Sv: [sdiy] Digital VCO update

[Michael Bacich]

On Apr 30, 2006, at 12:17 PM, Eric Brombaugh wrote:

> Due to limited resources an LFO will have to be a separate design.  
> It's come up often enough that I'd be interested in what folks  
> would like to see in an LFO. A few questions:
>
> * Frequency range?
>
> * Waveforms?

An update on my previous comments --

I agree with Paul Maddox that a good frequency range would be from  
0.01 Hz to 1000 Hz.  Hopefully, your chip can handle that kind of range.

I've been thinking about it a bit, and I'm not too worried that the  
stepping is going to be too audible at slow rates.  This is because  
you are likely to attenuate the LFO before it hits its destination,  
and thus, would also be attenuating any stepping noise.  Most pitch  
mod is less than one octave, which would mean attenuating your 10  
Volt output swing to just one Volt.  That would make the steps pretty  
insignificant.  Filter sweeps tend to span more octaves, but usually  
not ten octaves -- more like 3 to 5 octaves at the most.  It'll be  
interesting to see how audible the stepping will be for wide filter  
sweeps.

Regarding waveforms, I think it would be really cool if you could  
include a table for a so-called "hypertriangle" wave.  This is a  
triangle that has a kind of exponential change in amplitude over  
time.  That is, as it starts ramping upward, it ramps faster and  
faster as it approaches its peak, then as it ramps downward, it does  
the opposite:  it gradually slows down as it approaches its lowest  
point.  This type of LFO can be very musical, especially when using  
it to modulate flangers and phasers.  You can check out audio  
examples of this at Scott Stites' DIY 32-stage phaser website.  It's  
pretty dramatic, and after you hear the phaser with the hypertriangle  
modulation, you don't really want to hear it again with a normal  
triangle.  It just seems to make the sweep sound more smooth,  
natural, and musical at both the low and high ends of the sweep.   
There's probably a technical explanation for this, but I don't know  
what it is.  I'm going to guess that the hypertri LFO is also good  
for modulating highly resonant LP, HP, and BP filters at slow rates,  
although I haven't heard it done yet.

I don't know if an inverted hypertriangle would also be useful (start  
ramping up fast, then slow down as you approach the peak).  So far,  
all the applications I have seen use one that starts ramping slow,  
gradually speeding up towards the peak.  Still, I have a hunch that  
there might be some good apps for the inverted one, too.

I guess if you wanted to save lookup table ROM space, you could  
always invert externally via analog means (same goes for the inverted  
saw, too)

You can find Scott's most recent stuff at this URL:

http://mypeoplepc.com/members/scottnoanh/birthofasynth/id18.html

I'm sure Scott could also give you some tips as to how that wave  
should look.  I think he generates his hypertriangle by putting a  
nonlinear element (an FET that distorts the wave?) into a standard  
triangle wave oscillator circuit.  It can also be achieved with a  
voltage controlled LFO by outputting a standard linear-ramp triangle  
wave, and then feeding a small amount of the triangle output back  
into the LFO's CV input.  As the wave amplitude rises, so does this  
feedback CV, causing the wave to speed up faster and faster as the CV  
rises higher and higher, then reversing the process and slowing down  
on the way back down.

Mike B.


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://synth-diy.org/pipermail/synth-diy/attachments/20060502/d2aa5d89/attachment.htm>