[sdiy] Dual Log/Linear Irwin-Gallo VCA

[David G. Dixon]

All good points, Neil.  I'd just add the following comments:

1.  The Irwin linearization trick is not "a lot of effort"; it's actually
very easy, and it eliminates some of the not-so-nice features of the 2164
when used as a gain block (the negative CV-to-gain relationship, and the 5k
impedance on the CV input).

2.  I can buy a THAT 2180 dual trimless VCA for $5.95, a THAT 2159 single
deluxe VCA clone for $3.50, or an SSM2164 quad VCA clone for $3.50.  Hence,
the 2164 is far more economical.  Plus, I'd still need to do the Irwin trick
with the THAT VCAs if I wanted a linear response.

3.  Not all of us have the advantage of a fully developed modular with loads
of VCAs already in it, and years of experience with electronics to see all
the possibilities of a given chip straightaway.  I can attest, however, that
the Irwin-Gallo VCA is a very nice and very elegant circuit, and an
excellent way for us neophytes to get acquainted with the 2164.



> Andre Majorel wrote:
> > Interesting. I wonder how stable it is WRT temperature. The 2164
> > was designed to pass audio, not to drive VCOs.
> 
> The spec on the temperature dependence on the control input is -3300ppm/C,
> about the same as for the usual expo converter, and can be compensated for
> either with a temp.co resistor somewhere in the control path, or you can
> use one of the other gain cells as a compensator.
> 
> As for driving VCOs, it is perfectly suitable for that: my 18dB/oct filter
> oscillates quite nicely over the audio range, and I have a design on the
> bench for a multi-phase VCLFO which with some capacitor swaps it will
> easily go way up into the audio range.  As it is now it goes down to
> around 30 minutes.
> 
> Personally, I think its a bit odd to build a plain VCA out of an SSM2164.
> A THAT corp dedicated audio VCA would give better audio performance, and
> you'd get better separation between channels than you would with the
> SSM2164 (same package, shared supply rails, cross-coupling between cells
> on the silicon, etc).
> 
> Sure, Irwin spotted up a neat trick, by turning a '2164 into an LM13700
> through a lot of effort.
> 
> No, for me the real magic of the '2164 is having four easy-to-use gain
> cells in one package.  That opens up a lot of interesting possibilities...
> 
> Cheers,
> Neil