[sdiy] modular synth "standards"

[jh.]

> Using the same voltages for both control and audio signals is mighty
> compelling in theory -- one can remove preconceptions and assumptions
> about how a module or signal will be used.  A control signal can be
> used as an audio signal or a VCO can be used as an LFO.  Cool.  But
> I'll claim that in practice it's not all that useful.

I disagree with you here, Don. In all my modulars or semi-modulars, Audio
and CV are interchangable, and I would never want to miss this. Even
in a non-modular such as a Prophet 5,  the VCOs restricted to
being a "signal source" would mean no audio rate PWM. (Which is one
of the P5's best features, IMO.)

> Take for example a audio VCO with a range switch so it can also be
> used as an LFO.  Audio range VCO waveforms need to have a zero DC
> component to avoid VCA thumping while LFO waveforms need to "start" at
> zero to be useful as a modulation source.

AC coupling cures thumping. I have an extra AC coupling module in the JH-3
(Jack -> 1uF -> Jack) for that purpose, and often I'm using a resonant
HPF in front of the main VCF and VCA anyway. My VCOs have
symmetrical outputs, btw, but some other modules haven't. The Interpolating
Scanner is good for mangling CVs as well as Audio Waveshaping,
but the way it is built it produces a heavy DC offset at the output.
No problem to remove that with AC coupling. Way more convenient
than having 2 different level standards throughout my whole system.


> For a 10V p-p output, the audio sawtooth wave would then go from -5V
> to +5V volts while the LFO modulation sawtooth would go from 0V to
> +10V -- it's much more practical for sawtooth modulation to go from
> our starting point up than to start at some negative value, fly
> through our normalized point, and continue up to some positive value.
> Similiarly an audio square wave would go from -5V to +5V while an LFO
> square wave would go from 0V to +10V for trilling effects.

Why would I want the saw wave go to twice the positive voltage than the
Triangle ? I find it "natural" to preserve the voltage swing when I switch
modulation waveforms. Which leads to the same result as with audio:
+/-5V as a standard, other ranges in exception.


> A PWM square wave for audio needs to have some DC restoration (mixing
> some of the PWM control into the output) for a zero DC component,
> while that wouldn't be very practical as a modulation source.

This is a very interesting point. I wonder which is the "right" way to go.
Correction by mixing the CV in, or leaving it as it is (with or without
AC coupling). All three options will give different results when used
on an overdriven VCF input. Which one is the "right" one ?
Which one is used in most non-modular synths ?


> There are other examples -- a VCF might be optimized to sound good for
> audio and have a nonlinear transfer characteristic (like the Moog
> ladder) while that wouldn't be accurate enough for a control voltage
> portmento.

I agree. I wouldn't use a Moog cascade for portamento. The SEM-Filter
in its original form needs AC coupling at the input to prevent
"motorboating".  So you're right, a different SVF would be needed for
sub audio CV processing. But then again the "vintage" SEM SVF
has the AC coupling filter for removing DC offsets (see above) built
in.

> Now, you *could* build a deluxe VCO with a range switch that not only
> tuned it down to LFO frequencies but also adjusted all the output
> voltages, but that's getting pretty weird as a conceptual building
> block.

I don't see why the "normal" output range would be different for audio
and CV purposes. Of course there are always exceptions, but these can
always be covered with a simple extra CV processor. (I see your point,
but to me the ARP Chroma-like PWM correction is the exception. Most
synths with PWM work with simple AC coupling somewhere before the
VCF or VCA.)

> So I don't really see any practical advantages to using the same
> voltage range for audio and control signals.  But using standard
> studio audio signal levels mean that you can patch in studio effects
> and mixers without a problem.  I think that's a huge advantage.

I think it's all a question of the number of interface points needed.
Would I need more interfaces to convert different levels inside
my system (every time I cross that arificial audio vs. CV border),
or if I do the conversion when I go in or out of the system.
I'm heavily voting for the latter, because you can include more
useful features in these input/output modules. In the JH-3, my input
modules go from balanced and stereo to 5Vp mono, with oveload
indicators, multiples and a switch to bypass the system for one of
the balanced stereo output pairs. And (that's what is added because
of the different voltage standard) a potentiometer for variable gain.
On the output side, there's a stereo pot for variable attenuation,
and a stereo balanced line driver. The point is that just removing
the pots doesn't save much.

OTOH, I'm building a small MOTM system for signal processing
only (no VCOs, apart from the VCLFO which goes up into the
audio range as well), and here I have converted the 4-pole VCF
to work with smaller input levels (and still having overdrive
reserves). But this is really the exception of the rule, because
it will not become a full fledged modular - it's more like a bank
of outboard FX.

More options: Bringing the CV levels *down* to the audio levels,
rather than the other way round. Scaled 1V/Oct CV would possibly
be separated from the rest then. (I did this on the Synthi Clone.)
Brings some other difficulties. Resistor -> pnp with grounded base
for OTA control needs a certain voltage level to be fairly linear,
for instance.

It's always a compromise ...

JH. (back on the list again)