[sdiy] VCA response - continously variable between linear and exponential

rsdio at sounds.wa.com rsdio at sounds.wa.com
Sun Mar 3 03:11:22 CET 2013


Why would you need continuously variable curves? Are you just talking  
about shifting away from the standard 1 V per octave, but still being  
based on the octave standard?

In my experience, a VCO either responds to CV with a linear  
relationship between voltage and frequency (simpler circuits, easier  
to get precise), or it responds to CV with a logarithmic relationship  
between voltage and frequency (allowing a pitch bend CV to apply the  
same relative pitch change regardless of octave). The latter happens  
to have a linear relationship between voltage and octave or  
semitones. I cannot think of any reason to have a response other than  
one of those two. Note that the relationship between pitch (Hz or  
period) and octave is not linear.

As I mentioned, the linear response is more basic, and because it is  
linear you can make it as precise as you want. Unfortunately, a pitch  
bend CV applied in one octave will have a different relative pitch  
offset than in any other octave, making it impossible to pitch bend a  
chord on a polyphonic keyboard (unless you specifically desire  
inharmonious results). Thus, the only reason for non-linear CV of  
pitch is to allow mixing multiple CV sources (e.g. from keyboard,  
pitch bend, octave/footage selector, etc), and in that case the only  
logical choice is the specific curve of an octave relationship. The  
only disadvantage of the latter is that it is very difficult to  
create this non-linear curve. It's easiest to use, but hardest to  
create. Thus, the only two logical choices seem to be a Hz scale or  
an octave scale, e.g. V/Hz, Hz/V, V/octave, octave/V, V/period,  
period/V. There are lots of variations, but they still basically fall  
into just two categories in terms of the exponential or linear  
aspect, and I can't see the utility of anything "between" linear and  
exponential.

I imagine that the only useful variation would be the value for "n"  
in the expression, "n V per octave." The industry has mostly settled  
on 1 V per octave, with some using 1.2 V per octave, but my point is  
that it's always something per octave unless it's totally linear.

By the way, I'm sure everyone here understands the above, and I don't  
intend to insult anyone by spelling it all out, but I really want to  
make sure there's not something I'm missing. It seems like there was  
an implication that a cubic or quadratic or higher power exponential  
curves might be desirable, but I can't see why. I can see why 2 V per  
octave or 0.5 V per octave might be incredibly useful, because it  
would allow a different range of pitch for the same power supply. But  
unless I'm using the terminology incorrectly, that's not continuously  
variable between linear and exponential.

Brian Willoughby
Sound Consulting


----- "neil harper" <metadata at gmx.com>
> hey all,
> i'm looking for some advice/pointers on how to get a VCA response  
> curve
> to be continuously variable between linear and exponential. i can do
> switchable between the two, but is there any trick to getting  
> everything
> inbetween?
>
> if it helps, my VCA is lm13700 based but i don't mind starting with
> something else.




More information about the Synth-diy mailing list