[sdiy] Favorite VCO options

[cheater cheater]

On Wed, Apr 14, 2010 at 00:36, David G. Dixon <dixon at interchange.ubc.ca> wrote:
>> > The feed to all of them is the output saw, not the raw integrator
>> output.
>>
>> cool. I suggest adding a place just before the shapers where you can
>> cut the traces and add a header/connector.
>
> There's already headers which feed the raw (low output impedance) saw and
> triangle to the panel, so I can select between them for the pulse and sine
> shapers.  You can't route the output (1k output impedance) waves around on
> the board because you'll attenuate the output (a lesson I learned the hard
> way some time ago).

I'm not really talking about routing stuff around. There will be
traces going from the 1k output to whereever. Put some holes in those
traces. :-)

>
>> Sync out should be, from what I understand, the pulse from the
>> comparator when it resets.
>
> Oh, I get you now.  However, this pulse is -15V to ground to be compatible
> with the gate of an N-channel JFET -- not necessarily all that useful
> outside of this particular circuit.

Then it could use to be converted to adhere to the standard.

> In any case, any module which can be
> synced will have an input capacitor to differentiate the incoming master
> signal anyway,

Says who? :-)

> and other circuitry to make it compatible for that specific
> device, so there's really no need to bring out the comparator pulse.

There is. For one thing, the standard for triggers is generally a
short +5V pulse.
Compare: http://www.synthesizers.com/gates.html

This means that a +15V rising saw wave would trigger at the wrong point.

For another thing, what if the gear only works on rising-edge
detection? In that case, there's no rising edge in the positive-going
ramp (or falling edge, in case you're talking about a negative-going
saw). Nothing gets triggered. Sure, hardware can have a rectifier to
convert broken trigger signals, but a diode delays the trigger a bit
in that case, and the circuit becomes more complicated, so I shun away
from doing something like that.

More importantly: if you have two oscillators that are in sync, you
want them to start resetting at the same time. In your example, the
master oscillator would reset, and once it's halfway done we get a
(falling) edge that can trigger the slave oscillator to reset. This
means the two oscillators will not actually be in phase. This matters
especially at high frequencies - the higher the frequency, the bigger
part of its cycle is taken up for core resetting.

But most importantly: if the oscillator is FM'd, the saw output could
have steep rising edges without the integrator resetting. So, the saw
output is definitely not suitable.

>> Try and see. You'll never know!
>
> No, I probably won't!
>
>> Hey, some people are pedantic!
>
> Boy, you ain't kiddin'!  ;-)
>
>> Linear FM has a very different effect than modulating reset position:
>> it changes the slope of the wave keeping the peak-to-peak voltage the
>> same. Modulating the reset level, on the other hand, keeps a constant
>> slope but the peak-to-peak voltage changes. This creates a different
>> sound. It's an even bigger difference with other shapes: because of
>> the way the waveform is imperfect, the shaped waveforms will have
>> glitches and other stuff - which makes it real cool. This lets you get
>> a controllable version of one aspect of what makes vintage oscillators
>> sound so nice and fat.
>
> Yes, I understand what you're saying, but I wonder how different the two
> effects really are.

audibly very different

> The main effect is to add randomness to the period of
> the saw,

Nope, it isn't :-) it's to modify the period of the saw while at the
same time changing the peak to peak voltage. It really sounds
different, trust me!

> which either one does, albeit by a slightly different mechanism.  I
> believe that the magnitude change associated with a noisy reference voltage
> would be quite inaudible.  Since we perceive loudness on the log scale, tiny
> linear fluctuations in amplitude will go unnoticed.

Except they go through a multitude of linearities which expose it!

> Couple that with the
> fact that what you propose would be a major pain in the ass,

Why? What's involved? I'm asking because I'm wondering what's needed
other than being able to tap into the circuit and maybe sum voltages.
I could see myself kludging something up, but why do we want that? :-)

>> What I meant was the multi-core trapez VCO core, there was a thread on
>> it last year. A triangle core has less tracking error in the high
>> frequencies, and the trapez vco core even less. And there are other
>> things you can do :-)
>
> Again, that's a different animal entirely.  Stay tuned for the 2164 Expo VCO
> Tricore (rising like a phoenix from the ashes of my last failed attempt).

alright!

