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

Robin Whittle rw at firstpr.com.au
Tue Sep 17 03:58:03 CEST 2013

I have not tried this technique, but it looks good to me:

   Low-cost audio VCA has high performance
   Mike Sims, Lectrosonics Inc, Rio Rancho, NM
   EDN 1995-01-19

The original source of this no longer produces a proper article:


The page is archived at André Majorel's excellent site of Synth-DIY
resources: VCAs, transistor arrays etc.:

  http://www.teaser.fr/~amajorel/sims/     << the article

I don't know anywhere else which archives this page.  I have kept a copy
for myself in case André's site disappears.  I recall seeing another
article which involved the same system, but I can't find whatever copy I
made of it.

It works by using a feedback loop driving the signal into both the main
(2nd) and a first VCA, with both VCAs being closely matched due to being
on the same chip.  The first VCA has a fixed control current and the
feedback look drives whatever small voltage is required into both VCAs
to make the first one produce an exact replica of the audio input signal.

This is achieved by the first one's output current going back to the
input pins.  The audio input voltage is turned into a current by 30.1k
R1 (of course any suitable resistor value will do here) and most of that
current goes through the output of the first VCA.  This point remains
close to ground voltage compared to the input voltage.  Therefore, it is
best to use a relatively high input voltage range so R1's voltage and
therefore current is hardly affected by the small voltage changes at its
right end, due to the voltage required at the input of the VCAs to
produce the output current which brings this point close to ground.

In doing so, the feedback loop overcomes whatever distortion mechanisms
are inherent to both VCAs.  The first VCA is used otherwise normally.  I
think this means that the VCA's input pins can be driven harder than
normal, since ordinarily driving them hard would produce distortion in
the form of the output current of the VCA, for a given control current,
falling off at high input voltages.  The feedback network takes care of
this and would result in higher than normal drive voltages to the input
pins as required to get the desired output current.

Therefore, the second VCA, which is the one we are using for audio out,
is being driven hard, without distortion, which reduces (I guess,
depending on the mechanisms) the noise level at the output.

This would work for the LM13600 as described in the article and for the
LM13700s which are now still widely available.  See André's main page
for all the low-down on VCA availability, history etc.

The article reports that the technique results in:

   The circuit's maximum input voltage is +20 dBu (dBu=dB referred to
   775 mV rms). THD measures less than 0.015%; noise, -70 dBu; and
   control-voltage feedthrough, -70 dB.

  - Robin

On 2013-09-17 3:44 AM, Harald wrote:
> Yes  but i want to see what is possible with an LM13700 in the best case.
> I am down to 0.04% THD. Have to look at the noise though.
> Am Montag, 16. September 2013, 18:22:33 schrieb Colin f:
>>> Pardon me for stating the obvious but... If you're looking
>>> for high fidelity, wouldn't it make more sense to use one of
>>> the following ?
>>> - AD SSM2164 / Coolaudio V2164
>>> - THAT 2162 / Coolaudio V2162
>>> - THAT 2180
>>> - THAT 2181 / Coolaudio V2181
>> Seconded.
>> Having tried the THAT chips, I would only use a 13700 as a VCA for a noise
>> generator.
>> Cheers,
>> Colin f

More information about the Synth-diy mailing list