gain ratio with resistor RX66 and thus each filter pole 
can be stronger or weaker as required. I would actually call this a sort
of analog ROM where a filter type is selected based on the ABC address
inputs at the bottom of the 4051.

  Now, how the coefficient/weighting resistors were selected is the real 
key to this circuit.  I would suppose the numerical coefficients would 
show up in the math of each filter when when down in abstract form, but 
how these numbers were turned into resistor values--I don't know offhand.

  It might be possible with the 2044; I'll have a look at the datasheet.  
The only caveat is the 3372 also has two VCAs in it that would have to be 
made from OTAs or whatnot as well.

Crow
/**/