> However, I have been thinking (in my increasingly rare idle moments) about a
> sort of "Oscillator Helper" module -- kinda like Hamburger Helper, but with
> fewer calories

Never heard of this. Did the soviets print food tickets for those?

> which would provide a whole pile of outboard waveshaping
> and mixing functions without cluttering up the actual VCO panel or board.
> In know that "dotcom" has something similar, but mine would have more stuff
> on it.

More stuff is always better. I applaud this approach.

>> > I always assumed that the Moog slaves were just synced.  I didn't
> realize
>> > they used common current sources.
>>
>> I thought they did. Didn't they? I'm fairly sure they did :-) I should
>> check though..
>
> I've no idea.  I've never looked into it.  It would be a bit of a challenge
> to do well, though, I should think.  The only place where I've ever used
> current mirrors to replicate an expo current is in my 4-pole LPF, and there
> the acccuracy is probably not all that critical.

I think they have.

>> When you sync the oscillator, the wave should be at a phase between 0
>> and 2pi, settable. It is *very* useful.
>
> A variable sync does something like this by varying the height of the sync
> pulse, and therefore the point along the saw ramp where the sync can occur.
> Other than that, I have no idea how to do what you're talking about.

Well, you know how the oscillator stops resetting at some voltage?
Make that voltage higher for the duration of the sync reset, and there
you go. I guess that's a bit complicated, but I think you'd like the
results.

>> a linear fm offset pot is just for when you don't want to go all the
>> way connecting a dc source to a VCA/gain then to the lin FM input.
>> Instead of wasting 2 modules and 2 patch cords, you turn the pot :-)
>>
>> same with FM offset.
>
> If you're talking about input attenuators, they're already there.  Other
> than that, I'm not sure why you would route a dc source to an FM input.
> That's what the 1V/oct input is for!

Because it's 10x easier than using the 1V/oct input jack and sourcing
it from some other module :-)

> My one firm rule of module design is this:  Every input is attenuated, no
> outputs are attenuated.  To me, this gives maximum flexibility.
>
>> >> 8. CV scaling (it is useful to have this separate for each VCO) - to
>> >> make octaves larger or smaller than 1V
>> >
>> > Why?
>>
>> It can make the chords *huge*. Try it! If you have windows, try the
>> 'oatmeal' VST synth, it has a knob for stretch tuning (= size of
>> octave), I think somewhere on the upper right of the GUI. Just
>> pressing C3 and C4 together makes a nice phasing sound. So you start
>> with C-F-A and it's fairly inanimated and then you press the C an
>> octave higher, and it suddenly changes the sound completely.
>
> This can also be achieved by putting a VCA between the keyboard's CV output
> and the 1V/oct input.  I've done that.

I guess so. It's not something you'd see every day, I could see myself
doing that a lot though, just a little dab - and VCAs aren't the
perfect things for scaling pitch CV..

> You can then modulate the VCA with
> an LFO too, which is pretty groovy, or use an AR envelope to give "stretch"
> for realistic (or not) plucked sounds.  I will grant you, however, that
> having the tuning controls accessible from the front panel (even through a
> hole for a small screwdriver) would be nice (but, alas, mine aren't).

I was thinking of a ten-turn potentiometer with a 2-centimeter knob,
but ok. Sniff..

>> is the sigmoid going to be the same thing as sin(x) from -pi/2 to
>> pi/2, repeated over and over?
>
> Essentially, yes.  It is half a sine wave with twice the period which resets
> like a sawtooth (and sounds an awful lot like one too).

OK

>> I'm not sure what the centre/edge pwm switch does.
>
> It aligns the pulse either centre-to-centre with the triangle, or
> left-edge-to-left-edge with the saw.  This does very little unless you are
> combining different waveforms from the same VCO in a waveshaper (like a ring
> modulator), in which case it is pretty useful.  I got the inspiration for it
> by looking at the Oakley VCO front panel on their website, and I figured
> that it was such an easy thing to do, why not?

Makes sense.