[sdiy] Lowest distortion neede for VCA - linearizing the LM13600/13700 & other dual VCAs

Phil Macphail phil.macphail at liivatera.com
Tue Sep 17 17:24:26 CEST 2013


On 17/09/2013 17:45, "Dave Manley" <dlmanley at sonic.net> wrote:

>
>
>Phil Macphail <phil.macphail at liivatera.com> wrote:
>>There is also a further refinement to the Simms
>>design that really isn't for the faint-hearted that can improve the
>>distortion further - implementation is tricky so at that point the
>>V2164
>>is probably a better choice.
>>
>>Phil.
>
>That's a bit of a tease.  Please elaborate.
>
>-Dave

No tease intended at all, but I have had a few off-list mails so here goes
-

This is all from memory, so details are 'left as an exercise for the
reader'. 
The OTA with its output fed back to the input works like a resistor for
small offset voltages. As the signal level increases the resistor is
non-linear, but matches the gm of the second OTA, so the output
distortions cancel to produce low-distortion signal. This is all lovely if
the signal into the first OTA is a current, and this is achieved by using
a large resistor (I think the original was around 30k).
The current through the input resistor has a non-linearity as a result of
the tanh V/I relationship of the OTA input, although it is small (perhaps
50mV error with 10V across the resistor in the original design). This
error can be eliminated by using the 'first' OTA as the feedback element
in an inverting amplifier, where the input resistor now feeds a virtual
earth. 
As Garth said in Waynes World, I didn't expect it to be so easy. And it
isn't - the challenge is to stop the amplifier oscillating whilst still
having enough bandwidth to make a nice VCA.
It is possible to make a stable design, but if you don't know what you are
doing you will spend hours/ days trying to make this work. That's why I
said that you may prefer to just use the 2164 (or stick this the original
Sims design, as it does work very well).

Phil.





More information about the Synth-diy mailing list